See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228715406

Centrifuge Model Study of Laterally Loaded Pile Groups in Clay

Article in Journal of Geotechnical and Geoenvironmental Engineering · March 2004

DOI: 10.1061/(ASCE)1090-0241(2004)130:3(274)

CITATIONS READS
168 5,085

4 authors:

Tommy Ilyas C. F. Leung

University of Indonesia National University of Singapore
24 PUBLICATIONS 255 CITATIONS 211 PUBLICATIONS 4,533 CITATIONS

SEE PROFILE SEE PROFILE

Khuan Y Chow Setia Budi

Health Promotion Board Airlangga University
122 PUBLICATIONS 3,683 CITATIONS 16 PUBLICATIONS 270 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by C. F. Leung on 30 May 2014.

The user has requested enhancement of the downloaded file.

 Centrifuge Model Study of Laterally Loaded Pile Groups
in Clay
T. Ilyas1; C. F. Leung2; Y. K. Chow3; and S. S. Budi4

Abstract: A series of centrifuge model tests has been conducted to examine the behavior of laterally loaded pile groups in normally
consolidated and overconsolidated kaolin clay. The pile groups have a symmetrical plan layout consisting of 2, 2⫻2, 2⫻3, 3⫻3, and
4⫻4 piles with a center-to-center spacing of three or five times the pile width. The piles are connected by a solid aluminum pile cap
placed just above the ground level. The pile load test results are expressed in terms of lateral load–pile head displacement response of the
pile group, load experienced by individual piles in the group, and bending moment profile along individual pile shafts. It is established that
the pile group efficiency reduces significantly with increasing number of piles in a group. The tests also reveal the shadowing effect
phenomenon in which the front piles experience larger load and bending moment than that of the trailing piles. The shadowing effect is
most significant for the lead row piles and considerably less significant for subsequent rows of trailing piles. The approach adopted by
many researchers of taking the average performance of piles in the same row is found to be inappropriate for the middle rows, of piles for
large pile groups as the outer piles in the row carry significantly more load and experience considerably higher bending moment than those
of the inner piles.
DOI: 10.1061/共ASCE兲1090-0241共2004兲130:3共274兲
CE Database subject headings: Bending moments; Centrifuge model; Clays; Deflection; Efficiency; Pile groups; Pile lateral loads;
Soil-pile interaction.

Introduction present, much of the centrifuge model studies on laterally loaded

The performance of pile foundations subject to lateral load is of pile groups were carried out in sand 共for example, McVay et al.
considerable importance in geotechnical practice. In the field, 1994, 1995, 1996, 1998; Remaud et al. 1998兲. Relatively few
piles are often arranged in groups, and the behavior of a pile centrifuge model studies were conducted to investigate the per-
group may differ substantially from that of a single pile. The formance of laterally loaded piles in clay and majority of these
elasticity interaction factor approach 共Poulos and Davis 1980兲 is studies concentrated on the study of single piles 共for example,
commonly employed to evaluate the degree of interaction among Hamilton et al. 1991; Kitazume and Miyajima 1994兲. A series of
piles in a pile group. Several full-scale field tests on pile groups centrifuge model tests was, therefore, initiated to examine the
subject to lateral loads have also been reported. These include behavior of laterally loaded pile groups in normally consolidated
studies on piles in sand 共for example, Brown et al. 1988; Ruesta 共NC兲 and overconsolidated 共OC兲 clay. In addition, the test obser-
and Townsend 1997兲 and in clay 共for example, Matlock et al. vations are compared with those obtained from field studies. The
1980; Brown et al. 1987; Rollins et al. 1998兲. Some degree of observed pile–soil–pile interaction behavior is compared to that
understanding on the performance of laterally loaded pile groups obtained from the elasticity interaction factor approach.
has been gained from these field studies.
Laboratory studies such as centrifuge model tests could pro-
vide further insight on the behavior of deep foundations. At Experimental Setup

