Interpretation from large-scale shake table tests on piles
undergoing lateral spreading in liquefied soils
M. Cubrinovskia, *, T. Kokushob, K. Ishiharab
a
b
Kiso-Jiban Consultants Co. Ltd., Tokyo 102-8220, Japan
Department of Civil Engineering, Chuo University, Tokyo 112-8551, Japan
Abstract
Results from a benchmark test on full-scale piles are used to investigate the response of piles
to lateral spreading. In the experiment, two single piles, a relatively flexible pile that moves
together with the surrounding soil and a relatively stiff pile that does not follow the ground
movement have been subjected to large post-liquefaction ground displacement simulating
piles in laterally spreading soils. The observed response of the piles is first presented and then
the results are used to examine the lateral loads on the pile from a non-liquefied soil at the
ground surface and to evaluate the stiffness characteristics of the spreading soils. The
measured ultimate lateral pressure from the crust soil on the stiff pile was about 4.5 times the
Rankine passive pressure. The back-calculated stiffness of the liquefied soil was found to be
in the range between 1/30 and 1/80 of the initial stiffness of the soil showing gradual decrease
in the course of lateral spreading.
Keywords: Pile response; Liquefaction; Lateral spreading; Large-scale shake table test;
Reduction of soil stiffness; Lateral soil pressure
* Corresponding author. Fax: +81-3-5210-9405.
E-mail address: misko.cubrinovski@kiso.co.jp
(M. Cubrinovski)
1
1. Introduction
The most frequently encountered soil profile for piles in liquefied deposits consists of
three distinct layers, as illustrated in Fig. 1 where the liquefied layer is sandwiched between a
non-liquefied crust layer at the ground surface and non-liquefied base layer. Liquefaction
during strong ground shaking results in almost a complete loss of stiffness and strength of the
liquefied soil, and consequent large ground deformation. Particularly large and damaging for
piles can be post-liquefaction ground displacements due to lateral spreading [7]. During the
spreading, the non-liquefied surface layer is carried along with the underlying spreading soil,
and when driven against embedded piles, the crust layer is envisioned to exert large lateral
loads on the piles. Thus, the excessive lateral movement of the liquefied soil, lateral loads
from the surface layer and significant stiffness reduction in the liquefied layer are key features
that need to be considered when evaluating the pile response to lateral spreading.
In the light of the liquefaction characteristics and kinematic mechanism as above, a
three-layer soil model was adopted in a previous study [4] for a simplified analysis of piles. In
the adopted pseudo-static approach, the spreading is represented by a horizontal displacement
of the liquefied soil whereas effects of the non-liquefied surface layer are modeled by an earth
pressure and lateral force at the pile head, as shown in Fig. 1. Here, the earth pressure
represents the loads that act directly on the pile while the lateral force approximates the loads
that are transferred to the pile through the upper foundation. Using the secant-stiffness
approach, the interaction between the liquefied soil and the pile is assumed to be specified by
an equivalent linear spring (k) where k is the subgrade reaction coefficient representing the
initial stiffness of the soil while the reduction in stiffness due to liquefaction is taken into
account by way of the degradation factor .
Key parameters influencing the pile response are the magnitude of lateral ground
2
displacement (UG), ultimate pressure from the surface layer (pu) and stiffness reduction in the
liquefied layer (), as indicated in Fig. 1. These parameters are associated with intrinsic
uncertainties, however, and therefore one encounters difficulties in selecting their most
appropriate values. For this reason, great efforts have been made over the past decade either to
back-calculate these parameters from well-documented case histories of recent earthquakes or
to evaluate them using sophisticated experiments on scaled-down soil-pile models. In this
paper, results from a benchmark experiment on full-size piles are used to investigate the
lateral loads from a non-liquefied surface layer on the pile and stiffness characteristics of
liquefied soils undergoing spreading. The ultimate lateral pressure from the surface layer on a
single pile (pu) and characteristics of the stiffness reduction parameter () are examined in
detail.
2. Large-scale shake table experiment
The test was conducted by the Japan Electric Power Civil Engineering Association
(JEPOC) using the large-scale shake table of the National Research Institute for Earth Science
and Disaster Prevention (NIED) at Tsukuba, Japan. The test was specifically designed for
investigating the response of piles to large post-liquefaction ground displacements, as
described in the following.
2.1 Physical model
A prototype model of piles was prepared in a laminar box with dimensions of 12 m 
3.5 m  6 m (length-width-height), bottom-fixed at a large shake table. The model consisted
of two single piles embedded in a deposit of saturated sand with a crust layer of sand above
3
the water table, as illustrated in Fig. 2a.
A steel pile with an outer diameter of 31.8 cm and a pre-stressed high-strength concrete
pile (PHC-pile) 30 cm in outer-diameter were used in the test. The 4.9 m long piles were fixed
at the base and free at the top. The piles were installed at a distance of about 15 pile-diameters
and were considered therefore free of cross-interaction effects. Moment-curvature
relationships of the test piles are shown in Fig. 3 where point C denotes concrete cracking,
point Y indicates yielding of steel or reinforcement respectively and concrete crushing or the
ultimate level for the PHC pile is denoted by point U.
