Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

www.elsevier.com/locate/soildyn

Piles in liquefiable soils: seismic analysis and design issues

W.D.L. Finna,*, N. Fujitab
a
Department of Safety Systems Construction Engineering, Kagawa University, 22117-20 Hayashi-cho, Takamatsu 761-0396, Japan
b
Anbuki Komuten and Kagawa University, Takamatsu 761-0396, Japan

Abstract
A general picture of the current state of the art and the emerging technology for dealing effectively with the design and analysis of pile
foundations in liquefiable soils is presented. Two distinct design cases are considered and illustrated by case histories. One is pile response to
strong shaking accompanied by the development of high pore water pressures or liquefaction and the other is the response to the pressures
and displacements caused by lateral spreading of liquefied ground. Analyses of centrifuge tests on pile foundations are used to demonstrate
current capacity for reliable analysis of piles under strong shaking. An example from practice involving 1.5 m cast in place reinforced
concrete piles supporting a 14 storey apartment building located on reclaimed land in liquefiable soils is presented to illustrate the advantages
of a comprehensive analysis in evaluating a proposed design.
q 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Liquefaction; Piles; Seismic response; Lateral spreading; Centrifuge tests

1. Introduction from centrifuge tests. The progress in analysis has also

allowed more fundamental and comprehensive evaluations
Loose cohesionless sands and silts below the water table of case histories, leading to a greater appreciation of design
develop high porewater pressures or liquefy during strong problems.
earthquake shaking. The high pore water pressures may lead The objective of the paper is to convey a general picture
to a significant degradation of strength and stiffness. The of the current state of the art and the emerging technology
seismic design of pile foundations in these liquefiable soils for dealing effectively with the design and analysis of pile
poses very difficult problems in analysis and design. foundations in liquefiable soils taking into account both the
Depending on the occurrence of liquefaction, the pile effects of earthquake shaking and of the lateral pressures
foundation may undergo substantial shaking, while the soil from post-liquefaction displacements.
is in a fully liquefied state and soil stiffness is at a minimum.
During this shaking phase, the pile is prone to suffering
severe cracking or even fracture. Liquefaction leads also to 2. Behavior of pile foundations during earthquakes
substantial increases in pile cap displacements above those
for the non-liquefied case. After liquefaction, if the residual During liquefaction, large ground displacements can take
strength of the soil is less than the static shear stresses place on sloping ground or towards an open face such as a
caused by a sloping site or a free surface such as a river river bank. Displacements from lateral spreading during the
bank, significant lateral spreading or downslope displace- Niigata earthquake are shown in Fig. 1 [1]. Displacements
ments may occur. The moving soil can exert damaging as large as 10 m occurred towards the Shinano River. Such
pressures against the piles, leading to failure. Such failures displacements were very damaging to pile foundations and
were prevalent during the 1964 Niigata and the 1995 Kobe caused the failure of two major bridges. Damage to a pile
earthquakes. Lateral spreading is particularly damaging if a under a building in Niigata caused by about 2 m of ground
non-liquefied layer rides on top of the liquefied soil. It is displacement is shown in Fig. 2. The complete shearing of a
only in the last few years that the profession has begun to pile supporting a warehouse on Port Island near Kobe City
deal effectively with these two critical design issues. The by about 1.5 m of ground displacement during the 1995
progress is due to developments in analysis and findings Kobe earthquake is shown in Fig. 3. The function of these
piles was to control settlement. They were designed
* Corresponding author. Tel.: þ 81-87-864-2170; fax: þ81-87-864-2188. primarily for vertical loads and could not carry the moments
E-mail address: [email protected] (W.D.L. Finn). and shears caused by strong shaking and lateral spreading.
0267-7261/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 7 - 7 2 6 1 ( 0 2 ) 0 0 0 9 4 - 5
732 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

Fig. 1. Ground displacements in 1964 Niigata earthquake (adapted from Ref. [1]).