1 All the tests were conducted at 70 g on the National University of

Senior Lecturer, Civil Engineering Dept., Univ. of Indonesia, Depok,
Singapore 共NUS兲 Geotechnical Centrifuge, which has a radius of
16424 Indonesia. E-mail: t – [email protected]
2
Associate Professor, Centre for Soft Ground Engineering, Civil 2 m and comprises a balanced arm with dual swing platforms.
Engineering Dept., National Univ. of Singapore, Singapore, 117576. The centrifuge has a capacity of 40 g tones and a maximum
E-mail: [email protected] acceleration of 200 g. Details of the NUS Centrifuge are given in
3
Professor, Centre for Soft Ground Engineering, Civil Engineering Lee et al. 共1991兲. Fig. 1 shows a sketch of the model setup. The
Dept., National Univ. of Singapore, Singapore 117576. E-mail: cylindrical stainless steel model container has an internal diam-
[email protected] eter of 550 mm and a height of 375 mm. An inverted U-shaped
4
Professor, Civil Engineering, Dept., Univ. of Indonesia, Depok, metal frame consisting of vertical and horizontal actuators was
16424 Indonesia. E-mail: [email protected] bolted onto the model container. The vertical actuator was at-
Note. Discussion open until August 1, 2004. Separate discussions tached to the center of the top of the metal frame while the hori-
must be submitted for individual papers. To extend the closing date by
zontal actuator was attached to the left leg of the frame.
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos- The model pile is made of a hollow aluminum square tube
sible publication on May 3, 2002; approved on May 8, 2003. This paper instrumented with ten levels of strain gauges to enable bending
is part of the Journal of Geotechnical and Geoenvironmental Engineer- moment measurements along the pile shaft. At each level, two
ing, Vol. 130, No. 3, March 1, 2004. ©ASCE, ISSN 1090-0241/2004/3- active strain gauges were glued onto the opposite faces of the
274 –283/$18.00. model pile shaft and connected to two dummy gauges on a strain

274 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004

 achieve an average degree of consolidation of at least 95% result-
ing in a final clay thickness of about 245 mm.
T-bar penetrometer tests, which were first introduced by Stew-
art and Randolph 共1991兲, were carried out in-flight to determine
the soil strength profile after soil consolidation. The penetrometer
adopted in the present study has a cross bar of 25 mm long and 5
mm in diameter attached perpendicularly to the end of a vertical
shaft to form a T. The penetrometer was pushed into the soil at a
rate of 20 mm/s and its penetration resistance was monitored by a
load cell mounted immediately behind the penetrometer. Stewart
and Randolph 共1991兲 recommended a bearing capacity factor of
10.5 to derive the soil strength from the penetration resistance.
The test results revealed that the undrained shear strength of the
NC clay varies linearly with depth from 0 at the ground surface
increasing to about 20 kPa at about 15 m 共prototype depth兲 below
the ground level. Thus, the clay can be considered as soft clay.
For OC kaolin clay sample, a surcharge of 60 kPa was gradu-
ally placed on top of the clay slurry 共also at 150% LL兲 in the
model container at 1 g for about a week. The clay was then
further consolidated under its self-weight in the centrifuge for
about 6 h to achieve an average degree of consolidation of at least
95%. The final clay thickness was also about 245 mm. Results
from in-flight T-bar tests after soil consolidation reveal that the
soil strength increases from about 10 kPa at the ground surface to
about 25 kPa at 15 m below the ground surface. The overconsoli-
dation ratio gradually decreases from about 8 at the ground sur-
face to almost unity at about 8 m below the ground surface.

Fig. 1. Centrifuge model setup 共units in millimeters兲 Test Procedure and Configurations

After completion of consolidation for both NC and OC clays, the

meter mounted on board the centrifuge to form a Wheatstone full centrifuge was stopped to facilitate pile installation. The model
bridge circuit. The instrumented pile was protected by a thin layer pile/pile group was carefully jacked into the soil at 1 g by the
of epoxy having a final pile width of 12 mm that is equivalent to servocontrolled vertical actuator using a closed-loop displacement
840 mm in prototype scale. The flexural rigidity E m I m of the- feedback obtained from the vertical displacement transducer. The
model pile is 384 kN cm2 , resulting in a prototype E p I p of pile installation had to be done at 1 g, as it was very difficult to
922 kN m2 . The model pile is 260 mm long with the lower 210 control the pile penetration depth in soft clay under high-g con-
mm pile length to be inserted into the clay. At 70 g, the simulated dition. Craig 共1985兲 reported that unlike sand, the difference in
prototype pile embedment pile length is 14.7 m. For calibration, the pile capacity for piles installed in clay at 1 g and at high g is
the model pile was loaded as a cantilever beam at 1 g to calibrate relatively insignificant as the volume change during pile installa-
the pile bending moment against the strain gauge circuit output tion is relatively small. The pile installation was completed when
from the strain meter. The model pile cap is made of 20 mm 共1.4 the midheights of the pile cap and horizontal load cell were ex-
m prototype scale兲 thick solid aluminum and placed at 50 mm actly aligned at the same elevation. The centrifuge was then spun
共3.5 m prototype scale兲 above the ground level. The piles were up to 70 g again for about 1 h to enable the pore water pressure
tightly secured through openings in the pile cap using screws. developed during pile installation to dissipate completely. A lat-
For the soil sample preparation, a 30-mm-thick sand layer was eral load test was then conducted on the pile/pile group by dis-
first placed at the bottom of the model container before placement placing the pile cap using the servocontrolled horizontal actuator
of kaolin clay slurry. The kaolin clay has the following properties: with a closed-loop feedback from the horizontal displacement
average bulk unit weight ␥⫽16 kN/m3 ; average water content transducer. The pile was displaced using displacement-control
⫽66%; liquid limit LL⫽79.8%; plastic limit PL⫽35.1%; com- mode at a rate of 0.05 mm per second. The lateral load was
pression index C c ⫽0.55; recompression index C s ⫽0.14; and co- monitored by a load cell mounted at the tip of the horizontal
efficient of permeability k⫽2⫻10⫺8 m/s. The NC kaolin clay actuator. The lateral displacement of the pile cap was recorded by
sample was prepared in the following manner. First, kaolin clay a highly accurate noncontact microlaser displacement transducer.
slurry was thoroughly mixed at 150% of its LL and deaired for 24 The transducer used has a 40 mm measurement range and a sen-
h. The clay slurry was then placed in the model container to a sitivity of 0.02 mm. The lateral load and displacement of the
height of about 380 mm and then consolidated under its self- pile/pile group and strain gauge readings were recorded at regular
weight in the centrifuge at 70 g to allow the excess pore water intervals during the entire load test.
pressure in the clay to dissipate through the clay surface and the Pile groups of up to 16 piles were investigated in the present
openings located close to the bottom of the container via the study. Figs. 2共b–e兲 provide a plan view of the pile configuration
bottom sand layer. Ground settlements and pore water pressures for tests involving piles with center-to-center pile spacing of 3D
in the soil were measured to monitor the progress of soil consoli- (D⫽pile width), while Figs. 2共f and g兲 provide the pile group
dation. In general, 6 h soil consolidation time was required to configuration for tests with piles at 5D spacing. In most cases,