The sand deposit consisted of two horizontal layers of Kasumigaura sand (D50 = 0.265
mm, UC = 2.36 and FC = 3 %), both at a relative density of about 50 %. The lower saturated
sand layer was prepared by pouring sand into the laminar box through a water layer of about
50 cm to the prescribed height of 3.8 m from the base of the piles. The crust layer at the
ground surface was prepared by placing dry sand above the water table.
A large number of accelerometers, pore pressure transducers, displacement and pressure
gauges were installed to measure the response of the piles and ground. Pairs of strain gauges
were installed at a regular distance of 20 cm along the pile body for measuring bending strains
of the piles. In total, 227 channels were used for data acquisition in the experiment. Details of
the experimental setup and instrumentation are given in [16].
2.2 Dynamic excitation and lateral loading
The experiment was conducted in two phases, as illustrated in Figs. 2a and 2b. In the
first phase, the model was shaken with a sine wave excitation in the longitudinal direction
(Fig. 2a). The applied base-input motion had peak acceleration of 0.217 g, frequency of 2 Hz
and duration of the intensive part of 30 sec (60 cycles). The key objective in this phase of the
4
test was to induce liquefaction in the saturated sand deposit while keeping the response of the
piles in the range of elastic deformations.
Once liquefaction was induced and the dynamic phase of the test was concluded, a rigid
loading frame was attached to the outer-side of the laminar box and the liquefied portion of
the deposit was subjected forcibly to a lateral movement with a rate of 4.1 cm/sec at the top of
the layer, as illustrated in Fig. 2b. This second phase of the test aimed at subjecting the piles to
large post-liquefaction ground displacements simulating piles in laterally spreading soils. It is
apparent from the time scale of the applied accelerations and displacements in the subsequent
loading phases shown at the bottom of Figs. 2a and 2b that the lateral ground movement was
initiated about 6 seconds after the end of the shaking phase, which was approximately the
time required to attach the loading frame to the laminar box. The application of lateral ground
movement proceeded for approximately 22 seconds until eventually a permanent ground
displacement of about 84 cm was reached at the top of the liquefied layer. In the course of the
lateral loading, the model was subjected to a low-amplitude (amax = 0.027 g) high-frequency (f
= 10 Hz) base shaking, as indicated in Fig. 2b.
2.3 Excess pore pressures
Recorded excess pore pressures during the dynamic phase and lateral loading phase of
the test respectively are shown in Figs. 4a and 4b at five depths in the soil profile. The
saturated sand below approximately 1.4 m depth is shown to have completely liquefied after
few cycles of shaking, as evidenced by the pore pressures recorded at 2.15 m and 3.9 m depth
depicted in Fig. 4a. Partial build-up of the pore pressure is seen at depths slightly above or
below the water table of 0.8 m and 1.15 m respectively whereas no excess pore pressures have
been observed during the shaking in the top 50-60 cm of the crust layer. It is important to
mention that the response of both piles remained purely elastic during the shaking. Other
5
features of soil and pile responses induced during the dynamic phase of the test are beyond
the scope of this paper.
The lateral loading phase of the test was initiated about 6 seconds after the end of the
shaking allowing practically no time for dissipation of the pore pressures induced by the
shaking. As shown in Fig. 4b, the excess pore pressure throughout the depth of the soil profile
did not change significantly during the lateral loading though some oscillation in the pore
pressure at depths of 0.8 m and 1.15 m is seen and effects of lateral ground movement on the
pore pressure response are apparent immediately after the start and after the end of the lateral
loading. By and large, however, the pore pressure in the liquefied sand remained fairly stable
and nearly at its maximum level all the way through the application of lateral ground
movement.
2.4 Pile response induced by lateral ground movement
Measured maximum responses of the two piles in the course of increasing lateral
ground displacement are shown in Figs. 5a and 5b where horizontal displacements at the pile
head and bending moments near the base of the pile are shown respectively. Note that the
bending moment in Fig. 5b is expressed normalized to the yield moment of the pile. The
dashed lines in these figures indicate the ground displacement at the surface of the deposit
which for all practical purposes can be considered to be identical to the applied horizontal
displacement at the top of the liquefied layer.
It may be seen in Fig. 5a that the PHC pile basically followed the ground movement
showing steady increase in the deflection and bending deformation in the course of lateral
spreading. As a result, the bending moment near the base of the pile sharply increased from
the start of the lateral loading and reached the yield level at a fairly small ground
6
displacement of about 9 cm, as indicated in Fig. 5b. The pile failed soon upon further lateral
loading as the horizontal displacement at the ground surface reached approximately 17 cm.
The post-failure response of the PHC pile was affected by the plastic hinge that developed
near the base of the pile and therefore it was not considered in this study.
