Soil Dynamics and Earthquake Engineering 84 (2016) 251–268

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Effect of in-situ variability of soil on seismic design of piled raft

supported structure incorporating dynamic soil-structure-interaction
Biplab Das a, Rajib Saha b,n, Sumanta Haldar c
a
Government of Tripura, Tripura (N), India
b
Department of Civil Engineering, National Institute of Technology, Agartala 799046, Tripura (W), India
c
Civil Engineering, School of Infrastructure, Indian Institute of Technology Bhubaneswar, Bhubaneswar 751013, Odisha, India

art ic l e i nf o a b s t r a c t

Article history: Inherent variability of soil considerably affects the seismic design of piled raft supported structures.
Received 16 August 2015 Conventional design of such structure adopts fixity at base level of superstructure and pile head. How-
Received in revised form ever, soil–pile–superstructure interaction largely affects the fundamental frequency and design forces in
27 December 2015
columns and piles. In contrast, fixed base assumption cannot capture soil structure interaction (SSI)
Accepted 27 February 2016
effect. In addition, uncertainty in soil may further leads to a change in the dynamic behavior of the
system. This study examines the effect of inherent variability of undrained shear strength of soil in
Keywords: seismic design of structures supported by piled raft foundation embedded in soft clay. Superstructure is
Soil structure interaction modeled as lumped mass stick model and piled raft slab is modeled as rigid plate. Pile is modeled as
Dynamic analysis
Euler–Bernoulli beam element and soil resistance is modeled using linear Winkler springs attached to
Probability of failure
the pile. Dynamic analysis is carried out in time domain to estimate the responses. Monte Carlo simu-
Monte Carlo simulation
Uncertainty lation technique is used for probabilistic assessment of the fundamental frequency and forces at column
Piled raft and pile attributing a wide range of parametric variation of a representative soil–piled raft–super-
structure system. The study shows that the fundamental frequency and forces in column and pile
changes significantly due to soil variability.
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction the failure of pile foundation supported structure in various seis-

Sustainable design of structure is a prime issue to ensure safe,
economic and environmental friendly structure during any dis-
astrous event, especially during strong earthquake. Probabilistic
analysis of structural response incorporating different uncertain
parameters may help to arrive at a sustainable design solution.
Piled raft foundation is commonly used to support such heavy as
well as important structures such as tall towers, bridges, nuclear
structures etc. in soft to medium soil. Seismic design of structures
supported by piled raft foundation is a challenging and complex
problem as the mechanism of transfer of lateral loads to the col-
umn and soil from pile is essentially dependent on soil type and
pile, which is attributed by soil structure interaction (SSI) problem.
Traditionally, seismic design of such structure is performed
assuming fixed base condition. As a result, the complex soil–piled
raft–structure interaction is ignored which may have a serious
implication on the dynamic response of structure [1–4]. However,
n
Corresponding author.
E-mail addresses: erbiplab08@gmail.com (B. Das),
rajib.iitbbsr@gmail.com (R. Saha), sumanta.haldar@gmail.com,
sumanta@iitbbs.ac.in (S. Haldar).
mic events (e.g.1985 Mexico City earthquake, 1989 Loma Prieta
earthquake, 1995 Kobe earthquake) indicated the importance of
SSI in seismic design. On the other hand, a perception of beneficial
implication of SSI generally prevails in designer’s mind based on
few codal guidelines [5,6] and ignored in seismic design. Several
other studies have pointed out the importance of soil–pile foun-
dation–structure interaction to obtain the design response of the
structure [7,8,3,9–11].
It is observed that previous studies mostly focused on intricacy
of pile-soil modeling and method of analysis, whereas, a limited
effort has been rendered to see the effect of such interaction on
overall seismic behavior of structure [12,13,8,10,14]. A recent study
by Saha et al. [15] indicated that relative acceleration of heavy raft
and upper part of the pile with respect to the neighboring soil
attracts additional lateral force. They pointed that this additional
force may leads to considerable increase in pile head shear, which
results in unsafe design of pile and over-safe design of column and
contrary to the general notion of SSI that intuitively assume
reduction in shear forces in column and pile owing to the period
lengthening of structure and increased damping due to soil media
[6,2]. In reality in-situ properties of soil in a distinct geological
layer are highly uncertain [16]. Inherent variability of soil may lead

http://dx.doi.org/10.1016/j.soildyn.2016.02.015
0267-7261/& 2016 Elsevier Ltd. All rights reserved.
252 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
to uncertainty in the responses of such system which results in
change in column and pile shear forces as compared to fixed base
response. In order to address the uncertainty in soil properties
along with SSI, probabilistic assessment is required in the context
of current design paradigm.
Probabilistic assessment of piled raft supported structure
incorporating SSI addressing the seismic design implication is
rather limited [17]. In some studies, the effect of dynamic soil
structure interaction on structural response considering system
variability and uncertainty in ground motions were studied for
shallow foundation supported structure through comprehensive
numerical simulation [18,19]. Tandjiria et al. [20], Pula and
Rozanski [21] and Chan and Low [22] outlined probabilistic seis-
mic design of pile with an emphasis on different methods of
probabilistic analysis. Further, Pula and Rozanski [21] presented a
complete solution to the problem of random lateral bearing
capacity of rigid piles embedded into non-stratified homogeneous
soil. Haldar and Babu [23] presented the effect of in-situ variability
of soil and forces on seismic design of pile considering pseudo-
static analysis.
In this study, the effect of inherent variability of soil shear
strength and Young’s modulus parameters on dynamic response of
soil–piled raft–structure system using probabilistic analysis is
addressed. Beam on linear Winkler foundation (BWF) model is
used to model the soil-piled raft foundation system for the sake of
simplicity. Finite element analysis is performed using acceleration
time history of seismic loading to obtain the dynamic response of
representative soil–piled raft–structure system. Stiffness of soil
springs is modeled as random variable. Variability in soil stiffness
is defined by a coefficient of variation (COV) with respect to an
assumed mean value from literature. The variation of stiffness at
any depth is defined by Log Normally distributed function as soil
properties resemble non-negative values. Based on the previous
studies, it is observed that variability of density of soil does not
exceed 10% [24–26], hence it is considered as deterministic para-
meter. Monte Carlo (MC) simulation technique is adopted for
obtaining the probabilistic response of the system. The effect of
inherent variability on dynamic characteristics and elastic
response of the system is studied for various soil, pile and struc-
tural parameters, namely relative stiffness of pile with respect to
soil (Ep/Es), length to diameter ratio of pile (L/d), spacing to dia-
meter ratio of pile (S/d) and different period of superstructure
under fixed base case (Tfixed). In order to investigate the sole effect
of variability of soil properties, the dynamic response of structure
is obtained for three different cases, such as (a) fixed base condi-
tion, (b) deterministic SSI condition and (c) probabilistic SSI con-
dition. Further, the probability of failure of pile foundation is cal-
culated considering ultimate limit state and displacement criteria
in the present study. Finally design implications are also sug-
gested. This study may give a valuable input to modify the con-
ventional seismic design approach for piled raft supported
structure.
2. System modeling
The piled-raft supported structure is modeled using finite ele-
ment approach which is described in the following sections.
2.1. Superstructure modeling
Superstructure is idealized as a lumped mass stick model
having single degree of freedom (SDOF). The lumped mass model
is used to represent a three dimensional space frame super-
structure supported on soil–piled raft foundation. The elastic beam
column element is used to represent column of the stick model.
Two different periods of superstructure under fixed based condi-
tion (Tfixed), namely 0.4 s and 2.0 s, representing typical short and
long period structure are considered. The periods are determined
by adjusting the lumped mass and lateral stiffness of the column.
Column stiffness is attributed by assigning appropriate sectional
properties.
2.2. Foundation modeling
Foundation primarily consists of two major components,
namely raft and pile–soil system which are discussed as follows:
2.2.1. Raft-soil modeling
A 10 m
10 m raft is modeled as a four noded shell element,
each node having six degrees of freedom (3-rotations and
3-translations) and discretized into (40
40) small elements based
on a convergence study. Total mass of the raft is calculated and
assigned as a distributed mass to all nodes. Soil beneath the raft is
idealized as equivalent linear springs (Winkler soil spring ideali-
zation) connected with each node of raft in all translational
degrees of freedom. Dashpots are also attached in parallel with
these springs to incorporate the effect of soil damping. Stiffness of
distributed lateral springs in two mutually perpendicular hor-
izontal directions (lateral (Kx1) and longitudinal (Kx2)) are assigned
following Dutta et al. [27] as presented below,
K x1 ¼ K xG1 =n2 ð1aÞ
K x2 ¼ K xG2 =n2 ð1bÞ
where KxG1 and KxG2 are the overall lateral stiffness of soil spring
[28] attached beneath the raft in lateral and longitudinal direc-
tions, respectively and n is number of elements. Vertical stiffness
formulation suggested by Gazetas [28] takes care for the coupled
lateral-rocking mode of vibration. Hence, vertical spring stiffness
(Ky) values are compared with the values provided by Gazetas [28]
and are adapted in a distributed form by Dutta et al. [27],
5:4 Es LR
Ky ¼
U ð2Þ
n2 þ 2 2ð1  νÞð1 þ νÞ
where Es is Young’s modulus of soil, LR is the length of raft and ν is
the Poisson’s ratio of soil. Stiffness of all springs connected at
intermediate nodes of the raft assumed to be same. However, at
the corner and peripheral nodes, spring stiffness one-fourth or half
of the stiffness of the springs at intermediate depending on their
influence area. The expressions for K xG1 and K xG2 are presented in
Table 1.
2.2.2. Pile–soil modeling
Pile is modeled using an elastic beam column element having
6 degrees of freedom at each node (3-rotations and 3-translations)
which is further discretised into 20 divisions in vertical direction.
Beam on Winkler foundation (BWF) modeling approach is used to
model the pile-soil interaction due to horizontally applied
dynamic loading. Two horizontal springs are attached to each pile
in two mutually perpendicular horizontal directions (i.e. lateral
Table 1
Stiffness of equivalent springs along various degree of freedom [28].
Degrees of freedom Stiffness of equivalent soil spring
Horizontal (KxG1) (lateral direction) E s lð2 þ 2:54χ 0:85 Þ
ð1 þ νÞð2  νÞ
0:1Es l½1  ðB=lÞ
Horizontal (KxG2) (longitudinal direction) E s lð0:73 þ 1:54χ 0:75 Þ
 ð0:75
ð1 þ νÞð1  νÞ  νÞð1 þ νÞ
2
where χ ¼ Ab =4l , Ab is the area of the foundation considered, B and l are half width
and half length of a rectangular foundation, respectively, Es and ν are Young’s
modulus and Poisson’s ratio of soil respectively.
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268 253