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004 / 275

 tests were carried out in both NC and OC kaolin clay. The defi-
nition of lead row piles 共piles in the row furthest away from the
lateral pushing load兲, middle row piles 共middle second and third
rows in the case of 4⫻4 pile group兲 and rear row piles as well as
outer and center piles for individual pile row is also given in Fig.
2. Owing to limited number of recording channels in the strain
meters, only a maximum of five piles can be instrumented in one
test. The location of the instrumented piles for each test is also
shown in Fig. 2.

Load–Displacement Responses and Group

Efficiency
From this point onwards, all the test results are presented in pro-
totype scale unless otherwise stated. The lateral load test is termi-
nated once the pile head displacement exceeds one pile width of
840 mm. The lateral load–pile head displacement responses of a
single pile in NC and OC clay are represented by the top curve in
Figs. 3共a and b兲, respectively. The lateral pile head displacement
refers to the displacement of the pile cap at 3.5 m above the
ground surface. As expected, the pile displacement increases with
applied load and the rate of increase in displacement increases
significantly with load after a pile displacement of about 25 cm.
As the top part of the OC clay is considerably stiffer than that of
NC clay, the load–displacement response of a pile in OC clay is
significantly stiffer than that of NC clay.
To evaluate the performance of pile groups against that of a
single pile, the average load per pile, which is defined as the total
load on the pile group divided by the number of piles in the
group, is employed. Figs. 3共a and b兲 present the average load–
pile head displacement responses for the pile groups with center-
to-center pile spacing of 3D in NC and OC clay, respectively. At
relatively small pile head displacements, the displacement of the
single pile does not differ significantly with that of pile groups
under the same average lateral load. At larger pile head displace-
ments, the average pile group responses are considerably softer
than the single pile responses, and the degree of softening in-
creases with the pile group size.
Fig. 2. Plan view of pile group configuration (D⫽pile width兲 In order to examine the effect of spacing between piles in a
group, tests were also carried out on 2⫻3 and 3⫻3 pile groups
with center-to-center pile spacing of 5D. Figs. 4共a and b兲 present
the average lateral load versus pile head displacement from tests
in NC and OC clay, respectively. As the differences in the average

Fig. 3. Average lateral load–pile head displacement response for pile groups with 3D spacing (D⫽pile width兲

276 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004

 Fig. 4. Average lateral load–pile head displacement responses for pile groups with 5D spacing