However, piles can be designed to carry the moments and Therefore various approximate methods of analysis are
shears generated by earthquake shaking or post-liquefaction used.
ground displacements. Fig. 4 shows a bridge on pile The most common approach to the analysis of pile
foundations. The foundation soils liquefied during the foundations is to use Winkler springs and dashpots to
1983 Nihon-Kai-Chubu earthquake. This led to a failure simulate soil stiffness and damping. The springs may be
of the approach embankments by lateral spreading but the elastic or non-linear. Some organizations such as the
pile foundations survived without damage. A pile support- American Petroleum Institute [2] gives specific guidance
ing a crane rail on Port Island, just offshore of Kobe City, is for the development of non-linear load –deflection ( p– y )
shown in Fig. 5. The ground moved more than 1.0 m in this curves as a function of soil properties that can be used to
location after liquefaction occurred during the Kobe earth- represent non-linear springs. The API ( p – y ) curves,
quake. The relative motion between the ground and the pile which are the most widely used in engineering practice,
is clearly evident in Fig. 5. However, the pile was designed are based on data from static and slow cyclic loading tests
to carry significant shears and moments and survived in the field. The reliability of these ( p –y ) curves for the
without damage. analysis of pile foundations even under static and slow
cyclic loading has been shown to be relatively low [3].
Their performance under seismic loading is poorly
3. Seismic analysis of pile foundations established.
The review of case histories has clearly demonstrated the
design problems posed by pile foundations in liquefied soils.
To cope with these problems it is essential to have a reliable
method of calculating the effects of earthquake shaking and
post-liquefaction displacements on pile foundations. An
overview of the methods used in practice will be given
which will stress the advantages and limitations of the
various methods. The aim of the review is to present an
integrated up to date assessment of the state of the art in
general terms. The technical details will not be presented.
For these the reader is referred to the cited references.

4. Overview of fundamental concepts

The pile foundation –structure system vibrates during

earthquake shaking as a coupled system. Logically it should
be analyzed as a coupled system. However, this type of
analysis is not feasible in an engineering practice. Many of
the popular structural analysis programs cannot include the
pile foundation directly into a computational model. When Fig. 2. Damage to a pile by 2 m of ground displacement in Niigata
it can be done, the computational demands are excessive. earthquake, 1964.
 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742 733

Fig. 3. Shearing of a pile by ground displacement in Kobe earthquake, 1995. Fig. 5. Undamaged pile supporting a crane rail in ground which moved
more than 1.0 m.
A general Winkler model for seismic response
Since seismic response analysis is usually conducted
analysis is shown in Fig. 6 [4]. The near field interaction
assuming that the input motions are horizontally polarized
between pile and soil is modeled by springs and
shear waves propagating vertically, the PILE-3D model
dashpots. The near field pile –soil system, together with
retains only those parameters that have been shown to be
any structural mass included with the pile, are excited by
important in such analysis. These parameters are the shear
the seismic base motions and free field motions applied
stresses on vertical and horizontal planes and the normal
to the end of each Winkler spring. The free field motions
stresses in the direction of shaking. The soil is modeled by
at the desired elevations in the soil layer are computed
3-D finite elements as shown in Fig. 7. The pile is modeled
by 1-D dynamic analyses using a program such as SHAKE
using beam elements or block elements. The pile is assumed
[5] or DESRA-2C [6].
to remain elastic. This assumption is in keeping with the
An alternative to the Winkler type computational
design philosophy that the structural elements of the
model is to use a finite element continuum analysis
foundation should not yield. In the analysis of concrete
based on the actual soil properties. Dynamic non-linear
piles, the cracked section moduli are used, when defor-
finite element analysis in the time domain using the full
mations exceed the cracking limit.
3-D wave equations is not feasible for engineering
practice at present because of the time needed for the
computations. However, by relaxing some of the boundary
conditions associated with a full 3-D analysis, it is 5. Verification of methods of analysis
possible to get reliable solutions for non-linear response of
pile foundations with greatly reduced computational There is a very little quantitative data on the seismic
effort. The results are very accurate for excitation due to response of pile foundations in the field and what there is,
horizontally polarized shear waves propagating vertically. is not readily accessible. In recent years, seismic loading of
A full description of this method, including numerous model pile foundations in centrifuge tests has provided data
validation studies, has been presented by Wu and Finn that allows a more realistic evaluation of the reliability of
[7,8]. The method is incorporated in the computer various methods for the seismic analysis of pile foundations.
program PILE-3D . A typical example is given below.

