You are on page 1of 10

Proceedings of the 8

th
U.S. National Conference on Earthquake Engineering
April 18-22, 2006, San Francisco, California, USA









MODEL TESTS ON LATERAL EARTH PRESSURE ON LARGE GROUP PILE EXERTED
BY HORIZONTAL DISPLACEMENT OF LIQUEFIED SANDY GROUND


Ikuo Towhata
1
,Vlatko Sesov
2
, Ramin Motamed
3
& Marcelo Gonzales
4


Abstract
A series of shaking table test were performed to study the group pile behavior under
flow of liquefied ground. Two models of group piles i.e. 6 x 6 and 11 x 11 piles were
embedded into a gently sloped liquefiable soil deposit. Upon shaking, the kinematic
effects of liquefaction induced horizontal displacements and the influence of pile
spacing on pile response were chosen as the main problems to be investigated during
the tests. The results from the experiments show that the pile response upon
liquefaction-induced flow depends on the location of the pile in the pile group. The
distribution of the lateral loading in the pile group due to liquefaction-induced
ground flow shows that the front pile row in the pile group receives the main portion
of the total lateral force. This was more pronounced in the case of 2.5-diameter than
5-diameter pile spacing. The results of this study also show that piles can be used as
an efficient mitigation measure against lateral spreading of liquefied soil.

INTRODUCTION

Lateral spreading of liquefied soil in gentle sloping ground or near waterfront is a cause of
significant damage to underground structures and pile foundation of engineered structures. Cases of
damaged pile foundation due to these reasons have been reported in a number of seismic events.
Investigations carried out after the Niigata 1964 Earthquake showed that damages observed on pile
foundations were strongly related to large permanent ground displacement, which was induced by
intensive liquefaction [1]. A large number of pile foundations were deformed and damaged during
the Kobe Earthquake 1995 due to movement of quay walls and revetments toward the sea which
brought extensive flow of the ground behind them. These and also other case histories clearly
demonstrate that pile foundations are susceptible to seismic damages.
Most of the present information about seismic behavior of pile foundation in liquefied soil is
based on results of studies which represent the pile foundation as a single pile or small group pile
(2x2 or 3x3). Knowledge of the group pile response on lateral spreading of liquefied soil is very
limited. The present study investigated the behavior of a large group pile subjected to lateral flow of
liquefied soil by performing a series of shaking table tests.

SHAKING TABLE TESTS

Table 1 summarizes the characteristics of six experiments which were conducted on the shaking
table facility at the University of Tokyo, Civil Engineering Department. The tested models were
prepared in a rigid square container with a side of 2 m and a height of 0.60 m.

1
Professor, Dept. of Civil Engineering, The University of Tokyo, Japan. email:towhata@geot.t.u-tokyo.ac.jp
2
Postdoctoral researcher, Dept. of Civil Engineering, The University of Tokyo, Japan
3
Doctoral student, Dept. of Civil Engineering, The University of Tokyo, Japan
4
Former graduate student, Dept. of Civil Engineering, The University of Tokyo, Japan
Paper No. 1227
Pile groups of 6 x 6 and 11 x 11 were investigated in the present study. Fig.1a presents a plan
view of the pile configuration for tests (Test1 & Test2) involving piles with center-to-center pile
spacing of 5D (D-pile diameter) while Fig.1b presents pile configuration for tests (Test3 & Test4)
involving piles with center-to-center pile spacing of 2.5D. Fig.1c shows a plan view of the pile
configuration for tests (Test5 & Test6) involving piles with various spacing. This pile configuration
was formed by removing the non-instrumented piles from the 11x11 pile group model in Fig.1b.
Table 1. Tested models
Test Ground model Piles Spacing Shaking-direction
Test1
Single liquefied
layer
6 x 6 5 D
Lateral direction
Test2
Liquefied + top
non-liquefied
6 x 6 5 D
Lateral direction
Test3
Single liquefied
layer
11 x 11 2.5 D
Lateral direction
Test4
Liquefied + top
non-liquefied
11 x 11 2.5 D
Lateral direction
Test5
Single liquefied
layer
17 various
Lateral direction
Test6
Liquefied + top
non-liquefied
17 various
Lateral direction



