You are on page 1of 13

Computers and Geotechnics 40 (2012) 114126

Contents lists available at SciVerse ScienceDirect

Computers and Geotechnics


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

Dynamic behavior of pile foundations under cyclic loading in liqueable soils


Amin Rahmani , Ali Pak
Department of Civil Engineering, Sharif University of Technology, P.O. Box 11155-9313, Tehran, Iran

a r t i c l e

i n f o

Article history:
Received 5 March 2011
Received in revised form 12 September
2011
Accepted 13 September 2011
Available online 11 November 2011
Keywords:
Liquefaction
Pile foundations
Fully coupled three-dimensional dynamic
analysis
Dynamic behavior of pile

a b s t r a c t
In this paper, a fully coupled three-dimensional dynamic analysis is carried out to investigate the
dynamic behavior of pile foundations in liqueed ground. A critical state bounding surface plasticity
model is used to model soil skeleton, while a fully coupled (uP) formulation is employed to analyze soil
displacements and pore water pressures. Furthermore, in this study, variation of permeability coefcient
during liquefaction is taken into account; the permeability coefcient is related to excess pore water
pressure ratio. Results of a centrifuge test on pile foundations are used to demonstrate the capability
of the model for reliable analysis of piles under dynamic loading. Then, the veried model is used for a
parametric study. The parametric study is carried out by varying pile length, frequency of input motion,
xity of the pile head, thickness of the liquefying soil layer and relative density of liquefying soil layer.
Three different soil proles have been considered in this study. In general, parametric studies demonstrate that xity of the pile head, thickness of liquefying soil layer and frequency of input motion are
the most critical parameters which considerably affect piles performance in liqueed grounds.
Crown Copyright 2011 Published by Elsevier Ltd. All rights reserved.

1. Introduction
The behavior of pile foundations under earthquake loading is an
important issue that widely affects the performance of structures.
Design procedures have been developed for evaluating pile behavior under earthquake loading; however, application of these procedures to cases involving liqueable ground is uncertain since the
performance of piles in liqueed soil layers is much more complex
than that of non-liquefying soil layer not only because the superstructure and the surrounding soil exert different dynamic loads
on pile, but also because the stiffness and shear strength of the surrounding soil diminishes over time due to non-linear behavior of
soil and also pore water pressure generation.
Liquefaction represents one of the biggest contributors to damage of constructed facilities during earthquakes [1]. This phenomenon was reported as the main cause of damage to pile
foundations during the major earthquakes such as Alaska, 1964,
Loma-Prieta, 1989, Hyogoken-Nambu, 1995 [1]. Prediction of
seismic response of pile foundations in liquefying soil layers is
difcult, and there are many uncertainties in the mechanisms involved in soilpile-superstructure interaction. However, in recent
decades, a wide range of centrifuge and shaking table tests and
also various numerical methods have been employed in order to
provide better insights into the dynamic behavior of pile foundations in liqueable soils. These researches can be divided into

three categories: eld observations, laboratory tests, and numerical modeling.


1.1. Field observations
These studies mainly investigate the distribution of the failure
patterns, settlement and lateral displacement of piles. During
Niigata Earthquake, 1964, many pile foundations failed to support
structures due to the liquefaction of the surrounding soil. According to Hamada [2], the ground in the vicinity of a four-storey building moved approximately 1.1 m, and the maximum lateral
displacement of concrete piles with a diameter of 35 cm and length
of 69 m was around 70 cm. The large amount of lateral displacement caused severe damage to the pile at the interface of the liqueed and non-liqueed layers. Mori et al. [3] conducted an
excavation survey and internal inspection of the damaged piles
of a silo which suffered severe damage due to the Hokkaido Nansei-Oki Earthquake, 1993. They concluded that damage usually occurs at three different locations: at the pile head (for xed-head
piles), at a depth of 13 m below the pile cap (for free-head piles)
and at the interface of the liqueed and non-liqueed layers. This
observation has been conrmed by others such as Tachikawa
et al. [4], Shamoto et al. [5], and Onishi et al. [6].
1.2. Laboratory tests

Corresponding author. Tel.: +98 21 6616 4225; fax: +98 21 6601 4828.
E-mail address: aminrahmaani@gmail.com (A. Rahmani).

These studies include some dynamic centrifuge tests and


also shaking table tests of pile-supported structures in which

0266-352X/$ - see front matter Crown Copyright 2011 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2011.09.002

115

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

seismic response of pile, soil and superstructure are investigated.


Wilson et al. [7] conducted a series of centrifuge tests on single
piles and pile groups located in liqueable soils in order to observe the py behavior of piles embedded in liquefying sands.
The centrifuge tests results indicated that py curves were
highly time-dependent in liqueable soils; lateral resistance on
the pile decreased by increasing pore water pressure, and there
was very little lateral resistance on the pile even under large
relative displacements. Furthermore, Yao et al. [8] used large
shaking table tests and concluded that the transient state prior
to soil liquefaction was important in the design of piles because
dynamic earth pressure showed peak response in this state.
Other researchers such as Abdoun and Dobry [9], Suzuki et al.
[10], Dungca et al. [11], Bhattacharya et al. [12], Tamura and
Tokimatsu [13] and Han et al. [14] also investigated dynamic
behavior of pile foundations in liqueable soils using shaking
table tests.

1.3. Numerical modeling


The effectiveness of numerical simulation tools for analyzing
liquefaction problems have become more important and prominent in the light of potential disadvantages of physical models
used in experimental simulation [15]. Since two and three dimensional numerical modeling are computationally complex and
time-consuming, most of the researchers and designers prefer to
use one-dimensional Winkler method based on nite element or
nite difference methods for the seismic analysis of pile foundations. Kagawa [16], Yao and Negami [17], Fujii et al. [18], and
Liyanapathriana and Poulos [19] developed this method so that
liquefaction of surrounding soil was taken into account during
analyzing process. Miwa et al. [20], Liyanapathirana and Poulos
[19,21], and Chang et al. [22] showed that one-dimensional method is approximately capable of predicting maximum lateral displacement and maximum bending moment of pile foundations
in liqueed soils; however, it is obvious that Winkler models
are not able to simulate the prototype model accurately because
it is difcult to estimate the accurate values for the springs and
dashpots coefcients which considerably change over time. Finn
and Fujita [23], Klar et al. [24], Oka et al. [25], Uzuoka et al.
[26], Cheng and Jeremic [15] Comodromos et al. [27] used
three-dimensional nite element method in order to simulate
piles in liquefying soil layers. Each of these models possesses
varying prediction accuracy and certainty. In some of these papers
fully-coupled formulation (uP or uPU formulation) has been
employed; while in others the uncoupled formulation, in which
soil skeleton displacements and pore water pressure generation
are computed separately, has been used. According to the studies,
three-dimensional models are able to simulate most of the phenomena which have been observed in the eld more accurately
than that of one-dimensional models.
In general, considering previous studies on the performance of
pile foundations embedded in liqueable grounds, one may come
to the conclusion that there is a signicant lack of understanding
in involved mechanisms. Besides, it should be noted that in the
previous studies the variation of soil permeability during liquefaction has not been considered in the modeling. However, it is commonly demonstrated that soil permeability considerably changes
during liquefaction. Therefore, in the present study, it is intended
to take permeability variation into account in the soilpile-superstructure simulation. The work presented in this paper utilizes a
fully coupled three-dimensional dynamic analysis together with
a well-calibrated constitutive model and a veried numerical
methodology in order to simulate the behavior of piles embedded
in liqueable soils more accurately.