Fig. 4. Bridge on undamaged pile foundations with failed approaches;

Nihon-kai-chubu earthquake, 1983. Fig. 6. A Winkler spring model for pile foundation analysis.
734 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

344 kN m compared with a measured peak value of

325 kN m.

5.2. Analysis using API p –y curves

A dynamic analysis of the foundation-superstructure

system was also conducted using the p – y curves
prescribed by the American Petroleum Institute [2] to
model the soil –pile interaction. These p –y curves are
defined by the equation
 
kH
P ¼ 0:9pu tanh y ð1Þ
0:9pu

where pu is the ultimate bearing capacity at depth H, k the

initial modulus of subgrade reaction, y the lateral
deflection, and H is the depth. The relative density of the
sand surrounding the pile is Dr ¼ 38%. This corresponds to
Fig. 7. Quasi-3D model for analysis of pile foundations. k of approximately 15,000 kN/m3 according to the API
recommendations. The distribution of moments for a value
5.1. PILE-3D analysis
of k ¼ 15,000 kN/m3 is shown in Fig. 9. The Winkler
model predicts a maximum moment of 550 kN m that is
PILE-3D was used to analyze the seismic response of a much greater than the measured moment of 325 kN m. The
single pile in a centrifuge test conducted at the API p– y curves seem to be too stiff for this level of
California Institute of Technology. Details of the test shaking. To obtain a reasonable approximation to the peak
may be found in Finn and Gohl [9]. Fig. 8 shows the moment in the pile requires k ¼ 2500 kN/m3, which is only
soil –pile-structure system used in the test. The system 1/6 of the value recommended by API. In another test in
was subjected to a nominal centrifuge acceleration of the same sand run at very low acceleration levels with a
60g. A horizontal acceleration record with a peak peak acceleration of 0.04g, the API stiffness
acceleration of 0.158g is input at the base of the system. k ¼ 15,000 kN/m3 gives a very good approximation to
The distribution of shear moduli was measured prior to the measured bending moments (Fig. 10). The response in
shaking, using bender elements, while the centrifuge was this case is almost elastic and the initial stiffness controls
in flight. the response. These results suggest that the initial stiffness
The computed and measured moment distributions along of the API p– y curves is reasonable, but the curves do not
the pile at the instant of peak pile head deflection are shown bend away fast enough from the initial tangent at the origin
in Fig. 9. The moments computed by PILE-3D agree quite during strong shaking. Therefore, the stiffness close to
well with the measured moments. The peak moment
predicted by the quasi-3D finite element analysis is

Fig. 8. Instrumented pile for centrifuge test. Fig. 9. Comparison of measured and computed bending moments.
 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742 735

fact this is a nominal load capacity. Many buildings have a

capacity of about 1.5 –2.0 times the nominal value. The
multiple of the nominal strength is called the overstrength
factor. The overstrength of the superstructure needs to be
taken into account when designing the pile foundation, if the
intent of design is that yielding should take place in the
columns, not in the foundation piles.