lateral load/displacement responses between the single pile and Figs. 5共a and b兲 clearly show that the group efficiency for pile
the pile groups in both NC and OC clay are insignificant, inter- groups in OC clay is considerably greater than that of pile groups
action among piles for pile groups with 5D spacing is established in NC clay. This observation suggests that interaction among piles
to be significantly less than that for pile groups with 3D spacing. is more significant for pile groups in NC clay. It also appears that
This observation on piles in clay is similar to that reported on at larger pile head displacements, the magnitude of group effi-
piles in sand by McVay et al. 共1994兲, who established that group ciency remains fairly constant for pile groups with nine or more
interaction among piles reduces significantly when the center-to- piles. For NC clay, the group efficiency falls to about 45% at a
center pile spacing increases from 3 to 5 pile width. pile head displacement of 0.5D. This is in line with the finding of
To further evaluate the performance of the pile groups, the pile Rollins et al. 共1998兲 who reported that for a field case study in-
group load–displacement efficiency is employed. The group effi- volving a laterally loaded nine-pile group in clay, the group effi-
ciency, ␩, is defined as the average lateral load per pile divided by ciency falls to below 50%. For smaller pile groups, the group
lateral load on a single pile at the same pile head displacement. efficiency is considerably greater than that of larger pile groups,
Figs. 5共a and b兲 show the magnitude of group efficiency for pile being about 68% for a four-pile group in NC clay, 76% for a
groups in NC and OC clay, respectively, at various pile head two-pile group in NC clay, and 74% for a four-pile group in OC
displacements ranging from 0.1 to 0.5D. For all cases, the group clay; all cases refer to a pile head displacement of 0.5D.
efficiency decreases considerably with increasing pile head dis-
placement and number of piles in a group. Further, the group
efficiency reduces significantly when the pile spacing decreases Performance of Individual Piles
from 5 to 3D. This is in agreement with the finding of Rollins
et al. 共1998兲 who established that the pile group efficiency im- The bending moment along an instrumented pile can be deter-
proves significantly when the pile spacing increases from 3 to mined from the readings of the strain gauges installed along the
5D. pile. The strain gauge circuit setup and the calibration of the

Fig. 5. Pile group efficiency at various pile head displacements

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004 / 277

 Fig. 6. Bending moment profiles of single pile at various pile head displacements

model piles for bending moment have been described in an earlier The measured bending moment (M ) profiles are used to derive
section. Figs. 6共a and b兲 show the development of bending mo- the soil reaction ( p) profile by double differentiation and dis-
ment profile with pile head displacement for a single pile in NC placement (y) profile by double integration. The formulas are
and OC clay, respectively. It is worthy to note that the measured formulated using the elastic beam theory as follows:
bending moment at the ground level for each pile is equal to the
d2M
applied moment on the pile 共i.e., applied load multiplied by ec- p⫽ (1)
centricity of load above ground level兲, thus verifying the accuracy dz 2

冕冕
of the strain gauge responses. The depth to the maximum bending M
moment is observed to be at around 7 – 8D below the ground y⫽ dzdz (2)
EI
surface. The bending moment increases with applied load in all
cases. At the same pile head displacement, the pile in OC clay where z⫽depth below ground surface. The bending moment pro-
resists higher bending moment than that of the pile in NC clay. files are analyzed using curve fitting involving a seventh order
This is as expected as the pile in NC clay would resist consider- polynomial. Two boundary conditions are required to solve Eq.
ably less load as compared to the pile in OC clay at the same pile 共2兲. The first condition is the measured pile head displacement
head displacement, as illustrated in Fig. 3. and the second is the zero pile displacement elevation to be de-
The development of bending moment for piles in a group fol- termined from the soil reaction profile. Fig. 8共a兲 shows the soil
lows a similar trend as that for a single pile as the pile bending reaction profile of the lead and rear piles for a two-pile group in
moment increases with applied load. However, there are consid- NC clay at a pile head displacement of 0.5D derived using Eq.
erable differences in the bending moment profiles among indi- 共1兲. Fig. 8共b兲 presents the pile deflection profiles of the lead pile
vidual piles in a group. The bending moment profiles for the lead of the two-pile group at various pile head displacements derived
and rear piles of the two- and four-pile groups with center-to-
center pile spacing of 3D in NC clay at pile head displacement of
0.5D are shown in Figs. 7共a and b兲, respectively. For the two-pile
group, the maximum bending moment of the lead pile is slightly
larger 共about 5%兲 than that of the rear pile. However, for the
four-pile group, the maximum bending moment of the lead pile is
considerably larger 共about 15%兲 than that of the rear pile. This
observation is contrary to that assumed in the elasticity interaction
factor approach for laterally loaded pile group proposed by Pou-
los and Davis 共1980兲, whereby all piles in two- and four-pile
groups with symmetrical plan configuration are expected to expe-
rience identical bending moment and deflection profiles. The ten-
dency for a rear pile to exhibit less lateral resistance due to a pile
in front of it is commonly termed as the ‘‘shadowing’’ effect; see,
for example, Brown et al. 共1988兲. In the present study, the shad-
owing effect is noted to increase when the group size increases
Fig. 7. Bending moment profile at pile head displacement of 0.5D
from 2 to 4 piles.