In marked contrast to the response of the PHC pile, the steel pile exhibited large lateral
resistance and did not follow the ground movement. As shown in Fig. 5a, the displacement of
the steel pile gradually increased in the initial 6 seconds of lateral loading to the maximum
value of about 5 cm and then it remained nearly constant during the subsequent lateral loading
in spite of the increasing lateral displacement of the surrounding soils. The same response
pattern can be seen in the development of the bending moment of the steel pile in Fig. 5b. The
maximum bending moment of the steel pile induced by the lateral ground movement was
about 60 % of its yield moment.
It is to be noted that combined effects from the movement of the liquefied soil and
lateral pressure from the surface layer contributed to the observed responses of the piles as
above. These effects were evaluated separately, as described in the following sections, to gain
insight into the soil-pile interaction during the lateral spreading and to quantify the key
parameters influencing the pile response.
3. Lateral pressure from the surface layer
3.1 Experimental p- relations
The experimental bending-strain data were used to calculate the lateral load on the piles
through double differentiation and the pile deflection through double integration assuming
zero displacement at the pile tip and matching the measured deflection at the pile head [9, 18].
7
Two separate polynomials spliced together with continuity conditions were used for the
surface layer and liquefied layer respectively. A third-order polynomial was used for the crust
layer resulting in a linearly increasing reaction force with depth. Based on the observed pore
pressures and back-calculated lateral loads as above, the interface between the non-liquefied
crust layer and the completely liquefied sand has been identified to be located about 40 cm
below the water table, that is, at 1.4 m depth [18]. Thus, in what follows, the resultant pressure
from the crust layer actually represents the pressure from the top 1.4 m of the sand deposit.
The force per unit length of the pile obtained through double differentiation as above
was divided by the diameter of the pile to evaluate the net pressure (p). By combining the
lateral pressure and the displacement curves at discrete intervals, p- curves were developed
for different depths in the crust layer where p denotes the lateral soil pressure and  is the
relative displacement between the soil and the pile. Fig. 6 shows such p- curves developed
from the data of the steel pile at three depths in the crust layer. Here, the lateral pressure p is
expressed normalized to the effective overburden stress in order to observe the characteristic
shape and variations of p- curves with depth. It is to be mentioned that the soil pressure on
the pile was also directly measured in the test by means of pressure gauges, at three depths in
the crust layer. By and large, the directly measured pressure was found to be in agreement
with the net pressure back-calculated from the pile response. There were also clear differences
between these two pressures which are to be expected in view of the difficulties to precisely
measure the soil pressure in the test and because of differences between the localized pressure
measurement in the test and the average pressure distribution assumed in the calculation.
It is evident in Fig. 6 that the lateral pressure from the surface layer initially sharply
increased with the relative displacement until eventually the ultimate lateral pressure was
mobilized at a relative displacement of about 12 to 15 cm. Close inspection of Figs. 5 and 6
reveals that the development of lateral displacement and bending moment of the steel pile
8
including its maximum response is in very good accord with the development of the lateral
pressure from the crust layer on the pile. Thus, the bending moments of the steel pile steadily
increased with the increasing lateral pressure from the surface layer and eventually reached its
maximum value at the time when the ultimate pressure from the crust soil has been mobilized.
Once the ultimate lateral pressure has been mobilized, however, the pile response remained
nearly unchanged despite the significant increase in the ground displacement in the course of
the subsequent lateral loading. These response features indicate that the lateral load from the
surface layer was the key factor influencing the response of the steel pile.
3.2 Ultimate lateral pressure
It is well-known that the lateral pressure per unit width of a single pile is greater than
that of a continuous wall due to the shearing resistance on the vertical sides of the failure
wedge in the soil. In a number of experimental and theoretical studies, the three-dimensional
effects for a single pile have been approximately taken into account by multiplying the earth
pressure on a wall by a shape factor , i.e., P =  Pp where Pp is the Rankine passive pressure.
Following this reasoning, the ultimate lateral pressure from the crust soil on the test piles was
evaluated and compared to results of other experimental studies, as described below.
By integrating the lateral pressure throughout the depth of the surface layer, the
resultant pressure from the crust soil on the pile was obtained. The evaluated resultant
pressure P per unit width of the steel pile is shown with the square marks in Fig. 7 in terms of
a normalized value,  = P / Pp. Here, Pp is the resultant Rankine passive pressure defined as
Pp = 0.5 Kp  H2 in which H = 1.4 m,  = 17 kN/m3 and Kp = 3.69 were assumed with an angle
of internal friction of  = 35. It is seen in Fig. 7 that the measured ultimate lateral pressure
from the surface layer on the steel pile was about 4.5 times the Rankine passive pressure.