7. Dynamic analysis approach to foundation design

There are two alternatives to the pseudo-static approach

for evaluating the seismic response of structures: a modal
analysis using a response spectrum or a time history
analysis. Analyses for performance based design of new
structures or for the retrofit of existing structures demand a
more realistic computational model than the rigid base
model. The flexibility introduced by the foundations should
be included as noted above.
Fig. 10. Comparison of measured and computed bending moments for near
elastic response using API procedure.
The pile foundation restrains displacements and rotations
at the pile head. For a 2-D frame structure this means that
the initial value is being mobilized over a too large the pile foundation provides stiffness against lateral and
displacement range. These results suggest that empirical vertical displacements and rotation about an axis normal to
pressure – displacement relations describing soil pile inter- the frame. These stiffnesses are often represented by
action, which are based on static or slow cyclic loading test discrete springs as shown in Fig. 11(a). This is the simplest
data may prove to be rather unreliable when used in model of pile head stiffnesses. It assumes that all degrees of
dynamic response analysis. freedom are uncoupled. The spring stiffnesses are included
in the stiffness matrix of the structure and affect the periods
of the structure calculated from an eigenvalue analysis. In
6. Simplified pseudo-static analysis of building fact lateral displacement is coupled with rotation and this
foundations introduces two cross-coupling stiffnesses that can also be
represented by discrete springs. The complete spring (and
The simplest form of analysis is often used in the design dashpot) model for lateral translation and rotation is shown
of pile foundations for buildings. Although the dynamic in Fig. 11(b). This spring system corresponds to the pile
interaction of foundation and building affects the design head stiffness matrix in Fig. 11(c). This stiffness matrix can
loads, this effect is ignored. The dynamic loads on the be included in the stiffness matrix of the structure. If the
building are determined according to building code spring stiffnesses can be calculated, then the effects of the
regulations, without doing a dynamic analysis. The building foundation on the structural loads can be taken into account
is assumed to be on a rigid base. The computed shears and in a systematic manner.
moments, which correspond to yield occurring at the base of If the soil were elastic, the stiffnesses would be single-
the column, are then applied to the pile foundation and a valued. However, since soil shear modulus is a function of
static analysis is conducted to determine the resulting shears shear strain, it is dependent on the level of ground shaking
and moments in the pile. The pile head is often assumed which varies during the earthquake. Therefore under strong
fixed against rotation in this analysis. The analysis is usually shaking, the soil behaves non-linearly and the pile head
conducted assuming that the interaction between ground stiffnesses are time dependent. Sometimes the effects of
and pile can be modeled with Winkler springs as shown in non-linearity on stiffness is taken into account somewhat
Fig. 6. For the static analysis described here only the near arbitrarily by reducing the soil modulus by a factor of 2 or 3,
field portion of the general model in Fig. 6 is used. depending on the anticipated level of shaking.
Some designers use elastic springs to represent the Time dependent lateral pile head stiffnesses were calcu-
interaction between the pile and the soil. For conditions of lated using PILE-3D for the kind of pile–structure system
strong shaking, the non-linear response of the soil should be shown later in Fig. 22. The stiffnesses were computed with and
taken into account by replacing the elastic springs by non- without the inclusion of the inertial forces from the super-
linear ( p –y ) curves, such as those recommended by the structure. These conditions correspond to taking kinematic þ
American Petroleum Institute. inertial interaction and kinematic interaction only, respect-
The load applied to the foundation represents the yield ively, into account. The corresponding stiffnesses are given by
capacity of the column supported by the pile foundation. In the two curves shown in Fig. 12.
736 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

Fig. 12. Time dependent pile-head lateral stiffnesses with and without
inertial interaction.

interaction can be included in an approximate way using a

model proposed by Abghari and Chai [10]. In this model, a
mass is mounted on the pile that is equivalent to the static
reaction load carried by the pile. The mass is mounted on a
massless cantilever in a single degree of freedom system
(SDOF). The cantilever stiffness is selected so that the
period of the SDOF is the same as the rigid base
fundamental period of the structure. The model is shown
in Fig. 22. The lateral stiffness curves in Fig. 12 that
includes inertial mass were calculated using this model
which represents an actual building.
Clearly it is not an easy task to select the appropriate
single-valued springs required by current commercial
software.

8. Validation of pile response analysis for liquefiable sites

by centrifuge test data

Seismic pore water pressures and liquefaction can

severely reduce the modulus and strength of a soil and so
reduce the pile head stiffnesses. This effect is taken into
account when using the semi-empirical computational
models by reducing the coefficient of subgrade reaction,
if an elastic Winkler model is used, and the initial stiffness
Fig. 11. Various representations of pile head stiffnesses: (a) uncoupled and ultimate strength if ( p – y ) curves are used. Finn and
springs, (b) coupled springs and (c) pile-head stiffness matrix. Thavaraj [11] developed an effective stress version of the
program PILE-3D , called PILE-3DEFF that tracks the devel-
When the inertial forces from the structure act on the opment of seismic pore water pressures and continuously
piles, they cause additional ground strains over and above modifies the soil properties to be compatible with the
those caused by free field ground shaking. The increased current effective stress state. Increments in seismic pore
strains lead to a greater reduction in soil modulus and so a water pressures are generated in each individual element
greater reduction in lateral stiffness. It is evident from depending on the accumulated volumetric strain prevailing
Fig. 12 that the inertial forces from the super-structure in that element and the current increment in shear strain,
should be included in the computation of lateral stiffness using a modified version of the Martin– Finn – Seed pore
during strong shaking. water pressure generation model [12] that incorporates the
In theory the inclusion of inertial interaction would two-constant volume change expression proposed by Byrne
require a coupled analysis of the structure and all supporting [13]. The moduli and shear strengths of the foundation
pile foundations. This is not feasible. However, the inertial soils were modified continuously to account for the effects
 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742 737

