Three-Dimensional Response of Neighboring Buildings

Sitting on Pile Foundations to Seismic Pounding

Behzad Fatahi, Ph.D., CPEng.1; Quoc Van Nguyen2; Ruoshi Xu3; and Wen-jing Sun4
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Seismic pounding occurs when the separation gap between buildings and structures is not wide enough, particularly during major
earthquake events; this can cause them to collide, causing local damage or, in extreme cases, collapse. This study investigated the impact that
this separation gap has on the seismic response of midrise buildings supported on piles while considering seismic soil–pile–structure interac-
tion (SSPSI). To achieve this aim, three 15-story reinforced concrete buildings sitting on pile foundations and with five different separation
gaps under excitations from the 1994 Northridge and 1995 Kobe earthquakes were numerically simulated. This study used three-dimensional
numerical modeling to simultaneously capture the effects of seismic pounding and SSPSI. Because the considered structure, pile foundation,
and soil deposit are three-dimensional in nature, the adopted three-dimensional numerical modeling can provide a more realistic simulation to
capture the seismic behavior of the system. The nonlinear behavior of structural elements was included, and the dynamic soil properties were
obtained from field data and backbone curves. A contact pair interface with small-sliding surface-to-surface formulation between buildings was
used to capture possible seismic pounding, and contact interfaces with a finite-sliding formulation were used to simulate the interaction between
the piles and the soil. The results, including lateral building deflections, interstory drifts, structural shear forces, foundation rocking, lateral pile
deflections, and the distributions of bending moments and shear forces of the piles, are presented and discussed. The findings of this study will
give engineers a better insight into the possible effects of seismic pounding on the seismic performance of buildings, and the response of end-
bearing piles in soft soils. DOI: 10.1061/(ASCE)GM.1943-5622.0001093. © 2018 American Society of Civil Engineers.
Author keywords: Seismic pounding; Separation gap; Seismic soil–pile–structure interaction; End-bearing pile foundation; ABAQUS.

Introduction event. In recent earthquake events, such as the 2008 Wenchuan

It is not unusual for structures in metropolitan areas to be con-
structed very close to each other, similar to those presented in
Fig. 1. However, during a strong seismic event, it is highly likely
that they interact due to seismic pounding, leading to severe dam-
age and possible collapse. The lessons learned from previous
earthquake events revealed that seismic pounding is a common
phenomenon that can cause failure of nonstructural elements,
damage to structural elements, and as a consequence, partial or
total collapse of structures (Rosenblueth and Meli 1986; Park and
Hashash 2004; Song et al. 2008; Chouw and Hao 2012).
Rosenblueth and Meli (1986) reported that in the 1985 Mexico
City earthquake, pounding occurred in approximately 40% of the
buildings, which partially or entirely collapsed, whereas 15% of the
buildings collapsed primarily due to collisions during this seismic
1
Associate Professor of Civil and Geotechnical Engineering, School of
Civil and Environmental Engineering, Univ. of Technology Sydney, City
Campus P.O. Box 123 Broadway, Sydney, NSW 2007, Australia (corre-
sponding author). E-mail: behzad.fatahi@uts.edu.au
2
Ph.D. Candidate, School of Civil and Environmental Engineering,
Univ. of Technology Sydney, Sydney, NSW 2007, Australia. E-mail:
Quoc.V.Nguyen-1@student.uts.edu.au
3
Ph.D. Candidate, School of Civil and Environmental Engineering,
Univ. of Technology Sydney, Sydney, NSW 2007, Australia. E-mail:
Ruoshi.xu@uts.edu.au
4
Assistant Professor, Dept. of Civil Engineering, Shanghai Univ.,
Shanghai 200444, China. E-mail: wjsun@shu.edu.cn
Note. This manuscript was submitted on February 22, 2017; approved
on September 22, 2017; published online on January 17, 2018. Discussion
period open until June 17, 2018; separate discussions must be submitted
for individual papers. This paper is part of the International Journal of
Geomechanics, © ASCE, ISSN 1532-3641.
earthquake and the 2011 Christchurch earthquake, structural damage
due to seismic pounding has also been observed and reported (Song
et al. 2008; Chouw and Hao 2012).
Many researchers (Anagnostopoulos 1988; Hao et al. 2000;
Karayannis and Favvata 2005; Mahmoud et al. 2008; Hao 2015)
have pointed out that the effects of pounding should be consid-
ered in the structural design stage. According to Hao (2015), the
best way to avoid seismic pounding is to provide a sufficient sep-
aration gap (SG) between buildings. Many modern seismic
standards and codes make recommendations on the minimum
requirement of the SG (also known as the seismic gap) between
adjacent buildings. FEMA 356 (FEMA 2000) requires a mini-
mum distance (si ) between adjacent structures at any level i as
follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
si ¼ D2i1 þ D2i2 (1)
where Di1 and Di2 = lateral deflections of neighboring structures.
The European seismic design standard [Eurocode 8 (CEN 2005)]
also has a similar requirement for the size of the SG. The Chinese
seismic design code (GB50011 2010) requires a SG of more than
100 mm when the height of a reinforeced concrete building is within
15 m. When the building height exceeds 15 m, a 20-mm increment
in the SG for every 5-, 4-, 3-, and 2-m increment in height is
required in regions that correspond to Chinese seismic intensity lev-
els of 6, 7, 8, and 9, respectively. The Australian seismic design
standard [AS1170.4 (Standards Australia 2007)] requires a SG
greater than 1% of the height for structures higher than 15 m,
although existing buildings constructured before the introduction of
modern seismic provisions are usually excluded (Anagnostopoulos
1996). However, Anagnostopoulos (1988) stated that these require-
ments may not be easy to satisfy, even for new constructions due to