5D 5D
5
D
P6 P4 P5 P2
P12
P3 P1
P13 P14
P11 P15
P10 P9 P8 P7

P6 P1
P7
P8
P9 P10
P12
P11
P14
P13
P15
P17 P16
P2 P3 P4 P5
2
.
5
D
2.5D 2.5D

a) 6 x 6 pile group model b) 11 x 11 pile group model c) Only instrumented piles
Figure 1. Configuration of the pile group models

Piles whose characteristics are shown in Table 2 were fixed at the bottom and free at the top.
Such a simple pile model can eliminate the inertial effects of a super structure on pile response and
enable more clear insight into the kinematic effects of liquefaction-induced horizontal displacements,
which is the primary goal of this study. Toyoura sand was used to prepare the ground models, Table
3. The soil profile in Test1, Test3 and Test6 consisted of a gently inclined single layer of fully
saturated Toyoura sand (Fig.2). In order to investigate the effects of the top non-liquefiable layer on
the pile group response, the soil profile in Test 2, Test 4 & Test 6 was prepared with two sand layers
Soil flow Soil flow Soil flow
(Fig.3). The bottom layer was fully saturated and with uniform thickness of 30 cm. The top layer was
made of dry Toyoura sand with slope inclination of 5% representing non-liquefiable layer.

Table 2. Properties of pile Table 3. Material properties of Toyoura sand

Material Poly Carbonate
Height (cm) 53
Outer / Inner diameter (cm) 3.2 / 2.7
E (N/cm
2
) 270000
EI (Ncm
2
) 532810


5
0

c
m
5
3

c
m
5%
Dr=35%


Figure 2. Single layer ground model Test1, Test3 & Test5


5
0

c
m
5
3

c
m
195 cm
Dry sand
5%
3
0

c
m

Figure 3. Two layer ground model Test2, Test4 & Test6

The level of water inside the container was made equal to the highest elevation of the slope where
ground models consisted of a single layer, while in models with two layers, it was set at the surface
of the bottom layer. To prevent water migration into the top dry layer due to suction, impermeable
sheets were installed after finishing the bottom layer and 2 cm of fine gravel layer was placed above
to prevent possible capillary effects. The piles were instrumented with strain gauges, which were
attached to different locations along the pile height. Accelerations, excess pore water pressures, soil
displacements and pile bending strains were recorded during the tests. Also, colored sand markers
were used to observe the soil deformations during the shaking from the top as well as through a
transparent side window of the container. All the models were subjected to the same amplitude of
300 Gal and frequency of shaking f=10 Hz.
Property Toyoura
Maximum void ratio, e
max
0.93
Minimum void ratio, e
min
0.62
Mean particle diameter, D
50
, 0.16
Fines content,% 3
TEST RESULTS

Lateral spreading
The permanent ground displacements were measured by inclinometers. Each inclinometer was
made of seven acrylic plates connected by steel rings. These plates follow the lateral displacement of
the ground. An accelerometer was attached to each plate and the displacement was calculated from
the component of gravity acceleration parallel to the inclination. The estimated soil displacements
from the inclinometers were in good agreement with the soil displacements observed by the colored
markers on the transparent side of the wall and on the surface of the soil model.
During the early stage of shaking, excess pore water pressure rapidly increased reaching the
initial effective stresses. The soil then started to move under the gravitational forces (Fig.4). The
permanent soil displacement accumulated and reached its maximum value at the end of shaking.
5.0 7.5 10.0 12.5 15.0 17.5 20.0
-50
0
50
100
0
1
S
o
i
l

d
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Time(sec.)
Pore water pressure ratio
Soil displacements
Top non-Liquefied layer
Liquefied layer
r
u
=1.0
Soil starts to move
start of shaking
P
o
r
e

w
a
t
e
r

p
r
e
s
s
u
r
e

r
a
t
i
o

r
u
input acceleration

Figure 4. Development of excess pore water pressure and soil displacement Test2

Fig.5 presents the characteristic deformed shape of the soil profile in the ground models with a
single layer: Test1, Test3 & Test5. The soil displacements were measured at three locations: on the
upstream side, (22 cm in front of the first pile row); beside the pile group in the middle part of the
sloped ground and among the piles at the middle section. Maximum soil displacements, (Table 4),
were observed at the top portion of the soil profile with no soil movement at the bottom.