and longitudinal directions) and a spring is attached to each pile respect to analytical and numerical solutions reported in literature
node in vertical direction to simulate horizontal and rocking [32,29]. Suitability of linear idealization of soil resistance at piled
motion of piled raft foundation. Further, dashpots are also raft has also been conducted by the authors and their co-workers
attached in parallel to each of the springs to consider the damping [33]. The responses were estimated by the linear Winkler spring
effects of pile–soil system. Soil springs are assumed to be linear model and compared with the results using nonlinear Winkler soil
and lateral stiffness of such discrete horizontal springs is calcu- spring as proposed by Boulanger et al. [9] and Curras et al. [12].
lated [29] as, This study indicates marginal difference in elastic response of
kx ¼ 1:2Es ð3Þ structure for both the models particularly in low to moderate
where kx and Es are lateral spring stiffness exerted by projected range of ground motions. In general, it was observed that linear
contact area of pile with soil and Young’s modulus of soil respec- idealization of soil over-predict the shear forces at column and pile
tively. The variation of Es is considered uniform throughout the head than that of nonlinear soil idealization.
depth in this study.
Gapping action between soil and pile under tension is not Table 2
Superstructure, raft, pile and soil data assumed for analysis.
considered, since gapping effect was found to be marginal in case
of fixed head pile [30]. Stiffness of vertical springs connected to Superstructure Value
pile node except the end node is calculated from the frictional
resistance between soil and pile shaft, while, the stiffness of ver- Height of the column (m) 3.0
Lumped mass (kg) 173400 (0.40 s),1,180,000 (2.0 s)
tical springs connected to tip of pile is calculated from tip resis-
Sectional area of columns (m2) 0.60, 0.241
tance of pile. The stiffness of vertical spring is calculated using the Young's modulus of column (kN/m2) 21.78
106
standard formula of shaft friction [31], Raft
Size of raft 10 m
10 m
kvertical ¼ αC u d ð4Þ Thickness (m) 0.9 m (very soft), 1.1 m (soft),
and the stiffness of vertical spring connected to pile tip node 1.6 m (moderately stiff)
Pile
which takes care of the end bearing action of the pile can be given Pile diameter, d (m) 0.3, 0.5
by, Slenderness ratio (L/d) 40, 60 and 80
Pile length, L (m) 18, 20, 24
kendbearing ¼ N c cb d ð5Þ
Young’s modulus of pile, Ep(KN/m2) 21.78
106
where α ¼adhesion factor of soil, Cu ¼undrained shear strength at Poisson’s ratio of concrete 0.17
Section modulus, Z (m3) 0.0045, 0.020
periphery of pile and d ¼width of projected area of element which Flexural strength of pile, Fy (kN/m2) 3050
is equal to diameter of pile, Nc ¼bearing capacity factor, Flexural moment of pile section, My 13.725 (d ¼0.3 m), 63.541 (d ¼ 0.5 m)
cb ¼undrained shear strength of soil at base of pile. A schematic (kN-m)
diagram of the finite element model of soil–piled raft–structure Soil
Soil consistency Very soft, soft, moderately stiff
system is presented in Fig. 1. Further, the details of superstructure,
Undrained shear strength, Cu (kN/m2) 9.80, 14.50, 35.00
raft, pile and soil characteristics assumed for analysis is presented SPT (N) value 1, 3, 6
in Table 2. γ sat (kN/m3) 13.50, 17.00, 18.50
Idealization for soil–foundation system to behave as a linear Young’s modulus of soil, Es (kN/m2) 2500, 5000, 15000
Poisson’s ratio (ν) 0.40
elastic material during dynamic loading was validated with

Fig. 1. Finite element modeling of soil–pile raft–superstructure system.

254 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
2.3. Damping
Previous studies [34] suggest general range of damping of
building system varies within 2–8%. However, this may go up to as
high as 20% for the subgrade medium [28,35]. Consideration of 5%
material damping of soil could be a realistic assumption for
understanding the behavior of pile–raft system [36]. However, the
effect of combined material and radiation damping due to inter-
action between shallow foundation and soil with various feasible
choices of damping ratio of soil–foundation system, for instance
within a range of 2–30% of critical damping along with constant
structural damping of 5% of critical damping on overall response of
soil–foundation–structure system was investigated by Roy and
Dutta [37]. The objective of this previous study was to see the
effect of soil damping which comprises of both material as well as
radiation part on calculating overall flexible base damping para-
meter and most effectively on the maximum response of whole
system. The study indicates that consideration of high damping for
soil (including material and radiation) even up to the extent of 30%
of critical damping makes marginal difference in maximum
response of the system that of case with 5% of critical damping of
soil–foundation system as well as super-structure. Therefore, to
strike a balance between rigor and accuracy, 5% of critical damping
in each mode of vibration of Pile-raft-soil and superstructure
system is considered for the present study regardless of structural
support condition.
2.4. Ground motion
For the sake of simplicity, a sinusoidal motion having ampli-
tude of 0.02 g is applied at the base of structure, i.e. at the support
level of foundation for both deterministic and probabilistic ana-
lysis. Previous studies showed that resonating behavior of pile
supported structure in soft soil during past earthquakes that
resulted to failure of the system [38,39]. Hence, frequency of the
motion is assigned same as the fundamental frequency of soil–pile
raft–structure system in order to simulate the resonating response
of the system. Furthermore, harmonic base excitation replicates
earthquake motion reasonably good for linear elastic problem as
indicated in various other previous studies [7,29].
3. Probabilistic analysis
3.1. Modeling of uncertainty in soil parameters
In this study uncertainty in soil parameter is modeled con-
sidering undrained shear strength (Cu) and Young’s modulus (Es)
of soil as random variables. The property variations of soil is
considered by the mean value and coefficient of variation (COV) of
Es and Cu. The soil Young’s modulus and (Es) and undrained shear
strength (Cu) of soil are statistically characterized by the log-
normal distribution due to non-negativity of soil parameters [23].
The sample realizations of Cu and Es are obtained from the stan-
dard Gaussian random field using the following transformations,
 
Esi ¼ exp μln Es þ σ ln Es Gi ð6aÞ
 
C ui ¼ exp μln cu þ σ ln cu Gi
where Esi and cui are the ith realization of Es and Cu respectively,
and Gi is a zero mean, unit variance Gaussian random numbers,
μln Es , μln cu , σ ln Es and σ ln cu are described as,



μln Es ¼ ln μEs  0:5σ 2ln Es and μln cu ¼ ln μcu  0:5σ 2ln cu ð7aÞ
   
σ 2ln Es ¼ ln 1 þ COV2Es and σ 2ln cu ¼ ln 1 þCOV2cu ð7bÞ
where COVEs ¼ σ Es =μEs is the coefficient of variation of Es and
μEs and σ Es are the sample mean and standard deviation of Es,
respectively. Similarly, COVcu ¼ σ cu =μcu is the coefficient of varia-
tion of cu and μcu and σ cu are the sample mean and standard
deviation of cu, respectively.
The mean values of Young’s modulus ðμEs Þ and undrained shear
strength of soil ðμcu Þ are used in deterministic analysis as given in
Table 2. Present study considers COVcu and COVEs as 20%, 30% and
50% [16]. It is to be noted that the stiffness of soil springs are
estimated using Eqs. (4) and (5) from the realizations of Cu and Es.
It is assumed that stiffness of the soil springs follow log-normal
distribution due to linear relationship among spring stiffness, Cu
and Es.
3.2. Monte Carlo simulation
Monte Carlo simulation (MCS) is used to generate sample rea-
lizations of cu and Es for the selected ranges of mean and COV
values. The realizations of soil spring stiffness are estimated based
on each realization of Es and cu using Eqs. (4) and (5). For instance,
200 realizations of lateral soil spring stiffness, i.e., kx are calculated
using Eq. (4) considering 200 numbers of log-normally distributed
randomly generated Es values obtained from Eq. (6a). Fig. 2 presents
the realizations of lateral soil spring stiffnesses generated from
Monte Carlo simulation along with mean stiffness. It is observed
that minimum and maximum value of spring stiffness is obtained as
approximately 1000 kN/m and 13,000 kN/m respectively consider-
ing COVEs ¼ 50%. For each realization of soil stiffness, dynamic ana-
lysis of the whole structure is carried out using OpenSEES [40] code.
A convergence study is performed to determine the number of
sample realizations for MCS analysis. Realizations of fundamental
period of the structure considering interaction between piled raft–
soil–superstructure system are obtained for different number of
realizations of spring stiffness values considering COV¼20%,
Tfixed ¼ 0.40 s and very soft clay (Ep/Es ¼10,000). The mean and
standard deviation of the fundamental period of the entire system
(Tssi(mean)) is presented as a function of number of Monte Carlo
simulations (Fig. 3(a) and (b)). Fig. 3(a) and (b) indicates that the
fluctuation of (Tssi(mean)) is marginal for number of trials beyond 180.
Further, another plot of COVcu ¼ 50% L/d¼ 80 and S/d¼ 3 is presented
in Fig. 3(c) and (d) which will be the worst case in order to verify
convergence of results. Hence, a choice of 200 numbers of iterations
is adopted in this study for further analysis. However, 200 number
of MCS may not valid for nonlinear pile–soil system which may be
considered as limitation of present study.
ð6bÞ
Fig. 2. Realizations of lateral stiffness of soil springs attached to pile obtained from
Monte Carlo simulation for COVEs ¼ 50%.
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268 255