© ASCE 04018007-1 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007


Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
Fig. 1. Cases of neighboring buildings with SG: (a) tall buildings in
Hong Kong (image by Behzad Fatahi); (b) old and new adjacent build-
ings linked in Shanghai (image by Wen-jing Sun)
the economic, technical, and legal reasons put forward by property
owners, engineers, and developers.
According to Wolf and Skrikerud (1980), the SGs between
structures may still be insufficient if the plastic displacement of
structures and soil beneath the foundations develops during an
actual earthquake event. It is common practice to not consider the
underlying soil, particularly when structural pounding is analyzed,
thus the effect of soil–structure interaction (SSI) is ignored by
claiming it will produce a conservative design, and the effect of SSI
often decreases the fundamental frequency of a structure and
increases the effective damping compared to the same structure
under a fixed-base condition (Kramer 1996). Based on a compre-
hensive numerical study, Tavakoli et al. (2011), Tabatabaiefar and
Fatahi (2014), and Tabatabaiefar et al. (2014) concluded that as the
soil under the structure becomes softer, the influence of SSI on
amplifying lateral deflections and interstory drifts would be more
significant, and therefore, SSI should be considered for structures
supported by relatively soft soil. Furthermore, Carbonari et al.
(2011) and Hokmabadi et al. (2015) studied the effect of soil–pile–
structure interaction (SPSI) through a finite-element analysis and
concluded that the effects of SPSI should be considered in structural
design because it would amplify the global seismic response of
buildings, particularly the lateral displacement and interstory drifts.
On this basis, the effects of SSI should be considered when studying
seismic pounding because it is caused by the relative lateral move-
ment of two adjacent structures. Shakya and Wijeyewickrema
(2009) reported that when seismic pounding occurs, the location
and the impact force acting on the pounding area are influenced by
the dynamic characteristics of the structures and underlying soil,
and the distances between neighboring buildings.
© ASCE 04018007-2
Int. J. Geomech., 2018, 18(4): 04018007
Recently, the issue of coupling the effects of seismic pounding
and SSI has received a lot more attention. Rahman et al. (2001) per-
formed a series of time-history analyses to study the effect of seis-
mic pounding on two reinforced concrete buildings with different
heights while taking into account the effect of the underlying soil.
Chouw (2002) adopted the boundary-element method to investigate
two colliding buildings modeled by finite-element analysis while
considering the flexibility of soil in the Laplace and time domains.
Shakya and Wijeyewickrema (2009) conducted an analysis of two
buildings with different heights while considering SSI to study the
effects of midcolumn pounding on the building under the influence
of near-field and far-field earthquakes. Uz and Hadi (2011) devel-
oped a MATLAB program based on the fourth-order Runge-Kutta
method to carry out a response-history analysis on two adjacent
buildings while considering the combined effects of seismic pound-
ing and SSI. Moreover, Naserkhaki et al. (2012) developed an ana-
lytical model of nearby buildings resting on soil where the buildings
are connected by viscoelastic contact elements during pounding.
Mahmoud et al. (2013) studied the influence of SSI on the seismic
pounding of two buildings with the same height but different
dynamic characteristics; the buildings were represented by lumped-
mass systems, and dynamic soil behavior was considered by using
spring-dashpot elements. Furthermore, Zou et al. (2013) studied the
effect of SPSI on the pounding response of two closely spaced
buildings of different heights through analytical modeling. Madani
et al. (2015) developed a single-degree-of-freedom (SDOF) finite-
element model that considers SSI and the structure–soil–structure
interaction to investigate the effect of seismic pounding on the
dynamic response of structures in which structural and soil nonli-
nearity are considered.
Most of these aforementioned research studies only examined
the effects of interaction between shallow footings and supporting
soil; only a few considered deep foundations. Although most of
these studies investigated the dynamic responses of two conven-
tional buildings of equal or unequal heights using simple lumped-
mass models, only a few studied the response of retrofitted build-
ings, such as shear wall–braced buildings. As mentioned by Wolf
and Skrikerud (1980), the problem of insufficient SGs can also be
caused by retrofitting structures for increased seismic requirements,
because this process modifies their dynamic properties and their
subsequent seismic response. Moreover, because the effects of seis-
mic pounding on the responses of end-bearing piles has rarely been
examined, this study investigated how seismic pounding between
adjacent buildings supported by end-bearing pile foundations influ-
ences the seismic response of buildings and piles while considering
SPSI. To achieve these goals, Abaqus software, a three-dimensional
(3D) finite-element numerical simulation package, was utilized to
analyze a fully coupled soil–foundation–structure system by adopt-
ing the direct method. Two buildings that represent conventional
moment-resisting buildings and retrofitted buildings were consid-
ered, with the assumption that they are of equal height and with
aligned story levels. The nonlinear behavior of structural elements
was considered, and the dynamic soil properties were obtained
from field data and backbone curves. Contact surfaces were
assigned to interacting building models to capture possible seismic
pounding. To rigorously account for the effect of SPSI, the nonlin-
ear behavior of contact surfaces between piles and the surrounding
soil medium were captured. The numerical results, including the
response spectrum, shear distribution along the buildings, lateral
structural displacement, maximum drift envelope, maximum rock-
ing of the foundation slab, lateral pile displacements, and the bend-
ing moments and shear forces along the piles, are presented and
discussed.
Int. J. Geomech.
 Characteristics of a Soil–Pile–Structure System
Features of Adopted Buildings
A 15-story three-bay reinforced concrete moment-resisting building
representing conventional midrise buildings, and two 15-story
three-bay reinforced concrete buildings with shear wall systems
representing retrofitted buildings were modeled in this study. The
adopted shear wall–braced buildings have shear walls at the middle
bay in both directions and along their entire height. As Fig. 2 shows,
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
the conventional moment-resisting building is located in the middle
with two shear wall–braced buildings sitting on each side. Fig. 3
presents the numerical soil–pile–structure model developed in the
Abaqus software.
Both buildings are 45 m high (z-direction) and 12 m wide in the
x- and y-directions. Following relevant Australian building codes
[AS1170.1 (Standards Australia 2002); AS1170.4 (Standards
Australia 2007); AS3600 (Standards Australia 2009a)], a routine
design procedure was carried out for structural analysis and design
by utilizing SAP2000 14. During the design phase, P-delta effects
were taken into account by following AS3600 (Australia Standards
2009a). Fig. 4 shows that the structural sections designed for the
moment-resisting building [Fig. 4(a)] and the shear wall–braced
building [Fig. 4(b)] satisfy common engineering practice in metro-
politan areas. According to AS3600, Grade 32 concrete, whose
characteristic compressive strength (fc0 ) is 32 MPa and modulus of
elasticity (E) is 30.1 GPa, and Grade N500 steel reinforcing bars
with a characteristic yield strength (fsy ) of 500 MPa were used.
Moreover, the stiffness of all structural members (EI) was modi-
fied according to ACI 318-08 (ACI 2008) by using cracked sections
Separation gap (SG) =
0.1, 0.3, 0.7, 1.0, 1.2 m
Shear wall braced buildings
End bearing piles
x
Input earthquake direction
Elevation view
(a)
Soil infinite Separation gap (SG)
element CIN3D8
Soil finite
element C3D8R
y
15m 170m
x Plan view
(b)
Fig. 2. General setup of the soil–pile–structure system: (a) elevation view; (b) plan view
© ASCE 04018007-3
Int. J. Geomech., 2018, 18(4): 04018007
for reinforced concrete sections, as presented in Table 1. The natural
periods of the moment-resisting building and shear wall–braced
building in a fixed-base condition were 1.23 and 0.65 s, respec-
tively. The total mass for the moment-resisting building was 1,683 t,
and 2,495 t for the shear wall–braced building. As shown in Fig. 2,
the SGs between the adjacent buildings considered in this study
were 100, 300, 700, 1,000, and 1,200 mm; these gaps were applied
on both sides of the middle building to investigate the effects of
seismic pounding.
Characteristics of Pile Foundations
This study considered end-bearing pile foundations to support the
adopted buildings. The foundation is designed to fulfill the require-
ments of bearing capacity and settlement based on a conventional
engineering design (Poulos and Davis 1980; Bowles 1988;
Standards Australia 2009b), and the foundation material is concrete
by assuming a density of 2,400 kg/m3, Young’s modulus of 3.01
GPa, and Poisson’s ratio of 0.2. As Fig. 5 shows, the foundation
consists of a 12  15  1-m reinforced concrete foundation slab and
a group of 4  4 reinforced concrete piles 30 m long by 1.2 m in di-
ameter (D). The pile spacing (center to center) is 3.6 m (3D) along
the x-direction (the same direction as the applied earthquake) and
4 m (3.3D) along the y-direction (the orthogonal direction to the
applied earthquake direction); these figures agree with other
researchers (Small and Zhang 2002; Shelke and Patra 2008; Kumar
et al. 2016). It was also assumed that the pile toes were socketed
into the strong bedrock, thus creating an end-bearing load mecha-
nism for the piles.
Moment resisting building
45m
Soft soil
30m
Soil infinite
element CIN3D8
30m
15m
Int. J. Geomech.
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
Fig. 3. Abaqus 3D numerical model capturing both structural models and adopted soil–structure model
Characteristics of Soil
According to AS1170.4 (Standards Australia 2007), a site with
more than 10-m depth of soil and a shear-wave velocity of 150 m/s
or less should be classified as Class Ee soft soil. Also, the standard
evaluates the site effects based on the top 30 m of soil profile
because most of the amplification and attenuation occurs within the
top 30 m of soil. Therefore, in this study, a 30-m-deep soil deposit
containing soft clayey soil (CL) was used to represent a Class Ee
soft-soil site. Table 2 summarizes the parameters used in this model.
Note that the properties of soil were taken from actual in situ and
laboratory tests, which means they have merits over idealized prop-
erties that may be unrealistic.
Numerical Model
The governing dynamic equations of motion for the soil–foundation–
structure system are too complex to be solved by conventional ana-
lytical methods due to different vectors and matrices of the soil, the
foundation, and the structure, so the substructure method and the
direct method are used to tackle this problem. The substructure
method, which takes advantage of superposition, has been devel-
oped and is used widely by practicing engineers and researchers. As
Wolf (1989) reported, the approximations of soil nonlinearity
through iterative wave-propagation analyses may allow this method
to be applied to a moderately nonlinear system. However, the princi-
ple of superposition requires an assumption of linear soil and struc-
tural behavior (Kramer 1996), albeit it is not easy to consider the
exact soil and structural nonlinearities in a dynamic analysis.
However, the direct method allows for the soil, foundation, and
structure to be simulated simultaneously, and because an assump-
tion of superposition is not required, a correct and accurate model
and analysis can be achieved (Borja et al. 1994). Moreover, the SSI
has been studied by many researchers (Hayashi and Takahashi
© ASCE 04018007-4
Int. J. Geomech., 2018, 18(4): 04018007
2004; Carbonari et al. 2011), and accurate and realistic outcomes
have been achieved by adopting the direct method; therefore, it was
used in this study. The soil–pile–structure system was simulated
numerically using the finite-element software Abaqus, version 6.14.
This software can simulate complex problems that may require sub-
stantial computational memories. In fact, some researchers (Chu
and Truman 2004; Koskinen 2005; Matinmanesh and Asheghabadi
2011; Nguyen et al. 2016, 2017) have used Abaqus software to
study SSI and reported reasonable results. Because the considered
structure, pile foundation, and soil deposit are 3D in nature, the 3D
numerical modeling can provide a realistic simulation to capture the
seismic behavior of the soil–foundation–structure system more
appropriately. However, the 3D modeling is challenging and time-
consuming. The procedure utilized to perform a numerical simula-
tion while considering the combined effects of seismic pounding
and seismic SPSI is explained here.
Building and Foundation Models
In this numerical model, the beam elements (B31), as shown in
Fig. 6(a), are utilized to model the columns. The B31 beam element
is a first-order 3D beam element that considers a 3D continuum in
the fashion of one-dimensional approximation. The floor slabs and
shear walls are modeled by shell elements (S4R), and as Fig. 6(b)
shows, this S4R shell element is a four-node shell element that con-
siders a uniform large-strain formulation.
The elastic–perfectly plastic constitutive model is assigned to
structural elements to carry out an inelastic analysis by specifying
the yielding stress so that the structural elements will behave elasti-
cally until the defined yielding stress is reached. According to Shing
and Tanabe (2001), the yielding stress of concrete material is con-
sidered to be the same as the characteristic compressive strength (fc0 )
of concrete. The energy disspation of the adopted buildings during
earthquakes is accounted for by Rayleigh damping. Table 3 shows
Int. J. Geomech.
 y z

0.25m
0.25m
x x

0.4m
0.4m
0.4m
0.4m
Column IV Slab
SlabIV
IV
Column IV

y z

45 m
height1515xx3 3m==45m
0.45m
0.45m

0.25m
0.25m
x x

0.45m
0.45m
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

Column III
III Slab III
Slab III

Building height
Column
y z

0.25m
x x

0.5m
0.5m
0.5m
0.5m
Column SlabIIII
ColumnIIII Slab

y z
0.55m

0.25m
0.55m

0.25m
x x

0.55m
0.55m
Column I Slab II
Slab
Column I

length =
Pile length 30m
= 30m
z
y

Pile
x

(a)

y z y
0.25m

x x x
0.4m

0.4m

0.4m
Column IV Slab IV Shear wall IV

y z y Building height 15 x 3 m = 45 m
0.45m

0.45m
0.25m

x x x

0.45m
Column III Slab III Shear wall III

y z y
0.25m

x x x
0.5m

0.5m

0.5m
Column II Slab II Shear wall II

y y
z
0.25m
0.55m

0.55m

x x x

0.55m
Column I Slab I Shear wall I
Pile length = 30m

z
y
x
(b)

Fig. 4. Designed structural sections and reinforcement details for (a) moment-resisting building and (b) shear wall–braced building

© ASCE 04018007-5 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 Table 1. Designed Sections for Columns, Slabs, and Shear Walls

Dimensions Moment of inertia of Cracked factor Reduced moment of inertia of

Section type (m) plane area (m4) [according to ACI 318-08 (ACI 2008)] plan area [Icr (m4)]
Column
I 0.55  0.55 0.0076 0.7 0.00532
II 0.50  0.50 0.0052 0.7 0.00364
III 0.45  0.45 0.0034 0.7 0.00238
IV 0.40  0.40 0.0021 0.7 0.00147
Floor slab 0.25  1.0 0.0013 0.25 0.000325
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

Shear wall
I 0.55  4.0 2.933 0.7 2.0531
II 0.50  4.0 2.667 0.7 1.8669
III 0.45  4.0 2.400 0.7 1.6800
IV 0.4  4.0 2.133 0.7 1.4931

Earthquake direction

SL7 SL5 SL3 SL1 M1 M3 M5 M7 SR1 SR3 SR5 SR7

SL8 SL6 SL4 M2 M4 M6 SR2 SR4 SR6

4m
15 m

SL2 M8 SR8
4m
4m

x 3.6 m 3.6 m 3.6 m 3.6 m 3.6 m 3.6 m 3.6 m 3.6 m 3.6 m

12 m 12 m 12 m

Left shear wall building Moment resisting building Right shear wall building

Fig. 5. Adopted end-bearing foundation setup

Table 2. Characteristics of Adopted Soil (Data from Tabatabaiefar and
Fatahi 2014)
Soil property
Unit weight [ g (kN/m )]
3
Maximum small strain shear modulus [Gmax (MPa)]
Poisson’s ratio (
)
Effective friction angle [Ø0 (degrees)]
Effective cohesion [c0 (kPa)]
Plasticity index [PI (%)]
Shear-wave velocity [vs (m/s)]
the adopted mass and stiffness coefficients of Rayleigh damping
defined according to the first and second mode frequencies of each
structure, and the corresponding damping values.
As suggested by some researchers (Maheshwari et al. 2004;
Ghee and Guo 2011; Comodromos and Papadopoulou 2012;
Hokmabadi and Fatahi 2016), solid elements were utilized to model
the pile foundations in this study, and as Figs. 6(c and d) show,
C3D8R and C3D15 element types were used to model the founda-
tion slabs and piles, respectively. The C3D8R element is an eight-
node, first-order, reduced-integration linear element with an hourglass
control that ensures that the strains calculated at the integration
points are more reasonable. The C3D15 element is a 3D 15-node
Value continuum element. All piles have a rigid connection with the
foundation slab, and the pile toes are fixed at the bottom of the soil
14.42 deposit to account for the socket end of piles in strong bedrock.
33.1
0.4
12 Soil Model
20
As Fig. 6(c) shows, C3D8R elements were also used to model the
15
soil deposit, and although the locking phenomenon does not occur
150
due to reduced integration, these elements are not stiff enough in
bending, which is not critical when modeling soil; however, because
the integration points are located in the middle of this element, small
elements are needed to capture the concentration of stresses.
The nonlinearity of soil during a seismic event is critical in deter-
mining the dynamic response of piles (Maheshwari and Watanabe
2006) and so, too, is the response of the corresponding soil–pile–
structure system. The equivalent linear method is the most common
method adopted by practicing engineers to consider the site effect
and nonlinearity of soil, and therefore, it was adopted in this study.
To carry out the initial analysis, some values of the shear modulus
and damping ratio were assumed in various regions of the model,