of the changing seismic pore water pressures. PILE-3DEFF

has been validated in a cooperative study with the
University of California at Davis using data from
centrifuge tests [14,15].
Dynamic centrifuge tests of pile-supported structures in
liquefiable sand were performed on the large centrifuge at
University of California at Davis, California. The models
consisted of structures supported by single piles, one
structure supported by a 2 £ 2 pile group and one structure
supported by a 3 £ 3 pile group. The typical arrangement of
structures and instrumentation is shown in Fig. 13. Full
details of the centrifuge tests can be found in Wilson et al.
[16]. Data from a single pile test only will be presented here.
The model dimensions and the arrangement of bending
strain gauges are shown in Fig. 14. Model tests were
performed at a centrifugal acceleration of 30g.
The soil profile consists of two level layers of Nevada
sand, each approximately 10 m thick at prototype scale.
Nevada sand is a uniformly graded fine sand with a
coefficient of uniformity of 1.5 and mean grain size of
0.15 mm. Sand was air pluviated to relative densities of
75 –80% in the lower layer and 55% in the upper layer.
Prior to saturation, any entrapped air was carefully
removed. The container was then filled with a hydroxy-
Fig. 14. Instrumented pile for single pile test.
propyl methyl-cellulose and water mixture under vacuum.
The viscosity of this pore fluid is about 10 times greater
than pure water to ensure limited drainage during the
period of shaking. Saturation was confirmed by measuring beam elements are within the soil strata and 11 elements
the compressive wave velocity from the top to the bottom are used to model the free-standing length of the pile
of the soil profile. above the soil. The super-structure mass is treated as a
rigid body and is represented by a concentrated mass at
the center of gravity. A rigid beam element is used to
9. Effective stress dynamic analysis of connect the superstructure to the pile head.
pile-superstructure system

The finite element mesh used in the analysis is shown

in Fig. 15. The finite element model consists of 1649
nodes and 1200 soil elements. The upper sand layer is
9.1 m thick and is divided into 11 layers; the lower sand
layer is 11.4 m thick and is divided into nine layers. The
single pile is modeled with 28 beam elements. Seventeen

Fig. 13. Layout of models for centrifuge tests. Fig. 15. Finite element mesh for single pile.
738 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

Fig. 16. Input acceleration time history.

9.1. Soil and pile properties

The small strain shear moduli Gmax, were estimated using

the well-known formula proposed by Seed and Idriss [17]

Gmax ¼ 21:7kmax Pa ðs0m =Pa Þ0:5 ð2Þ

in which kmax is a constant which depends on the relative

0
density of the soil, sm the initial mean effective stress and Pa
is the atmospheric pressure. The program PILE-3DEFF
accounts for the changes in shear moduli and damping
ratios due to dynamic shear strains at the end of each time Fig. 18. Comparison of measured and computed pore water pressure time
half cycle. The shear strain dependencies of the shear histories at three depths.
modulus and damping ratio of the soil were defined by the
curves suggested by Seed and Idriss [17] for sand. The
friction angles of the upper and the lower layers were taken
as 35 and 408, respectively. Fig. 19 shows the measured and computed bending
moment time histories at two different depths; 0.76 and
9.2. Seismic response analysis 1.52 m. Generally there is a very good agreement between
the measured and computed time histories. Fig. 20 shows
The Santa Cruz acceleration record was scaled to the profiles of measured and computed maximum bending
0.49 g and used as input to the shake table. The base moments with depth. The comparison between the measured
input acceleration is shown in Fig. 16. Results of and computed maximum moments is also fairly good.
analysis are given at prototype scale. The measured
and computed acceleration responses of the super-
structure are shown in Fig. 17. There is generally a
good agreement between them, especially in the time
period of peak response. Fig. 18 shows the comparisons
between the measured and computed pore water
pressures at three different depths; 1.14, 4.56, and
6.78 m in the free field. There is generally a good
agreement between the measured and computed pore
water pressures.