In front of piles
Front pile row Last pile row
5%
Dr=35%
y
max
y
max
y
max
Between piles Middle free field

Figure 5. Permanent ground displacement in single layer models Test1, Test3 & Test5

Table 4 Maximum ground displacements

Soil Displacement
(cm)
Test1 Test2 Test3 Test4 Test5 Test6
In front of piles 8.0 5.5 (0.5*) 7.0 5.0 (0.2*) 6.5 6.0 (0.5*)
Beside the pile group 18.0 6.0 (2.0*) 20.0 6.0 (2.5*) 16.0 8.5 (3.0*)
Between piles 11.0 5.5 (0.5*) 10.0 3.0 (0.2*) 10.5 6.0 (0.7*)
(*) max. soil displacement measured in a non-liquefiable layer

Figure 6 shows the distribution of permanent soil displacement along the height of the soil profile
in the ground models with two layers: Test2, Test4 & Test6. The soil displacement presented in this
figure shows a distribution pattern, which is different from that in single layer models. As seen, the
maximum soil displacements were observed at the middle height of the liquefied layer, decreasing
toward the top non-liquefied layer. The maximum values of soil displacements presented in Table 4
for two-layer models refer to the displacements recorded in the liquefiable layer.

Dry sand
5%
3
0

c
m
In front of piles
Front pile row Last pile row
y
max
Between piles Middle free field
y
max
y
max


Figure 6. Permanent ground displacement in two layers models Test2, Test4 & Test6

Figures 5 and 6 present several important issues concerning the permanent soil displacement of
gently sloped ground models. Firstly, the elevation at which the maximum soil displacement occurs
is different between two types of models. Secondly, there occurred no slippage between two layers.
Another issue that should be stressed out is that the overall development of soil displacements in all
the performed tests was influenced by the presence of the piles. Maximum soil displacements were
observed besides the pile group, while the soil displacements measured in front of the pile were
restrained by the front row of piles. Soil flow among the piles in the pile group was found to be
related to the pile spacing and the presence of the top non-liquefiable layer (Table 4). Smaller pile
spacing contributes toward decreasing of soil flow rate among the piles in the pile group.

Pile Response

The piles were instrumented with strain gauges attached to different locations along the pile
height. Pile bending strains were directly measured by strain gauges during the tests. Knowing the
flexural stiffness of the pile, the bending moments were directly estimated from the recordings of the
strain gauges. Interpolating polynomials were used to fit moment data. In the case of the two-layer
ground model, two separate polynomials were used to fit the moment data in the top non-liquefiable
layer and the bottom liquefiable layer. Continuity conditions were added at the contact area between
the two layers. Shear forces and lateral loading acting on piles were obtained by differentiation,
while rotation and pile displacement were obtained by integration of the bending moment polynomial
curve at each time step during the shaking.
In the light of the performance of piles subjected to lateral spreading of liquefiable soils, there is a
necessity for clearly understanding the soil-pile interaction during the earthquake loading. The
mechanism of soil-pile interaction during the performed tests can be clarified in Fig.7 and Fig.8.
Fig.8a and Fig.8b present the time histories of soil and pile displacement in a two layer ground model.
As shown in Fig.8a soil displacement in the liquefied layer is much larger than the pile displacement
throughout the shaking. The soil displacement in the top non-liquefied layer is also larger than the
pile displacement (Fig.8b), but not to the extent as it is in the liquefied layer.

5 10 15 20
-10
0
10
20
30
40
50
60
pile displacement
soil displacement at
liquefied layer
D
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Time(sec.)

5 10 15 20
-1
0
1
2
3
4
5
pile displacement
soil displacement at
non-liquefied layer
D
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Time(sec.)