© ASCE 04018007-6 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 and then the maximum cyclic shear strain of each element was
recorded after the analysis. The new values of the shear modulus
and corresponding damping ratio for the subsequent analysis were
determined by referring the maximum cyclic shear strain to the
modulus degradation curves and corresponding damping ratio
curves. According to Vucetic and Dobry (1991), the plasticity index
(PI) is one of the main factors controlling the variations of the shear
modulus reduction and damping ratio against the cyclic strain
curve. For a given level of shear strain, as the PI increases, the shear
modulus degradation factor increases while the damping ratio
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
modulus and damping ratio were used to obtain the best prediction
of the real behavior of soil and to capture soil nonlinearity. For the
1994 Northridge and the 1995 Kobe earthquakes, the dominant fre-
quencies are 0.6 and 1.2 Hz, respectively, whereas the durations are
30 and 50 s, respectively, and both earthquakes were chosen by the
International Association for Structural Control and Monitoring for
benchmark seismic studies (Karamodin and Kazemi 2010). Hence,
because of the characteristics of each earthquake, the induced
shear-strain level differs in the soil medium, which is why the
adopted shear modulus reduction factor and the corresponding

decreases. In this study, ready-to-use backbone curves, which damping ratios were changing with the corresponding earthquake
included the influence of the PI on the modulus degradation curves motion inputs to include the appropriate soil nonlinearity (Table 4).
and corresponding damping ratio, as provided by Vucetic and The most common mechanism used to represent the dissipation
Dobry (1991), were adopted. Fig. 7 shows the particular curves of of energy is viscous damping, where the dissipative force is assumed
modulus degradation and damping ratio selected based on the soil to be a function of particle velocity (Das 1983). Soil damping is one
properties (see Table 2). An iterative procedure was carried out until of the essentials needed to simulate the soil–foundation–structure
the difference between the computed values of shear modulus and system because it modifies the input motions and influences the
the damping ratio in two successive analyses was less than a certain
dynamic response of structures. Park and Hashash (2004) concluded
value in the model. Finally, the acquired values of the shear
that Rayleigh damping could provide acceptable results for many
applications as long as the parameters for soil media are selected
3 accurately. Thus, in the present study, Rayleigh damping was used to
simulate the nonlinear variations of energy losses in the soil during
2 earthquake excitation. The two frequencies that cover the range with
4 2 a significant amount of input motion were used to define the mass
and stiffness coefficient of Rayleigh damping, following the recom-
1
mendation made by Park and Hashash (2004). Table 4 contains the
1 corresponding parameters of Rayleigh damping for each earthquake.
(a) (b)

12 Contact Surfaces for Pounding Simulation

3
4 6 7 Contact surfaces were applied where collisions between the build-
11
10 4 ings might have been possible. Abaqus provides more than one
13 5 8 approach, including general contact, contact pairs, and contact ele-
15
14 2 ments, to model the interacting surfaces. In this study, a contact pair
9 6 with small-sliding surface-to-surface contact was used. The contact
1 3
1 pair approach needs two surfaces to make contact during analysis,
8
7 to be defined as a master surface and a slave surface. Here, master
2 5
surfaces were applied on each side of the floor of the middle build-
(c) (d)
ing (i.e., the conventional moment-resisting building), and the cor-
3 7 responding slave surfaces were applied on the same level of the
other buildings (i.e., the retrofitted buildings). Note that based on
4 the authors’ assessment, the location of the master surface would
8
not influence the seismic response of the adopted buildings, which
2 means the analysis would yield the same results irrespective of
6
which master surface in the side or the middle building was
1 selected. A small-sliding tracking formulation was used to capture
the behavior of elements when there was relatively little sliding of
one surface over another. Surface-to-surface contact discretization
5 was used to consider contact interaction as a surface facet rather
(e) than a surface node. Specifically, an opening or a penetration dis-
tance was considered in the normal direction of the slave surface
Fig. 6. Elements used by adopted finite-element model: (a) column
facet, whereas the sliding distance was measured perpendicular to
element (B31); (b) slab element (S4R); (c) foundation slab element and
the normal direction of the slave surface facet. Based on the distan-
soil element (C3D8R); (d) pile element (C3D15); (e) infinite element
ces measured and mechanical properties assigned, the behavior of
(CIN3D8)
these interacting surfaces could be simulated accurately.

Table 3. Mass and Stiffness Coefficients of Rayleigh Damping for Adopted Buildings

First mode Second mode Mass damping Stiffness damping

Building type frequency (Hz) frequency (Hz) coefficient (a) coefficient ( b ) Damping ratio (%)
Moment-resisting building 0.830 2.341 0.3850 0.0050 5
Shear wall–braced building 1.522 6.110 0.7656 0.0021 5

© ASCE 04018007-7 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 1.0 25

0.9

Gmax
0.8 20

G
Modulus degradation ratio,
0.7

(%)
0.6 Modulus degradation ratio 15

Damping ratio,
0.5 Damping ratio

0.4 10
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

0.3

0.2 5

0.1

0 0
0.0001 0.001 0.01 0.1 1
Cyclic shear strain,
c
(%)

Fig. 7. Adopted soil modulus degradation curve and corresponding damping ratio curve for cohesive soils with PI = 15% (data from Vucetic and
Dobry 1991)

Table 4. Rayleigh Damping for Adopted Soil Deposit Considering Different Earthquake Excitations

Maximum shear Modulus reduction Mass damping Stiffness damping Damping

Earthquake strain (%) factor coefficient coefficient ratio (%)
Northridge 1994 0.14 0.35 1.0694 0.0055 12.8
Kobe 1995 0.10 0.41 1.0405 0.0046 11.5

Fig. 8 shows the three essential components needed to account

for seismic pounding during dynamic analysis (i.e., a gap to ensure
there is no stress transmitted across the interface before contact is Gap
made). The SGs in this study are 100, 300, 700, 1,000, and Moment Shear wall
resisting building braced building
1,200 mm. The normal stresses transferred across the interface
include a spring with stiffness k placed in the normal direction, and
the frictional forces transmitted through the interface include a
slider in the tangential direction.
This means a hard-contact pressure–overclosure relationship,
which defines the properties of gaps and normal springs, is used for
normal contact interaction between buildings, as shown in Fig. 8.
This particular contact relationship means that no contact stresses
will be transmitted until the facet of the slave surface makes contact
with the facet of the master surface (i.e., when the gap between the
buildings is closed and seismic pounding occurs, contact forces are
transmitted across the interface). This implies that no tensile stress
can be transmitted across the contact surface, and no compressive
stress can be transmitted before any contact is made.
Technically, this hard contact assumes that no penetration will
occur between the slave surface and the master surface when the
two corresponding surfaces make contact, but this assumption may
lead to overconstraining. To mitigate this possibility, a stiff approxi-
mation of hard contact known as the penalty method was used in
this study. As Fig. 9 shows, using the penalty method means allow-
ing for a small degree of penetration for the interacting surfaces, an
approach that resulted in a degree of numerical softening, which
can alleviate overconstraining and reduce the burden of iterations.
Tangential behavior was considered based on the Coulomb fric-
Fig. 8. Demonstration of adopted mechanical model of contact inter-
tion model and a penalty formulation that specifies the properties of action to capture possible seismic pounding
frictional sliders, as shown in Fig. 8. The Coulomb friction model

© ASCE 04018007-8 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 Separation, h
(when h 0)
No pressure, p = 0 p
when separation Penalty stiffness, k =
appears, h 0 -h
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
Contact pressure, p
k
Insignificant penetration
Penetration allowed
(when h 0)
Fig. 9. Representation of adopted penalty method for numerical soft-
ening of hard-contact simulation
correlates the critical shear stress (t cr ) and the contact pressure (p)
between the contacting surfaces by the coefficient of friction ( m ), as
presented in Eq. (2)
t cr ¼ m p (2)
In this study, m = tan 30 was used for the concrete material and
was also assumed to be the same in all frictional directions. The crit-
ical stress (t cr ) indicates the states of two contacting surfaces. As
shown in Fig. 10, this state is known as sticking when the equivalent
shear stress (t eq ) carried by the two contacting surfaces is below
the critical stress (t cr ), and it is called slipping as the equivalent
shear stress (t eq ) reaches the critical stress (t cr ) and the two con-
tacting surfaces start to slide relative to each other. The equivalent
shear stress (t eq ) is a combination of two orthogonal components
of shear stress (t 1 and t 2 ) by following Eq. (3)
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
t eq ¼ t 1 2 þ t 2 2 (3)
Interfaces between Piles and Soil
The interfaces between the piles and surrounding soil are essential
in the numerical simulation of a soil–pile–structure system to incor-
porate the different properties of these materials and capture any
possible separation and sliding that may occur between the piles
and surrounding soil during earthquake excitations.
In this study, a contact algorithm that includes finite-sliding and
surface-to-surface contact formulation between the outer perimeter
of each pile and the contacting surface of the surrounding soil was
adopted, whereas general contact requires defining the master surfa-
ces on the soil contacting facets and slave surfaces on the outer pe-
rimeter of the piles. This finite-sliding tracking formulation
accounts for the relative motion of two interacting surfaces and cap-
tures the arbitrary relative motion of contacting surfaces, thus mak-
ing it possible to simulate the gapping and sliding between piles and
soil during excitation. As Fig. 11 shows, the surface-to-surface dis-
cretization formulation enforces the contact conditions by including
the shapes of the slave and master surfaces, which means each con-
tact constraint will include an individual slave node and the nearby
slave nodes, thus the stress and pressure across the interface are
dealt with more accurately. This geometric correction was also
© ASCE 04018007-9
Int. J. Geomech., 2018, 18(4): 04018007
eq
Equivalent shear stress,
Critical shear
stress, cr
ng
pi Friction
ip
Sl coefficient,
Stick region
Contact pressure, p
Fig. 10. Demonstration of adopted Coulomb friction model
applied to alleviate the difficulties with accuracy and convergence
that are induced by approximated facets representing an actual
curved geometric surface.
The normal interaction between piles and soil was modeled as
hard contact due to the penalty method discussed in the preceding
section, whereas the tangential behavior of the interface follows the
classical Mohr-Coulomb model, which includes the slip rate–
dependent coefficient of friction.
Based on Oden and Martins (1985), a model assuming the expo-
nential decay of the coefficient of friction from the peak to the resid-
ual coefficient of friction shown in Fig. 12 was utilized to simulate
the frictional behavior between the piles and surrounding soil.
The exponential decay function of the coefficient of friction is
presented in Eq. (4)
m ¼ m r þ ð m p  m r Þedc g_ eq (4)
where m r = residual coefficient of friction; m p = peak coefficient of
friction; dc = decay coefficient; and g_ eq = equivalent slip rate. Thus,
the peak coefficient of friction ( m p ) is defined based on the classical
Mohr-Coulomb model by adopting the following equation:
 