2. Numerical formulation
In this study, a uP fully coupled formulation, presented by Zienkiewicz and Shiomi [28], is used for modeling of soil skeleton and
pore uid. The uP formulation captures the movements of the soil
skeleton (u) and the change of the pore pressure (P). This formulation is applicable for dynamic problems in which high-frequency
oscillations are not important, such as soil deposit under earthquake loading. Using the nite element method for spatial discretization, the uP formulation is as follows [29]:

MU

BT r0 dV  Q P  f

_ HP SP_  f p 0
QTU

where M is the mass matrix, U is the solid displacement vector, B is


the straindisplacement matrix, r0 is the effective stress tensor, Q
indicates the discrete gradient operator coupling the motion and
ow equations, P is the pore pressure vector, S is the compressibility matrix, and H is the permeability matrix. The vectors f(s) and f(p)
include the effects of body forces, external loads and uid uxes.
Numerical integration of the above-mentioned equations is
done using Newmark algorithm, and implementation of these procedures was performed using OpenSees [30] framework, which is
an object-oriented program for nite element analysis. The Open
System for Earthquake Engineering Simulation (OpenSees) is a
comprehensive and continually developing software that is used
in the simulation of seismic response of structural and geotechnical systems. In this study, a number of elements and material models from UCD computational geomechanics toolset, available in this
software, are employed. The employed elements and material
models are discussed in the following sections.
3. Constitutive modeling of sand behavior
Material model is one of the most important parts of numerical
simulation of the dynamic behavior of liqueable soils. Using a
comprehensive constitutive model which possesses the simulative
ability to model the behavior of drained or undrained saturated
sands under monotonic and cyclic loadings leads to accurate modeling of problems where liquefaction is involved. Accordingly, in
this research a critical state two-surface plasticity model developed by Dafalias and Manzari [31] is used. The most striking feature of this model is its capability to utilize a single set of
material parameters for a wide range of void ratios and initial
stress states for the same soil. It should be noted that initial stress
states, void ratio and fabric evolve through all stages of loading
(see Ref. [29] for more details about the employed material model).
This model possesses 15 parameters divided into 6 categories
based on their functions. These parameters are calibrated for Nevada sand by Shahir [32,33] using tests performed by Earth Technology Corporation in the course of the VELACS project [34].The
calibrated parameters are listed in Table 1.
4. Variation of soil permeability during liquefaction
Many studies have demonstrated that permeability coefcient
signicantly increases during liquefaction phenomenon due to
structural change in soil skeleton. At the onset of liquefaction, soil
particles lose full contact with each other, and this change creates
additional pathways for water. The creation of such new, larger
ow pathways reduces the pore shape factor and tortuosity
parameters, and consequently leads to a signicant increase in permeability coefcient [32]. The amount of increase in permeability
coefcient during liquefaction has been reported in some

116

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126


Table 1
Material parameters used for DafaliasManzari Model [32,33].
Parameter function

Parameter index

Value

Elasticity

G0

150.0
0.05

Critical state

M
c
kc
e0
n

1.14
0.78
0.027
0.83
0.45

Yield surface

0.02

Plastic modulus

h0
ch
nb

9.7
1.02
2.56

Dilatancy

A0
nd

0.81
1.05

Fabric-dilatancy

zmax
cz

5.0
800.0

5. Pilesoil model development

investigations. Arulanandan and Sybico [35], based on the measurement of changes in the electrical resistance of saturated sand
deposit during liquefaction in the centrifuge tests, concluded that
in-ight permeability of saturated sand during liquefaction increases up to 67 times greater than its initial value. Jafarzadeh
and Yanagisawa [36] by measurement of the volume of the expelled water from saturated sand columns in shaking table model
tests indicated that the average permeability coefcient during
excitation is 56 times greater than its static value. Manzari and
Arulanandan [37] used variable permeability in their numerical
simulation. In their study, predictions of excess pore pressure
and settlement were satisfactory, but lateral displacements were
not simulated reasonably well. Balakrishnan [38] employed a factor of 10 for increasing the permeability coefcient in numerical
model in order to adjust the results of the simulation with the centrifuge test measurements for the soil settlement during liquefaction. Also, according to Taiebat et al. [39] and Shahir and Pak
[40] using a constant value of permeability coefcient in numerical
analysis results in a much smaller value of soil settlement
compared to the measured value. Shahir and Pak [40] concluded
that incorporation of permeability variation in the numerical
model is necessary for capturing both pore pressure and settlement responses of a liqueable soil mass.
Therefore, in this study, variation of permeability coefcient has
been considered in numerical modeling of liqueable layers using
a formulation suggested by Shahir and Pak [40] in which a direct
relationship between the permeability coefcient and excess pore
water pressure ratio (ru) was proposed. This relationship is as
follows:
kp
ki

1 a  1r bu1

During PWP build up phaser u < 1

kp
li

During liquefied stater u 1

kp
ki

1 a 

1r bu2

During consolidation phaser u < 1

where ki is initial permeability coefcient, kb is permeability coefcient during excitation, ru is dened as the ratio of the difference of
current pore pressure and hydrostatic pore pressure over the initial
effective vertical stress (ru Du=r0v 0 ). a; b1 ; b2 are positive material
constant. These parameters are 20, 1.0 and 8.9, respectively for Nevada Sand [40]. This basically means that the permeability coefcient increases up to 20 times during the initial liquefaction. It is
to be noted that the proposed value for a is consistent with the reported value by Balakrishnan [38] because the peak value of 20 is
nearly equivalent with average value of 10. Validity and efciency
of the proposed formulation can be found in Refs. [33] and [40].

In this research, soil layers are modeled by cubic eight-node


elements with uP formulation (called EightNodeBrick_u_p element in OpenSees framework) in which each node has four degrees of freedom: three for soil skeleton displacements and one
for pore water pressure. Pile is modeled by beam-column elements which have six degrees of freedom for each node: three
for displacements and three for rotations. A lumped mass on the
pile head represents the superstructure. The nite element mesh
is presented in Fig. 1.
One of the important and difcult steps in numerical simulation
of pile foundations in the soil media is the connection of pile elements to the surrounding soil elements. In the present work, the
connection is provided by means of rigid beam-column elements
which possess the same physical properties of pile elements. As
shown in Fig. 2, these elements connect each pile node to surrounding nodes of soil elements at an equal depth. At the connection point, soil element nodes slaved to the connection element
node for three translational DOFs, while the three rotational DOFs
of the connection element are left unconnected. Furthermore, slippage between pile elements and surrounding soil elements is feasible in all directions by using zero-length interface elements,
which are dened by two nodes at the same location where the
connection beam-column element is connected to the surrounding
soil (as shown in Fig. 2). An elastic-perfectly plastic material model
is used for the interface elements. Some static eld tests and also a
centrifuge test, discussed later in this paper, were simulated in order to obtain suitable mechanical properties of the interface element. Based on these studies, Youngs modulus and yielding
strain of this material are selected to be 2000 kPa and 0.04, respectively and are used throughout the main simulations of pile behavior in liqueable soil layer. It is to be noted that values considered
for these parameters lead to immediate yielding of interface element i.e. slippage at soilpile interface can take place.
However, further studies and simulations revealed that the effect of interface elements was not that signicant for the centrifuge
test that has been simulated in this study. Comparing the numerical results of the model with and without employing interface elements revealed a maximum difference of 5%. This can be specic to
the kind of the problem that is studied in this research where the
saturated liqueable ground is horizontal and the soil is cohesionless. Nevertheless, the above mentioned interface element properties were employed to improve the quality of simulations and
match the numerical results with the experimental values as much
as possible.
Due to the symmetry of the model, the model is halved at the
line of symmetry along the center-line of the pile, and all applied
static loads are halved. Soil elements are coarser far from the pile
and ner around the pile (see Fig. 1).
Boundary conditions are set in the following way:
 Base of the mesh is fully xed in all directions.
 At the side planes, parallel to the excitation direction, nodes are
restrained from movement in the y direction, and at the ones,
perpendicular to the excitation direction, the nodes at equal
depths are constrained to have equal displacements in the x
direction to simulate free-eld ground motion.
 All other internal nodes are free to move in any direction.
 Pore water pressures are free to develop for all nodes except the
ones at the level corresponding to the ground water table
elevation.
Simulations are carried out in three loading stages. In the rst
stage of loading where pile elements are not installed, self-weight,