Fig. 17. Comparison of measured and computed superstructure acceleration Fig. 19. Comparison of measured and computed bending moment time
time histories. histories at two depths.
 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742 739

Fig. 20. Comparison of measured and computed maximum bending

moment profiles along the pile.

10. Example from practice

The columns of a 14 storey apartment building located on Fig. 21. Site in reclaimed land.
reclaimed land are carried on cast in place reinforced
Fig. 24. When the ground does not liquefy, there is a
concrete piles, 1.5 in diameter. Results from the preliminary
fairly smooth distribution of deflections along the pile.
studies of the pile structure system will be presented here to
illustrate some of the concepts discussed earlier and to show When the surface layers liquefy, the pile deflections in the
the kind of results that can be obtained with the state of the unliquefied material remain small. The significant deflec-
analyses. The soil conditions and pile are shown in Fig. 21. tions occur in the liquefied materials. This deflection
Idealized site conditions for a demonstration analysis are pattern creates sharp curvature in the pile at the interface
shown in Fig. 22. The upper 10 m of reclaimed soil are
expected to liquefy during the design earthquake. The mass
mounted on the pile in Fig. 22 represents the mass
equivalent of the force carried by the pile. The purpose in
placing the mass on the pile is to model approximately the
inertial interaction between the super-structure and the pile
foundation. It is mounted on the pile head by a flexible
support that gives the mass a period of vibration of 1.4 s that
is the estimated fundamental period of the prototype
structure.
Two kinds of analyses were conducted; total stress
dynamic analysis in which seismic pore water pressures
and liquefaction are ignored and effective stress analysis
that automatically takes the seismic pore water pressures
into account. In general, soil properties are adjusted
continuously for current pore water pressures and shear
strains. The peak acceleration of the input acceleration
record is 0.25g and is amplified to 0.4g at the surface.
The surface accelerations become negligible after lique-
faction has occurred.
The maximum pile deflections for both the liquefied
and non-liquefied cases are shown in Fig. 23 for the case
in which the pile head is fixed against rotation. The
distributions of maximum bending moments are shown in Fig. 22. Model of soil–pile-structure system.
740 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

Fig. 23. Maximum deflections along the pile; pile head fixed against
Fig. 25. Maximum deflections along the pile; pile head not fixed against
rotation.
rotation.

between the liquefied and unliquefied zones and hence of the piles is reduced to a very low value, so that lateral
large bending moments at the interface as shown in deflections become much larger and the distribution of
Fig. 24. These moments need to be considered in design. deflections changes sharply.
The piles may need a local increase in reinforcement at Rotational restraint has a big effect on the pile head
this level to cope with the moments. Since the analysis moments and deflections, especially in the case where
was conducted with the assumption that the pile head is the surface layers liquefy. In the non-liquefied case, even if
fixed against rotation, in both the liquefied and non- the pile head is not fixed, the ground still provides
liquefied cases, the largest moments still occur at the pile significant restraint and there is only a modest increase in
head (Fig. 24). deflection at the pile head and a small reduction in bending
The maximum pile deflections for both the liquefied and moment. However, in the liquefied case the deflections
non-liquefied cases are shown in Fig. 25 when the pile head doubled at the pile head and the pile head moment dropped
is not fixed against rotation. The distributions of maximum by a factor of 3. The critical design section with the
bending moments are shown in Fig. 26. When the surface maximum moment is now the interface between the
layers liquefy, the soil restraint on the lateral displacement liquefied and non-liquefied regions. This section must now

Fig. 26. Maximum moments along the pile; pile head not fixed against
Fig. 24. Maximum moments along the pile; pile head fixed against rotation. rotation.
 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742 741

be reinforced to almost the level of the pile head section in

the case when no liquefaction occurs. As the rotational
restraint at the pile head has such a major effect on the
amount of deflection and location of maximum moment in
the case of liquefied soils, realistic values of rotational
stiffness at the pile head should be included in any seismic
response analysis. The rotational restraint is a function of
the rotational stiffness of both the pile head, the column
connected to the pile cap, and any grade beams connecting
one column footing with another.