Rint t c0
m p ¼ 0 ¼ Rint tan w 0 þ 0 (5)
s s
where Rint (= 0.75) = reduction factor commonly used by practicing
engineers; t = shear strength; s 0 = effective normal stress; c0 =
effective cohesion intercept of the failure envelope; and w 0 = slope
of the failure envelope or the effective internal friction angle. As
Fig. 12 shows, the coefficient of friction changes faster against the
equivalent slip rate ( g_ eq ) as the decay coefficient (dc ) increases. By
referring to Randolph et al. (1994), the residual coefficient of fric-
tion ( m r ) and the decay coefficient (dc ) are assumed to be 0:2 m p
and 0.05, respectively, to simulate friction between the piles and the
soil. Moreover, the equivalent slip rate ( g_ eq ) is calculated by com-
bining the two local slip velocity components ( g_ 1 and g_ 2 ) along
the interface between two bodies by using Eq. (6)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
g_ eq ¼ g_ 1 2 þ g_ 2 2 (6)
However, simulating an ideal frictional behavior can be very dif-
ficult because it may cause convergence problems, which is why the
use of a penalty formulation allows for a small amount of relative
movement when soil and corresponding pile surfaces should ideally
stick together, as shown in Fig. 13, so that the numerical iterations
are greatly reduced.
Int. J. Geomech.
 Soil elements

Master surface

Considered
surface area
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

Pile elements

Element nodes
Slave surface

Fig. 11. Demonstration of surface-to-surface discretization applied to enforce soil–pile contact conditions

Boundary Conditions

-dc
. As mentioned by Semblat (2011), the boundaries at the sides of the
eq
= r+ ( p- r) e model must account for free-field motion. In this study, eight-node
Friction coefficient for
pile-soil interaction,

p linear one-way infinite brick (CIN3D8) elements were used to

d c= 0.025 model the infinite elements acting as an unbounded soil domain
d c= 0.050 because plane wave propagation was also modeled. Fig. 6(e) shows
that the infinite elements defined the orientations, so Nodes 1, 2, 3,
d c= 0.075
and 4 are connected to the solid elements (soil medium), whereas
the other nodes (Nodes 5, 6, 7, and 8) are oriented outward from the
r soil medium. The soil deposit is 5 times larger than the adopted
buildings in terms of the dimensions in the x-direction, as suggested
. by other researchers (Kumar et al. 2015), to minimize wave reflec-
Shear strain rate between pile and soil, eq tion induced by artificial boundaries.
This study simulated a soft-soil deposit sitting on a hard bedrock,
Fig. 12. Representation of adopted exponential decay behavior of fric- so a rigid boundary was used to represent low-velocity sediment
tion coefficient over a high-velocity bedrock.

Input Earthquakes
Fig. 14 shows two near-field earthquakes, including the 1994
Equivalent shear stress, eq Northridge earthquake and the 1995 Kobe earthquake, which were
e used in this study to investigate how seismic pounding affects the
slipping
cr response of midrise buildings. The moment magnitude scale (MW)
and peak ground acceleration (PGA) are 6.7 and 0.843g, respec-
ing

tively, for the 1994 Northridge earthquake, whereas the correspond-

ck
sti

ing values are 6.8 and 0.833g, respectively, for the 1995 Kobe earth-
quake. To conduct a time-history analysis, accelerograms of
Equivalent shear strain, eq earthquake records were applied at the bedrock level.
ing

A baseline correction was also applied to modify the accelero-

ck

grams, because when they are used to deduce the velocity and dis-
sti

slipping placement time histories via integration, particularly for strong

ground motions, a significant baseline drift that does not match the
field measurements may be observed. This baseline drift can occur
e
due to field conditions, background noise, and the hysteresis behav-
Fig. 13. Demonstration of adopted friction behavior with penalty
ior of accelerometers, which is why a baseline correction procedure
method of soil–pile contact interaction
is often utilized to compensate for baseline drift. This study used the
correction technique proposed by Newmark (1973) such that an

© ASCE 04018007-10 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 1
0.8 1994 Northridge earthquake
0.6 Mw=6.7, PGA=0.843g

Acceleration (g)
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0 10 20 30 40 50 60
Time (second)
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

(a)

1
0.8 1995 Kobe earthquake
0.6 Mw=6.8, PGA=0.833g
Acceleration (g)

0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0 10 20 30 40 50 60
(b) Time (second)

Fig. 14. Considered earthquake accelerogram: (a) 1994 Northridge earthquake; (b) 1995 Kobe earthquake

acceleration correction [a0 ðtÞ] is added to the raw data record [aðtÞ]
to produce a corrected acceleration record ac ðtÞ ¼ a þ a0 ] to mini-
mize the mean square velocity over the time of the event. This cor-
rected acceleration is parabolic over any number of the time inter-
vals during the event
   
t  T1 t  T1 2
a0 ðtÞ ¼ C1 þ C2 þ C3 ; T1 < t < T2
T2 T2
where T1 and T2 = limits of a time interval; and Ck ðk ¼ 1;
2; and 3Þ = constants obtained from the velocity minimization, as
shown in Eq. (8)
Tð2
∂ energy at a certain natural period. The response spectra adopted in
½vc ðtÞ2 dt
∂Ck
T1
where vc ðtÞ = corrected velocity record obtained by integrating the
corrected acceleration record [ac ðtÞ]. Note that these velocities are
obtained by assuming that the uncorrected and corrected accelera-
tions vary linearly over each time increment of the original accelera-
tion history. This is not exact for the corrected acceleration record
(because of the parabolic variation of the correction in time), but it
is assumed that the acceleration history is discretized at small-
enough time increments to ensure this is an insignificant error.
Results and Discussion
The results are presented in terms of the response spectrum at the
foundation slab level, the maximum envelope of floor shear, the
maximum rocking angle of foundation slabs, the transient maxi-
mum lateral building deflection, the maximum interstory drifts,
the maximum lateral pile deflection and corresponding bending
moments, and the shear forces distributed along the piles. The
results for both types of buildings are presented and discussed in
the following sections.
Response Spectra and Natural Frequencies
In most modern seismic codes and standards, a response spectrum
is needed to calculate the lateral force demand (e.g., base shear) of a
structure in a pseudostatic analysis. The higher spectral acceleration
indicates that more seismic energy would be transmitted to the sys-
(7) tem in a certain natural period, and thus the superstructure would
attract more base shear. The response spectrum is a function of the
natural period of a SDOF system with a certain damping ratio,
which is presented by plotting the maximum spectral acceleration
experienced by the SDOF system. A higher value of spectral accel-
eration implies that the structural system will attract more seismic
(8) this study utilized 5% for a system damping ratio under the influ-
ence of adopted ground motions.
Fig. 15 shows the response spectra of the ground motions for a
variety of SGs; these ground motions were recorded at the founda-
tion slabs during the 1994 Northridge earthquake [Fig. 14(a)] and
the 1995 Kobe earthquake [Fig. 14(b)] excitations. This spectrum
can help design engineers evaluate the foundation input motion
(FIM) and allow them to carry out seismic soil–pile–structure inter-
action (SSPSI) analysis using the substructure method. Referring to
Fig. 15, by considering SSPSI, the response spectrum shows an
amplification at the longer period range, particularly for the period
between 1 and 2 s, and attenuation at the shorter period range, such
as when the natural period is less than 0.5 s. It is also clear that the
response spectrum curves are almost identical regardless of the
changes in the SG, so both shear wall–braced buildings, on the left
and right sides, delivered a similar response spectrum. Referring
to Table 4, the adopted soil dynamic properties were different
depending on the applied earthquakes, which led to the differen-
ces in the response spectra, although the two earthquakes shared
similar moment magnitude scales and PGAs. A comparison
between the first and second mode frequencies of both types of
buildings with different foundations is summarized in Table 5. A

© ASCE 04018007-11 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 3 3
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
2.5 SG = 700 mm 2.5 SG = 700 mm
SG = 1000 mm SG = 1000 mm
SG = 1200 mm SG = 1200 mm
Spectral acceleration Sa (g)

Spectral acceleration Sa (g)

2 Bedrock record 2 Bedrock record
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

1.5 1.5

1 1

0.5 0.5
Moment resisting building Shear wall braced building
Northridge earthquake (1994) Northridge earthquake (1994)
Mw=6.7, PGA=0.843g Mw=6.7, PGA=0.843g
0 0
0 1 2 3 4 0 1 2 3 4
(a) Period (second) (b) Period (second)

3 3
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
2.5 SG = 700 mm 2.5 SG = 700 mm
SG = 1000 mm SG = 1000 mm
SG = 1200 mm SG = 1200 mm
Spectral acceleration Sa (g)

Bedrock record
Spectral acceleration Sa (g)

2 Bedrock record 2

1.5 1.5

1 1

0.5 0.5
Moment resisting building Shear wall braced building
Kobe earthquake (1995) Kobe earthquake (1995)
Mw=6.8, PGA=0.833g Mw=6.8, PGA=0.833g
0 0
0 1 2 3 4 0 1 2 3 4

(c) Period (second) (d) Period (second)

Fig. 15. Response spectrum of ground motions of (a) the moment-resisting building and (b) the shear wall–braced building during the 1994
Northridge earthquake, and (c) the moment-resisting building and (d) the shear wall–braced building during the 1995 Kobe earthquake in conjunction
with different SGs

linear perturbation and Lanczos method available in Abaqus were bearing pile foundation, the first mode frequency was 0.815 Hz.
used to extract eigenvalues to calculate the natural frequencies This decrease in natural frequency stemmed from the extra
and corresponding mode shapes of the soil–pile–structure system. degrees of freedom induced by introducing pile foundations and
Fig. 16 shows the typical first mode shapes of the soil–foundation– a soil medium. Moreover, this variation in the dynamic charac-
structure system where the fundamental natural frequency was teristics of the system was the primary governing factor influ-
0.830 Hz for the moment-resisting building in a fixed-base con- encing the amount of seismic energy absorbed by the structure
dition, and when the building was sitting on a 30-m-long end- subjected to strong earthquakes.

© ASCE 04018007-12 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 Shear Forces
Fig. 17 shows the envelope of shear forces in the building used to
study the SG and the impact of possible seismic pounding on a
moment-resisting building and shear wall–braced buildings. This
envelope resulted from a 3D numerical analysis of both types of
buildings under applied earthquakes. The maximum shear force at
each building level, regardless of the direction, is the absolute maxi-
mum value of story shear force, and is the summation of shear
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
forces acting on the columns and shear walls (if existing) at every
time interval. The envelope of the shear force was then determined
by plotting the maximum absolute shear forces against the corre-
sponding story levels.
As Fig. 17 shows, under both earthquake excitations, the enve-
lopes of shear forces are almost identical when the SGs are 1,000
and 1,200 mm, thus implying that no seismic pounding occurred
during these analyses. However, seismic pounding did occur when
the SGs were equal to or less than 700 mm because the correspond-
ing envelopes differ markedly from the cases with no pounding.

As plotted in Fig. 17(a), the moment-resisting building under

Table 5. Variation of Natural Frequency of Adopted Moment-Resisting
Building and Shear Wall–Braced Building the 1994 Northridge earthquake experienced 6.3 MN of shear
force at Level 10 when the SG was 100 mm, which is 37% more
Frequency (Hz) than the shear force at the same level when the SG was 1,200 mm.
Foundation
Building type type Mode 1 Mode 2 Moreover, where SG = 700 mm, the building experienced an in-
crement of shear force at the rooftop almost 3 times more than the
Moment-resisting building 30-m-long pile 0.815 2.280 cases with no seismic pounding. Obviously, seismic pounding
Fixed base 0.830 2.341 occurred at Levels 10 and 15 when the SGs were 100 and
Shear wall–braced building 30-m-long pile 1.145 4.642 300 mm, and at the rooftop when SG = 700 mm during the 1994
Fixed base 1.522 6.110 Northridge earthquake.