117

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

)
(m

Beam-Column Element

21.0 (m)

21.0

17.0

ZP

u-P Elements

3.8

Lumped Mass
.4
42

z
y

4.0

Y
X

X : 3@5(m)+1@2.68(m)+2@1.34(m)+
1@0.5(m)+1@0.335(m)
Y : 2@2.68(m)+1@0.5(m)+1@0.335(m)
Z : 5@2(m) +4@1.5(m)+5@1(m)
ZP: 3@1(m)+1@0.8(m)

42.4
z

Input Motion
x

Fig. 1. Finite element mesh.

Surrounding Soil Elements

y
x
Connection Elements

Interface Elements
Pile Elements

Fig. 2. Outline of Connections between pile elements and surrounding soil


elements.
Fig. 3. Layout of the model for centrifuge test by Wilson et al. [7].

including both the soil skeleton and the pore water weight, are applied on soil elements. In this stage the initial stress state, void ratio and soil fabric evolve. These values are used as initial values for
the next stage of loadings. The second stage includes pile installation and application of its self-weight and the superstructure
weight. Then, at the nal stage, an acceleration time history is applied to the model as an input motion, and dynamic analysis are
performed for the soilpile-superstructure system.

6. Verication and validation of the numerical model


Verication is meant to identify and remove errors in computer
coding and verify numerical algorithms and is desirable in quantifying numerical errors in computed solution [15]. Accordingly, in
this study, due to the sophisticated methodology used to develop
a pile in a soil deposit together with an advanced constitutive
model for soil skeleton and the applied three-dimensional uP
formulation, a detailed verication is done. In the rst step, the
numerical model is veried against some closed form elastic solutions and benchmark problems in which a single pile is modeled in
non-liquefying soil layers under static vertical or horizontal loadings, e.g. Poulos and Davis [41], Aristonous et al. [42], and Kuchukarsalan [43]. Comparison of results demonstrated the capability of
the numerical model to predict the result by the maximum error of
4%. In the next step, results of a centrifuge test on pile foundations
are used to demonstrate the capability of the model for reliable
analysis of piles under dynamic loading. For this purpose, the

dynamic centrifuge test of pile-supported structure in liqueable


sand performed by Wilson et al. [7] is simulated.
The soil prole consists of two horizontal layers of saturated,
ne and uniformly graded Nevada sand (D50 = 0.15 mm, Cu = 1.5).
The lower dense layer (Dr = 80%) is 11.4 m thick, and the upper
medium dense layer (Dr = 55%) is 9.1 m thick at the prototype scale
(see Fig. 3). Furthermore, the single pile is equivalent to a steel pipe
pile with a diameter of 0.67 m and wall thickness of 19 mm at the
prototype scale. The pile is extended 3.8 m above the ground level
and carries superstructure load of 480 kN; the embedded length of
pile is about 16.8 m. The container is lled with a hydroxyl-propyl
metyl-cellulose and water mixture whose viscosity is about 10
times greater than pure water.
The soilpile-superstructure system was spun at a centrifugal
acceleration of 30 g, and the pile remained elastic during earthquake loading.
As shown in Fig. 1, the nite element mesh consists of 896 cubic
eight-node soil elements and 16 beam-column elements, four elements are used to model the free-standing length of the pile and
the rest are within the soil strata. Also, the pile head is free to move
in all directions. Properties of Nevada Sand used in the numerical
model are presented in Table 2. According to the laws of centrifuge
modeling, permeability coefcient is three times greater than the
value at the prototype scale.
The Kobe 1995 acceleration record is scaled to 0.22 g and used
as an input motion to shake the model; the base input acceleration

Table 2
Material parameters for Nevada sand, Popescu and Prevost [44].
Parameter

Unit

Porosity (n)
Saturated unit weight
Permeability coefcient
Permeability coefcient
in the prototype scale

kN/m3
m/s
m/s

Value for
Dr = 55%

Value for
Dr = 80%

0.409
19.87
6.05  105
1.815  104

0.377
20.41
3.7  105
1.11  104

Acceleration (g)

0.3
0.2
0.1
0
-0.1

10
12
Time (sec)

14

16

18

20

Excess Pore Pressure


Ratio

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

1.2

Depth = 1 m

1
0.8
0.6
0.4
Centrifuge Test
Simulation

0.2
0
-0.2 0

10

15

20

25

20

25

Time (sec)
Depth = 4.5 m

Excess Pore Pressure


Ratio

118

0.8
0.6
0.4
Centrifuge Test
Simulation

0.2
0
-0.2 0

-0.2

10

15

Time (sec)

-0.4

-0.3

7. Parametric study
In order to provide better insights into the dynamic behavior of
piles embedded in liqueable soil layers, a parametric study has
been carried out on three different soil proles by varying

Excess Pore Pressure


Ratio

Depth = 21 m

0.8
0.6
0.4
0.2

Centrifuge Test
Simulation

0
-0.2

15

10

20

25

Time (sec)

Fig. 5. Comparison of time histories of excess pore pressure ratio in the free eld at
the depths of 1, 4.5, 21 m with the centrifuge test by Wilson et al. [7].

Acceleration (g)

1.5
1

Centrifuge Test

0.5

Simulation

0
-0.5

10

15

20

25

Time (sec)

-1
-1.5
Fig. 6. Comparison of time histories of superstructure acceleration with the
centrifuge test by Wilson et al. [7].