9
Before comparing the ultimate lateral pressure measured in this test with results of other
experimental studies, it is important to draw a distinction between two types of lateral loading
of piles, namely, the active-pile-loading and passive-pile-loading. Majority of the
experimental studies to date have involved active-pile-loading or active piles [5] in which, as
illustrated in Fig. 8a, a horizontal force that is applied at the pile head is the causative load for
the pile deformation while the resulting movement of the surrounding soil is its consequence.
In the case of passive-pile-loading, on the other hand, the pile is subjected to lateral loading
along its shaft by the lateral movement of the surrounding soils, as depicted in Fig. 8b.
Apparently, the passive-pile-loading is representative of piles subjected to lateral spreading. In
fact, the displacement field of the soil near the ground surface shown in Fig. 8b schematically
illustrates the distortion of soils surrounding the steel pile as observed in the test. Since the
steel pile moved merely 5 cm in the direction of the applied lateral ground movement, the soil
behind the pile could not follow the large horizontal displacements that were induced in the
far field soil. Instead, the crust soil behind the pile gradually moved upwards and the soil near
the ground surface eventually moved over the top of the pile.
Previous analytical studies by Bransby [2] have suggested that the pile-load-transfer
curves due to active loading (p-y curves) are different from the p- curves in passive loading.
Another important difference to be noted is that the earth pressure mobilized in active loading
provides the resisting force while the pressure mobilized in passive loading provides the
driving force for the pile deformation. This feature needs to be considered when estimating
the value of the ultimate lateral pressure in the analysis of piles subjected to lateral spreading.
Fig. 9 shows the summary of the data obtained by several investigators regarding the
shape factor for the ultimate pressure u = Pu / Pp, for both active and passive piles. The data
shown in the figure are from experimental studies on model piles in sand deposits and the
large-scale test presented herein is the only test involving full-size piles in liquefied soils. It
10
may be seen in Fig. 9 that the value of approximately u = 4.5 obtained for the steel pile in the
test presented herein is nearly coincident with that reported by Poulos [13] for passive piles. It
is important to mention that the test data used by Broms [3] yielded mostly values of u = 3-6
and that Broms adopted the lower-bound value of u = 3 for active piles.
3.3 Relative displacement required to mobilize the ultimate pressure
As discussed in the preceding sections and shown in Fig. 7, the ultimate lateral pressure
from the surface layer on the pile gradually increased with the increase in the relative
displacement between the soil and the pile. In the case of the steel pile, the ultimate lateral
pressure was mobilized at a relative displacement of about 12 to 15 cm. On the other hand, as
indicated by the circular symbols in Fig. 7 showing the resultant pressure from the surface
layer on the PHC pile prior its failure, the ultimate lateral pressure has not been mobilized in
the case of the PHC pile because of the small relative displacements. Clearly, the relative
displacement required to mobilize the full passive pressure would be an important parameter
to consider in the assessment of pile response to lateral spreading.
Rollins [15] compiled data from several investigators that have conducted laboratory
and field tests on sands to evaluate the relative displacement required to develop the passive
pressure. These data are summarized in Fig. 10 in terms of a normalized value u / H where u
is the relative displacement at which the ultimate pressure is mobilized, as depicted in the
inset of the figure, while H denotes the height of the model wall or pile cap used in the test. It
is evident in Fig. 10 that for dense sands with Dr = 70 % to 80 %, the ultimate pressure was
mobilized at a relative displacement of about u = 0.02H to 0.08H and that larger movement
was needed to mobilize the passive pressure in loose sand. Rollins also suggested that the
presence of a low strength layer below the surface layer may increase the required deflection
11
to mobilize the passive pressure which appears to be an important feature for piles in liquefied
deposits. Superimposed in Fig. 10 is the result for the steel pile obtained in the present study
where a ratio of u / H = 0.15 / 1.4 = 0.107 was obtained. Here, a value of 15 cm was adopted
for u while the thickness of the crust layer of 1.4 m has been used for H.
4. Stiffness of liquefied soils undergoing spreading
With reference to the three key parameters introduced in the analytical model shown in
Fig. 1, the only parameter remaining unknown from the large-scale test is the stiffness
degradation of the liquefied soil, as defined by the degradation factor . Using the analytical
model, the value of  can be readily back-calculated from the test results, as described below.
Fig. 11 shows the analytical model used for back-calculating the value of . In the
employed method of simplified analysis [4], the spreading is represented by a horizontal
displacement of the liquefied soil while effects of the non-liquefied surface layer are modeled
by an earth pressure at the pile head. The nonlinear behavior of the pile and soil is
approximately modeled using the equivalent linear approach within an iterative calculation
procedure. Thus, the nonlinear p- curve in the liquefied soil is approximated by the secant
stiffness k where k is the horizontal subgrade reaction coefficient representing the initial
stiffness while  specifies the reduction in stiffness due to liquefaction and subsequent
spreading. Note that the degradation parameter  incorporates both effects from the reduction
of the effective stress level and nonlinear effects due to large ground deformation.