Fig. 16. First mode shapes of adopted (a) moment-resisting building and (b) shear wall–braced buildings

© ASCE 04018007-13 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 15 15
Moment resisting building 14
14
Northridge earthquake (1994),
13 Mw = 6.7, PGA = 0.843g 13

12 12

11 11

10 10

9 9

Storey level
Storey level
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

8 8

7 7 Shear wall braced building

Northridge earthquake (1994),
6 6 Mw = 6.7, PGA = 0.843g
5 5 SG = 100 mm
SG = 100 mm
4 4 SG = 300 mm
SG = 300 mm
3 3 SG = 700 mm
SG = 700 mm
2 2 SG = 1000 mm
SG = 1000 mm
1 1 SG = 1200 mm
SG = 1200 mm
0 0
0 2 4 6 8 10 12 0 2 4 6 8 10 12 14 16
(a) Total leveling shear force (MN) (b) Total leveling shear force (MN)

15 15

14 14

13 13

12 12

11 11

10 10

9 9
Storey level
Storey level

8 8
Shear wall braced building
7 SG = 100 mm 7 Kobe earthquake (1995),
6 6 Mw = 6.8, PGA = 0.833g
SG = 300 mm
5 SG = 700 mm 5 SG = 100 mm
4 SG = 1000 mm 4 SG = 300 mm
3 SG = 1200 mm 3 SG = 700 mm

2 Moment resisting building 2 SG = 1000 mm

Kobe earthquake (1995),
1 Mw = 6.8, PGA = 0.833g 1 SG = 1200 mm

0 0
0 2 4 6 8 10 0 2 4 6 8 10 12 14 16
(c) Total leveling shear force (MN) (d) Total leveling shear force (MN)

Fig. 17. Total leveling shear forces of (a) the moment-resisting building and (b) the shear wall–braced building under the 1994 Northridge earthquake,
and (c) the moment-resisting building and (d) the shear wall–braced building under the 1995 Kobe earthquake in conjunction with different SGs

Fig. 17(c) summarizes the shear forces developed in the 15-story
moment-resisting building during the 1995 Kobe earthquake. Note
that the envelopes of shear forces are identical when the SGs are
1,000 and 1,200 mm, indicating no pounding, whereas the increase
of shear forces in some particular story levels with reduced SGs
(i.e., SG = 700, 300, and 100 mm) implies that the neighboring
buildings collided. When the SG = 100 mm, the moment-resisting
building experienced 50 and 400% more shear forces at Levels 10
and 15, respectively, compared to the corresponding values for
cases with no pounding (i.e., SG = 1,200 mm). Furthermore, when
the SGs were 300 and 700 mm, the moment-resisting building expe-
rienced 5.5 and 3.6 MN of shear forces at Level 14, respectively. In
contrast, taking the case of SG = 1,000 mm as an example, the shear
force at the same level was only 1.8 MN, and therefore, the colli-
sions induced by the 1995 Kobe earthquake occurred at Levels 10
and 15 when the SG was 100 mm, and at Level 14 when the SGs
were 300 and 700 mm. These values show that seismic pounding
will increase the shear force demand at particular levels, which may

© ASCE 04018007-14 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 threaten column integrity if seismic pounding is not included in the
design phase. Referring to Figs. 17(b and d), seismic pounding
occurred at the same level of the shear wall–braced buildings for the
corresponding cases.
In general, the base shear for the moment-resisting building (i.e.,
the middle building) increases when separation decreases, as shown
in Figs. 17(a and c). For example, the base shear could have
increased by up to 18 and 12% for the 1994 Northridge and the
1995 Kobe earthquakes, respectively. In the present study, the shear
forces in the structures stemmed from structural inertia induced by
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
earthquake is shown in Fig. 18; it also includes the response of
Levels 15 and 10 for the moment-resisting building and the left
shear wall–braced building. Indeed, Fig. 18 shows that at the begin-
ning of dynamic analysis (when the time is equal to 0), the differ-
ence in displacement between two buildings is 300 mm, which indi-
cates a 300-mm SG. Fig. 18(a) indicates that seismic pounding
occurred five times at Level 15, whereas Fig. 18(b) shows that seis-
mic pounding occurred twice because higher story levels experi-
ence higher accumulated lateral displacement, and thus seismic
pounding is more likely to happen at higher story levels. Moreover,

the earthquakes and the collision caused by seismic pounding. In
fact, seismic pounding implies that confinement can be induced by
the presence of two shear wall–braced buildings (i.e., the side build-
ings), which alters their natural periods and changes the dynamic
characteristics of the system. Moreover, the natural periods of both
types of buildings decrease as the SG decreases, which means that
an increase in the base shear is due to seismic pounding, and this
decline in the natural period may attract more seismic energy, as
shown in Fig. 15.
In contrast, Fig. 17(d) shows that during the 1995 Kobe earth-
quake, the base shear force acting on the shear wall–braced struc-
ture increased with the SG. This is because the subsequent reduction
in base shear was observed as the natural period of the building sys-
tem shifted to a shorter period, and therefore, the distribution of
shear forces along the building was influenced more significantly,
even though seismic pounding occurred when the SG was less than
700 mm. An example of lateral building displacement and pounding
force time history where SG = 300 mm due to the 1994 Northridge
the maximum impact forces induced by seismic pounding are sig-
nificant but with short durations, and these impact forces are mainly
resisted by the mass of the building. Therefore, the impact forces
significantly influence the global response of the building and are
obviously detrimental to the structural elements around the impact
area. Therefore, the coupling effects of seismic pounding and
SSPSI should be considered by practicing engineers because they
can be detrimental to the local and global safety of a building.
It could also be concluded that the minimum SG required by
some seismic standards (e.g., AS1170.4 and GB50011) is not suffi-
cient because seismic pounding occurred when SG = 700 mm,
which is greater than the standard requirement. Basically, the stand-
ard SG might be sufficient for neighboring buildings with similar
dynamic characteristics, but not for neighboring buildings with dif-
ferent dynamic characteristics, such as retrofitted buildings standing
next to each other. Retrofitting alters the dynamic characteristics of
a building, usually without changing the SG, and therefore, the
pounding issue needs to be reanalyzed after retrofitting.

1200 60

900 45
Lateral displacement (mm)

Pounding force (MN)

600 30

300 15

0 0

-300 Level 15 -15

Northridge earthquake (1994),
-600 Mw = 6.7, PGA = 0.843g -30

-900 -45
Moment resisting building Left shear wall braced building Impact force
-1200 -60
0 3 6 9 12 15
(a) Time (s)

1200 60

900 45
Lateral displacement (mm)

Pounding force (MN)

600 30

300 15

0 0

-300 Level 10 -15

Northridge earthquake (1994),
-600 Mw = 6.7, PGA = 0.843g -30

-900 -45
Moment resisting building Left shear wall braced building Impact force
-1200 -60
0 3 6 9 12 15
(b) Time (s)

Fig. 18. Lateral building displacement of the moment-resisting building and left shear wall–braced building and pounding force time histories of
(a) Level 15 and (b) Level 10 during the 1994 Northridge earthquake

© ASCE 04018007-15 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 Rocking of Foundation Slabs
Fig. 19 shows the maximum rocking of a foundation slab for both
types of buildings in conjunction with five different SGs. Maximum
rocking is taken as the absolute maximum value of slab rotation
during seismic excitation. Foundation slabs rock when the iner-
tial forces generated in the superstructure during earthquakes
cause compression on one side and tension on the other side; this
causes settlement on the compression side and possible uplift or
upward movement on the other side. In this study, confinement
induced by two side buildings is another factor that causes a
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
foundation slab to rock, as well as the inertial force generated
and the dynamic properties of the underlying soil. Therefore,
because the weight of the moment-resisting building is less than
the weight of the shear wall–braced building, and because of the
confinement induced by neighboring buildings on the lateral
movement of the moment-resisting building, it is expected that
the rocking of the foundation slab of the moment-resisting build-
ing is less than the rocking of the foundation slab of the shear
wall–braced building.
During the 1994 Northridge earthquake [Fig. 19(a)], as the SG
decreased, the maximum rocking of the foundation slab of the
moment-resisting building decreased, but it increased for the shear
wall–braced building. The increase in base shear with the reduction
in the SG can be seen in Fig. 17(a) for the moment-resisting build-
ing, and it should result in the foundation slab rocking more.
However, because of the confinement caused by neighboring build-
ings, foundation movement is restrained, and is even more
restrained as the SG decreases. Shear wall–braced buildings that are
free to move on one side indicate the pattern in which the founda-
tion slab experiences maximum rocking varies with the base shear,
thus indicating an increase in the rocking angle as the SG decreases;
this is shown in in Fig. 17(b).
Fig. 19(b) shows the trend of maximum rocking of the founda-
tion slab during the 1995 Kobe earthquake, and shows that the
maximum rocking of the foundation slab increases with the SG
Northridge earthquake (1994)
Mw = 6.7, PGA = 0.843g
0.30
Shear wall braced building
Moment resisting building
0.25
Maximum rocking (degree)
0.20
0.15
0.10
0.05
0.00
100 300 700 1000
(a) Separation gap (mm)
Fig. 19. Maximum rocking of foundation slab of the moment-resisting building and shear wall–braced building in conjunction with different SGs
during the (a) 1994 Northridge and (b) 1995 Kobe earthquakes
© ASCE
Int. J. Geomech., 2018, 18(4): 04018007
for the shear wall–braced buildings, whereas the trend for the
moment-resisting building was opposite, mainly due to variations
in the base shear for the moment-resisting and shear wall–braced
building, as shown in Figs. 17(c and d), respectively. Thus, the
variation of inertial forces had a more pronounced influence on
the rocking of the foundation slab.
Lateral Deflection and Interstory Drift
Fig. 20 shows the lateral deflections of both types of buildings when
the roof reached its maximum deflection. As Hokmabadi et al.
(2012) stated, recording the lateral deflection of a building in this
way tends to provide a more reasonable structural deformation than
recording the absolute maximum values, regardless of when they
occurred. In this study, the lateral deflections are relative to the
movement of the foundation slab at ground level by subtracting the
lateral displacement of the foundation slab from the lateral deflec-
tion of each story level.
Fig. 20(a) shows that the lateral deflections of the moment-
resisting building under excitation of the 1994 Northridge earth-
quake decreased with the decrease in the SG. For instance, when
SG = 1,200 mm, the corresponding maximum lateral deflection was
780 mm, and a maximum lateral deflection of 528 mm was obtained
when SG = 100 mm. This decrease in lateral deflection is directly
related to the confinement induced by the closely spaced buildings,
despite the increases observed in the base shear with the decrease in
the SG. Moreover, the shear wall–braced buildings experienced
more lateral displacement as the SG decreased due to the extra shear
forces induced by seismic pounding and the free side of buildings.
Fig. 20 also shows that the maximum lateral deflection of a moment-
resisting building is greater than the deflection of a shear wall–braced
building because a shear wall system induces high structural stiffness
in the latter. Thus, the variation in the results of lateral deflections for
shear wall–braced buildings is not as significant as the corresponding
deflections in the moment-resisting buildings.
Kobe earthquake (1995)
Mw = 6.8, PGA = 0.833g
0.35
Shear wall braced building
Moment resisting building
0.30
0.25
Maximum rocking (degree)
0.20
0.15
0.10
0.05
0.00
1200 100 300 700 1000 1200
(b) Separation gap (mm)
04018007-16 Int. J. Geomech.
 15 15
SG = 100 mm SG = 100 mm
14 14
SG = 300 mm SG = 300 mm
13 13
SG = 700 mm SG = 700 mm
12 12
SG = 1000 mm SG = 1000 mm
11 11
SG = 1200 mm SG = 1200 mm
10 10
9 9
Storey level