278 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004

 Fig. 10. Bending moment profile of individual piles at pile head
Fig. 8. Soil reaction and pile deflection profiles of two-pile group in displacement of 0.5D
NC clay

is 200 kN, and 165 kN for the 2 and 2⫻2 pile groups, which are
using Eq. 共2兲. Fig. 8共b兲 reveals that the point of rotation for the 87 and 72%, respectively, of the lateral load of a single pile. For
pile is close to the pile tip at a depth of about 12.5 m below the the two-pile group, the lead pile experiences 96% of the load of a
ground surface. single pile while the rear pile only experiences 78% of that of a
Upon application of lateral load on the pile cap at 3.5 m above single pile at the same pile head displacement. For the 2⫻2 pile
the ground level, the relatively rigid pile cap is observed to dis- group, assuming the left and right piles behave identically, each
place and tilt slightly as a rigid body, as revealed by the video of the front row piles experiences only 80% of the load of a single
camera during the centrifuge tests. It is believed that this causes pile, while each of the rear row pile only experiences 63% of that
the rotation of the piles at the pile head as illustrated in Fig. 8共b兲. of a single pile. This analysis further reinforces the shadowing
To further evaluate the performance of individual piles in a effect phenomenon reported earlier, and the shadowing effect
group, the lateral load taken by a pile in the group can be deduced increases considerably when the pile group size increases from 2
by integrating the corresponding soil reaction response profile to 4.
关see, for example, Fig. 8共a兲 for a two-pile group in NC clay兴. The Fig. 10共a兲 shows the bending moment profiles of the five in-
load–displacement response of the lead and rear piles for the 2 strumented piles 关see Fig. 2共d兲 for location of the instrumented
and 2⫻2 pile groups in NC clay are, hence, determined and piles兴 of a 3⫻3 pile group 共pile spacing 3D) in NC clay at a pile
plotted in Figs. 9共a and b兲, respectively. The average applied load head displacement of 0.5D. The outer lead pile again resists the
per pile for the pile groups, as determined by the applied load on largest maximum bending moment among piles in the group. In
the pile cap divided by the number of piles in the group, are also contrast, the outer rear pile 共which is supposed to experience the
shown in Figs. 9共a and b兲 for comparison purpose. It is noted that same bending moment as the outer lead pile due to geometrical
the sum of individual pile loads is reasonably close to the total symmetry according to interaction factor approach兲 exhibits about
applied load for both pile groups verifying the accuracy of the 27% less maximum bending moment than the outer lead pile. The
measured data. At a pile head displacement of 0.5D, the loads difference in the magnitude of the maximum bending moment is
carried by the lead piles are 22 and 28% larger than that by the larger than that of 2 and 2⫻2 pile groups. As the outer rear pile
rear piles for the 2 and 2⫻2 pile groups, respectively. Fig. 3共a兲 is located two rows behind the outer lead pile, the shadowing
shows that the corresponding load on a single pile is 230 kN. effect becomes even more significant. The center middle pile ex-
Thus, the average lateral load per pile at 0.5D pile displacement periences the least maximum bending moment, being 46% less