Using results from lateral loading tests on the piles, the subgrade reaction coefficient
was estimated to be k = 15 MN/m3, which is a value similar to that obtained using
conventional empirical expressions for k based on the SPT blow count [1, 8]. Note that k is
the stiffness corresponding to a relative displacement of about 0.5 to 1 cm rather than the
12
initial stiffness at very small displacements. Another modeling feature important to mention is
that the subgrade reaction coefficient was assumed to be constant throughout the depth of the
liquefied layer. This is equivalent to assuming an average SPT N value for the liquefied layer
and that conventional expressions for k based on the SPT blow count are applicable. Effects of
the overburden stress on k have been ignored in this model in order to simplify the analysis
and based on the reasoning that the stiffness of laterally spreading soils is predominantly
controlled by the nearly complete reduction in the effective stress level and by the large
ground deformation.
The ground displacement applied in the experiment at a given time during the lateral
loading and the corresponding lateral load from the surface layer evaluated from the measured
pile response as described in the preceding sections were applied as input loads in the
pseudo-static analysis, as depicted in Fig. 11. Since  was the only unknown parameter in the
analytical model, the value of  was assumed and an analysis was conducted using the
simplified model. The pile response computed in the analysis was then compared to the
response observed in the experiment and such analyses were repeated until eventually a value
of  was identified that provides the best agreement between the computed and measured pile
displacements and bending moments. In this way, a best-fit value for  was back-calculated
for a given lateral ground displacement. Note that in the case of the steel pile the problem was
essentially linear thus making the back-calculation of  straightforward.
For example, the model shown in Fig. 11 represents the loading condition for the steel
pile at the time when the ground displacement applied in the test was 45 cm at the top of the
liquefied layer. Using this loading condition, a series of analyses were conducted assuming
different value for  in each analysis, in the range between 1/1000 and 1/10. Results from
these analyses are shown in Figs. 12a and 12b where pile displacements and bending
moments are displayed respectively. It is evident in Fig. 12 that the best agreement between
13
the computed and measured pile response was obtained for a value of  = 1/55, which was
thus adopted as the best-fit value of for the spreading displacement of 45 cm.
In this way, best-fit values of  were back-calculated for different lateral ground
displacements applied in the course of the lateral loading as summarized in Fig. 13. The
shaded area in this figure marks the range of  values for which the computed pile response
was within ± 10 % deviation from the measured response of the steel pile while the solid line
with symbols indicates the estimated best-fit values of  for which the computed pile response
was practically coincident with the measured one, as demonstrated in Fig. 12 for the value of
 = 1/55. By and large,  takes values in the range between 1/30 and 1/80 and shows gradual
decrease with the increasing ground displacement or relative displacement between the pile
and the soil. In relation to the possible viscous effects on the pile response and hence on the
value of , it is to be mentioned that the ground deformation induced in the test had nearly
constant strain rate of about 1 %/sec. The above values of  are generally in agreement with
results of other investigators though one should acknowledge the relatively large range in the
variation of . Thus, for example, Orense [11] estimated the stiffness of spreading soils from
large-scale tests to be in the range between 1/10 and 1/500 of the initial stiffness while
O’Rourke [12] back-calculated a value of 1/60 in a simulation of a well-known case history of
damaged piles in the 1964 Niigata earthquake. When comparing  values from different
studies, one should recognize that, exactly speaking, the value of  is more or less affected by
the uncertainty in the value of k as well as by the assumptions and particular features of the
analytical model used in the back-calculation.
One important observation from the results presented in Fig. 13 is that the sensitivity of
the analytical pile response on the value of , as indicated by the vertical size of the shaded
area, depends on the relative displacement between the soil and the pile. Thus, at small
relative displacements or when the pile basically follows the ground movement, the pile
14
response is relatively insensitive to the value of . Conversely, the response involving large
relative displacements between the pile and the soil is very sensitive to the value of , as
clearly demonstrated by the analytical results presented in Fig. 12. The former finding is in
agreement with results of analytical studies conducted by Tokimatsu [17] who in simulations
of case histories from the 1995 Kobe earthquake involving cyclic liquefaction and small
relative displacements between the pile and soil indicated that the value of  does not have
significant influence on the computed pile response.
5. Discussion
Combined effects from the movement of the liquefied soil and lateral pressure from the
surface layer contributed to the measured response of the piles in the lateral spreading test. As
described in the preceding sections and directly evidenced in the marked difference between
the responses of the steel pile and PHC pile, these effects are significantly influenced by the
stiffness of the pile relative to the soil (relative stiffness of the pile). In order to compare the
individual contributions from the liquefied soil and surface layer in the total pile response and
illustrate their dependence on the relative stiffness of the pile, numerical analyses of the steel
pile were carried out using three analytical models, as shown in Fig. 14, with the pile length
and thickness of the liquefied layer being the only differences among the models. Properties
of the pile and soil in these models are identical to those of the steel pile in the test, with
Model S shown in Fig. 14a being in effect the exact analytical model for the steel pile in the
test. Note that the relative stiffness of the pile is different in the three analytical models.