Storey level
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

8 8
7 7
6 6
5 5
4 4
3 3
2 Moment resisting building 2 Shear wall braced building
Northridge earthquake (1994), Northridge earthquake (1994),
1 1
Mw = 6.7, PGA = 0.843g Mw = 6.7, PGA = 0.843g
0 0
0 200 400 600 800 0 100 200 300 400 500
(a) Maximum lateral deflection (mm) (b) Maximum lateral deflection (mm)

15 15
SG = 100 mm
14 14
SG = 300 mm
13 13
SG = 700 mm
12 12
SG = 1000 mm
11 11
SG = 1200 mm
10 10
9 9
Storey level

Storey level

8 8
7 7 SG = 100 mm

6 6 SG = 300 mm

5 5 SG = 700 mm

4 4 SG = 1000 mm

3 3 SG = 1200 mm

2 Moment resisting building 2 Shear wall braced building

Kobe earthquake (1995), Kobe earthquake (1995),
1 Mw = 6.8, PGA = 0.833g 1 Mw = 6.8, PGA = 0.833g
0 0
0 200 400 600 800 0 200 400 600 800
(c) Maximum lateral deflection (mm) (d) Maximum lateral deflection (mm)

Fig. 20. Maximum lateral structural deflections of (a) the moment-resisting building and (b) the shear wall–braced building during the 1994
Northridge earthquake, and (c) the moment-resisting building and (d) the shear wall–braced building during the 1995 Kobe earthquake in conjunction
with different SGs

Fig. 20(c) shows the lateral deflection of the moment-resisting
building under excitation during the 1995 Kobe earthquake where
maximum lateral deflections increased as the SG decreased.
However, Fig. 20(d) shows that the lateral deflections of shear
wall–braced buildings decreased with the decrease in the SG.
This means that lateral building deflection is influenced by three
factors: (1) shear force–induced building distortion, (2) foundation
slab rocking, and (3) confinement embraced by the neighboring
structures. In those cases that considered the excitation of the 1995
Kobe earthquake, the induced shear forces in the building contrib-
uted more significantly to lateral deformation; indeed, Fig. 17(c)
shows that the story shear forces generally increased as the SG
decreased, and the shear forces at some particular levels increased
due to seismic pounding. As a consquence, the moment-resisting

© ASCE 04018007-17 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 15 15
Moment resisting building Shear wall braced building
14 Northridge earthquake (1994), 14 Northridge earthquake (1994),
Mw = 6.7, PGA = 0.843g Mw = 6.7, PGA = 0.843g
13 13
12 12
SG = 100 mm
11 11
SG = 300 mm
10 10
SG = 700 mm
9 9
SG = 1000 mm
Storey level

Storey level
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

8 SG = 1200 mm 8 SG = 100 mm
7 7 SG = 300 mm
6 6 SG = 700 mm
5 5 SG = 1000 mm
4 4 SG = 1200 mm
3 3
2 2
1 1
0 0
0.0 1.0 2.0 3.0 4.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2
(a) Inter-storey drift (%) (b) Inter-storey drift (%)

15 15
Moment resisting building
14 Kobe earthquake (1995), 14
Mw = 6.8, PGA = 0.833g
13 13
12 12
11 11
10 10
9 9
Storey level
Storey level

8 8
7 7 SG = 100 mm
SG = 100 mm
6 6 SG = 300 mm
SG = 300 mm
5 5 SG = 700 mm
SG = 700 mm
4 4 SG = 1000 mm
SG = 1000 mm
3 SG = 1200 mm
3 SG = 1200 mm
2 2
Shear wall braced building
1 1 Kobe earthquake (1995),
Mw = 6.8, PGA = 0.833g
0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
(c) Inter-storey drift (%) (d) Inter-storey drift (%)

Fig. 21. Maximum interstory drifts of (a) the moment-resisting building and (b) the shear wall–braced building during the 1994 Northridge earthquake,
and (c) the moment-resisting building and (d) the shear wall–braced building during the 1995 Kobe earthquake in conjunction with different SGs

building experienced more lateral deflection as the SG decreased, deflections decreased with the decrease in the sg, as shown in
even though lateral deflection was restricted due to confinement Fig. 20(d).
induced by the neighboring buildings. As Fig. 20(c) shows, the var- Fig. 21 presents the maximum interstory drifts of both types of
iation of maximum lateral deflection was only 20%. The shear wall buildings. According to the Australian seismic standard [AS1170.4
–braced building in Fig. 17(d) shows that, although seismic pound- (Standards Australia 2007)], Eq. (9) is used to determine the maxi-
ing amplified the shear forces at some levels, the base shear mum interstory drifts
attracted by the building decreased as the SG decreased due to
changes in the dynamic characteristics. Thus, lateral building drift ¼ ðdiþ1  di Þ=h (9)

© ASCE 04018007-18 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 where di = deflection at the ith level; diþ1 = deflection at the
ði þ 1Þth level; and h = story height. Interstory drifts generally
follow the same pattern of variation as the lateral deflections for
corresponding cases. For instance, during the 1994 Northridge
earthquake when the SG increased from 100 to 1,200 mm, inter-
story drift for the moment-resisting building increased from 2.08
to 2.94% [Fig. 21(a)], and decreased from 1.11 to 0.94% for the
shear wall–braced building [Fig. 21(b)]. As shown in Figs. 21(c
and d), during the 1995 Kobe earthquake, the interstory drifts
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
decreased by 17% and increased by 107% for the moment-
resisting building and shear wall braced building, respectively,
when the SG increased from 100 to 1,200 mm. Therefore, seismic
pounding can increase interstory drift, and thus seismic pounding
interferes with the building’s seismic performance level because
interstory drift is an important parameter for assessing the seismic
performance level. In summary, a minimum SG of 1.75% of the
structural height is recommended to avoid structural pounding
between midrise buildings on pile foundations.

0 100 200 300 0 100 200 300 400

0 0
Moment resisting building Shear wall braced building
Northridge earthquake (1994), Northridge earthquake (1994),
Mw = 6.7, PGA = 0.843g Mw = 6.7, PGA = 0.843g
-5 -5
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
-10 SG = 700 mm -10 SG = 700 mm
SG = 1000 mm SG = 1000 mm
Pile depth (m)

Pile depth (m)

SG = 1200 mm SG = 1200 mm
-15 -15

-20 -20

-25 -25

-30 -30
(a) Lateral pile deflection (mm) (b) Lateral pile deflection (mm)

0 100 200 300 0 100 200 300

0 0
Moment resisting building Shear wall braced building
Kobe earthquake (1995), Kobe earthquake (1995),
Mw = 6.8, PGA = 0.833g Mw = 6.8, PGA = 0.833g
-5 -5
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
-10 SG = 700 mm -10 SG = 700 mm
SG = 1000 mm SG = 1000 mm
Pile depth (m)
Pile depth (m)

SG = 1200 mm SG = 1200 mm
-15 -15

-20 -20

-25 -25

-30 -30
(c) Lateral pile deflection (mm) (d) Lateral pile deflection (mm)

Fig. 22. Maximum lateral pile deflections of (a) the moment-resisting building and (b) the shear wall–braced building during the 1994
Northridge earthquake, and (c) the moment-resisting building and (d) the shear wall–braced building during the 1995 Kobe earthquake in con-
junction with different SGs

© ASCE 04018007-19 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 Lateral Pile Deflections
Fig. 22 presents the lateral deflection of piles along their depth
under an earthquake loading for both types of buildings, in conjunc-
tion with different SGs. The numerical model allows the time his-
tory of the relative lateral movement [uðz; tÞ] of any point along the
pile to be accessed, where z is the depth from the bottom of the foun-
dation slab to the considered points, and t is the time during excita-
tion. The time when maximum lateral deflection occurred at the
head of a pile is used to plot lateral deflection along the pile [uðzÞ].
Fig. 23 shows the deformation of a soil–pile–structure system dur-
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
ing a dynamic analysis.
For the moment-resisting building, a pile labeled M1 (Fig. 5)
is used as an example because all the piles shared a similar
amount of lateral deflections in the group of end-bearing piles.
The lateral deflections of Pile M1 during the 1994 Northridge
and 1995 Kobe earthquakes are plotted in Figs. 22(a and c),
respectively, whereas the results of Pile SL1 (Fig. 5), which
supported the left shear wall–braced building, are shown in
Figs. 22(b and d).
Fig. 22 shows that where no seismic pounding occurred during
the analysis (i.e., SG = 1,000 and 1,200 mm), the lateral deflections
plotted along the pile are very close to each other, but where the SG
was equal to or less than 700 mm, the lateral deflections differed
compared to the results where no seismic pounding occurred. This
shows that lateral pile deflection can be influenced significantly by
the seismic pounding.
Left shear wall
braced building
Moment resisting
building
(a)
Quiet
(b) boundary
Fig. 23. Deformation of structure–pile–soil system: (a) deformed piles; (b) section view of deformed structure–pile–soil system
© ASCE 04018007-20
Int. J. Geomech., 2018, 18(4): 04018007
Figs. 22(a and c) show that the lateral deflection of piles increase
as the SG decreases where piles are supporting the moment-resisting
building. For instance, during the 1994 Northridge earthquake,
when SG = 1,000 mm, the maximum lateral pile deflection was
112 mm, whereas the maximum lateral pile deflection increased by
150% when SG = 100 mm. Figs. 22(b and d) reveal that the lateral
pile deflections of the shear wall–braced building increased as the
SG decreased under excitation by the 1994 Northridge earthquake,
whereas the opposite occurred under excitation during the 1995
Kobe earthquake. The foundation slab transmits the inertial forces
developed in superstructures from the superstructure to the pile
group, and then to the competent bedrock. Thus, the base shear
attracted by the superstructure influences the lateral deflection of
piles, and as a consequence, it is expected that the lateral deflection
of piles and the base shears share the same trend of variation seen in
Figs. 17 and 22.
Bending Moments along Piles
Fig. 24 presents the bending moments along piles under the influ-
ence of two seismic excitations and five different-size SGs. The
elastic beam theory (Euler-Bernoulli’s equation) is used to produce
the bending moment along the depth of each pile [MðzÞ] as follows:
d2 ½uðzÞ
MðzÞ ¼ EI (10)
dz2
Right shear wall
braced building
Deformed
piles
Bedrock
location
Section view of deformed
soil-pile-structurel system
Quiet
boundary
Int. J. Geomech.
 -4 -2 0 2 4 6 -4 -2 0 2 4 6 8
0 0
Moment resisting building Shear wall braced building
Northridge earthquake (1994), Northridge earthquake (1994),
Mw = 6.7, PGA = 0.843g Mw = 6.7, PGA = 0.843g
-5 -5
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
-10 SG = 700 mm -10 SG = 700 mm
SG = 1000 mm SG = 1000 mm

Pile depth (m)

Pile depth (m)
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.