Figure 7. Time histories of soil displacements and pile displacements in the two layer ground model

Schematic views of soil and pile deformed shape in the single and two-layer ground models are
presented in Fig.8a and Fig.8b. From the results presented in Fig.7 and Fig.8 it can be concluded that
permanent soil displacement is the cause and lateral loads, which develop along the piles are the
consequences. This is the characteristic behavior of the so called passive piles [3].

d
soil
pile
displacement
d
e
p
t
h
liq.layer

d
soil
displacement
non-liq.
d
e
p
t
h
pile
liq.layer


Figure 8. Soil-pile interaction in the single and two layer ground model

Figure 9 shows the time histories of soil displacement, the soil flow velocity calculated as the
first derivative of displacement and the lateral load acting on the pile in the liquefiable layer. Fig.10a
and Fig.10b present the time histories of lateral load acting on the pile in the non-liquefiable layer
and the soil-pile relative displacement. The time history of lateral load acting on a pile shows better
correlation with soil velocity than displacement. The results from these experiments show that the
a) b)
a) b)
top non-liquefiable layer does not develop large displacement like the bottom liquefiable layer.
Several reasons could be pointed out for such a phenomenon. First, due to the different thickness of
the top layer the initial vertical pressure acting on the liquefiable layer is different, being bigger at
the upstream part or the left side (Fig.3) than downstream or right side of the two-layer ground
model. As liquefaction takes place, difference in vertical pressure results in development of
horizontal displacements and vertical settlements in the bottom liquefied layer. Soil subsidence is
evident at the upstream part, whilst uplift of the bottom liquefied layer appears at the downstream
part of the two layer ground model and restrains the movement of the top layer. The rigid end wall
condition does not allow development of large shear deformation in soil in its vicinity. In such
conditions, the top layer maintains its shear stiffness during the shaking and the axial compression in
the horizontal direction is the predominant mode of deformation. The top liquefiable layer acts like a
horizontal column, the behavior of which affects the distribution of lateral loads in the pile group.
The piles tend to follow the soil displacements of the top layer to some extent during which lateral
force develops and acts on the pile in the top non-liquefiable layer. The soil-pile relative
displacement is insufficient to develop full passive pressure [4]. Also, low overburden pressure and
low relative density contribute towards minor development of passive pressure from the top layer.
The maximum passive pressure during the tests is mobilized at a soil-pile relative displacement of
=0.01H
NL
to 0.02H
NL,
where H
NL
is the thickness of the top non-liquefiable layer.
-30
0
30
60
90
120
5 10 15 20
-0.5
0.0
0.5
1.0
1.5
2.0
S
o
i
l

r
a
t
e

f
l
o
w


(
m
m

/

s
)
Soil flow rate
Lateral load
Time (sec.)


L
a
t
e
r
a
l

l
o
a
d

(
N
/
c
m
)

Test1-Pile1
Soil displacement


Figure 9. Lateral load, soil displacement and soil flow velocity in the liquefiable layer- Test1, Pile1

6 7 8 9 10 11 12 13 14 15
0
2
4
6
8
0.0
0.2
0.4
0.6
pile displacement
soil displacement at
non-liquefied layer
D
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
Time(sec.)
d
relative displacement
Lateral load from top
non-liquefied layer


L
a
t
e
r
a
l

l
o
a
d

(
N
/
c
m
)
x x

Figure 10. Lateral load and soil-pile relative displacement in the non-liquefiable layer- Test2, Pile1
a)
b)
Pile group behavior

The behavior of an isolated single pile under lateral spreading is manly governed by the stiffness
of the pile, the stiffness of the soil along the length of the pile and the permanent ground
displacement. In contrast, in the field, piles are often arranged in groups and behavior of a pile group
may differ substantially from that of a single pile. It is known that, for closely spaced piles, less than
5 pile diameters, the interaction of one pile with the others in the pile group cannot be neglected and
the pile-soil-pile interaction should be taken into consideration. Most of the present information
about pile group behavior is based on the pile-soil-pile interaction in medium to dense sand, clay and
other type of soil conditions which is substantially different from pile group effects in liquefiable and
spreading ground.
The distribution of the maximum bending moment in each instrumented pile in the pile groups is
presented in Fig.11. The results for pile bending moments in a 6x6 pile group with 5D pile spacing,
Fig11a, show higher values compared with those for a 11x11 pile group with 2.5D pile spacing,
Fig.11b. Piles located in the corner of the pile group (pile11, see Fig.1) develop the largest bending
moment in the pile group. In both pile groups, 6x6 and 11x11, there can be seen clearly the
decreasing tendency of the values of pile bending moments from the front toward the middle pile
rows. These are increased again at the last (rear) pile row. The piles in the middle rows of the pile
groups are shadowed by the front pile row, and the bending moments in these pile rows are much
smaller than those front piles row. In the case of 5D pile spacing, Fig.11a, the maximum bending
moments in the middle pile rows are larger than maximum bending moments in the middle pile rows
in the case of 2.5D pile spacing, Fig11b. It can be observed that, in the case of 2.5D pile spacing, the
bending moments in middle pile rows, show similar maximum values, for example the bending
moment in the 3
rd
pile row does not differ so much compared to those measured in the 9
th
pile row
(Fig.11b).