11. Analysis of effects of lateral spreading on pile

foundations

11.1. Displacement analysis

In the case histories section, it was shown that large post-

liquefaction displacements can occur and that these can be
very damaging to pile foundations. These potential Fig. 28. Pile displacements due to post-liquefaction ground displacements.
deformations often control design in weak highly liquefiable
soils. It is very difficult to predict these displacements
reliably. In engineering practice, the displacements at the applied to the near field Winkler model of Fig. 6 and the
top of the liquefied layer are estimated by empirical displacements, bending moments and shears are calculated
formulae based on field data from past earthquakes. The using a static analysis. The deformed shape of a pile
first predictor equation was developed in Japan by Hamada foundation caused by post-liquefaction displacements from
et al. [1]. A version of this has been adopted by the Japan lateral spreading is shown schematically in Fig. 27.
Water Works Association [18]. Youd [19] has developed the Displacements and bending moments in the 1.5 m cast in
most comprehensive predictor equations that include the place pile in Fig. 22 due to lateral spreading displacements of
effects of earthquake magnitude. 15 and 25 cm at the ground surface are shown in Fig. 28 and
The displacements are assumed to vary linearly from top Fig. 29, respectively. The pile head is fixed against rotation.
to bottom of the liquefied layer. These displacements are A force based analysis is recommended by JWWA [18]
for analysis of piles in liquefied soils. An unliquefied surface
layer is assumed to apply passive pressure to a pile.
Liquefied layers are assumed to apply a lateral pressure
corresponding to a lateral pressure coefficient of 0.3. This
force based approach has been verified by Dobry [20] using
centrifuge tests. Abdoun [21] has demonstrated very clearly
the interaction of piles and soil during lateral spreading by
animating data from centrifuge tests involving lateral
spreading against piles.

12. Closing remarks

This paper gives a general overview of the important

factors that affect the seismic design of piles to resist
earthquake loading in liquefiable soils during earthquakes.
Case histories show that damage occurs primarily in
liquefiable soils and is concentrated in critical areas such
as at the pile head when it is fixed against rotation and the
boundary between liquefied and non-liquefied layers. When
the soil liquefies, there are large losses in stiffness and
strength that can result in substantial moments from
increased displacements. The case histories also show that
Fig. 27. Distortion of pile foundation by moving soil. when piles are properly designed they can successfully
742 W.D.L. Finn, N. Fujita / Soil Dynamics and Earthquake Engineering 22 (2002) 731–742