For each model, a set of analyses were conducted assuming different lateral ground
displacement in each computation. In the analyses, the lateral load from the surface layer and
reduction of stiffness in the liquefied soil were defined according to the experimental
15
relationships shown in Figs. 7 and 13 respectively. Results of the analyses are summarized in
Figs. 15a through 15d.
Fig. 15a displays the computed displacements at the pile head as a function of the
applied ground displacement at the top of the liquefied layer. The response of Model S is
practically identical to the observed response of the steel pile in the test and is representative
of the behavior of a relatively stiff pile that does not follow the lateral ground movement. The
response of Model F, on the other hand, is typical for a relatively flexible pile that moves
together with the surrounding soil whereas Model I shows an intermediate response between
these two cases.
As mentioned earlier, in the employed simplified analysis the nonlinear behavior is
approximately modeled using iterative equivalent linear calculations. Thus, even though the
analysis permits to estimate the inelastic deformation and damage to piles, the pile response is
eventually computed by way of a linear analysis [4]. In this analysis, the pile response is
calculated separately for each load component and then the total response is obtained by the
superposition of all components. For this reason, it is possible to divide the computed pile
response into two components, one due to the load from the liquefied layer and the other due
to the load from the surface layer. Utilizing this feature, in Figs. 15b through 15d, the
computed displacement at the pile head normalized to its maximum value is plotted so that
the effects from the liquefied layer and non-liquefied surface layer are separated and thus their
contributions in the total pile displacement are indicated. Comparing Figs. 15b, 15c and 15d,
it is apparent that the contribution of the surface layer in the total response of the pile
decreases with the increasing flexibility of the pile. Thus, the displacement of the stiff pile
shown in Fig. 15b (analytical Model S) was predominantly due to the lateral load from the
surface layer with the ultimate lateral pressure pu being the parameter that effectively
controlled the maximum lateral load and hence the maximum response of the stiff pile.
16
Conversely, the displacement of the flexible pile depicted in Fig. 15d (Model F), was
governed by the displacement of the liquefied soil. In this case, the full lateral pressure from
the crust soil could not be mobilized because of the small relative displacement between the
crust soil and the pile. Note however that the non-liquefied surface layer still influenced the
pile response and increased the displacement of the flexible pile. The results described above
clearly illustrate that the proportion of the effects from the liquefied layer and non-liquefied
crust layer in the total pile response depends on the relative stiffness of the pile. In this
context, the value of  is critically important in the evaluation of pile response because it
affects the relative stiffness of the pile assumed in the calculation.
6. Conclusions
The large-scale test on single piles presented herein provided extremely valuable data
on the behavior of piles in laterally spreading soils. In this paper, effects from the
non-liquefied crust soil at the ground surface and reduction in stiffness due to liquefaction and
subsequent spreading have been thoroughly examined and discussed in relation to the relative
stiffness of piles.
The PHC pile followed the lateral movement of the spreading soils exhibiting response
typical for a relatively flexible pile. The response of the PHC pile was practically controlled
by the magnitude of the ground displacement.
On the other hand, the steel pile did not follow the ground movement exhibiting strong
lateral resistance and behavior that is representative of a relatively stiff pile. The response of
the steel pile was effectively controlled by the lateral load from the crust layer. Thus, the
maximum response of the pile was achieved at the time when the maximum pressure was
mobilized in the crust soil. In the course of the subsequent lateral loading, the pile response
remained unchanged in spite of the significant increase in the lateral ground displacement.
17
The ultimate lateral pressure from the crust soil per unit width of the steel pile was about 4.5
times the Rankine passive pressure.
The back-calculated stiffness of the liquefied soil undergoing spreading was in the range
lateral spreading. The sensitivity of the analytical pile response on the value of  was found to
be negligible at small relative displacements between the soil and the pile whereas it was quite
significant at large relative displacements.
The above findings have important bearing for the analysis of piles subjected to lateral
spreading. They suggest that, for relatively stiff piles, the ultimate lateral pressure from the
surface layer pu and stiffness degradation factor  of liquefied soil are the key parameters
influencing the pile response. The magnitude of lateral ground displacement may not be
critically important for these piles. Conversely, the response of flexible piles is predominantly
controlled by the magnitude of spreading displacements. For flexible piles, the values of pu
and  are less important since the response of these piles involves small relative displacement
between the soil and the pile.
Acknowledgments
The large-scale experiment presented in this paper has been sponsored and carried out
by the Japan Electric Power Civil Engineering Association (JEPOC) using the large-scale
shake table of the National Research Institute for Earth Science and Disaster Prevention
(NIED), Tsukuba, Japan. The authors would like to acknowledge the permission of these
agencies to use the test data and thank Mr. Y. Suda (JEPOC) for the cooperation.
18
References
[1]
Architectural Institute of Japan. Recommendations for design of building foundations. 2001;
(in Japanese).