Bending Moment (MN.m)

is shown in Fig. 4. Time history of the measured and computed excess pore water pressure ratio at three different depths in the free
eld: 1, 4.5, 21 m, are presented in Fig. 5. It is important to note
that in this study, excess pore water pressure ratio (ru) is dened
as the ratio of the difference of current pore pressure and hydrostatic pore pressure over the initial effective vertical stress
ru Du=r0v 0 . The results indicate that there is generally a good
agreement between measured and computed pore water pressure.
Fig. 6 shows the computed and measured acceleration time histories of the superstructure. It is concluded that the applied method is also capable of predicting the acceleration values. It is
important to note that sharp acceleration spikes can be seen corresponding to sharp excess pore pressure ratio decrease at the depth
of 1 m. This is due to the temporary increase in stiffness of the soil
which results in large acceleration spikes transmission from the
ground to the superstructure.
Fig. 7 shows the measured and computed bending moment
time histories at two different depths; 1 and 2 m. It can be seen
that the results obtained from the numerical model agree reasonably well with the values recorded during the centrifuge test.
According to the computed and measured results, the maximum
bending moment at the depth of 1 m occurs at the time t = 3.5 s.
As depicted in Fig. 5, this time corresponds to sudden increase of
pore water pressure which results in the softening of surrounding
soil; and also as shown in Fig. 6, the time t = 3.5 s corresponds to
the peak value of superstructure acceleration which results in a
large amount of inertial forces induced to the pile shaft. Therefore,
it can be concluded that the maximum value of bending moment
recorded and computed at t = 3.5 s is the consequence of the surrounding soil softening and the large amount of inertial forces
developed at the pile head.
Finally, soil displacements are compared with those recorded
during the centrifuge test. Fig. 8 shows the time history of ground
surface settlement at the distance of 3 m from the pile. It is observed that there is a good agreement between the computed
and measured values.

Bending Moment (MN.m)

Fig. 4. Input earthquake ground motion (Acceleration record of Kobe (1995)


earthquake scaled to 0.22 g) [7].

Depth Z = 1 m

Centrifuge Test

Simulation

0
-1

10

15

20

25

Time (sec)

-2
-3
3

Depth Z = 2 m

Centrifuge Test

Simulation

0
-1
-2

10

15

20

25

Time (sec)

-3

Fig. 7. Comparison of time histories of bending moment with the centrifuge test by
Wilson et al. [7].

119

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

20

Time (sec)
Settlement (mm)

0
0

10

15

20

25

-20
Centrifuge Test
Simulation

-40

-60

-80
Fig. 8. Comparison of time histories of settlement at the distance of 3 m from the
pile.

boundary condition of pile head, pile length (L), thickness of liquefying soil layer (HL), relative density of liquefying soil layer (Dr)
and frequency of input motion (f). In the rst prole, the ground
consists of one homogenous and liqueable soil layer. In the second prole, the ground is two-layered: the upper layer is liqueable while the lower layer is not, and in the third one, the
ground is two-layered: the upper layer is dry and the lower layer
is saturated and liqueable. In all cases, the ground is level, and lateral spreading phenomenon is not plausible. Fig. 9 shows these
proles. It is important to note that the amplitude of input acceleration is larger for the third case due to the larger values of initial
effective stress at lower layer. Amplitude of the sinusoidal input
motion is 0.15 g for the rst and second soil proles and 0.5 g for
the third soil prole.

The nite element mesh used for simulation of the proposed


soil proles is shown in Fig. 10. It consists of 1024 cubic eight-node
soil elements. The concrete pile cross section is assumed to be
square with sides (B) of 50 cm, and it is also assumed to remain
elastic during the excitation. Two different boundary conditions
are assumed at pile head: free head boundary condition in which
pile head is free to move and rotate in any direction and xed head
boundary condition in which pile head is free to move in any direction but constrained against any rotation. The material properties
of the pile and Nevada sand are shown in Table 3. In all cases,
the superstructure is simulated by a single lumped mass with a
load of 1000 kN. The input acceleration record used for the analysis
is a sinusoidal acceleration time history with 10 s duration. The
analysis has been repeated for pile lengths of 15B, 25B and 40B,
and acceleration frequencies of 1, 3, 5 and 10 Hz. For the sake of
comparison, the fully xed pile head (against displacement and
rotation both) has also been simulated, but the results will not
be demonstrated here. The interested reader may refer to [45].
Before going into the study of the effects of mentioned
parameters on piles dynamic behavior, it is important to study
the dynamic performance of piles in each soil prole.

7.1. Pile response


The analysis has been repeated for pile lengths of 15B, 25B and
40B. Fig. 11 shows maximum lateral displacement of pile and maximum bending moment envelops for the 40B length pile for cases I,
II and III, respectively (due to the similar results obtained for each
pile length, only the results of the pile length of 40B are presented

Case I

Case II

Case III

W = 1000 kN

W = 1000 kN

W = 1000 kN
Dry Soil

Liquefiable Soil

Dr = 40 %

Dr = 30,40,50 %

25 m
Liquefiable Soil
Dr = 40 %

Nonliquefiable Soil
Dr = 85 %

PGA = 0.15g

PGA = 0.15g

Liquefiable Soil
Dr = 40 %

PGA = 0.5g

Fig. 9. Schematic of soil proles in cases I, II, and III.

u-P Elements

40

.5

Lumped Mass

Beam-Column Element

)
(m

25.0 (m)

25.0 (m)

z
x
Y

40.5 (m)
X

z
y

X : 3@5.0(m)+1@2.0(m)+3@1.0(m)+1@0.25(m)
Y : 2@2.0(m)+1@1.0(m)+1@0.25(m)
Z : 7@2.0(m) +4@1.5(m)+5@1.0(m)

Input Motion

Fig. 10. Finite element mesh. (Dark zone represents the pile.)

120

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126


Table 3
The material properties of the pile and Nevada sand used in the numerical model.
Material Parameter

Value

Pile
Youngs Modulus, (kPa)
Density, (ton/m3)
Poissons Ratio

Ep

qp
mp

Material Parameter

Value

Soil
Soil density (ton/m3)
Fluid density (ton/m3)
Porosity
Permeability (m/s)

3.0  107
2.40
0.2

qsat
qf
n
k

Dr = 30%

Dr = 40%

Dr = 50%

Dr = 85%

1.938
1.0
0.438
7. 5  105

1.957
1.0
0.427
6.6  105

1.976
1.0
0.415
5.9  105

2.052
1.0
0.370
3.7  105

here). A number of observations can be made about these results.


Firstly, for case I, the maximum bending moment invariably develops at the depth of about 2 m for free-head pile, and at the pile
head for xed-head pile (only xed against rotation). For case II,
the maximum bending moment develops at two locations; the rst
one is at the place explained for case I, and the other one is observed at the depth corresponding to the interface of liqueable
and non-liqueable layers; however, for the third case, the second
maximum bending moments are attained inside the lower liqueable layer. Secondly, by investigating time histories of lateral displacements and bending moment, it can be concluded that for all
three cases, the maximum lateral displacement of pile is attained
after when the soil layers liquefaction takes place. This is true for
the maximum bending moment which is attained at the bottom
of liqueable layer; however, the maximum bending moment near
the pile top, develops at the rst moments of dynamic loadings before liquefaction.
Ishihara [46] indicated that inertial forces which are the predominant forces before liquefaction are mainly responsible for
development of maximum bending moment near the pile head,
and kinematic forces which are predominant after liquefaction
are responsible for the maximum bending moment observed at
the interface of liqueable and non-liqueable layers. This is conrmed by the numerical method; the results are shown in Fig. 12
in which the maximum bending moment envelops of 25B and
40B length piles are obtained for piles with and without superstructure mass. It is interesting to observe that when the superstructure is removed, the maximum bending moment near the
pile head decreases signicantly, and no peak values are observed
at that location; while values at depths below 5 m are approximately unchanged. In other words, the same kinematic forces have
been developed in piles with and without superstructure.
For more clarication of the issue, the performance of a pile
embedded in a dry ground is compared with the dynamic performance of a pile embedded in a saturated (i.e. liqueable) ground.
The obtained results for a free-head and xed-head pile in Case II
are shown in Fig. 13. It is concluded that in dry grounds, the lateral
displacements of the pile are signicantly less than the values calculated in a saturated ground. This is due to the fact that lateral
displacement of dry soil is far less than lateral displacement of saturated soil; so little kinematic forces are exerted to the pile embedded in a dry ground.
7.2. Soil response
To investigate the effect of pile foundations on the surrounding
soil response, excess pore water pressure time histories computed
near the pile are compared with those computed in the free eld at
two different depths. The results are shown in Fig. 14 for a pile
with a length of 40B (21 m) in case I. It can be seen that at the
depth of 1 m (near the pile head), the time history of excess pore