Fig. 3. Variation of mean and standard deviation of SSI period of the structure with respect to the number of MCS trials for Tfixed ¼0.40 s in very soft soil supported by pile
group of 3
3 with S/d ¼3: (a) and (b) for the case of L/d¼ 60 and COVcu (or COVEs) ¼ 20%, (c) and (d) for the case of L/d ¼80 with COVcu (or COVEs) ¼ 50%.

4. Modeling parameters and parametric cases
A parametric study considering two representative natural
periods of vibration at fixed base condition (Tfixed), such as 0.40 s
and 2.0 s are considered, which represents relatively short and
long period structures, respectively. The structures supported by
piled raft foundation having three different L/d ratios, namely 40,
60 and 80 which are embedded in clay is considered in this study.
Three different types of soil stiffnesses, namely very soft, soft and
moderately stiff are considered which represents three relative
stiffness values of pile and soil (Ep/Es) ,namely 10,000, 5000 and
1500 respectively. The thickness of raft is calculated considering a
moderately rigid raft having relative stiffness (krs)¼ 1.0, as defined
in Clancy and Randolph (1996). Pile spacing (S) to diameter (d)
ratio (S/d) is kept constant to be 3 for three different L/d ratios.
However, limited cases with S/d ¼5 and 7 are considered for pile
group of L/d ¼40 supporting structure of Tfixed ¼0.40 s in order to
study the effect of variability of soil stiffness for different S/d ratios.
In fact, adoption of S/d ¼5 and 7 found to be feasible for gravity
loading induced by Tfixed ¼0.40 s structure with pile group of
L/d ¼40.
The superstructure is idealized as lumped mass stick model.
Floor dimension of 8.5 m
8.5 m is assumed for superstructure
which is supported by four columns having single bay. Multistoried
structures are idealized as single storey system having equivalent
natural period. To have an overview on the behavior of different
multistory structures with various heights, a 4storied and 16 storied
building for representative fundamental period of 0.4 s and 2 s
respectively are considered representing typical short and long
period structure. Two different column sections are chosen (Table 2)
in order to satisfy two different Tfixed values (i.e. 0.4 s and 2 s). Mass
of the stick model is calculated considering total gravity load of
8.0 kN/m2 per floor which consists of dead and live load. A piled raft
foundation having plan area of 10 m
10 m is considered for
256 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268

Table 3
Summary of parametric case studies.

Case study identification Subsoil condition Ep/Es (Tfixed) s Raft thickness (krs ¼1.0)(m) Pile group designed (m
m) L/d d (m) S/d

A-1 Very soft 10,000 0.40 0.9 4
4 40 0.5 3, 5, 7

A-2 3
3 60 0.3 3
A-3 2
3 80 0.3
A-4 2.0 10
10 60 0.3
A-5 9
9 80 0.3
B-1 Soft 5000 2.0 1.1 6
6 40 0.5 3
B-2 7
7 60 0.3
B-3 5
6 80 0.3
C-1 Moderately stiff 1500 2.0 1.6 3
3 40 0.5 3
C-2 4
5 60 0.3
C-3 4
4 80 0.3

supporting the superstructure load. Floating pile group is designed Table 4

based on loads from superstructure, raft and undrained shear Probabilistic parameters.
strength of soil. It is assumed that a part of the superstructure
Parameter Values
weight is taken by the raft based on the allowable bearing capacity
of soil. The remaining weight is considered to be carried by piles. Mean Young’s modulus ðμEs Þ, kPa 2500, 5000, 15,000
Summary of all the parameters pertaining to soil, piled raft and Mean value of soil undrained shear strength ðμcu Þ, kPa 9.80, 14.50, 35.00
superstructure is given in Table 2. It is worth to mention piled raft Coefficient of variation, COVCu (%) 20, 30, 50
foundation is a feasible option in case of very soft clay and multi- Probability distribution function, pdf Log-normal

storey structure having long period of vibration. In few exceptional

cases, short period structures, for instance, nuclear power plant
structures, bridge piers, abutments and few important buildings
supported on piled raft foundation owing to heavy load from
superstructure. Hence, in this study short period structure
(Tfixed ¼0.40 s) founded in very soft soil and long period structure
(Tfixed ¼2.0 s) founded in soft and moderately stiff soil are con-
sidered for completeness. However, piled raft foundation for soft to
moderately stiff soil could be an uneconomical solution, hence not
included in this study. Summary of all the cases considered in this
study is presented in Table 3 and uncertainty parameters associated
with these parametric models is presented in Table 4.
Results (denoted as Parameter which are time period of structure
and shear force at column and pile) are obtained for three different
conditions: (a) deterministic analysis of superstructure which is sub-
jected to dynamic loading under fixed base assumption where in-situ
variability cannot be considered, (b) deterministic analysis considering
superstructure supported by piled raft foundation with SSI condition
and (c) probabilistic analysis considering superstructure supported by
piled raft foundation with SSI condition and in-situ variability of soil.
For design purpose, Paramterfixed/Parameterssi(det) shows change in
response compared to conventional fixed base case. Whereas,
Parmaterssi(mean)/Parameterssi(det) or (or Parameter,ssi(mean)/Parameter,fixed)
shows the effect of in-situ variability of soil on responses incorporating
SSI condition which may help to identify the importance of soil
variability separately. The main objective of presenting Parameter,ssi
(mean)/Parameter,fixed and Paramterfixed/Parameterssi(det) in the same figure
is to compare the effect of SSI and the effect of in-situ variability on
design of piled raft supported structure.
mean values of the soil parameters as given in Table 2. Deterministic
results are obtained in order to normalize the probabilistic response
of the same soil–piled raft–structure model incorporating inherent
variability of soil.
5.2. Effect of variability on probabilistic responses
The ensemble average of the natural period of vibration and
responses of the soil–piled raft–superstructure system are esti-
mated from the realizations of Monte Carlo simulations. The rea-
lizations of natural period of vibration (Tssi), maximum shear force
at pile head (VB,pile)ssi, maximum displacement at pile head (δpile )ssi
and maximum bending moment at pile head (Mpile)ssi are recorded
for all the simulations for Tfixed ¼0.40 s embedded in very soft soil
(i.e. case A-2 from Table 3). The cumulative frequency of all the
parameters are estimated and it was found, using chi-square
goodness of fit tests, that there is good agreement between the
simulated data and assumed normal distribution for Tssi (Fig. 4a).
Similarly for (VB,pile)ssi,
(δpile )ssi and (Mpile)ssi, it is observed that there is a good agree-
ment between the simulated data and assumed log-normal dis-
tribution (Fig. 4b–d). Further, the COVs of output response para-
meters, i.e., Tssi, (VB,pile)ssi and (Mpile)ssi are calculated to be 7.2%,
9.7% and 8.4% respectively with respect to variability of undrained
shear strength as well as Young’s modulus of soil as 20% which is
considerably marginal as compared to input variability. While, the
COV of (δpile )ssi is found to be as high as 54.60%.
5.3. Effect of variability on Tssi