Fig. 9. Load–displacement responses for lead and rear piles

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004 / 279

 Fig. 11. Load–displacement responses of individual piles

than that of the outer lead pile and 22% less than that of the outer middle second row pile. Hence, it can be deduced that for both
pile in the same row. In many reported studies on laterally loaded middle second and middle third row piles, the maximum bending
pile groups 共see, for example, Rollins et al. 1998兲, the average moment experienced by the outer piles is considerably greater
performance of an individual row of piles in a group is usually than that by the center piles. This is similar to that observed for
adopted for simplicity in the analysis. For the 3⫻3 pile group in the 3⫻3 pile group and further illustrates that the simplification
the present study, such simplification is acceptable for the rear of all piles in the same row behave similarly is not appropriate for
row pile where the outer and center piles experience similar maxi- the middle pile rows.
mum bending moment magnitudes. However, the simplification is The distribution of loads among the five instrumented piles in
erroneous for the middle row piles as the center middle pile ex- the 4⫻4 pile group are determined and shown in Fig. 11共b兲. The
periences 22% less maximum bending moment as compared to outer lead pile carries the largest lateral load, followed in de-
that of the outer middle pile. This matter will be further evaluated scending order by the center lead pile, the outer middle second
using the p-multiplier concept later. row pile, the center rear pile and the center middle third row pile.
The load carried by each of the five instrumented piles in the In addition, very large variations in the load distribution among
3⫻3 pile group can be determined using the same approach piles are noted. Thus, for large pile groups, the effect of shadow-
adopted for the 2 and 2⫻2 pile groups and the results are shown ing of front piles over rear piles is more significant as there are
in Fig. 11共a兲. As expected, the outer lead pile exhibits the largest more rows of piles and pile–soil–pile interaction among piles
lateral load as compared to other piles within the group. The load also increase significantly due to larger number of piles.
carried by the outer middle and center rear piles lies in between To further evaluate the performance of single pile and pile
that by the outer lead and outer rear piles. The center middle pile groups of different sizes, Fig. 12共a兲 compares the maximum bend-
carries the least load. Comparing Fig. 11共a兲 with Fig. 7 for the 2 ing moments in the outer lead piles from all tests in NC clay. It is
and 2⫻2 pile groups, the larger difference in the magnitude of evident that the bending moment in the outer lead pile decreases
load carried by the piles in the 3⫻3 pile group reveals that both as the number of piles in a group increases. This provides further
shadowing and pile–soil–pile interaction effects increase with the evidence that the shadowing and pile–soil–pile interaction effects
number of piles in a group. increase with increasing number of piles in a group.
The bending moment profiles of the five instrumented piles The behavior of pile groups in OC clay essentially follows a
关see Fig. 2共e兲 for location of the instrumented piles兴 for the similar trend as that of pile groups in NC clay in terms of bending
4⫻4 pile group 共pile spacing 3D) in NC clay at a pile displace- moment profiles and load distribution among piles in a group.
ment of 0.5D are shown in Fig. 10共b兲. As before, the outer lead However, the pile–soil–pile interaction appears to be less signifi-
pile also exhibits the largest bending moment among all piles in cant than that in NC clay. This appears to suggest that group
the group. The variation in the pile bending moment is significant interaction decreases with increase in strength of the top soil.
as the center lead pile resists almost 100% higher maximum Owing to limitations of space, only the maximum bending
bending moment than that of center rear pile, although the two moment-pile head displacement responses of the outer lead pile
piles are expected to experience the same bending moment ac- for the pile groups in OC clay are presented here. The results
cording to the interaction factor approach. This can be attributed shown in Fig. 12共b兲 also reveal that a pile group with larger
to the center rear pile being located three rows behind the center number of piles would experience a smaller bending moment at
lead pile resulting in a more significant shadowing effect as com- the same pile head displacement as the load carried by the outer
pared to smaller pile groups. Among all piles, the center middle lead pile decreases with increasing number of piles in a group.
third row pile experiences the least bending moment. On the other
hand, the outer second row pile experiences a maximum bending
moment of 480 kN. As the shadowing effect does not affect the Experimental p -Multiplier for Piles in NC Clay
trailing row piles as significantly as the lead row piles, the outer
middle third row pile would, hence, experience only slightly less To account for the pile group effect, Brown et al. 共1987兲 em-
maximum bending moment than that of the outer middle second ployed the concept of ‘‘p-multiplier’’ to correlate the
row pile. Similarly, the center middle third row pile would, expe- p(lateral soil reaction) – y(lateral displacement) curve for an indi-
rience slightly less maximum bending moment than that of center vidual pile in a group to that of a single pile, see Fig. 13. The

280 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004

 Fig. 12. Maximum bending moment 共BM兲–pile head displacement responses of outer lead pile

experimental p-multipliers in the present study can thus be deter- mostly the lead row piles and that it is considerably less signifi-
mined by comparing the lateral soil resistance between an indi- cant in all other trailing pile rows. This finding is in reasonably
vidual pile in a group and a single pile under the same pile head good agreement with that reported by Rollins et al. 共1998兲 on
displacement. As an illustrative example, the distribution of pile laterally loaded pile group in clay in the field. However, the find-
load and the experimental p-multiplier for each individual pile ing differs from that on piles in sand as Brown et al. 共1988兲 found
obtained for the 3⫻3 pile group in NC clay with 3D pile spacing that the difference in the average load carried by each pile row is
at pile head displacement of 0.5D are shown in Fig. 14. In this distinguishable.
test, only five piles were instrumented 关see Fig. 2共d兲兴 to physi- A further examination of individual pile performance in Fig.
cally monitor the lateral load carried by each of the five piles. The 14 reveals that the difference between the load carried by the
load carried by the other three outer piles on the right-hand side
of the pile group may be taken to be the same as those of the outer
piles on the left-hand side due to symmetry. The load carried by
the center lead pile can hence be back-calculated as the total
lateral load on the pile group is known. The average p-multiplier
for each of the three pile rows is also given in Fig. 14. The results
essentially confirm the earlier observation that the average load of
the lead row piles is 30% higher than that of middle row piles. As
the average load of the middle row piles is only 4% higher than
that of the rear row piles, it can be deduced that shadowing affects

Fig. 14. Load distribution of piles and p-multiplier for 3⫻3 pile
group in normally consolidated clay 共pile head displacement
Fig. 13. P-multiplier concept ⫽0.5D)