0
100
200
300
400
500
600


M
a
x
i
m
u
m

P
i
l
e

b
e
n
d
i
n
g

m
o
m
e
n
t
s

(
N
c
m
) Test1
Test2
6th 5th 4th 3rd 2nd
PILE ROW
1st
corner pile-Pile11

0
100
200
300
400
500
600

M
a
x
i
m
u
m

P
i
l
e

b
e
n
d
i
n
g

m
o
m
e
n
t
s

(
N
c
m
)
Test3
Test4
PILE ROW
corner pile- Pile11
11th 10th 9th 8th 7th 6th 5th 4th 2nd 1st 3rd

0
250
500
750
1000
1250
1500

M
a
x
i
m
u
m

P
i
l
e

b
e
n
d
i
n
g

m
o
m
e
n
t
s

(
N
c
m
)
Test5
Test6
PILE ROW
11th 10th 9th 8th 7th 6th 5th 4th 2nd 1st 3rd

Figure 11 Distribution of maximum pile bending moments in the pile group
a) 5D pile spacing b) 2.5D pile spacing
c) Individual piles
Soil flow
Soil flow
Soil flow
Corner pile
Corner pile
Another issue that should be stressed out is that the results from the tests where the top non-
liquefied layer was applied Fig.11a (Test2) and Fig.11b (Test4) show bigger bending moments in
piles compared with those of a single liquefiable layer, Fig.11a (Test1) and Fig.11b (Test3). The
existing of the top non-liquefiable layer in the 6x6 pile group with 5D pile spacing Fig.11a (Test2)
makes the maximum bending moments to be more uniformly distributed in the pile group. Fig.11c
presents the results of maximum bending moments in models which consisted only of instrumented
piles distributed at various spacing. The pile bending moments in this case were larger compared to
those in the 6x6 and 11x11 pile group presented in Fig11a and 11c. The reason for such a pile
response is the lack of shadowing effects and the smaller pile-to-pile interaction. A similar
tendency of shadowing effect as in the case of 11x11 pile group can be observed in the pile column
starting with pile 1 to pile 3 and then a slight increase in bending moments can be observed (Fig11c,
row7) due to the increase of pile spacing from 2.5D to 5D at pile4 (see Fig.1). The bending moments
when top non-liquefiable layer is introduced, Fig.11c (Test6) are much larger then in the case of a
single liquefiable layer Fig11c (Test6).
To evaluate the performance of the pile group undergoing lateral spreading of liquefied ground,
the lateral force per pile row, which is estimated as the sum of the lateral force acting on each pile in
the pile row, is employed. Fig.12a and Fig12b present the lateral force per pile row divided by the
total lateral force acting on the pile group in the cases of 5D and 2.5D pile spacing.