5. Results and discussion
5.1. Deterministic analysis
Time history analysis is performed based on single value deter-
ministic parameters under sinusoidal motion for obtaining the SSI
period and dynamic response of the finite element based soil–piled
raft–structure model. Newmark’s β  γ method (β ¼ 0:5,γ ¼ 0:25)
considering a small time step of T/100 is used for solving the
dynamic motion of equation, where T is the fundamental time period
of the structure. Dynamic responses are obtained for fixed base
condition and soil–piled raft–superstructure system considering the
The correct estimation of the natural period of vibration is an
important parameter in seismic design of structure. The effect of
variability of undrained shear strength and Young’s modulus of
soil is studied on the natural period of vibration of structure
supported by piled raft foundation considering different values of
COVcu (or COVEs) and due to variation of other influential para-
meters, such as, relative stiffness of pile and soil (Ep/Es), slender-
ness ratio (L/d), spacing to diameter ratio (S/d) and Tfixed.
5.3.1. Influence of Ep/Es
The effect of variability of soil undrained shear strength and
Young’s modulus of soil is studied for three different Ep/Es ratios,
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
Fig. 4. Best fit CDF for MCS response parameters (a) Tssi (b) (VB,pile)ssi (c) (Mpile)ssi and (d) (δpile )ssi. for structure having Tfixed ¼0.4 s in very soft soil with COVcu (or COVEs) ¼20%.
namely 10,000 (i.e. very soft soil), 5000 (i.e. soft soil) and 1500 (i.e.
stiff soil). Structure having Tfixed ¼ 2.0 s and pile group having
L/d ¼60, 80 and S/d ¼3 are considered for this study. Tfixed ¼0.40 s
is found to be unsuitable for studying the variation of Ep/Es effect
as piled raft foundation is infeasible as well as uneconomical
solution in soft and moderately stiff clay. Hence, Tfixed ¼ 0.40 s case
is not considered in Ep/Es ¼ 5000 and 1500 condition. Fig. 5(a)
and (b) presents the normalized natural period of vibration of
structure as a function of COVcu (or COVEs) considering determi-
nistic and probabilistic analyses for SSI and fixed-base structure
for different Ep/Es ratios in L/d ¼60 and 80 respectively. In order to
examine the effect of Ep/Es on sample mean of Tssi(Tssi(mean)) the
following cases are considered: (a) the Tssi is obtained considering
undrained shear strength and Young’s modulus of soil as deter-
ministic parameter and (b) Tssi(mean) estimated from Monte Carlo
simulations considering undrained shear strength and Young’s
modulus of soil as an uncertain parameter. Both deterministic and
probabilistic results are presented in Fig. 5(a) in terms of two
normalized periods, i.e. (Tfixed/Tssi(det)) and (Tssi(mean)/Tssi(det)) for
Ep/Es ¼10,000, 5000 and 1500 in order to assess the effect of in-situ
257
variability of cu and Es. It is observed that normalized period of
vibration obtained from deterministic analysis (i.e. Tfixed/Tssi(det))
exhibit marginal variation and the values are less than one. This
indicates SSI results in lengthening of period of structure. On the
other hand, a significant change in the ratio of lengthening of
period for different Ep/Es ratios is observed due to the effect of
variability of soil stiffness which is presented in terms of Tssi(mean)/
Tssi(det) for Ep/Es ratios. For example, in case of Ep/Es ¼10,000 (i.e.
very soft clay), the value of Tssi(mean)/Tssi(det) attains a maximum
value of 1.2 for COVcu (or COVEs) of 20% which is about 20% higher
than the deterministic values. Increase in Tssi(mean) with respect to
deterministic values remains same for COVcu (or COVEs) of 30% and
50%, which indicates that increase in effect of soil variability is
marginal on the natural period of vibration of soil–piled raft–
superstructure system. Increase in Tssi(mean) as compared to
deterministic Tssi(det) is observed to be significant in case of very
soft soil (Ep/Es ¼10,000) as depicted in Fig. 5(a). The results indi-
cate that COVcu (or COVEs) has a considerable effect on the natural
period of vibration of the system. For lower Ep/Es ratio which
indicates pile group embedded in moderately stiff soil, the effect of
258 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
1.3
Tssi, (mean) / Tssi,(det.)
1.25
Tfixed / Tssi,(det.)
Ep/ Es=10000
Normalized Time Period
1.2
1.15
Ep/ Es=5000
1.1
1.05
Ep/ Es=1500
1 Ep/ Es=10000,5000,1500
0.95
20 30 40
COVCu or COVEs(%)
Fig. 5. Variation of normalized period of structure having Tfixed ¼ 2.0 s with respect to different values of COVcu (or COVEs) and Ep/Es ratios, S/d ¼3 for (a) L/d¼ 60 and
(b) L/d¼80.
soil variability is considerably subdued and the variation of nor-
malized period compared to deterministic prediction is within
25%. This clearly indicates that the in-situ variability of soil design
parameters may have significant effect on period of piled raft
supported structure embedded in soft clay. Further, a similar trend
is also observed for Tfixed ¼2.0 s and L/d ¼80 as shown in Fig. 5(b).
5.3.2. Influence of slenderness ratio (L/d)
The effect of COVcu (or COVEs) on the natural period of vibration
of structure incorporating SSI is studied for different L/d ratios of
flexible piled raft foundation. Fig. 6 presents normalized period of
structure as a function of COVcu (or COVEs) for structure having
Tfixed ¼0.40 s embedded in very soft (Ep/Es ¼10,000) and 2.0 s period
of structure embedded in soft (Ep/Es ¼5000) and moderate stiff soil
(Ep/Es ¼1500) for L/d¼40, 60 and 80. It is observed that the deter-
ministic results for normalized period (i.e.Tfixed/Tssi(det)) varies mar-
ginally for three different L/d ratios irrespective of Tfixed values as
shown in Fig. 6a–c. However, significant change in lengthening of
period for different L/d ratios is observed attributing the effect of
COVcu (or COVEs) which is presented as Tssi(mean)/Tssi(det) (Fig. 6). For
example, in case of structure havingTfixed ¼0.40 s and supported by
pile group with L/d¼40, the value of Tssi(mean)/Tssi(det) attains about
2.1 for COVcu (or COVEs) of 20% which is around 250% higher than
the deterministic results. However, this value slightly decreases for
COVcu (or COVEs)¼30% and almost remains constant for COVcu (or
COVEs)¼50%. This indicates that the increase in COVcu (or COVEs) has
marginal effect on Tssi(mean) for a particular L/d ratio. The effect of
variability is also evidenced for L/d¼ 60 and L/d¼80. Further, it is
also observed that in case of L/d¼ 40, the lengthening of period is
relatively higher than that of the case of L/d¼ 60 and 80. However,
in case of Tfixed ¼2.0 s founded on piled raft foundation with
L/d¼40, Tssi(mean)/Tssi(det) attains a maximum value of 1.1 and 1.03 for
soft and moderately stiff soil respectively, when COVcu (or COVEs)¼
20% - which is around 12% and 4% higher than the deterministic
values. This increase remains constant for COVcu (or COVEs)¼30%
and 50% and for the structures embedded in moderately stiff soil.
While, Tssi(mean)/Tssi(det) marginally increases with COVcu (or COVEs) in
case of structures embedded in soft soil. These results indicate that
COVcu (or COVEs) has a considerable effect on dynamic character-
istics of structure supported on piled raft foundation pertaining to
1.3
Tssi, (mean) / Tssi,(det.)
1.25 Tfixed T
/ ssi,(det.)
EP / ES =10000
Normalized Time Period
1.2
1.15
EP / ES =5000
1.1
1.05
EP / ES =1500
EP / ES =10000
EP / ES =5000
1 EP / ES =1500
0.95
50 20 30 40 50
COVCu or COVEs(%)
different L/d ratios. Pile group with lower L/d ratio, the lengthening
of period significantly increases with respect to deterministic pre-
dictions for short and long period of structures. For short period
structure with relatively short pile, the effect is higher as compared
to long period structure. It is observed that pile with higher L/d
ratio, the lengthening of period is less due to incorporation of
uncertainty in cu and Es for both short and long period of structures.
This is because, longer embedded depth of pile improves the
averaging effect and reduces the uncertainty in Tssi(det) of soil–piled
raft–superstructure system.
5.3.3. Influence of S/d of pile
The effect of COVcu (or COVEs) on Tssi(mean) is studied for differ-
ent S/d ratios of pile supporting structure having Tfixed ¼0.40 s.
S/d ¼5 and 7 leads to an uneconomical size of raft for structure of
Tfixed ¼2.0 s and hence discarded for analysis. In fact, present study
considers a 10 m
10 m size of raft which is designed to support
different height of building structure which represents different
period of superstructure. Further, it is observed that selection of
pile group with S/d ¼5 and 7 in case of Tfixed ¼2.0 s structure
requires relatively higher area as compared to the considered
10 m
10 m size of raft which may be an uneconomical solution.
Fig. 7 presents normalized period of vibration of structure as a
function of COVcu (or COVEs) for Tfixed ¼0.40 s supported on pile
group having L/d ¼40. It is observed that the deterministic results
for normalized period of vibration (i.e. Tfixed/Tssi(det)) exhibits mar-
ginal variation for three different S/d ratios. A considerable change
in lengthening of period is observed while considering three dif-
ferent values of COVcu (or COVEs) (Fig. 7). For instance, in case of
S/d ¼7, the value of Tssi(mean)/Tssi(det) attains a maximum value of
1.6 for COVcu (or COVEs)¼ 20% which is around 60% higher than the
deterministic prediction. However, this ratio almost remains same
for COVcu (or COVEs)¼ 30% and 50%. The effect of COVcu (or COVEs) is
also evidenced for other two S/d ratios i.e., S/d ¼3 and 5. It is
observed that the lengthening of period incorporating SSI is higher
in the case of S/d ¼7 and lower in case of S/d ¼ 3, which indicate
that the effect of variability of soil is subdued in the case of S/d ¼3.
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268 259

3 1.2

Tssi, (mean) / Tssi,(det.) Tssi, (mean) / Tssi,(det.)

2.5 Tfixed / Tssi,(det.) Tfixed / Tssi,(det.)

1.15 L/ d=40
Normalized Time Period

Normalized Time Period

L / d=40 L/ d=60
L/ d=80
2
L / d=60
1.1
L / d=80
1.5

L / d=40 1.05
1 L / d=60
L / d=80 L / d=40
L/ d=60
1 L/ d=80
0.5

0 0.95
20 30 40 50 20 30 40 50
COVCu or COVEs(%) COVCu or COVEs(%)

1.1

Tssi, (mean) / Tssi,(det.)