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004 / 281

Table 1. Comparison of p-Multiplier Values from Various Experimental and Field Studies 共All Pile Groups with Pile Center-to-Center Spacing
of 3 Pile Widths兲
Average p-multiplier
Size of
Author/soil type and shear strength pile group Lead row Second row Third row Fourth row
Clay
Present study/normally consolidated clay: undrained shear strength⫽0–20 kPa 2⫻1 0.80 0.63
2⫻2 0.96 0.78
3⫻3 0.65 0.50 0.48
4⫻4 0.65 0.49 0.42 0.46
Brown et al. 共1987兲/overconsolidated clay: strength⫽70– 180 kPa 3⫻3 0.7 0.5 0.4
Meimom et al. 共1986兲/silty clay: strength⫽25 kPa 2⫻2 0.9 0.5
Rollins et al. 共1998兲/clayey silt: strength⫽50– 75 kPa 3⫻3 0.6 0.4 0.4
Sand
Brown et al. 共1988兲/clean medium sand: 3⫻3 0.8 0.4 0.3
friction angle ␾⫽38°
McVay et al. 共1995兲/medium dense sand 3⫻3 0.8 0.4 0.3
McVay et al. 共1998兲/medium dense sand 4⫻3 0.8 0.4 0.4 0.3
Ruesta and Townsend 共1997兲/loose find sand: ␾⫽32° 4⫻4 0.8 0.7 0.3 0.3

center and the outer piles for both the lead and rear row is rather 1. The average lateral load per pile decreases with increasing
small. On the other hand, the outer middle pile carries 51% more number of piles in the group. The reduction in pile group
load than the center middle pile. This is also reflected by the large efficiency is less severe for piles installed in OC clay than
difference in the maximum bending moment experienced by the that of piles installed in NC clay.
two piles 关Fig. 10共a兲兴. Although the average load carried by the 2. Group efficiency for pile groups with a center-to-center pile
middle and rear row piles is similar, appropriate caution should be spacing of 3D decreases as the number of piles in a group
taken when designing piles in the middle row in view of the large increases for both NC and OC clay. However, when the
difference in the load and bending moment between the outer and center-to-center pile spacing increases to 5D, the group in-
center piles. This is particularly so for the case of 4⫻4 pile teraction effect becomes insignificant.
groups as there would be two center piles having very small bend- 3. In line with the findings of Brown et al. 共1988兲 and Rollins
ing moment 关Fig. 10共b兲兴 and carry an estimated 60% less load et al. 共1998兲, the shadowing effect of lead piles over trailing
than that of the two outer piles 关Fig. 11共b兲兴 in the same row. piles is observed and such effect increases with increasing
Brown and Shie 共1991兲 compared the average p-multiplier of number of piles in a group. This results in a higher lateral
the lead, middle, and rear row piles for pile groups in sand and load for the lead row piles as compared to that on the trailing
clay from several field studies. Their comparison is further ex- piles. The shadowing effect is most significant for the lead
tended to include the average experimental p-multiplier obtained row piles and less significant on subsequent rows of trailing
from the present study and shown in Table 1. For the 4⫻4 pile piles. Hence, if uniform pile type and size have been adopted
group in the present study, owing to limited number of instru- for a pile group, appropriate checks should be conducted to
mented piles, simplifications have been made by assuming that evaluate whether the lateral load carrying and bending mo-
the load carried by the outer middle second pile and outer middle ment capacity of the lead row piles, in particular the outer
third pile as well as the center middle second pile and center lead piles, would be exceeded.
middle third pile are the same. Such simplifications are consid- 4. In many previous research studies, the average performance
ered reasonable as the shadowing effect of the second row pile of piles in the same row rather than the performance of in-
over the third row pile is relatively insignificant. The results of dividual piles is employed in the analysis. This simplification
Rollins et al. 共1998兲 on laterally loaded pile groups in clay and is found to be acceptable for the lead and rear row piles
those of Ruesta and Townsend 共1997兲 and McVay et al. 共1995, where the behavior of the outer and center piles do not differ
1998兲 on pile groups in sand are also included in Table 1 for much. However, for the middle pile rows 共middle row in the
comparison. Considering the variation in soil properties in various case of the 3⫻3 pile group and middle second and third
field and laboratory studies, the experimental p-multipliers ob- rows in the case of the 4⫻4 pile group兲, the present study
served in the present study are reasonably consistent with those reveals that the center pile共s兲 often carries much less load
and bending moment than those of the outer piles in the
obtained from the early studies.
same row. Thus, appropriate caution should be taken when
designing piles in the middle row共s兲 of a large pile group in
view of the large difference in the load and bending moment
Conclusions between the outer and center piles.
Centrifuge model tests have been performed to investigate the
behavior of laterally pile groups in normally consolidated and Acknowledgments
overconsolidated kaolin clay. The tests involved single piles and
pile groups with the number of piles ranging from 2 to 16 with The present study is part of the research collaboration between
center-to-center pile spacing of 3 or 5D. The following conclu- the National University of Singapore and University of Indonesia.
sions can be obtained from the present study: The writers would like to express their gratitude to the technical