SG = 1200 mm SG = 1200 mm
-15 -15

-20 -20

-25 -25

-30 -30
(a) Bending moment (MNm) (b) Bending moment (MNm)

-10 -8 -6 -4 -2 0 2 4 -8 -6 -4 -2 0 2 4
0 0
Moment resisting building Shear wall braced building
Kobe earthquake (1995), Kobe earthquake (1995),
Mw = 6.8, PGA = 0.833g Mw = 6.8, PGA = 0.833g
-5 -5
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
-10 SG = 700 mm -10 SG = 700 mm
SG = 1000 mm SG = 1000 mm
Pile depth (m)
Pile depth (m)

SG = 1200 mm SG = 1200 mm
-15 -15

-20 -20

-25 -25

-30 -30
(c) Bending moment (MNm) (d) Bending moment (MNm)

Fig. 24. Bending moments of piles of (a) the moment-resisting building and (b) the shear wall–braced building during the 1994 Northridge
earthquake, and (c) the moment-resisting building and (d) the shear wall–braced building during the 1995 Kobe earthquake in conjunction
with different SGs

where E = Young’s modulus of the pile material, which is assumed
to be 30.1 Gpa; and I = second moment of the cross section of piles,
which is 0.1018 m4. The sixth-order polynomial fitting curves were
used to reproduce the most accurate lateral deflections, and then the
second differential of the polynomial was obtained to determine the
bending moments along the pile.
Because of the load-bearing mechanism of an end-bearing foun-
dation, piles experience their maximum bending moment at the toe,
where they are connected to the strong bedrock, as shown in Fig. 24.
Overall, the reported pile bending moments for the moment-resisting
building, as shown in Figs. 24(a and c), indicate that the bending
moments at the heads of the piles increased as the SG decreased. By
using Pile M1 under excitation in the 1994 Northridge earthquake as
an example, the bending moment at the head increased by 486% as
the SG increased from 100 to 1,200 mm, whereas Figs. 24(b and d)
show that where Pile SL1 supported the shear wall–braced building

© ASCE 04018007-21 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 under excitation from the 1994 Northridge earthquake, the bending
moment at the pile head increased as the SG decreased, but this trend
was the exact opposite under excitation from the 1995 Kobe earth-
quake. As expected, and by comparing Figs. 24 and 22, the maximum
mobilized bending moment in a pile increased as deflection in the
head increased. Moreover, as the structural shear forces increased,
the bending moment generated at the head of a pile increased (com-
pare Figs. 17 and 24). Furthermore, bending moments along the pile
vary considerably due to seismic pounding, which means that seismic
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
pounding can alter the bending moment along a pile. Therefore, the
effect of seismic pounding must be considered when designing the
foundations of closely spaced structures.
Shear Forces along Piles
Fig. 25 shows the shear forces in pile elements for different cases.
The shear force imposed on piles during excitation is another im-
portant parameter that governs the design of a foundation. The shear

-0.5 0 0.5 1 1.5 -0.5 0 0.5 1 1.5 2

0 0
Shear wall braced building
Moment resisting building
Northridge earthquake (1994),
Northridge earthquake (1994),
Mw = 6.7, PGA = 0.843g
Mw = 6.7, PGA = 0.843g
-5 -5
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
-10 SG = 700 mm -10 SG = 700 mm
SG = 1000 mm SG = 1000 mm

Pile depth (m)

Pile depth (m)

SG = 1200 mm SG = 1200 mm
-15 -15

-20 -20

-25 -25

-30 -30
(a) Shear force (MN) (b) Shear force (MN)

-1.5 -1 -0.5 0 0.5 -1 -0.8 -0.6 -0.4 -0.2 0

0 0
Moment resisting building Shear wall braced building
Kobe earthquake (1995), Kobe earthquake (1995),
Mw = 6.8, PGA = 0.833g Mw = 6.8, PGA = 0.833g
-5 -5
SG = 100 mm SG = 100 mm
SG = 300 mm SG = 300 mm
-10 SG = 700 mm -10 SG = 700 mm
SG = 1000 mm SG = 1000 mm
Pile depth (m)

Pile depth (m)

SG = 1200 mm SG = 1200 mm
-15 -15

-20 -20

-25 -25

-30 -30
(c) Shear force (MN) (d) Shear force (MN)

Fig. 25. Shear forces of piles of (a) the moment-resisting building and (b) the shear wall–braced building during the 1994 Northridge earth-
quake, and (c) the moment-resisting building and (d) the shear wall–braced building during the 1995 Kobe earthquake in conjunction with
different SGs