[2]
Bransby MF. Difference between load-transfer relationships for laterally loaded pile groups:
active p-y or passive p-. J Geotech Engng, ASCE, 1996; 122(12):1015-1018.
[3]
Broms B. Lateral resistance of piles in cohesionless soils. J Soil Mech and Found Engng,
ASCE, 1964; 90(SM3):123-156.
[4]
Cubrinovski M, Ishihara K. Simplified method for analysis of piles undergoing lateral
spreading in liquefied soils. Soils and Foundations, 2004; 44(5):119-133.
[5]
De Beer E. Piles subjected to static lateral loads. Proc. Specialty Session 10, 9th Int. Conf. Soil
Mech and Found Engng; Tokyo, 1977; p. 1-14.
[6]
Feng Y-S, Ho Y-C, Chen T-J. Passive earth pressure with critical state concept. J Geotech
Geoenv Engng, ASCE, 2002; 128(8):651-659.
[7]
Japanese Geotechnical Society. Special issue on geotechnical aspects of the January 17, 1995
Hyogoken-Nambu earthquake. Soils and Foundations, 1998.
[8]
Japan Road Association. Specifications for road bridges. 1980; (in Japanese).
[9]
Kamei M, Morimoto I, Suda Y, Hayashi H, Yoshisako K, Kokusho T, Ishihara K. Investigation
of lateral loads applied to piles during lateral spreading: Experimental results. Proc. 57th Conf.
of Japanese Society of Civil Engineers, 2002; 3:1059-1060 (in Japanese).
[10]
Meyerhof GG, Mathur SK, Valsangkar AJ. Lateral resistance and deflection of rigid walls and
piles in layered soils. Can Geotech J, 1981; 18:159-170.
[11]
Orense R, Ishihara K, Yasuda S, Morimoto I, Takagi M. Soil spring constants during lateral
flow of liquefied ground. Proc. 12th World Conf. on Earthquake Engng, Auckland, 2000;
CD-ROM: Paper 2099.
[12]
O’Rourke TD, Meyersohn WD, Shiba Y, Chaudhuri D. Evaluation of pile response to
liquefaction-induced lateral spread. Proc. 5th U.S.-Japan Workshop on Earthquake Resistant
Design of Lifeline Facilities and Countermeasures against Soil Liquefaction, Tech. Report
NCEER-94-0026; 1994:457-479.
[13]
Poulos HG, Chen LT, Hull TS. Model tests on single piles subjected to lateral soil movement.
Soils and Foundations, 1995; 35(4):85-92.
[14]
Prasad YVSN, Chari TR. Lateral capacity of model rigid piles in cohesionless soils. Soils and
Foundations, 1999; 39(2):21-29.
[15]
Rollins KM, Sparks A. Lateral resistance of full-scale pile cap with gravel backfill. J Geotech
Geoenv Engng, ASCE, 2002; 128(9):711-723.
[16]
Suda Y, Hayashi H, Yoshisako K, Morimoto I, Yamamoto Y, Kokusho T, Ishihara K.
19
Investigation of lateral loads applied to piles during lateral spreading: Description of the
experiment. Proc. 57th Conf. of Japanese Society of Civil Engineers, 2002; 3:1057-1058 (in
Japanese).
[17]
Tokimatsu K, Asaka, Y. Effects of liquefaction-induced ground displacements on pile
performances in the 1995 Hyogoken-Nambu earthquake, Special Issue of Soils and
Foundations; 1998; 163-177.
[18]
Yamamoto Y, Morimoto I, Suda Y, Hayashi H, Yoshisako K, Kokusho T, Ishihara K.
Investigation of lateral loads applied to piles during lateral spreading: Evaluation of lateral
loads. Proc. 57th Conf. of Japanese Society of Civil Engineers, 2002; 3:1061-1062 (in
Japanese).