Tfixed / Tssi,(det.)
L / d=40
L / d=60
Normalized Time Period

1.05
L / d=80

L / d=40
L / d=60
1
L/ d=80

0.95

0.9
20 30 40 50
COVCu or COVEs(%)
Fig. 6. Variation of normalized period of structure with respect to different values of COVcu (or COVEs) and L/d ratios of pile for: (a) Tfixed ¼ 0.4 s embedded in very soft soil,
(b) Tfixed ¼ 2.0 s embedded in soft soil and (c) Tfixed ¼ 2.0 s embedded in moderately stiff soil.

5.3.4. Influence of period of superstructure (Tfixed)
An attempt is also made to study the effect of soil variability on
SSI period of structure for long (Tfixed ¼2.0 s) and short period
(Tfixed ¼0.40 s) structure. It is observed from Fig. 8 that short period
structures (i.e. Tfixed ¼0.40 s) supported on piled raft foundation
exhibits considerable lengthening of period due to SSI for deter-
ministic case as compared to long period structure (i.e. Tfixed ¼2.0 s).
However, consideration of variability may further change in
lengthening of period of stiff structures significant as can been seen
from Fig. 8. Such effect is also observed in long period structure, i.e.,
Tfixed ¼2.0 s, but relatively subdued in magnitude. For instance, in
case of Tfixed ¼0.40 s, the value of Tssi(mean)/Tssi(det) attains a max-
imum value of 1.6 for COVcu (or COVEs)¼50%, which is around 60%
higher than the deterministic value. Moreover, in case of Tfixed ¼2.0 s
period of structure, the value of Tssi(mean)/Tssi(det) attains a maximum
value of near 1.2 for COVcu (or COVEs)¼50%, which is around 20%
higher than the deterministic prediction. It is observed that the
lengthening of period incorporating SSI is higher in case of short
period of structure than that of long period structure.
5.4. Effect of variability on shear force at column and pile
The effect of COVcu (or COVEs) on dynamic response of structure
supported on piled raft foundation is studied considering different
values of COVcu (or COVEs) for different values of Ep/Es, L/d, S/d and
Tfixed. Maximum shear force and bending moment at column and
pile head is presented as dynamic response quantity in the
present study.
5.4.1. Influence of Ep/Es
Normalized maximum shear force at column and pile head for
three different values of COVcu (or COVEs), three different Ep/Es
ratios and Tfixed ¼2.0 s are presented in Fig. 9. Normalization is
made with respect to fixed base condition is presented under
Fig. 9. Shear force at column and pile head is obtained for three
260 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268

ssi(mean)/VB,fixed ranges 0.05–0.2, which indicates an opposite trend in

Tssi, (mean) / Tssi,(det.)

comparison to the observation made in case of column shear.
S / d =7
Tfixed / Tssi,(det.)
Similar trend of observation for normalized shear force at col-
2 S / d =5
umn and pile head observed in Fig. 10(a) and (b) clearly admits the
S / d =3
Normalized Time Period

importance of probabilistic based SSI design. In Fig. 10(a) and (b),

normalization was made with respect to shear force obtained from
1.5 deterministic SSI analysis which make a slight difference from
Fig. 9(a) and (b). However, it is observed that incorporation of SSI
modeled deterministically may not always give safe or over-
S / d =3 estimation of response in column and pile foundation in a piled
S / d =5
1 raft supported structural system during a seismic event. Moreover,
S / d =7
the effect of soil in-situ variability considerably varies with Ep/Es
ratio of pile, while, in deterministic SSI case, variation of relative
stiffness of pile to soil (Ep/Es) has a negligible effect as shown in
0.5
Figs. 9 and 10. However, it is observed that the variability effect is
maximum in increase of column shear for lower value of Ep/Es and
upper bound response is obtained in pile head for relatively higher
0 value of Ep/Es.
20 30 40 50
COVCu or COVEs(%) 5.4.2. Influence of L/d
The effect of COVcu (or COVEs) on shear force at column and pile
Fig. 7. Variation of normalized time period for structure having Tfixed ¼0.4 s,
head is examined for different L/d ratios of pile. Figs. 11 and 12
L/d¼ 40 and Ep/Es ¼ 10,000, with respect to different values of COVcu (or COVEs) and
S/d ratios. present the variation of normalized shear at column and pile head
considering both probabilistic and deterministic analysis for dif-
2.5 ferent L/d ratio as a function of COVcu (or COVEs). It is observed from
Fig. 11(a)–(f) that the influence of SSI attributing to deterministic
T,ssi, (mean) / T,ssi,(det.) modeling on column shear is marginal, while, such influence is
T,fixed / T,ssi,(det.) found to be significant in case of shear force transferred to pile
2 head. However, effect of L/d of pile has marginal effect on deter-
Normalized Time Period

Tfixed=0.40 sec
1.5
Tfixed=2.0 sec
Tfixed=2.0 sec
1 dition. Fig. 11(b), (d) and (f) shows that pile head shear reduces as
Tfixed=0.40 sec
0.5
0 ded in very soft clay which primarily indicates the effect of in-situ
20 30 40
COVCu or COVEs(%)
Fig. 8. Variation of normalized time period of structure having Tfixed ¼0.4 s and
2.0 s for L/d¼ 60 and Ep/Es ¼ 10,000 for different values of COVcu (or COVEs).
different cases: (a) sample mean of maximum shear force obtained
from MCS, which are presented as VB,col,ssi(mean) and VB,pile,ssi(mean),
(b) results from deterministic SSI analysis, i.e. VB,col,ssi(det) and VB,pile,
ssi(det) and (c) base shear obtained under fixed base condition, i.e.
VB,fixed.
It is observed from Fig. 9(a) and (b) that deterministic SSI analysis
yields a value of normalized shear at column (VB,col,ssi(det)/VB,fixed) in
Fig. 9(a) to be equal to 1 and the value of (VB,pile,ssi(det)/VB,fixed) is found
to be about 2.5 irrespective of Ep/Es ratio (Fig. 9(b)). This indicates pile
head shear increases 150% as compared to shear force obtained in
fixed base condition, while shear force at column remains same as
that of shear force obtained in fixed base condition. Interestingly,
VB,col,ssi(mean)/VB,fixed is increased and VB,pile,ssi(mean)/VB,fixed decreased
significantly than that of the deterministic case due to incorporation
of in-situ soil variability. Normalized column shear (VB,col,ssi(mean)/VB,
fixed) may go as high as 2.8–4 depending on Ep/Es ratio as compared to
what is obtained under fixed base condition. It is observed that VB,pile,
ministic results. Fig. 11(a), (c) and (e) indicate that shear force at
column is increased due to incorporation of soil variability effect
and may go upto an exceptionally high value of 45–60 times in
very soft soil, 4.0–8.5 times in soft soil and 3.4–4.6 times in
moderately stiff soil considering a variation of L/d ¼40, 60 and 80
respectively of response what is obtained under fixed base con-
compared to shear obtained from fixed base condition with an
exceptional case for L/d ¼40, 60. Column shear is found to increase
with increase in L/d ratio of pile group. On the other hand, an
opposite trend is observed for pile head shear. However, the
exceptional observation exhibiting a high value of column shear
and increased shear at pile head is found in Tfixed ¼0.40 s embed-
50
variability of soil may be extremely high for short period of
structures supported by shorter L/d ratio of pile group. Such high
response in structure as well as foundation member may be
inferred with failure of different stiff period of structures, such as,
bridges, flyovers during past earthquake (Meymand, 1997). How-
ever, in case of long period of structures (i.e. Tfixed ¼2.0 s)
embedded in soft and moderate stiff soil, the effect of in-situ
variability of soil may significantly reduce the shear force trans-
ferred to pile head with respect to the shear obtained under fixed
base condition. It indicates that deterministic SSI effect may
overestimate the response. Further, the effect of in-situ variability
of soil stiffness is further verified by normalizing the probabilistic
response by deterministic SSI results which are presented in
Fig. 12. A similar trend is observed in shear force at column and
pile head.
5.4.3. Influence of S/d
The effect of in-situ variability of soil stiffness is examined for
different S/d ratio of pile group which are presented in Fig. 13.
VB,col,ssi(mean)/VB,fixed and VB,pile,ssi(mean)/VB,fixed for different S/d ratio for
different values of COVcu (or COVEs) ispresented in Fig. 13(a)
and (c) indicates that shear force at column may be increased due to
incorporation of in-situ soil variability and may significantly go upto
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
Fig. 9. Variation of normalized shear force for structure having Tfixed ¼ 2.0 s with different Ep/Es ratio of pile for different values of COVcu (or COVEs) considering: (a) and
(b) L/d ¼ 60, (c) and (d) L/d¼ 80.
within a range of 45–72 times depending on S/d ratio of pile group
as compared to what is obtained under fixed base condition. It is
observed that increase in S/d ratio makes an increase in column
shear when the effect of in-situ soil variability is considered. While,
deterministic response (VB,col,ssi(det)/VB,fixed) indicates a subdued
effect of SSI on column shear and variation of S/d ratio of pile group
has also marginal effect on response at column. Further, it is
observed that the effect of in-situ variability of soil may significantly
increase the shear force transferred to pile head as compared to
fixed base condition. For instance, Fig. 13(b) and (d) indicates that
for S/d¼ 3.0 and 5.0, VB,pile,ssi(mean)/VB,fixed attains a maximum value
of 1.8, whereas, deterministic results, i.e.VB,pile,ssi(det)/VB,fixed shows a
value nearly 1.0 and variation of S/d ratio has marginal effect on
response. However, it may also be inferred that pile head shear goes
maximum for lesser S/d ratio which is completely opposite to shear
261
force transferred to column of piled raft supported structural sys-
tem considered in present study.
5.4.4. Influence of period of superstructure
The effect of soil variability on shear force at column and pile
head is studied for long (Tfixed ¼ 2.0 s) and short period
(Tfixed ¼0.4 s) structure. Fig. 14 represent normalized shear at col-
umn and pile head incorporating SSI condition as a function of
COVcu (or COVEs) for Tfixed ¼0.4 s and 2.0 s period of structure
supported with flexible group of pile (i.e. L/d ¼ 60) embedded in
very soft soil (Ep/Es ¼10,000). It is observed that the deterministic
results for normalized shear at column (i.e. VB,col,ssi(det)/VB,fixed) are
marginally varies with variation of period of structure. Whereas,
the deterministic results for normalized shear at pile head (i.e.
VB,pile,ssi(det)/VB,fixed) varies significantly with variation of period of
structure. VB,col,ssi(det)/VB,fixed for different COVcu (or COVEs) values is
262 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
Fig. 10. Variation of normalized shear force (normalization made w.r.t SSI condition) for structure having Tfixed ¼ 2.0 s with different Ep/Es ratio of pile for different values of
COVcu (or COVEs) considering: (a) and (b) L/d¼ 60, (c) and (d) L/d¼ 80.
presented in Fig. 14, which indicates that shear force at column is
relatively higher due to incorporation of in-situ soil variability for
the case of short period structure as compared to long period
structure. However, in case of long period structures embedded in
very soft soil consideration of in-situ variability of soil may sig-
nificantly reduce the shear force transferred to pile head as com-
pared to fixed base condition which indicates that deterministic
SSI effect may overestimates the response. However, the effect of
in-situ variability of soil may significantly increase the shear force
transferred to pile head as compared to fixed base condition in
case of short period structure embedded in very soft soil as evi-
denced from Fig. 14(b).
5.5. Design implication
Effect of in-situ variability of soil in estimating design response
may lead to a significant increase in dynamic response of the load
resisting member as compared to deterministically modeled soil–
piled raft–structural system. Hence, an attempt is also made to
examine the design implication in terms of probability of failure
(POF) of pile foundation due to the effect of variability of soil
design parameters which may be an important consideration from
the view point of reliability based design. Estimated failure prob-
ability based on MCS may not be exact due to limited sample size
and moreover, calculation of failure probability using MCS is
mainly recommended for lower values of probability of failure
[41]. In such cases, the failure probability can be obtained by using
First Order Reliability Method (FORM) from the estimated sample
mean, standard deviation and probability density function. A pile
is considered to be unserviceable if the mean maximum deflection
μδ; max exceeds the specified limits. Likewise, a pile may lead to
flexural failure if μBM; max exceed the corresponding yield capa-
cities. Where μδ; max and μBM; max are the sample mean of max-
imum lateral displacement and maximum bending moment at pile
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268 263