0
10
20
30
40
50

Test1
Test2

L
a
t
e
r
a
l

f
o
r
c
e

p
e
r

p
i
l
e

r
o
w

/

T
o
t
a
l

L
a
t
e
r
a
l

f
o
r
c
e


(
%
)
6th 5th 4th 3rd 2nd
PILE ROW
1st

0
10
20
30
40
50


L
a
t
e
r
a
l

f
o
r
c
e

p
e
r

p
i
l
e

r
o
w

/

T
o
t
a
l

L
a
t
e
r
a
l

f
o
r
c
e




(
%
)
Test3
Test4
PILE ROW
11th 10th 9th 8th 7th 6th 5th 4th 2nd 1st 3rd

Figure 12. Distribution of the lateral force per pile row in the pile groups

The front pile row receives the largest portion of the total lateral force in both pile configurations.
The inner pile rows were subjected to much smaller portion of the total lateral force that is acting on
the pile group. The top non-liquefiable layer in the case of 5D pile spacing in Fig.12a (Test2)
contributes to a more uniform distribution of lateral loads in the pile group. The inner pile rows were
subjected to bigger lateral forces compared with those in the case of the single layer model. In the
case of 2.5D pile spacing, Fig.12b (Test4), with a top non-liquefiable layer, the front row receives a
larger portion of the total lateral force compared with the single layer model. Table 5 summarizes the
results on the total lateral force acting on the pile groups as estimated from the shaking table tests
and the force calculated in accordance with the formulas given in the Specifications for Highway
Bridges Design Code PartV Seismic design JRA (Japan Road Association), [2].
Table 5 Total lateral force acting on the pile group
Lateral force (N) Test1 Test2 Test3 Test4
Shaking table tests 280~320 440~460 340~360 700~730
JRA design codes 420 730 420 730
a) 5D pile spacing b) 2.5D pile spacing
Soil flow Soil flow

The calculation of the total lateral force acting on the pile group by the JRA seismic design codes
gives higher values than the results from the shaking table tests as expected, because these seismic
design codes are mostly based on the back-calculation of the damaged pile foundations during the
Kobe earthquake where the ultimate lateral pressure was mobilized.

CONCLUDING REMARKS

Shaking table tests were conducted to investigate the pile group behavior in a sloping liquefied
ground with and without top non-liquefied layer. The main findings can be summarized as follows:
1) Pile response in liquefied ground correlates more closely with the soil flow velocity than soil
displacement, indicating the viscose nature of lateral flow of liquefied soil.
2) The top non-liquefiable layer acts as a horizontal column subjected to compression. The small
displacements developed in this layer were insufficient to mobilize the full passive pressure.
3) The piles in the pile group were subjected to different lateral force due to ground flow of
liquefied soil. Pile spacing plays a significant role in distribution of lateral force. Smaller pile spacing
creates larger pile to pile interaction.
4) The front pile row receives the largest portion of total lateral force while the middle rows receive
much smaller portion. The last (rear) pile row was subjected to similar magnitude of lateral force as
the front pile row.
5) The front pile row protects the pile rows that follow. This finding gives an idea to use the row of
piles with appropriate pile spacing as a mitigation measure against liquefaction induced lateral
spreading in the case of existing and new engineering structures.
6) The total lateral force acting on the pile groups is larger when a top non-liquefied layer is
introduced in the ground model.
7) The results from the shaking table tests were compared with the Specifications for Highway
Bridges Seismic Design (Japan Road Association). It can be suggested that these design codes
reasonably well estimate the total lateral force, which is acting on the pile groups due to lateral
spreading of the liquefied ground. However, this total lateral force cannot be uniformly distributed
over the piles in the pile group.

ACKNOWLEDGMENT

The investigations presented in this paper were supported by Grant-in-Aid for Scientific Research
as well as the JSPS (Japanese Society for the Promotion of Science) program. The authors gratefully
acknowledge these supports.

REFERENCES

[1] Hamada, M., Yasuda,S., Isoyama, R., & Emoto, K., Study on liquefaction induced permanent
ground displacements (1986), Report of Association for The Development of Earthquake
Prediction
[2] Specifications for Highway Bridges in Japan, Part V, Seismic design, Japan Road Association
[3] De Beer, E., Piles subjected to static lateral loads, The effect of horizontal loads on piles, due to
surcharge or seismic effects,. Ninth International Conference on Soil Mechanics and
Foundation Engineering, Tokyo, 1977
[4] Rollins, K. M. & Sparks, Andrew, Lateral Resistance of full scale pile cap with gravel backfill,
Journal of Geotechnical and Geoenvironmental Engineering September, 2002, Vol. 128, No.9,
711-723

You might also like