water pressure near the pile is signicantly different from the


one in the free eld during excitation period, while after excitation
(from 10 to 25 s), time histories are nearly identical. Since the difference between time histories decreases in the depth far from pile
head (i.e. depth of 15 m), the dynamic response of superstructure
seems to be responsible for the observed behavior. In other words,
inertial forces from the superstructure caused the pile top to vibrate in a completely different nature from the surrounding soil,
and this leads to larger shear strains exerted to the soil which leads
to temporary dilation and contraction behavior. It is to be noted
that as seen in Fig. 12 inertial forces only affect pile top sections.
For this, time histories of excess pore water pressure near the pile
and in the free eld are approximately identical at deeper depths.
Furthermore, lateral displacement of soil near the pile is compared with the lateral displacement of soil in the free eld before
and after liquefaction of the ground. The results are shown in
Fig. 15 for case I and III (The results obtained for case I is same
as case II). It is noted that the lateral displacement envelope of
the soil near pile and the soil in the free eld have an identical
trend before the ground liquees while there is a signicant difference between the lateral displacement of soil near the pile and the
soil in the free eld after liquefaction occurs. Accordingly, it seems
that dynamic response of pile entirely differs from dynamic response of soil after liquefaction.
7.3. Effect of pile length on pile performance
In this section, results obtained from repetitive analysis for pile
lengths of 15B, 25B and 40B for three different soil proles are discussed (B = pile width). Due to the similarity of results for three
cases, only the results of case II are presented in Fig. 16. It is concluded that for all three proles and any pile head boundary conditions, pile length has no effect on maximum lateral
displacement of pile during excitation (i.e. from 0 to 10 s) while
signicant changes are observed after excitation period (i.e. from
10 to 25 s); a change of about 15B in pile length results in a change
of 2040% in maximum value of pile lateral displacement. Aristonous et al. [42] and many other researchers reported that for piles
embedded in dry soil layers, pile length has a little effect on pile
lateral displacements. However, in this study, it is concluded that
pile length signicantly affects pile lateral displacements after
excitation. From t = 10 s to t = 25 s, inertial forces are minimum
not only because the excitation has nished but also because soil
layers have liqueed so the kinematic forces, which are predominant, are exerted to the longer pile since the longer pile is in touch
with greater amount of the liqueed soil.
Fig. 17 shows the maximum bending moment envelops of freehead and xed-head piles with lengths of 15B, 25B and 40B. In all
cases, it can be concluded that pile length has no effect on the place
of maximum bending moment, and the maximum value always attained at pile head or at the depth of about 2 m.

121

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

Maximum Bending Moment


(kN.m)

Maximum Lateral Disp. (cm)


0

0
0

0.5

1.5

100

150

-15

-10
-15
-20

Fixed Head
Free Head

Maximum Lateral Disp. (cm)


1.5

Fixed Head

Maximum Bending Moment


(kN.m)
0

-5

50

100

150

-15

Depth (m)

Depth (m)

-10

-10
-15
-20

Fixed Head
Free Head

Maximum Lateral Disp. (cm)

Fixed Head

Maximum Bending Moment


(kN.m)

Maximum Bending Moment


(kN.m)
0

0
0

-5

100

200

300

-20

Fixed Head
Free Head

-25

Depth (m)

Depth (m)

-15

200

400

600

-5

-5

-10

-15

-25

(b)

0
6

300

-20

-25

200

-10

Free Head

-25

100

-5

-5

-15

Maximum Bending Moment


(kN.m)
0

-20

-10

-25

(a)

0
1

300

-20

-25

0.5

200

Free Head

-25

100

-5

Depth (m)

-10

-20

Depth (m)

50

-5

Depth (m)

Depth (m)

0
0

-5

Depth (m)

Maximum Bending Moment


(kN.m)

-10
-15
-20

-10
-15
-20

Free Head

-25

Fixed Head
-25

(c)

Fig. 11. Maximum lateral displacement and maximum bending moment envelops for a free-head and xed-head (xed against rotation) pile in (a) Case I (b) Case II
(Thickness of liqueable layer is 11 m.) (c) Case III (Thickness of dry layer is 5 m.).

Maximum Bending Moment (kN.m)

0
0

50

100

150

Without SS Mass
With SS Mass

-12

Depth (m)

Depth (m)

50

100

150

-5

-4

-8

Maximum Bending Moment (kN.m)

-10
Without SS Mass

-15
With SS Mass

-20
Pile Length = 25B

-16

Pile Length = 40B

-25

Fig. 12. Maximum bending moment envelops for pile lengths of 25B and 40B (B: pile width) with and without superstructure mass (Case I).

7.4. Effect of frequency of excitation on pile performance


Fig. 18 shows maximum lateral displacement of pile, maximum
bending moment and pile head settlement time histories for different frequencies of excitation; 1, 3, 5 and 10 Hz. It is noted that the

frequency has a signicant effect on pile response. For example, by


increasing the frequency from 3 Hz to 10 Hz, maximum lateral displacement, maximum bending moment and settlement decreases
about 75%, 90% and 70%, respectively. Since acceleration amplitude
of the input motion is the same in all analysis, displacement

122

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

Maximum Lateral Disp. (cm)


0
0

0.5

1.5

-5

0.5

1.5

-5

Depth (m)

Depth (m)

Maximum Lateral Disp. (cm)

-10

-15

-10
-15
Fixed Head

Free Head
-20

-20

Dry Soil
Liquefiable Soil

Dry soil
Liquefiable Soil

-25

-25

amplitude of the soil becomes smaller as frequency increases so


small displacement amplitude is responsible for the observed
behavior.

Depth = 1 (m)

10
0
0

10

-10

15

20

Time (sec)

7.5. Effect of the boundary condition of pile head on pile performance

25
Free Field
Near Pile

-20

Two different pile head conditions are investigated in this


study; free-head and xed-head condition. The obtained results
indicate that if the pile top is xed against rotations, maximum lateral displacement decreases about 1015% for case I, and 1520%
for cases II and III. It is to be noted that pile head xity leads to
nearly 50% decrease in maximum lateral displacement in the case
where the ground is dry. Therefore, it can be concluded that for dry
soil, boundary condition of pile head reduces maximum lateral displacement of pile more than that of saturated and liqueable soils.
This is due to the fact that in the saturated ground, the pile is entirely under the control of kinematic forces after liquefaction so the
xity of pile head does not decrease displacement as much as it
does in dry condition.
On the contrary, in saturated ground, the xity of pile head
leads to much more increase in maximum bending moment

-30
200

Depth = 15 (m)

150
100

Free Field
Near Pile

50
0
0

10

-50

15

20

25

Time (sec)

Fig. 14. Comparison of time histories of excess pore water pressure generated near
pile and at the free eld (Case I, pile length of 40B).