80 2.4

VB,col, ssi, (mean) / VB,fixed VB,pile, ssi, (mean) / VB,fixed

70 L / d=40
VB,col, ssi,(det.) / VB,fixed VB,pile, ssi,(det.) / VB,fixed
2

Normalized Shear at Pile head

Normalized Shear at Column L / d=80
60

1.6 L / d=60
50 L / d=60
L / d=40
L / d=40 L / d=60
40 1.2

30
0.8 L / d=80
L / d=80
20

0.4
10

L / d=40,60,80
0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

12 4

VB,col, ssi, (mean) / VB,fixed VB,pile, ssi, (mean) / VB,fixed

VB,col, ssi,(det.) / VB,fixed VB,pile, ssi,(det.) / VB,fixed
10
Normalized Shear at Pile head
Normalized Shear at Column

3
L / d =80 L / d =40
8
L / d =60

L / d =80
6 2
L / d =60

4 L / d =40

L / d =80
1
L / d =60
2 L / d =40
L / d=80,60,40

0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

6 3

VB,col, ssi, (mean) / VB,fixed VB,pile, ssi, (mean) / VB,fixed

VB,col, ssi,(det.) / VB,fixed VB,pile, ssi,(det.) / VB,fixed
5
Normalized Shear at Pile head
Normalized Shear at Column

L / d =80 L / d =40

L / d =60
4 L / d =40 2 L / d =60

3 L / d =80

2 1

L / d =80
L / d=80,60,40 L / d =60
1
L / d =40

0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

Fig. 11. Variation of normalized shear force (normalization made w.r.t to fixed base condition) at column and pile head for different L/d ratio of pile for different values of
COVcu (or COVEs): (a) and (b) Tfixed ¼ 0.4 s embedded in very soft soil, (c) and (d) Tfixed ¼ 2.0 s embedded in soft soil and (e) and (f) Tfixed ¼ 2.0 s embedded in moderately stiff soil
respectively.
264 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268

160 2.4

VB,col ,ssi, (mean) / VB,col,ssi,(det.) VB,pile ,ssi, (mean) / VB,pile,ssi,(det.)

140
VB,fixed / VB,col,ssi,(det.) 2 VB,fixed / VB,pile,ssi,(det.)

Normalized Shear at Pile head

Normalized Shear at Column
L / d =80
120
L / d =80
1.6 L / d =40
100
L / d =60
L / d =80
80 L / d =60
1.2
L / d =40
L / d =60
60 L / d =40
0.8

40

0.4
20

L / d=80,60,40
0 0
20 30 40 50 20 30 40 50
COVCu or COVEs(%) COVCu or COVEs(%)

10 0.6

VB,col ,ssi, (mean) / VB,col,ssi,(det.) VB,pile ,ssi, (mean) / VB,pile,ssi,(det.)

VB,fixed / VB,col,ssi,(det.) VB,fixed / VB,pile,ssi,(det.)
0.5
Normalized Shear at Pile head
Normalized Shear at Column

8 L / d =80 L / d =80

L / d =60
0.4
6 L / d =40

L / d =60 0.3

4 L / d =40 L / d =80
0.2 L / d =60
L / d =40

2
0.1
L / d=80,60,40

0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

6 1

VB,col ,ssi, (mean) / VB,col,ssi,(det.) VB,pile ,ssi, (mean) / VB,pile,ssi,(det.)

5 VB,fixed / VB,col,ssi,(det.) VB,fixed / VB,pile,ssi,(det.)
Normalized Shear at Pile head
Normalized Shear at Column

0.8
L / d =80
L / d =80

4 L / d =60
L / d =40 0.6
L / d =60
3
L / d =40
0.4
2

L / d =80
L / d =60
L / d=80,60,40 0.2
1 L / d =40

0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

Fig. 12. Variation of normalized shear force (normalization made w.r.t to deterministic SSI effect) at column and pile head for different L/d ratio of pile for different values of
COVcu (or COVEs): (a) and (b) Tfixed ¼0.4 s embedded in very soft soil, (c) and (d) Tfixed ¼ 2.0 s embedded in soft soil and (e) and (f) Tfixed ¼2.0 s embedded in moderately stiff soil
respectively.
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268 265

100 2.5

VB,col,ssi, (mean) / VB,fixed

VB,pile,ssi, (mean) / VB,fixed
VB,col,ssi,(det.) / VB,fixed VB,pile,ssi,(det.) / V
Normalized Shear at Pile head

Normalized Shear at Pile head

80 2 B,fixed
S / d =7 S / d =3

S / d =3
60 1.5 S / d =5
S / d =5 S / d =5
S / d =7
S / d =3

40 1 S / d =7

20 0.5

S / d =7,5,3

0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

100 2.5

VB,pile,ssi, (mean) / VB,pile,ssi,(det.)

VB,col,ssi, (mean) / VB,col,ssi,(det.)
VB,fixed / VB,pile,ssi,(det.)
Normalized Shear at Pile head

VB,fixed / VB,col,ssi,(det.) 2
Normalized Shear at Column

80
S / d =7
S / d =5 S / d =3
S / d =3
S / d =5
60 1.5
S / d =5
S / d =3 S / d =7
S / d =7

40 1

20 0.5

S / d =3,5,7
0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

Fig. 13. Variation of normalized shear force at column and pile head for structure having Tfixed ¼ 0.4 s embedded in very soft soil with different S/d ratio of pile and COVcu (or
COVEs): (a) and (b) presents for normalization made with respect to fixed base condition, (c) and (d) presents for normalization made with respect to SSI condition.