© ASCE 04018007-22 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007

 force along the pile element [QðzÞ] was also obtained by adopting
the elastic beam theory and Eq. (11)
d3 ½uðzÞ
QðzÞ ¼ EI (11)
dz3
Obviously, the third differential of a polynomial was utilized to
calculate the shear forces along the pile.
It is evident from Fig. 25 that a reduction in the SG alters the
shear force distribution and maximum shear force along a pile.
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
This is an important observation because seismic pounding has a
significant influence on the force developed in the piles. As Figs.
17 and 25 show, seismic pounding changes the shear forces
mobilized in the structure and, consequently, in a pile, and as the
base shear in a superstructure increases, the maximum shear
forces in piles also increase. It is therefore essential to treat
superstructures, foundations, and soil as a whole while consider-
ing possible seismic pounding to obtain a safe and rational
design.
The loads applied to the soil–pile–structure system are a com-
bination of inertial and kinematic loadings. The inertial loading
is induced by the effect of earthquake acceleration on the struc-
tural mass. This inertial loading, stemming from the superstruc-
ture, will be transferred to the foundation and can be represented
using forces and moments on top of the pile foundation, and sub-
sequently, the pile foundation will move and rotate (generating
forces and bending moments in the piles) in an attempt to coun-
teract the applied inertial loading. Kinematic loading, however,
is induced as a result of soil movement in the ground surround-
ing the piles during the earthquake wave propogation from the
deep layers to the ground surface. This kinematic loading, stem-
ming from the soil surrounding the piles, will induce forces and
bending moments in the piles. Effects of inertial forces usually
decay rapidly with depth as the source of the lateral loading is
near the pile head, whereas kinematic loading can cause signifi-
cant stresses and bending moments in the piles deep inside the
ground. Referring to Figs. 24 and 25, the significant bending
moments and shear forces in the piles at a depth of 30 m (where
the piles are socketed in the strong rock, with significant imped-
ance contrasts between soft soil and rock) are mainly due to ki-
nematic loading, whereas contributions of the inertial loading to
the bending moments and shear forces in the piles are more dom-
inant near the ground surface. Moreover, as shown in Figs. 24
and 25, structural pounding significantly influenced the inertial
loading acting on pile heads so that the bending moments and
shear forces generated on the piles near the ground surface were
notably different when pounding occurred (i.e., changes in the
inertial loading due to pounding). In contrast, the difference in
moments and forces developed near pile tips may be due to the
changes in the kinematic loading due to the presence of nearby
pile foundations. Therefore, it is evident that ignoring kinematic
loading in the seismic analysis of a soil–pile–structure system
can result in unrealistic and unsafe prediction of the forces,
bending moments, and the overall performance of the structural
system.
Conclusions
To investigate the impact that seismic pounding has on the
response of midrise buildings sitting on end-bearing piles, a con-
ventional moment-resisting building and two shear wall–braced
buildings resting on end-bearing pile foundations and with
© ASCE 04018007-23
Int. J. Geomech., 2018, 18(4): 04018007
different seismic gaps were numerically simulated. Details of
how this numerical model accounted for the combined effects of
seismic pounding and SPSI were presented. The general contact
interfaces were simulated to consider the interaction between
piles and soil, whereas the contact pair interfaces between build-
ings were used to capture possible seismic pounding. The direct
calculation method was used to carry out a fully nonlinear time-
history dynamic analysis to simulate the dynamic behavior of
soil, pile foundations, and structures, and to capture the effects of
pounding under seismic excitations accurately.
The results show that seismic pounding influences the distri-
bution of shear force along the buildings and increases the story
shear forces at the collision level, which poses a significant threat
to column safety and the overall stability of buildings. Moreover,
neighboring buildings can have confinement effects that could al-
ter the dynamic characteristics of buildings, and help to increase
the shear forces in structural columns. Therefore, the combined
effects of seismic pounding and SSPSI should be considered by
practicing engineers because they can be detrimental to the safety
of buildings, both locally and globally. In this study, lateral build-
ing deflection was influenced by the structural distortion directly
induced by the shear force developed in the building, a rocking
foundation slab, and confinement caused by neighboring build-
ings. The corresponding results reveal that lateral deflection may
decrease due to confinement or increase due to variations in the
dynamic characteristics.
Because of the load-transmission mechanism of an end-
bearing foundation, lateral pile deflection can be influenced by
the base shear developed in the superstructure, which further indi-
cates that the occurrence of seismic pounding can increase lateral
pile deflection. Therefore, the effects of seismic pounding must be
considered when designing an end-bearing foundation for buildings.
Moreover, the maximum shear forces and their distribution along an
end-bearing pile can be impacted by the seismic pounding effect,
which can be overlooked in common design practice where super-
structures and foundations are treated separately. Thus, to obtain a
safe and rational design, the superstructure, foundation, and soil
must be considered with regard to the possibility of simultaneous
seismic pounding. The findings of this paper can provide engineers
with a better insight into how seismic pounding affects the seismic
performance of buildings and the response of end-bearing founda-
tions in soft soil. According to the results of this study, a minimum
SG of 1.75% of the structural height would be required to avoid
structural pounding between midrise buildings on pile foundations.
This minimum required SG might be more for buildings with
shallow foundations. The authors recommend conducting time-
history SPSI analysis to assess the possibility of structural pound-
ing and designing structural and foundation elements accord-
ingly. Designers may have the choice of avoiding structural
pounding by choosing a large-enough SG or considering the
pounding effects in the design when the structural pounding can-
not be avoided (e.g., existing retrofitted structures).
References
Abaqus [Computer software]. Dassault Systèmes Simulia Corp.,
Minneapolis.
ACI (American Concrete Institute). (2008). Building code requirements for
structural concrete and commentary (ACI 318-08), Washington, DC.
Anagnostopoulos, S. A. (1988). “Pounding of buildings in series during
earthquakes.” Earthquake Eng. Struct. Dyn., 16(3), 443–456.
Anagnostopoulos, S. A. (1996). “Building pounding re-examined: How se-
rious a problem is it?” Proc., Eleventh World Conf. of Earthquake
Engineering, Elsevier, Pergamon, Turkey, p. 2108.
Int. J. Geomech.
 Borja, R. I., Wu, W.-H., Amies, A. P., and Smith, H. A. (1994). “Nonlinear
lateral, rocking, and torsional vibration of rigid foundations.” J.
Geotech. Eng., 10.1061/(ASCE)0733-9410(1994)120:3(491), 491–513.
Bowles, J. E. (1988). Foundation analysis and design, McGraw-Hill,
New York.
Carbonari, S., Dezi, F., and Leoni, G. (2011). “Linear soil–structure interac-
tion of coupled wall–frame structures on pile foundations.” Soil Dyn.
Earthquake Eng., 31(9), 1296–1309.
CEN (European Committee for Standardization). (2005). “Design of struc-
tures for earthquake resistance.” Eurocode 8, Brussels, Belgium.
Chouw, N. (2002). “Influence of soil-structure interaction on pounding
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
response of adjacent buildings due to near-source earthquakes.” J. Appl.
Mech., 5, 543–553.
Chouw, N., and Hao, H. (2012). “Pounding damage to buildings and bridges
in the 22 February 2011 Christchurch earthquake.” Int. J. Protective
Struct., 3(2), 123–140.
Chu, D., and Truman, K. Z. (2004). “Effects of pile foundation configura-
tions in seismic soil-pile-structure interaction.” Proc., 13th World Conf.
on Earthquake Engineering, Canadian Association for Earthquake
Engineering, Vancouver, BC, Canada, 1551.
Comodromos, E., and Papadopoulou, M. (2012). “Response evaluation
of horizontally loaded pile groups in clayey soils.” Geotechnique,
62(4), 329.
Das, B. M. (1983). Fundamentals of soil dynamics, Elsevier, New York.
FEMA. (2000). “Prestandard and commentary for the seismic rehabilitation
of buildings.” FEMA 356, Washington, DC.
GB50011. (2010). Code for seismic design of buildings, China Building
Industry Press, Beijing (in Chinese).
Ghee, E. H., and Guo, W. D. (2011). “FLAC3D analysis on soil moving
through piles.” Proc., 2nd Int. Symp. on Frontiers in Offshore
Geotechnics, Taylor & Francis Group, Boca Raton, FL.
Hao, H. (2015). “Analysis of seismic pounding between adjacent build-
ings.” Aust. J. Struct. Eng., 16(3), 208–225.
Hao, H., Liu, X. Y., and Shen, J. (2000). “Pounding response of adjacent
buildings subjected to spatial earthquake ground excitations.” Adv.
Struct. Eng., 3(2), 145–162.
Hayashi, Y., and Takahashi, I. (2004). “Soil-structure interaction effects on
building response in recent earthquakes.” Proc., Third UJNR Workshop
on Soil-Structure Interaction, University of Southern California Press,
Menlo Park, CA.
Hokmabadi, A. S., and Fatahi, B. (2016). “Influence of foundation type on
seismic performance of buildings considering soil–structure interac-
tion.” Int. J. Struct. Stab. Dyn., 16(8), 1550043.
Hokmabadi, A. S., Fatahi, B., and Samali, B. (2012). “Recording inter-
storey drifts of structures in time-history approach for seismic design of
building frames.” Aust. J. Struct. Eng., 13(2), 175–179.
Hokmabadi, A. S., Fatahi, B., and Samali, B. (2015). “Physical model-
ing of seismic soil-pile-structure interaction for buildings on soft
soils.” Int. J. Geomech., 10.1061/(ASCE)GM.1943-5622.0000396,
04014046.
Karamodin, A., and Kazemi, H. (2010). “Semi-active control of structures
using neuro-predictive algorithm for MR dampers.” Struct. Control
Health Monit., 17(3), 237–253.
Karayannis, C. G., and Favvata, M. J. (2005). “Earthquake induced interac-
tion between adjacent reinforced concrete structures with non-equal
heights.” Earthquake Eng. Struct. Dyn., 34(1), 1–20.
Koskinen, M. (2005). “Modeling of soil-structure interaction between rail-
way bridge and soil.” Proc., ABAQUS Users’ Conf., Simulia-Dassault
Systèmes Press, Stockholm, Sweden.
Kramer, S. L. (1996). Geotechnical earthquake engineering, Pearson
Education India, Uttar Pradesh, India.
Kumar, A., Choudhury, D., and Katzenbach, R. (2016). “Effect of earth-
quake on combined pile–raft foundation.” Int. J. Geomech., 10.1061
/(ASCE)GM.1943-5622.0000637, 04016013.
Kumar, A., Choudhury, D., Shukla, J., and Shah, D. (2015). “Seismic
design of pile foundation for oil tank by using PLAXIS3D.” Disaster
Adv., 8, 33–42.
Madani, B., Behnamfar, F., and Riahi, H. T. (2015). “Dynamic response of
structures subjected to pounding and structure–soil–structure interac-
tion.” Soil Dyn. Earthquake Eng., 78, 46–60.
© ASCE 04018007-24
Int. J. Geomech., 2018, 18(4): 04018007
Maheshwari, B., Truman, K., El Naggar, M., and Gould, P. (2004). “Three-
dimensional nonlinear analysis for seismic soil–pile-structure interac-
tion.” Soil Dyn. Earthquake Eng., 24(4), 343–356.
Maheshwari, B., and Watanabe, H. (2006). “Nonlinear dynamic behavior of
pile foundations: Effects of separation at the soil-pile interface.” Soils
Found., 46(4), 437–448.
Mahmoud, S., Abd-Elhamed, A., and Jankowski, R. (2013).
“Earthquake-induced pounding between equal height multi-storey
buildings considering soil-structure interaction.” Bull. Earthquake
Eng., 11(4), 1021–1048.
Mahmoud, S., Chen, X., and Jankowski, R. (2008). “Structural pounding
models with Hertz spring and nonlinear damper.” J. Appl. Sci., 8(10),
1850–1858.
Matinmanesh, H., and Asheghabadi, M. S. (2011). “Seismic analysis on
soil-structure interaction of buildings over sandy soil.” Procedia Eng.,
14, 1737–1743.
Naserkhaki, S., Aziz, F. N. A., and Pourmohammad, H. (2012). “Earthquake
induced pounding between adjacent buildings considering soil-structure
interaction.” Earthquake Eng. Eng. Vib., 11(3), 343–358.
Newmark, N. M. (1973). “An exact correction for accelerometer error in
dynamic seismic analysis.” Bull. Seismol. Soc. Am., 59(1), 705–715.
Nguyen, Q. V., Fatahi, B., and Hokmabadi, A. S. (2016). “The effects of
foundation size on the seismic performance of buildings considering
the soil-foundation-structure interaction.” Struct. Eng. Mech., 58(6),
1045–1075.
Nguyen, Q. V., Fatahi, B., and Hokmabadi, A. S. (2017). “Influence of
size and load-bearing mechanism of piles on seismic performance of
buildings considering soil–pile–structure interaction.” Int. J. Geomech.,
10.1061/(ASCE)GM.1943-5622.0000869, 04017007.
Oden, J., and Martins, J. (1985). “Models and computational methods for
dynamic friction phenomena.” Comput. Methods Appl. Mech. Eng.,
52(1–3), 527–634.
Park, D., and Hashash, Y. M. (2004). “Soil damping formulation in nonlinear
time domain site response analysis.” J. Earthquake Eng., 8(2), 249–274.
Poulos, H. G., and Davis, E. H. (1980). Pile foundation analysis and design,
John Wiley and Sons, Hoboken, NJ.
Rahman, A., Carr, A., and Moss, P. (2001). “Seismic pounding of a case of
adjacent multiple-storey buildings of differing total heights considering
soil flexibility effects.” Bull. N. Z. National Soc. Earthquake Eng., 34,
40–59.
Randolph, M., Dolwin, J., and Beck, R. (1994). “Design of driven piles in
sand.” Geotechnique, 44(3), 427–448.
Rosenblueth, E., and Meli, R. (1986). “The 1985 earthquake: Causes and
effects in Mexico City.” Concr. J., Am. Conc. Inst., 8, 23–24.
SAP2000 14 [Computer software]. Computers and Structures Inc.,
Berkeley, CA.
Semblat, J.-F. (2011). “Modeling seismic wave propagation and amplifica-
tion in 1D/2D/3D linear and nonlinear unbounded media.” Int. J.
Geomech., 10.1061/(ASCE)GM.1943-5622.0000023, 440–448.
Shakya, K., and Wijeyewickrema, A. C. (2009). “Mid-column pounding of
multi-story reinforced concrete buildings considering soil effects.” Adv.
Struct. Eng., 12(1), 71–85.
Shelke, A., and Patra, N. (2008). “Effect of arching on uplift capacity of pile
groups in sand.” Int. J. Geomech., 10.1061/(ASCE)1532-3641(2008)8:
6(347), 347–354.
Shing, P. B., and Tanabe, T. (2001). Modeling of inelastic behavior of RC
structures under seismic loads, ASCE, Reston, VA.
Small, J., and Zhang, H. (2002). “Behavior of piled raft foundations under
lateral and vertical loading.” Int. J. Geomech., 10.1061/(ASCE)1532
-3641(2002)2:1(29), 29–45.
Song, E., Ye, L., Zhao, S., and Yang, Q. (2008). “Analysis on seismic dam-
age of buildings in the Wenchuan earthquake.” J. Build. Struct., 29, 1–9
(in Chinese).
Standards Australia. (2002). “Structural design actions—Permanent,
imposed and other actions standards Australia.” AS1170.1, Sydney,
Australia.
Standards Australia. (2007). “Structural design actions; Part 4: Earthquake
actions in Australia.” AS1170.4, Sydney, Australia.
Standards Australia. (2009a). “Concrete structures.” AS3600, Sydney,
Australia.
Int. J. Geomech.
 Standards Australia. (2009b). “Piling design and installation.” AS2159,
Sydney, Australia.
Tabatabaiefar, H. R., and Fatahi, B. (2014). “Idealisation of soil–structure
system to determine inelastic seismic response of mid-rise building
frames.” Soil Dyn. Earthquake Eng., 66, 339–351.
Tabatabaiefar, H. R., Fatahi, B., and Samali, B. (2014). “An empirical rela-
tionship to determine lateral seismic response of mid-rise building
frames under influence of soil–structure interaction.” Struct. Des. Tall
Special Build., 23, 526–548.
Tavakoli, H., Naeej, M., and Salari, A. (2011). “Response of RC structures
subjected to near-fault and far-fault earthquake motions considering
Downloaded from ascelibrary.org by "National Institute of Technology, Meghalaya" on 08/01/19. Copyright ASCE. For personal use only; all rights reserved.
Uz, M., and Hadi, M. N. (2011). “Seismic history analysis of asymmetrical
adjacent buildings with soil-structure interaction consideration.”
Earthquake Resistant Eng. Struct. VIII, 120, 225–236.
Vucetic, M., and Dobry, R. (1991). “Effect of soil plasticity on cyclic response.”
J. Geotech. Eng., 10.1061/(ASCE)0733-9410(1991)117:1(89), 89–107.
Wolf, J., and Skrikerud, P. (1980). “Mutual pounding of adjacent structures
during earthquakes.” Nucl. Eng. Des., 57(2), 253–275.
Wolf, J. P. (1989). “Soil-structure-interaction analysis in time domain.”
Nucl. Eng. Des., 111(3), 381–393.
Zou, L., Huang, K., Wang, L., and Fang, L. (2013). “Pounding of adjacent
buildings considering pile-soil-structure interaction.” J. Vibroeng., 15,

soil-structure interaction.” Int. J. Civ. Struct. Eng., 1, 881–896. 905–918.

© ASCE 04018007-25 Int. J. Geomech.

Int. J. Geomech., 2018, 18(4): 04018007