Lateral Displacement (cm)

Lateral Displacement (cm)

Case I
-1

-0.8

-0.6

-0.4

-0.2

-5

0
0

0.2

-1.5

-1

-0.5

-5

-10

-20

Near Pile
Free Field

-15

-25

-30

Lateral Displacement (cm)

Lateral Displacement (cm)


0

0
-0.5

-5
-10
-15

Near Pile
Free Field

-20
-25

Before liquefaction

0.5

-3

-2

-1

-5
-10

Depth (m)

-1

-25

After liquefaction

-30

-1.5

-20

Near Pile
Free Field

Before liquefaction

Case III

0.5

-10

Depth (m)

-15

Depth (m)

20

-15
Near Pile
Free Field

Depth (m)

Excess Pore Water Pressure


(kPa)

Excess Pore Water Pressure


(kPa)

Fig. 13. Comparison of maximum lateral displacement of the free-head and xed-head pile embedded in dry soil and liqueable soil in Case II (Thickness of liqueable layer
is 11 m.).

-20
-25

After liquefaction

-30

-30

Fig. 15. Comparison of lateral displacement of soil near pile and at the free eld before and after liquefaction in Case I and III.

123

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

Maximum Lateral Displacement (cm)


0

0.5

1.5

-5

Depth (m)

-5

Depth (m)

Maximum Lateral Displacement (cm)

-10
-15

-10
-15

L/B= 15 ( HL = 4 m)
-20

(a)

-25

L/B= 15 ( HL = 4 m)
-20

L/B= 25 ( HL = 6.5 m)
L/B= 40 ( HL = 11 m)

L/B= 25 ( HL = 6.5 m)

(b)

-25

L/B= 40 ( HL = 11 m)

Fig. 16. Comparison of pile maximum lateral displacement for different pile lengths in Case II (a) during excitation (t = 010 s) and (b) after excitation (t = 1025 s) (L: pile
length, B: pile width, HL: thickness of liqueable layer).

Maximum Bending Moment (kN.m)

Maximum Bending Moment (kN.m)


0

0
0

50

100

150

200

Depth (m)

Depth (m)

100

200

300

400

-5

-5
-10
-15

-10
-15
L/B=15 (HL = 4 m)

L/B= 15 ( HL = 4 m)
-20

L/B= 25 ( HL = 6.5 m)

-20

L/B= 40 ( HL = 11 m)

Free Head

L/B=25 (HL = 6.5 m)


L/B=40 (HL = 11 m)

Fixed Head

-25

-25

Fig. 17. Comparison of maximum bending moment for different pile lengths for free-head and xed-head(against rotation) pile in Case II (L: pile length, B: pile width, HL:
thickness of liqueable layer).

Maximum Lateral Disp. (cm)

Maximum Bending Moment (kN.m)

10

15

f = 1 Hz
f = 3 Hz
f = 5 Hz
f = 10 Hz

Depth (m)

Depth (m)

-25

200

400

600

800

-5

-10

-20

-10

-15

-20

-25

f = 1 Hz
f = 3 Hz
f = 5 Hz
f = 10 Hz

Pile Head Settlement (m)

0
-5

-15

Time (sec)

10

20

-0.05

-0.1

-0.15

f= 3 Hz
f=5 Hz
f= 10 Hz

-0.2

Fig. 18. Variation of maximum lateral displacement, maximum bending moment and pile head settlement for different frequencies of input excitation in Case I.

compared to that corresponding to dry soils. According to the results, if the pile head is xed against rotations, maximum bending
moment increases about 95% for case I, 130% for case II and 95% for
case III. However, in dry ground, pile head xity leads to nearly 20%
increase in maximum bending moment. In liquefying soils, when
the pile head is restrained rotationally, relative lateral displacement of pile at upper and lower regions signicantly increases
due to liquefaction of surrounding soil (it can be seen in Fig. 13);
larger relative lateral displacements lead to larger bending moment so boundary condition of pile head has a big effect on bending moment in the cases where the ground liquees.
7.6. Effect of thickness of liqueable soil layer on pile performance
In this section, results obtained from repetitive analysis for
thickness of liqueable layers(HL) of 5 m, 11 m and 15 m for a
pile length of 40B (i.e. 21 m) are discussed. This parameter is
investigated only for the Case II. Fig. 19 shows maximum lateral

displacement of the pile for various thicknesses of liqueable layers. It is concluded that the thickness of liqueable soil layer has a
little effect on maximum lateral displacement of pile during excitation (i.e. from 0 to 10 s) while signicant changes are observed
after excitation period (i.e. from 10 to 25 s). According to the results, twice increase in the thickness of liqueable layer leads to
about twice increase in the maximum lateral displacement of pile.
As mentioned before, after excitation period, piles are intensely under the control of the surrounding liqueed soil, so thicker liqueable layer considerably affects pile lateral displacements which can
be very important in performance-based design approaches.
Fig. 20 shows maximum bending moment envelops of freehead and xed-head pile for thickness of liqueable layers(HL) of
5 m, 11 m and 15 m for a pile length of 40B (i.e. 21 m). The maximum bending moment developed at the interface is investigated. It
is concluded that when the thickness is 5 m, there is 40% difference
between the value obtained for free-head pile and the value for
xed-head pile. However, for the thickness of 11 m and 15 m the

124

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

Maximum Lateral Displacement (cm)

Maximum Lateral Displacement (cm)


0

0
0

0.5

1.5

Depth (m)

Depth (m)

-10
-15

HL = 5 (m)
HL = 11 (m)
HL = 15 (m)

-20
-25

-5

-5

(a)

-10
-15

HL = 5 (m)
HL = 11 (m)
HL = 15 (m)

-20
-25

(b)

Fig. 19. Variation of maximum lateral displacement for various thicknesses of liqueable layers in Case II (a) during excitation (t = 010 s) and (b) after excitation (t = 10
25 s).

Maximum Bending Moment (kN.m)

Maximum Bending Moment (kN.m)

0
0

100

200

300

-10
-15

Fixed Head

-20

100

200

300

-5

HL= 5 m

Depth (m)

Depth (m)

-5

HL= 11 m

-10
-15
-20

Fixed Head
Free Head

Free Head
-25

-25

Maximum Bending Moment (kN.m)

0
0

100

200

300

Depth (m)

-5
-10
HL= 15 m

-15
-20

Fixed Head
Free Head

-25

Pile Head Settlement (cm)

Fig. 20. Bending moment envelops for a 40B length (B: pile width) free-head and xed-head pile in Case II.

25
20
15
10
5

L = 13 (m)
L = 21 (m)

0
0

10

15

20

25

30

Thickness of Liquefiable Layer (m)


Fig. 21. Variation of pile head settlement by the increase of thickness of liquefying
layer for the pile lengths(L) of 13 m and 21 m (Case II).

difference is nearly 14% and 0%, respectively. Results of numerical


analysis indicate that when the thickness of the liquefying layer
exceeds about one-fourth of the pile length, bending moment at
the interface does not depend on the boundary condition at the
pile head. It is worth mentioning that Liyanapathirana and Poulos
[19] reported this conclusion for the thickness of one-third of the
total thickness of the soil deposit using one-dimensional Winkler
Model.