Lateral
Earth force
pressure
Non-liquefied
surface layer
p
Liquefied
layer
p
U
G
u
k

Pile
Non-liquefied
base layer
Free field
displacement
of liquefied soil
20
Fig. 1
(a) Dynamic loading phase
Laminar box
Steel pile
PHC pile
1.0
Loading
frame
Crust layer
3.8 m
Saturated
sand
D = 31.8 cm
D = 30 cm
Shake table
Acceleration
(g)
12.0 m
Dynamic excitation: a
max
0.2
0
-0.2
0
10
= 0.217 g; f = 2 Hz;
20
Time (sec)
30
40
21
(b) Lateral spreading phase
4.1 cm/sec
Crust layer
Liquefied
sand
Lateral loading
Horizontal
acceleration
a = 0.027g
Lateral ground
displacement
max
f = 10 Hz
40
U
68.7
46.7
50
Time
60
(sec)
G-max
= 84 cm
70
Fig. 2
22
(a) Steel pile
1000
Y
MY = 894 kN-m
500
EI = 89.2 MN-m2
0
0
0.02
0.04
Curvature, 
0.06
(1/m)
(b) PHC pile (C-type: 10 MPa prestress)
150
Bending moment, M (kN-m)
Bending moment, M (kN-m)
1500
U
100
Y
MY = 75 kN-m
50
C
EI = 14.2 MN-m2
0
0
0.02
Curvature, 
0.04
0.06
(1/m)
Fig. 3
23
(a) During shaking
Pore pressure ratio,
u /  v'
Dynamic base-excitation
3.9 m
1
2.15 m
1.15 m
0.5
0.8 m
z = 0.5 m
0
0
10
20
Time
30
40
(sec)
(b) During lateral loading (spreading)
u /  v'
Lateral spreading
2.15 m
1
Pore pressure ratio,
3.9 m
1.15 m
0.5
0.8 m
z = 0.5 m
0
40
50
60
Time
70
80
(sec)
24
100
(a)
80
PHC pile
60
Ground surface
40
Failure
Yielding
Steel pile
20
0
1.2
45
(b)
1
0.8
50
Ultimate
moment
55
60
65
70
Failure
Ground surface
displacement
PHC pile
Yield
moment
120
100
80
0.6
60
0.4
40
Steel pile
20
U = 17 cm
G
9 cm
0
45
50
55
60
Time
65
70
0
(sec)
25
(cm)
0.2
Ground displacement
Normalized bending moment, M / M
Y
Horizontal displacement (cm)
Fig. 4
Normalized lateral pressure, p /  'v
Fig. 5
25
Steel pile
20
0.50 m
15
1.0 m
z = 1.4 m
10
5
0
0
20
40
60
80
Relative displacement between soil and pile,
 = U (z) - U (z)
g
p
(cm)
Fig. 6
26
Measured resultant pressure
P

P
Rankine passive pressure
p

5
4
3
2
Steel pile
1
0
PHC pile
0
20
40
60
80
Relative displacement between soil and pile,
 = U (z) - U (z)
(cm)
g
p
Fig. 7
27
(a) Active pile loading
Lateral
load
. .. . . . . .
..
.
Movement
of soil
(b) Passive pile loading
Ground
displacement
. .
. . .. . .. . . .
. . . . . .. . .. . .
..
. . .. .
. . ..
.. . .. . . . .
... . ....
. . . ...
. . .. . . .
.
Movement
of soil
Passive
earth pressure
Passive
earth pressure
Fig. 8
28
Measured ultimate pressure
Active piles
Single piles in sand
7
Meyerhof (1981)
Prasad (1999)
Broms (1964)
Passive piles
6
5
Passive piles
4
Poulos (1995)
Large-scale test
(this study)
p
3
P

u
P
u

Rankine passive pressure
8
2
1
30
35
40
45
Angle of internal friction,

50
(degree)
Fig. 9
29
Summary of experiments
from 5 studies (Rollins, 2002)
Fang (2002)
Large-scale test (this study)
0.2
u / H
p
u

80
90
0.1
0
30
40
50
60
70
Relative density, D
r
(%)
30
Fig. 10
Depth
0m
Lateral pressure
from surface
layer
45 cm
1.4 m
316 kPa
p
k

Applied
displacement of
liquefied soil
 is assumed
4.8 m
Fig. 11
31
0
(a)
1/1000
(b)
0
Measured
1/100
1
2
2
1/20
Depth
(m)
Computed
1/55
1

3
3

4
Steel pile
U = 45 cm
4 1/1000
1/20
1/100
G
5
0
1/55
5
2
4
6
Pile displacement (cm)
0
200 400 600 800
Bending moment (kN-m)
Fig. 12
32

0.05
Stiffness degradation factor,
0.1
0.01
Range of
 values
Best-fit
 value
Steel pile
0.001
0
20
40
60
Ground displacement at top of liquefied layer
80
(cm)
33
Fig. 13
A
A
(a) Model S
(b) Model I
(c) Model F
0m
Crust layer
Liquefied
layer
1.4 m
L = 4.8m
L = 7m
4.8 m
L = 10m
Fig. 14
34
Pile head displacement (cm)
Ground displacement
50
40
Yielding
30
Yielding
20
Model I (L=7m)
10
0
Model S (L=4.8m)
0
U / U (max)
20
0.75
p
0.25
0
0
20
40
60
(c)
0.75
Crust layer
0.5
P
p
(b)
Due to displacement
of liquefied soil
Model I (L=7m) Yielding
1
U / U (max)
60
Due to lateral pressure
from crust layer
0.5
P
40
Model S (L = 4.8 m; Steel pile in test)
1
Normalized pile displacement,
(a)
Model F
(L=10m)
0.25
0
Liquefied layer
0
20
40
60
35
P
(d)
Model F (L=10m)
Yielding
0.75
Crust layer
p
U / U (max)
1
0.5
0.25
0
0
Liquefied layer
20
40
60
Ground displacement at top of liquefied layer, (cm)
Fig. 15
36