head obtained from MCS, respectively. If δmax and BM max are modulus of pile as 21.78
109 N/m2. A tolerable displacement
assumed to be log-normal random variables (refer Fig. 4), the ðδallowable Þ is considered as 30 mm [42]. As an example, for case
probability of each failure mode may be expressed as, study A-1 (Table 3), the estimated value of μδ; max ¼ 0:43 m and the
 corresponding standard deviation σ δ; max is 0.26 m for pile head
ln δa  μln δmax
Pðδmax Z δa Þ ¼ 1  Pðδmax r δa Þ ¼ ¼ 1  Φ ð8Þ considering COV of 20%. Tables 5 and 6 present the probability of
σ ln δmax failure of pile foundation based on ultimate limit state and servi-
 ceability criteria respectively for very soft soil.
ln BM y  μln BMmax
PðBM max Z BM y Þ ¼ 1  PðBM max r BM y Þ ¼ ¼ 1  Φ ð9Þ Results are presented in Table 5, which indicates a significant
σ ln BMmax
variation in the probability of failure of pile foundation which
where Φð:Þ is the cumulative standard normal function, δa is the supports structure having Tfixed ¼0.4 s (denoted as A-1, A-2 and
allowable displacement and BMy is the yield bending moment. In A-3 cases) and 2.0 s (denoted as A-4 case) in very soft soil
Eqs. (8) and (9), μln δmax ¼ lnðμδmax Þ  0:5σ 2ln δmax , σ ln δmax ¼ (Ep/Es ¼10,000). It appears that pile group supporting short period
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n h io of structures exhibits a high risk of flexural failure due to devel-
ln 1 þ ðσ δmax =μδmax Þ2 ,σ δmax and σ BMmax are the standard devia- opment of higher bending moment at pile head as compared to
tion of δmax and BM max , respectively. The resisting moment of the long period of structure. Hence, the effect of in-situ variability of
pile section is calculated as 13,725 N-m and 63,541 N-m for a pile soil must be considered while designing pile group supporting
diameter of 300 mm and 500 mm respectively assuming Young’s short period structures mainly in very soft soil. Further, pile
266 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268

70 3.5

VB,col ,ssi, (mean) / VB,fixed

3
VB,pile ,ssi, (mean) / VB,fixed (
60 VB,col ,ssi,(det.) / VB,fixed

Normalized Shear at Pile head

VB,pile ,ssi,(det.) / VB,fixed
Normalized Shear at Column

Tfixed=0.40 sec
50 2.5 Tfixed=2.0 sec

40 2

Tfixed=0.40 sec

30 1.5
Tfixed=0.40 sec

20 1

Tfixed=0.40,2.0 sec
Tfixed=2.0 sec
10 0.5

Tfixed=2.0 sec

0 0
20 30 40 50 20 30 40 50

COVCu or COVEs(%) COVCu or COVEs(%)

Fig. 14. Variation of normalized shear at column and pile head for structure having Tfixed ¼ 0.4 s and 2.0 s, L/d¼ 60 and Ep/Es ¼ 10,000 for different values of COVcu (or COVEs).