Also, the effect of thickness of liqueable soil layer on pile head


settlement is investigated in this study. Fig. 21 shows variation of
pile head settlement by the increase of thickness of liquefying layer
for the pile lengths of 13 m and 21 m. It is noted that the increase
of the thickness of the liquefying layer has a little inuence on pile
settlement when bottom of the liqueable layer is far from the pile
toe while the settlement sharply increases by approaching the liqueable layer to the pile toe. It is commonly known that total settlement of piles is mainly due to the settlement of the soil in the
vicinity of pile toe. Therefore, when the liqueable layer gets closer
to the pile toe, settlement of pile head considerably increases.

7.7. Effect of relative density of liqueable soil layer on pile


performance
In this section, results obtained from repetitive analysis for
three different relative densities; 30%, 40% and 50% for a pile length
of 40B are discussed. Fig. 22 shows the maximum lateral deection
and maximum bending moment envelops for different relative
densities. It is observed that for the pile embedded in Nevada sand
layers, about 10% increase of relative density causes 15% decrease
in maximum value of lateral displacement and an average value of

125

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

Maximum Lateral Disp. (cm)

0
0

100

200

300

-5

Depth (m)

Depth (m)

-5
-10
-15
-20

Maximum Benidng Moment (kN.m)

Dr = 30 %
Dr = 40 %
Dr = 50 %

-25

-10
-15
Dr = 30 %
-20

Dr = 40 %
Dr = 50 %

-25

Fig. 22. Variation of pile maximum lateral displacement and maximum bending moment by the increase of relative density of liqueable soil layer in Case II (pile head: xed
against rotation and HL = 11 m).

30% decrease in the value of the bending moment. This is due to


the increase of the liqueable soil stiffness and smaller values of
soil displacement.
8. Conclusions
This paper tries to provide better insight into the performance
of pile foundations embedded in liqueable soil deposits. For this,
an accurate methodology for numerical modeling of piles in liqueable soils is employed. A fully coupled uP formulation is used to
analyze soil displacements and pore water pressures, and a bounding surface critical state elasticplastic model that accounts for
fabric change is employed to model the soil skeleton. Moreover,
the permeability coefcient is updated in each time step as a function of excess pore water pressure ratio. Results from a centrifuge
test are simulated and the obtained results demonstrate that the
numerical model has the ability to simulate pile behavior in liquefying soil reasonably well.
Parametric studies are carried out for three different soil proles. For each prole, the effect of pile length, xity of the pile
head, frequency of input motion, the thickness of liquefying soil
layer and the effect of relative density of liquefying soil layer on
pile performance are investigated. A summary of the ndings are
mentioned below:
It is found that for any values of the mentioned parameters and
for all three soil proles, the maximum lateral displacement of
free-head pile develops at pile head. Also, it is concluded that in
all cases the maximum bending moment develops at about 2 m
from pile top for free-head piles and at pile head for xed-head
piles. Besides, for case II and III there is other peak values of bending moment at the interface of liqueable and non-liqueable layers and inside the liquefying layer, respectively.
Due to the fact that pile dynamic response near the ground surface is approximately under the control of dynamic response of
superstructure, the pile has a signicant inuence on the seismic
response of the surrounding soil near the ground surface. Furthermore, the results of the numerical model demonstrate that pile response is different from the response of soil in the free eld after
liquefaction while before liquefaction there is a little difference.
For all cases, it is shown that pile length and the thickness of liqueable layer have little inuences on the maximum lateral displacements during excitation; however, when the input
excitation nishes there are considerable differences in the maximum values of the lateral displacements for different pile lengths.
It is also shown that pile length has no effect on the location of the
maximum bending moment.
For all cases, it is concluded that natural frequency of earthquake highly affects pile performance in liqueable soil deposits.
It is shown that if the frequency of the input motion increases
while the amplitude of acceleration remains constant, lateral

displacement, bending moment and pile head settlement signicantly decrease.


When the pile head rotation is xed in all directions, the maximum lateral displacement of pile embedded in dry soil layers reduces much more compared to that in saturated liqueable soil
layers. But maximum bending moment in saturated soil deposits
signicantly increases. Generally, it is concluded that in liqueable
ground, restraining pile head results in the development of larger
bending moment at the pile head although it decreases pile lateral
displacements; therefore, it is recommended that if the allowable
lateral displacement of superstructure and pile embedded in liqueable ground is large enough, the application of hinge joint at pile
head is much more efcient than the application of restrained
joint.
It is also shown that if the thickness of the liquefying layer
exceeds one-fourth of pile length, the bending moment at the
interface does not depend on the boundary condition at the pile
head. Moreover, it is found that 10% increase in relative density
of liqueable soil layer results in approximately 1530% decrease
in the maximum lateral displacement and maximum bending
moment.
In general, parametric studies have shown that frequency of
excitation, pile head xity and the thickness of liqueable layer
have higher effects on pile response compared to the other parameters. Therefore, the key to good designs is reliable estimates of
these parameters.

References
[1] Kramer SL. Geotechnical earthquake engineering. New Jersey: Prentice Hall
Inc.; 1996. p. 348422.
[2] Hamada M. Large ground deformation and their effects on lifelines: 1983
Nihokai-Chubu earthquake. In: Hamada M, ORourke T, editors. Case studies of
liquefaction and lifeline performance during past earthquake, (I): Japanese
case studies, vol. 4-1; 1992. p. 485 [chapter 4].
[3] Mori S, Namuta A, Miwa S. Feature of liquefaction damage during the 1993
Hokkaido Nanseioki earthquake. In: Proceedings of the 29th annual conference
of Japanese Society of soil mechanics and foundation engineering; 1994. p.
100508.
[4] Tachikawa H, Fujii S, Onishi K, Suzuki Y, Isemoto N, Shirahama M. Investigation
and analysis of pile foundation located on Kobe Port Island. In: Proceedings of
the 33rd Japan national conference on geotechnology engineering, vol. 1; 1998.
p. 8112.
[5] Shamoto Y, Sato M, Futaki M, Shimazu S. Site investigation of post liquefaction
lateral displacement of pile foundation in reclaimed land. Tsuchi to Kiso
1996;44(3):257.
[6] Onishi K, Namba S, Sento N, Horii K, Tatsumi Y, Oh-Oka H. Investigation of
failure and deformation modes of piles throughout overall length. Tsuchi to
Kiso 1996(45):246.
[7] Wilson DW. Soil pile superstructure interaction in liquefying sand and soft
clay, Ph.D. Dissertation, University of California at Davis; 1998.
[8] Yao S, Kobayashi K, Yoshida N, Matsuo H. Interactive behavior of soilpilesuperstructure system in transient state to liquefaction by means of large
shake table tests. Soil Dyn Earthquake Eng 2004;24:397409.
[9] Abdoun T, Dobry R. Evaluation of pile foundation response to lateral spreading.
Soil Dyn Earthquake Eng 2002;22:10518.