[5] Schnabel PB, Lysmer J, Seed HB. SHAKE : a computer program for
earthquake response analysis of horizontally layered sites. Report
EERC 71-12, University of California at Berkeley; 1972.
[6] Finn WDL, Lee MKW, Yoshida N. DESRA-2C, dynamic effective
stress response analysis of soil deposits with energy transmitting
boundary including assessment of liquefaction potential. Vancouver,
BC, Canada: Department of Civil Engineering, University of British
Columbia; 1997. p. 50.
[7] Wu G, Finn WDL. Dynamic elastic analysis of pile foundations using
the finite element method in the frequency domain. Can Geotech J
1997;34:34–43.
[8] Wu G, Finn WDL. Dynamic nonlinear analysis of pile foundations
using the finite element method in the time domain. Can Geotech J
1997;34:144–52.
[9] Finn WDL, Gohl WB. Centrifuge model studies of piles under
simulated earthquake loading from dynamic response of pile
foundations—experiment, analysis and observation. ASCE Conven-
tion, Atlantic City, New Jersey, Geotech Spec Publ 1987;11:21– 38.
[10] Abghari A, Chai J. Modeling of soil–pile superstructure interaction
for bridge foundations. Performance of deep foundations under
seismic loading. ASCE Geotech Spec Publ 1995;51:45– 59.
[11] Finn WDL, Thavaraj T. PILE-3D-EFF . A program for nonlinear
dynamic effective stress analysis of pile foundations. Anabuki Chair
Fig. 29. Displacements induced bending moments along the pile in of Foundation Geodynamics, Kagawa University, Japan; 1999.
liquefied ground after the earthquake. [12] Martin GR, Finn WDL, Seed HB. Fundamentals of liquefaction under
cyclic loading. J Geotech Engng Div. ASCE 1975;101(GT5):423–38.
[13] Byrne PM. A cyclic shear-volume coupling and pore pressure model
resist strong shaking and large displacements. The keys to for sand, vol. 1. Proceedings of the Second International Conference
good design are reliable estimates of environmental loads, on Recent Advances in Geotechnical Earthquake Engineering and
realistic assessments of pile head fixity and the methods of Soil Dynamics, St Louis; 1991, Report 1.24. p. 47–56.
[14] Finn WDL, Thavaraj T, Wilson DW, Boulanger RW, Kutter
analysis that can adequately assess the response of the pile-
B. Seismic analysis of piles and pile groups in liquefiable sand.
soil-structure system to strong shaking that may lead to Proceedings of the Seventh International Symposium on Numerical
liquefaction in complex layered systems. Models in Geomechanics, NUMOG VI, Graz, Austria; 1999. p. 287–
92.
[15] Finn WDL, Thavaraj T. Deep foundations in liquefiable soils: case
histories, centrifuge tests and methods of analysis. CD-ROM
Acknowledgments
Proceedings of the Fourth International Conference on Recent
Advances in Geotechnical Earthquake Engineering and Soil
Anabuki Komuten, Takamatsu, Japan funds the research Dynamics, San Diego, CA; March 26– 31 2001.
work on the seismic design and analysis of pile foundations. [16] Wilson DW, Boulanger RW, Kutter BL. Soil –pile-superstructure
The support of the company is gratefully acknowledged. interaction at soft or liquefiable sites. Centrifuge data report for CSP1-
5, Report No. UCD/CGMDR- 97/01-05, Center for Geotechnical
Modeling, CA: Department of Civil and Environmental Engineering,
University of California at Davis; 1997.
References [17] Seed HB, Idriss IM. Soil moduli and damping factors for dynamic
response analysis. Report #EERC 70-10, Earthquake Engineering
[1] Hamada M, Yasuda S, Isoyama R, Emoto K. Study on liquefaction Research Center, Berkeley, CA; 1970.
induced permanent ground displacements. Report for the development [18] JWWA, Seismic design and construction guidelines for water supply
of earthquake prediction, Tokyo Japan; 1986. facilities. Tokyo, Japan: Japan Water Works Association; 1997. 150 pp.
[2] API, Recommended practice for planning, designing, and construct- [19] Youd TL. Liquefaction induced lateral spread displacements.
ing fixed offshore platform. Washington, DC: American Petroleum ReportTN-1862, Naval Civil Engineering Laboratory, Port Hueneme,
Institute; 1995. California; 1993. 44 pp.
[3] Murchison JM, O’Neill MW. In: Meyer JR, editor. An evaluation of [20] Dobry R, Abdoun T. Post-triggering response of liquefied soil in the
p–y relationships in cohesionless soils. Proceedings of the ASCE free field and near foundations, vol. 1. State-of-the-Art Report,
Symposium on Analysis and Design of Pile Foundations, ASCE Proceedings of the ASCE 1998 Specialty Conference on Geotechnical
National Convention, San Francisco, California; 1996. p. 174– 91. Earthquake Engineering and Soil Dynamics, Seattle, WA; 3–6
October 1–5. August 1998. p. 270–300.
[4] Thavaraj T, Finn WDL. A program for dynamic analysis of a single [21] Abdoun T. Visualization of the seismic response of soil systems
pile using a Winkler model and p–y curves. Anabuki Chair of accepted. Proceedings of the International Conference on Physical
Foundation Geodynamics, Kagawa University, Japan; 2000. Modeling in Geotechnics, St Johns, Newfoundland, Canada; July 2002.