282 / JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004

staff of the Geotechnical Centrifuge Laboratory at National Uni- tests of the lateral load behavior of pile groups in soft clay.’’ Proc.,
versity of Singapore for their assistance and contributions in con- 12th Annual Offshore Tech. Conf., OTC, Houston, 163–174.
ducting the centrifuge model tests. The assistance of R.-F. Shen in McVay, M., Bloomquist, D., Vanderlaine, D., and Clausen, J. 共1994兲.
preparing some of the figures is gratefully appreciated. ‘‘Centrifuge modeling of laterally loaded pile group in sand.’’ Geo-
tech. Test. J., 17共2兲, 29–137.
McVay, M., Casper, R., and Shang, T. 共1995兲. ‘‘Lateral response of three-
row groups in loose to dense sands at 3D and 5D pile spacing.’’ J.
References
Geotech. Eng., 121共5兲, 436 – 441.
McVay, M., Shang, T., and Casper, R. 共1996兲. ‘‘Centrifuge testing of fixed
Brown, D. A., Resse, L. C., and O’Neill, M. W. 共1987兲. ‘‘Cyclic lateral
head laterally loaded battered and plumb pile groups in sand.’’ Geo-
loading of a large-scale pile group.’’ J. Geotech. Eng., 113共11兲, 1326 –
tech. Test. J., 19共1兲, 41–50.
1343.
McVay, M., Zhang, L., Molnit, T., and Lai, P. 共1998兲. ‘‘Centrifuge testing
Brown, D. A., Morrison, C., and Reese, L. C. 共1988兲. ‘‘Lateral load
of large laterally loaded pile groups in sands.’’ J. Geotech. Geoenvi-
behavior of pile group in sand.’’ J. Geotech. Eng., 114共11兲, 1261–
ron. Eng., 124共10兲, 1019–1026.
1276.
Brown, D. A., and Shie, C. F. 共1991兲. ‘‘Modifications of p – y curves to Meimon, Y., Baguelin, F., and Jezequel, J. F. 共1986兲. ‘‘Pile group behav-
account for group effects on laterally loaded piles.’’ Proc., ASCE Geo- ior under long time lateral monotonic and cyclic loading.’’ Proc., 3rd
technical Congress 1991, ASCE, New York, 1, 479– 490. Int. Conf. on Numerical Methods in Offshore Piling, Inst. Francais Du
Craig, W. H. 共1985兲. ‘‘Modeling pile installation in centrifuge experi- Petrole, Nantes, France, 286 –302.
ments.’’ Proc., 11th Int. Conf. on Soil Mechanics and Foundation Poulos, H. G., and Davis, E. H. 共1980兲. Pile foundation analysis and
Engineering, San Francisco, Calif., 2, 1101–1104. design, Wiley, New York.
Hamilton, J. M., Dunnavant, T. W., Murff, J. D., and Philips, R. 共1991兲. Remaud, D., Garnier, J., and Frank, R. 共1998兲. ‘‘Laterally loaded piles in
‘‘Centrifuge study of laterally loaded pile behavior in clay.’’ Proc., Int. dense sand: Group effects.’’ Proc., Int. Conf. Centrifuge 98, T.
Conf. Centrifuge 91, H. Y. Ko and F. McLean, eds., Balkema, Rotter- Kimura, O. Kusakabe, and J. Takemura, eds., Balkema, Rotterdam,
dam, The Netherlands, 285–292. The Netherlands, 533–538.
Kitazume, M., and Miyajima, S. 共1994兲. ‘‘Lateral resistance of a long pile Rollins, K. M., Peterson, K. T., and Weaver, T. J. 共1998兲. ‘‘Lateral load
in soft clay.’’ Proc., Int. Conf. Centrifuge 94, C. F. Leung, F. H. Lee, behavior of full-scale pile group in clay.’’ J. Geotech. Geoenviron.
and T. S. Tan, eds., Balkema, Rotterdam, The Netherlands, 485– 490. Eng., 124共6兲, 468 – 478.
Lee, F. H., Tan, T. S., Leung, C. F., Yong, K. Y., Karunaratne, G. P., and Ruesta, P. F., and Townsend, F. C. 共1997兲. ‘‘Evaluation of laterally loaded
Lee, S. L. 共1991兲. ‘‘Development of geotechnical centrifuge facility at pile group at Roosevelt Bridge.’’ J. Geotech. Geoenviron. Eng.,
the National University of Singapore.’’ Proc., Int. Conf. Centrifuge 123共12兲, 1153–1161.
91, H. Y. Ko and F. McLean, eds., Balkema, Rotterdam, The Nether- Stewart, D. P., and Randolph, M. F. 共1991兲. ‘‘A new site investigation tool
lands, 11–17. for the centrifuge.’’ Proc., Int. Conf. Centrifuge 91, H. Y. Ko and F.
Matlock, H., Ingram, W. B., Kelley, A. E., and Bogard, D. 共1980兲. ‘‘Field McLean, eds., Balkema, Rotterdam, The Netherlands, 531–538.

JOURNAL OF GEOTECHNICAL AND GEOENVIRONMENTAL ENGINEERING © ASCE / MARCH 2004 / 283

View publication stats