involved in the design of pile foundation under seismic loading

Table 5
Probability of flexural failure of pile using FORM based technique. and it is also provides the basis for predicting the reliability of
foundations. However, a case example is presented below in order
Cases MR (N-m) Probability of failure (in percentage) to understand the importance of in-situ variability of soil para-
meters in design implications
COVcu (or COVcu (or COVcu (or COVEs) ¼
COVEs) ¼ 20% COVEs) ¼30% 50%
5.5.1. Case example
A-1 63,541 37.10 41.33 51.79 A 5
6 group of piled raft foundation with 300 mm diameter
A-2 13,725 80.98 81.86 84.78 and 24.0 m length of piles embedded in homogeneous clay deposit
A-3 13,725 83.37 87.65 89.42
A-4 13,725 13.0 14.0 17.0
having undrained cohesion (Cu) of 14.5 kPa and Young’s modulus
A-5 13,725 11.0 12.0 15.0 (Es) of 5000 kPa. The piled raft foundation is designed to support a
single bay superstructure having a period of 2.0 s (total mass of
superstructure is considered as 118,000 kg considering an
Table 6
approximate height of structure is 16 storey). The modulus of
Probability of failure based on serviceability limit state using FORM. elasticity and flexural strength of pile section are 21.78
106 kPa
and 3.049
103 kPa, respectively. The section modulus of pile
Cases δallowable Probability of failure in percentage section is calculated as 4.5
10  3 m3. The resisting bending
(mm)
COVcu (or COVcu (or COVcu (or
moment of the pile section is the section modulus of pile section
COVEs) ¼ 20% COVEs) ¼30% COVEs) ¼ 50% times the flexural strength of pile material which is 13.73 kN-m.
Design ground acceleration (ag) specified for the site¼0.2 m/s2.
A-1 99.9 99.9 99.9 The coefficient of lateral subgrade reaction (Kh)/ ¼6000 kN/m2.
A-2 99.9 99.9 99.9
A-3 30 99.9 99.9 99.9
The allowable lateral displacement is set to a value of 30 mm.
A-4 41.6 42.7 43.8
A-5 32.7 33.5 34.8 5.5.1.1. Conventional design of pile foundation. The spectral ordinate
of IS 1893 [43] for soft site category corresponding to fundamental
period of superstructure 2.0 s is about 0.018. The base shear is
diameter 300 mm (A-2 and A-3 cases) exhibits relatively higher
calculated as (1,180,000 kg
9.81
0.00324)¼ 37.5 kN. The equiva-
failure probability as compared to 500 mm diameter (A-1 case).
lent depth of fixity obtained as per IS 2911 [44] is 4.0 mm. The
However, a gradual increase in percentage of failure probability is bending moment and lateral displacement at a single pile induced
observed with increase in COVcu (or COVEs) percentage from due to base shear are computed as 5.0 kN-m and 0.00077 m as per
Table 4. In case of limit state of serviceability, the probability of IS 2911-part 1-(2008) which is lesser than the resisting moment of
pile failure is found to be around 100% for pile group supported the pile section of 13.75 kN-m and allowable displacement of
structure having Tfixed ¼0.4 s, whereas, structure having 30 mm. Hence, the pile diameter is sufficient for the specified
Tfixed ¼0.4 s indicates a subdued failure probability of around 44% design earthquake and the design pile diameter is 0.3 m for the
irrespective of COVcu (or COVEs). However, it is observed that seismic design acceleration for deterministic case.
probability of failure with respect to displacement criteria is
higher than bending moment criteria which indicate that servi- 5.5.1.2. Probabilistic design. The conventional method of design
ceability criteria for seismic design of pile must be verified. does not have the provision to include the variability involved in
Moreover, the in-situ variability of soil, allowable lateral dis- the soil strength parameter. The variability of undrained cohesion
placement and resisting moment are considered to quantify risk (Cu) and Young’s modulus of soil (Es) is considered as 20%. Target
 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
reliability index is 2–3 as suggested in Haldar and Babu [23] which
probability of failure value of 2.3
10  2–1.3
10  3. The prob-
ability of failures is evaluated based on maximum bending
moment and displacement failure criterion from Eqs. (8) and (9).
The probability of failure based on bending moment and dis-
placement criterion are computed as 0.635 and 0.797 respec-
tively following FORM technique for pile diameter 0.3 m. Both the
values are above the target probability of failure values. Hence, pile
diameter 300 mm is unsatisfactory. Therefore, pile diameter is
increased to 0.6 m and the POFs are obtained as 0 and 0.085
respectively. It indicates that probability of failure is nil in respect
to bending moment criterion but still higher than minimum target
probability of failure considering displacement criterion. Further,
the pile diameter is increased to 0.8 m and the POFs consider-
ing bending moment and displacement criterion are obtained as
0 and 0.0083 which are lesser than minimum target probability of
failure value. Hence, the pile diameter must be obtained for a
specified design seismic force variability of the site.
6. Conclusions
This study assesses the effect of in-situ variability of undrained
shear strength and Young’s modulus of soil on seismic response of
soil–piled raft-structure system within elastic limit encompassing
a wide range of parametric variation of relative stiffness of pile
with respect to soil (Ep/Es), length to diameter ratio (L/d) and
spacing to diameter ratio (S/d) and for short and long period of
superstructure. Probability of failure of pile foundation is esti-
mated in order to investigate the effect of inherent variability of
soil parameters in design. Based on the study following conclu-
sions can be made:
(1) Fundamental period of soil–piled raft-structure system incor-
porating SSI (Tssi) is found to be best fitted with normal
distribution function. However, lognormal distribution best
fits for shear force, bending moment and displacement at
column and pile head.
(2) Inherent variability of soil undrained shear strength and
Young’s modulus of soil leads to a considerable increase in
fundamental lateral period of structure incorporating SSI
condition as compared to deterministic SSI model. However,
such increase is observed to vary with Ep/Es ratio of pile and it
is observed that pile group attributing higher Ep/Es ratio may
cause maximum increase in period of structure due to the
effect of soil variability. Whereas, for lower Ep/Es ratio, the
effect is subdued. Further, the variability effect is found to be
significant considering different L/d and S/d ratio of pile. It is
observed that smaller value of L/d of flexible pile and higher
choice of S/d ratio may lead to maximum increase in period of
structure.
(3) A significant increase in shear force at column is observed due
to the influence of inherent variability of soil. However, such
effect varies with different parametric variations of Ep/Es, L/d,
S/d ratio and stiff as well as flexible period of structure as
compared to the deterministic SSI condition. It is observed
that maximum increase in shear force at column may be
occurred in piled raft foundation supporting structure
embedded in stiffer soil (Ep/Es ¼ 1500) and a minimum
increase occur in softer soil (Ep/Es ¼10,000). Further, max-
imum increase in column shear may also occur for higher L/d
as well as S/d ratio. The effect of inherent variability is
observed to be relatively higher for short period of structure
supported by piled raft foundation as compared to a long
period of superstructure.
267
(4) Shear force at pile head significantly decreases due to incor-
poration of inherent variability of soil with few exceptions as
compared to shear force obtained at fixed base condition and
deterministic SSI model. While, deterministic SSI model
exhibits an increase in shear force at pile head as compared to
base shear at fixed base condition. This may lead to con-
servative pile design if designed considering deterministic SSI
condition.
(5) The effect of in-situ variability of undrained cohesion is found
to be influenced by the variation of Ep/Es, L/d, S/d and different
period of superstructure. The effect of in situ variability in
distribution of column shear is found to be increasing trend
with decrease in Ep/Es. An opposite trend is observed in
distribution of pile head shear. A similar trend of transfer of
column shear is observed for L/d and S/d variation with few
exceptions. Furthermore, the effect of in-situ variability of
undrained cohesion of soil may relatively higher in shorter
period structure as compared to longer period structure.
(6) Probability of failure for pile foundation embedded in very soft
soil indicates a possibility of high risk of flexural as well as
serviceability failure criteria for pile foundation supporting
shorter/stiff period of structure under seismic loading. It is
also observed that serviceability criteria may be the governing
criteria for pile design.
Summarily, a case study example attempts to show the
importance of variability of soil parameters in design of structures
supported by piled raft foundation in clay type soil. This study is a
limited effort to assess the effect of inherent variability of soil on
design forces of a soil–piled raft–structure system attributing a
wide range of parametric variation of different influential para-
meters. The study indicates that there is a need of further study in
this direction considering variability parameters of soil, SSI mod-
eling and uncertainty of ground motion which may give a crucial
input in seismic design of such heavy structures supported on
piled raft foundation.
References
[1] FEMA 440. Improvement of Nonlinear Static Seismic Analysis Procedures.
Washington, DC, U.S.A: Federal Emergency Management Agency; 2005.
[2] Gazetas G, Mylonakis G. Seismic soil-structure interaction: New evidence and
emerging issues. In: Dakoulas P, Yegian MK, Holtz RD, editors. Geotechnical
EarthqEng and Soil Dynamics III, II. ASCE; 1998. p. 1119–74.
[3] Tokimatsu K, Suzuki K, Sato M. Effects of inertial and kinematic interaction on
seismic behavior of pile with embedded foundation. Soil Dyn Earthq Eng
2005;25:753–62.
[4] Meymand PJ. Shaking Table Scale Model Tests of Nonlinear Soil–Pile–Super-
structure Interaction In Soft Clay. Berkeley: University of California; 1998.
[5] Applied Technology Council. Tentative provisions for the development of
seismic regulations of buildings. Washington, DC: National Bureau of Stan-
dards, Supt. of Docs; 1978.
[6] NEHRP. Recommended provisions for seismic regulations for new buildings
and other structures: Parts 1 and 2. Washington, DC: Building Seismic Safety
Council; 1997.
[7] Gazetas G. Seismic response of end-bearing single piles. Soil Dyn Earthq Eng
1984;3(2):82–93.
[8] Rovithis E, Pitilakis NKD, Mylonakis GE. Seismic analysis of coupled soil–pile–
structure systems leading to the definition of a pseudo-natural SSI frequency.
Soil Dyn Earthq Eng 2009;29:1005–15.
[9] Boulanger RW, Curras CJ, Kutter BL, Wilson WD, Abghari AA. Seismic soil–pile–
structure interaction experiments and analyses. J Geotech Geoenviron Eng,
ASCE 1999;125(9):750–9.
[10] Jeremic B, Kunnath S, Xiong F. Influence of soil–foundation–structure inter-
action on seismic response of the I-880 Viaduct. Int J EngStruct 2008;26
(3):391–402.
[11] Kaynia MA, Mahzoon PS. Forces in pile foundations under seismic loading.
J Eng Mech 1996;122(1):46–53.
[12] Curras JC, Boulanger WR, Kutter BL, Wilson DW. Dynamic experiments and
analysis of a pile-group-supported structure. J Geotech Geoenviron Eng. ASCE
2001;127(7):585–96.
268 B. Das et al. / Soil Dynamics and Earthquake Engineering 84 (2016) 251–268
[13] Yangcai H. Seismic response of tall building considering soil–pile–structure
interaction. Earthq Eng Struct Dyn 2002;1(1):57–64.
[14] Guin J, Banerjee PK. Coupled soil–pile–structure interaction analysis under
seismic excitation. J Struct Eng. ASCE 1998;124(4):434–44.
[15] Saha R, Dutta SC, Haldar S. Seismic response of soil–pile foundation–structure
system. J Civil Eng Manag 2015;21(2):144–64.
[16] Phoon KK, Kulhawy FH. Characterization of geotechnical variability. Can
Geotech J 1999;36:210–38.
[17] Niandou H, Breysse D. Reliability analysis of a piled raft accounting for soil
horizontal variability. Comput Geotech 2007;34:71–80.
[18] Moghaddasi M, Cubrinovski M, Chase JG, Pampanin S, Carr A. Probabilistic
evaluation of soil–foundation–structure interaction effects on seismic struc-
tural response. Earthq Eng Struct Dyn 2011;40:135–54.
[19] Barcena A, Esteva L. Influence of dynamic soil–structure interaction on the
nonlinear response and seismic reliability of multistory systems. Earthq Eng
Struct Dyn 2007;36:327–46.
[20] Tandjiria V, Teh CI, Low BK. Reliability analysis of laterally loaded piles using
response surface methods. Struct Saf 2000;22:335–55.
[21] Pula W, Rozanski A. Reliability of rigid piles subjected to lateral loads. Struct
Saf 2012;12:205–18.
[22] Chan CL, Low BK. Probabilistic analysis of laterally loaded piles using response
surface and neural network approaches. Comput Geotech 2012;43:101–10.
[23] Haldar S, Babu GLS. Effect of soil spatial variability on the response of laterally
loaded pile in undrained clay. Comput Geotech 2008;35:537–47.
[24] Lee, S., Budkowska, B.B., and Balachandar, R., 2007. Reliability of strength of a
free head laterally loaded pile embedded in soft clay. In: Proceedings of
Numerical Models in Geomechanics, Rhodes, Greece.
[25] Lumb P. The variability of natural soils. Can Geotech Journal 1966;3(2):74–97.
[26] Baecher GB, Christian JT. Reliability and statistics in geotechnical engineering.
London and New York: John Wiley and Sons; 2003.
[27] Dutta SC, Bhattacharya K, Roy R. Effect of flexibility of foundations on its
seismic stress distribution. J Earthq Eng 2009;13:22–49.
[28] Gazetas G. Formulas and charts for impedances of surface and embedded
foundations. J Geotech Eng, ASCE 1991;117(9):1363–81.
[29] Makris N, Gazetas G. Dynamic pile–soil–pile interaction. Part II: lateral and
seismic response. Earthq Eng Struct Dyn 1992;21:145–62.
[30] Pender, J. M., and Satyawan, P., 1996. Gapping Effects during Cyclic Lateral
Loading of Piles in Clay, In: Proceedings of the of the World Conference on
Earthquake Engineering, Mexico, pp. 23–28.
[31] Bowles JE. Foundation Analysis and Design. 6th edition. Singapore: McGraw-
Hill International Editions; 1997.
[32] Gazetas G, Fan K, Kaynia A. Dynamic response of pile groups with different
configurations. Soil Dyn Earthq En 1993;12:239–57.
[33] Saha, R., Tayal, S., Kumar, R., Dutta, S.C., and Haldar, S. (2012) Effect of Non-
linear behavior of soil on seismic response of soil–pile foundation–structure
system. In: Proceedings of 9th International Conference on Urban Earthquake
Engineering and 4th Asian Conference of Earthquake Engineering, 6th to 8th
March , 2012, Tokyo.
[34] Satake N, Suda K, Arakawa T, Saraki A, Tamura Y. Damping evaluation using
full-scale data of buildings in Japan. J Struct Eng ASCE 2003;129(4):470–7.
[35] Veletsos AS. Dynamics of Structure-foundation Systems. Structural and Geo-
technical Mechanics, WJ Hall. New Jersey: Prentice Hall; 1977.
[36] Velez A, Gazetas G, Krishnan R. Lateral dynamic response of constrained head
piles. J Geotech Eng, ASCE 1984;109(8):1063–81.
[37] Roy R, Dutta SC. Inelastic seismic demand of low-rise buildings with soil
flexibility. Int J Nonlin Mech 2010;45:419–32.
[38] Yashinsky, M., 1998. The Loma Prieta, California, Earthquake of October 17,
1989 – Highway Systems. Professional Paper 1552-B, USGS, Washington.
[39] Ghosh B, Lubkowski Z. Modeling Dynamic Soil Structure Interaction Under
Seismic Loads. In: Bhattacharjee S, editor. Design of Foundations in Seismic
Areas: Principles and Applications. IIT Kanpur. India: NICEE; 2007. p. 118–98.
[40] Mazzoni, S., Mckenna, F., Scott, M.H., Gregory, L., and Fenvas, G., OpenSees,
Open System for Earthquake Engineering Simulation User command- Lan-
guage manual 2007. Available online: 〈〈http://opensees.berkeley.edu〉〉.
[41] Haldar S, Basu D. Response of Euler–Bernoulli beam on spatially random
elastic soil. Comput Geotech 2013;50:110–28.
[42] Salgado R. The Engineering of Foundations. New York: Tata McGraw-Hill;
2008.
[43] IS: 1893 part I. Indian Standard Criteria for Earthquake Resistant Design of
structures. New Delhi, India: Bureau of Indian standards; 2000.
[44] IS: 2911 (Part 1/Section 4). Indian Standard Code of Practice for Design and
Construction of Pile Foundation: Part 1: Concrete Piles, Section 4 Bored Pre-
cast Concrete Piles. New Delhi, India: Bureau of Indian Standards; 1984.