126

A. Rahmani, A. Pak / Computers and Geotechnics 40 (2012) 114126

[10] Suzuki H, Tokimatsu K, Sato M, Abe A. Factor affecting horizontal subgrade


reaction of piles during soil liquefaction and lateral spreading, Geotechnical SP
No.145, ASCE; 2005. p. 19.
[11] Dungca JR, Kuwano J, Takahashi A, Saruwatari T, Izawa J, Suzuki H, et al.
Shaking table tests on the lateral response of a pile buried in liqueed sand.
Soil Dyn Earthquake Eng 2006;26:28795.
[12] Bhattacharya S, Madabhushi PG, Bolton MD. An alternative mechanism of pile
failure in liqueable deposits during earthquakes. Geotechnique 2004(3):
20313.
[13] Tamura S, Tokimatsu K. Seismic earth pressure acting on embedded footing
based on large-scale shaking table tests, Geotechnical SP No.145, ASCE; 2005.
p. 8395.
[14] Han J, Kim SR, Hwang JI, Kim MM. Evaluation of the dynamic characteristics of
soilpile system in liqueable ground by shaking table tests. In: 4th
International conference on earthquake geotechnical engineering.
Thessaloniki, Greece; 2007. p. 1340.
[15] Cheng Zh, Jeremic B. Numerical modeling and simulation of pile in liqueable
soil. Soil Dyn Earthquake Eng 2009;29:140416.
[16] Kagawa T. Lateral pile response in liquefying sand. In: Proceedings of the 10th
world conference on earthquake engineering, Madrid, Spain; 1992, Paper No.
1761.
[17] Yao S, Nogami T. Lateral cyclic response of piles in viscoelastic Winkler
subgrade. J Eng Mech 1994;120(4):75875.
[18] Fuji S, Cubrinovski M, Tokimatsu K, Hayashi T. Analyses of damaged and
undamaged pile foundations in liqueed soils during the 1995 Kobe
Earthquake. In: Proceedings of the 1998 conference on geotechnical
earthquake engineering and soil dynamics III, Seattle, Wash., vol. 2; 1998. p.
118798.
[19] Liyanapathirana DS, Poulos HG. Seismic lateral response of piles in liquefying
soil. J Geotech Geoenviron Eng 2005(1311):146679.
[20] Miwa S, Ikeda T, Sato T. Damage process of pile foundation in liqueed ground
during strong ground motion. Soil Dyn Earthquake Eng 2005;26:32536.
[21] Liyanapathirana DS, Poulos HG. Analysis of pile behaviour in liquefying sloping
ground. J Comput Geotech 2009;37:11524.
[22] Chang DW, Lin BS, Cheng SH. Dynamic pile behaviors affecting by liquefaction
from EQWEAP analysis. In: 4th International conference on earthquake
geotechnical engineering. Thessaloniki Greece; 2007. p. 1336.
[23] Finn WDL, Fujita N. Pile in liqueable soils: seismic analysis and design issues.
Soil Dyn Earthquake Eng 2002;22:73142.
[24] Klar A, Baker R, Frydman S. Seismic soilpile interaction in liqueable soil. Soil
Dyn Earthquake Eng 2004;24:55164.
[25] Oka F, Lu CW, Uzuoka R, Zhang F. Numerical study of structure-soil-group pile
foundations using an effective stress based liquefaction analysis method. In:
13th World conference on earthquake engineering. Canada: Vancouver; 2004.
p. 3338.
[26] Uzuoka R, Sento N, Kazama M. Three-dimensional numerical simulation of
earthquake damage to group-piles in a liqueed ground. Soil Dyn Earthquake
Eng 2007;27:395413.
[27] Comodromos EM, Papadopoulou MC, Rentzepris IK. Pile foundation analysis
and design using experimental data and 3-D numerical analysis. Comput
Geotech 2009;36:81936.

[28] Zienkiewicz OC, Shiomi T. Dynamic behavior of saturated porous media; the
generalized Biot formulation and its numerical solution. Int J Numer Methods
Eng 1984;8:7196.
[29] Jeremic B. Development of geotechnical capabilities in OpenSees, Report no.
PEER2001/12, Pacic Earthquake Engineering Research Center; 2001.
[30] Mazzoni Silvia, McKenna Frank, Fenves Gregory L. Open system for earthquake
engineering simulation user manual. Berkeley: Pacic Earthquake Engineering
Research Center, University of California; 1999. <http://OpenSees.berkeley.
edu/>.
[31] Dafalias YF, Manzari MT. Simple plasticity sand model accounting for fabric
change effects. J Eng Mech 2004;130(6):62234.
[32] Shahir H. A performance-based approach for design of ground densication for
mitigation of liquefaction, Ph.D. Dissertation, Sharif University of Technology;
2009.
[33] Shahir H, Pak A. Estimation of liquefaction-induced settlement of shallow
foundations by numerical approach. Comput Geotech 2010;37:26779.
[34] Dafalias YF. Overview of constitutive model used in VELACS. In: Arulanandan,
Scott, editors. Verication of numerical procedures for the analysis of soil
liquefaction problems, vol. II, Rotterdam, Balkema; 1993.
[35] Arulanandan K, Sybico Jr J., Post liquefaction settlement of sand. In: Proceeding
of the wroth memorial symposium. England: Oxford University; 1992.
[36] Jafarzadeh F, Yanagisawa E. Settlement of sand models under unidirectional
shaking. In: Ishihara K, editor. First international conference on earthquake
geotechnical engineering, IS-Tokyo; 1995. p. 693698.
[37] Manzari MT, Arulanandan K. Numerical predictions for Model No. 1. In:
Arulanandan K, Scott RF, editors. Verication of numerical procedures for the
analysis of soil liquefaction problems. Rotterdam: A.A. Balkema; 1993. p.179
85.
[38] Balakrishnan A. Liquefaction remediation at a bridge site, Ph.D. Dissertation.
Davis: University of California; 2000.
[39] Taiebat M, Shahir H, Pak A. Study of pore pressure variation during
liquefaction using two constitutive models for sand. Soil Dyn Earthquake
Eng 2007;27(1):6072.
[40] Shahir H, Pak A. Variation of permeability during liquefaction and its effects on
seismic response of saturated sand deposits. In: 8th International Congress on
Civil Engineering, Shiraz, Iran; 2009.
[41] Polos HG, Davis EH. Pile foundation analysis and design. New York, NY: John
Wiley and Sons; 1980 [chapters 8 and 15].
[42] Trochanis Aristonous M, Bielak Jacobo, Christiano Paul. Three-dimensional
nonlinear study of piles. J Geotech Geoenviron Eng 1991;117(3):42947.
[43] Kucukarsalan S. Linear and non-linear soilpile-structure interaction under
static and transient impact loading, PhD Dissertation, State University of New
York at Buffalo; 1999. p. 18898 [Chapter 6].
[44] Popescu R, Prevost JH. Centrifuge validation of a numerical model for dynamic
soil liquefaction. Soil Dyn Earthquake Eng 1993;12:7390.
[45] Rahmani A. Dynamic analysis of pile foundations embedded in liqueable
soils, Masters Dissertation, Sharif University of Technology; 2010.
[46] Ishihara K., Terzaghi oration: geotechnical aspects of the 1995 Kobe
earthquake. In: Proceedings of ICSMFE, Hamburg, 1997, pp. 20472073.

You might also like