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

net/publication/226863579

Laboratory modeling of laterally loaded pile groups in sand

Article  in  KSCE Journal of Civil Engineering · August 2011

DOI: 10.1007/s12205-011-0924-3

CITATIONS READS

18 2,611

2 authors:

Byung-Tak Kim Gil-Lim Yoon

GS Construction & Engineering Korean Institute of Ocean Science and Technology
17 PUBLICATIONS   594 CITATIONS    68 PUBLICATIONS   702 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reliability based design of offshore wind turbine foundation View project

Research of the behavior of Laterally Loaded Pile View project

All content following this page was uploaded by Gil-Lim Yoon on 04 March 2014.

The user has requested enhancement of the downloaded file.

KSCE Journal of Civil Engineering (2011) 15(1):65-75 Geotechnical Engineering
DOI 10.1007/s12205-011-0924-3
www.springer.com/12205

Laboratory Modeling of Laterally Loaded Pile Groups in Sand

Byung Tak Kim* and Gil Lim Yoon**
Received July 7, 2009/Accepted March 22, 2010

···································································································································································································································

Abstract
Extensive laboratory model pile tests were performed to investigate pile group interaction effects. Six types of configurations of
pile group with 3 to 8-diameter pile spacing were embedded into loose to medium dense sands. The pile group deflected over two
times more than the single pile under the same average load. Group effects significantly reduced load carrying capacity for all rows
relative to single pile behavior. The p-y multiplier factor depends upon the configuration of pile group and the pile spacing. The p-
multiplier factor for varying the configuration of the pile groups was from 0.2 to 0.5 at 3-diameter spacing in medium dense sand. It
was also found that the p-multiplier factor for Pile Spacing Ratio (PSR), greater than unity is higher than that for PSR smaller than
unity. Good agreement between the measured and computed pile group responses was obtained from the p-multiplier approach.
Single pile tests were also performed for comparison purposes.
Keywords: lateral load, pile group, sand, p-multiplier, group effect, model test
···································································································································································································································

1. Introduction ly, the ability to characterize both the axial and lateral behaviors
of piles is required to model rationally pile group responses
The response of a single pile subjected to lateral loads has been subjected to lateral loads.
successfully modeled by the p-y approach, such as that imple- So far, several methods have been proposed and implemented
mented in the program COM624P for silica sand. However, the to model lateral pile group response. For example, a number of
response of a laterally loaded pile group is much more computer codes make use of lateral springs based on Mindlin’s
complicated and still remains a fruitful research area. The theory (Williams, 1979). However, these methods give similar
behavior of an individual pile in group piles is controlled to a shear distributions for all the rows within the group, which
major extent by its location within a pile group and its pile head disagrees with both field data and the pile shadowing theory.
fixity condition. It is well known that the lateral resistance of a Focht and Koch (1973) proposed another method combining the
pile in a pile group is strongly influenced by the “shadowing p-y method for single piles with Poulos’s approach (1971) for
effects” as shown in Fig. 1 (Brown et al., 1988). For instance, pile groups. Based on Focht and Koch’s (1973) procedure and
both the lateral subgrade modulus for piles and the ultimate the characteristic load method by Duncan et al. (1994), Ooi and
lateral resistance within a pile group are reduced because of the Duncan (1994) developed a group amplification procedure.
overlapping of the stress zones in the surrounding soil. It is a However, they are neither able to estimate the distribution of
general conclusion from the limited field tests (Peterson et al., loads among piles within a group nor take into account the pile
1998; Townsend and Ruesta, 1997; Ng et al., 2001) and some group arrangement. Davisson (1970) proposed a group reduction
small scale model studies (Brown et al., 1988; Kong et al., 2007; factor method based on field pile group tests, in which a single
McVay et al., 1995; Pise, 1983; Randolph ,1981; Williams, subgrade modulus is used to model the whole pile group beha-
1979; Zhang et al., 1999) that, given the same pile head fixity, viors. More recently Ashour et al. (2004) proposed the assess-
the pile group will undergo significantly more displacement for a ment of the response of a laterally loaded pile group based on
given load per pile than a single isolated pile does. soil-pile interaction (Strain Wedge Model Approach). They sug-
For the fixed head pile groups, which is most common cases in gested that the interaction among the piles grows with the in-
the field, normally significant bending moments and shears crease in lateral loading, and the increasing depth and fan angle
develop at the pile heads and within the cap. The significance of of the developing wedge using the SW model results.
such restraints has been reported for both battered pile and plumb The group effects in pile groups are taken account of by
3×3 pile groups in loose sand (McVay et al., 1996). Consequent- reducing the subgrade modulus. Pise and Patra (2001) carried

*Ph.D., Senior Manager, T/K Design Team, Civil Business Division, GS Construction & Engineering, Seoul 100-722, Korea (E-mail: btkim@gsconst.co.kr)
**Member, Chief Researcher, Coastal & Harbor Engrgineering Division, Korea Ocean Research & Development Institute, Ansan 426-744, Korea (Corre-
sponding Author, E-mail: glyoon@kordi.re.kr)

− 65 −
 Byung Tak Kim and Gil Lim Yoon

Fig. 2. Side View of Layout for Static Fixed-head Test

Fig. 1. Overlapping of Failure Zones (Shadowing) and Gap For-
mation behind Piles (Brown et al., 1988)
with internal dimensions of 0.8 m in length, 0.8 m in width and
0.6 m in depth. The thickness of soil at least six times the pile
out lateral loading model (1-g) tests on fixed head model piles diameter was kept below the pile tip to minimize the influence of
and small pile groups in medium dense sand to investigate the the model box base.
ultimate lateral resistance of a pile group and group efficiency.
They concluded that for 12-diameter pile length (L/d = 12) the
isolation spacing appears to be at a spacing about six times of
pile diameter (6d) and above. Similar to Ooi and Duncan’s pro-
cedure (1994), this method is not able to calculate the distri-
bution of loads among piles in a group. Meanwhile, Brown et al.
(1988) suggested a method based on the p-multiplier concept; all
the p-y curves for a single pile within the pile group are multi-
plied by the p-multiplier Pm. Each row of piles within the pile
group is assigned a different p-multiplier Pm, which is used for all
the p-y curves along the piles within that pile row. The varied p-
multipliers will result in each row’s pile head shears to be dif-
ferent. Brown et al. (1988) and McVay et al. (1994, 1995) suc-
cessfully predicted the response of 3×3 (s/d=3.0) laterally loaded
free head pile groups using p-multipliers concepts.
Reported herein is a detailed investigation on group efficiency,
load - deflection behavior of a series of model pile group results.
Emphasis is focused on the validity of the p-y multiplier concept
due to the change in group size, pile spacing, and soil density.
Subsequently, appropriate p-multipliers for small to large plumb
pile groups in both medium sand and medium dense sand are
proposed.

2. Pile Group Model Test and Their Instrumenta-

tions

2.1 Model Pile Group Test

A side view of the model pile group test during the static fixed-
head testing is shown in Fig. 2. The piles in the group were
embedded into a sand layer with six types of pile layout with
piles spaced from three to six-pile diameters center to center. The
model tests were conducted in laboratory strongbox containers Fig. 3. Schematic Drawing of the Instrumented Pile Cap

− 66 − KSCE Journal of Civil Engineering

 Laboratory Modeling of Laterally Loaded Pile Groups in Sand

The dimensions of the model pile testing were determined by a The model pile used for the lateral loading tests was 12 mm in
dimensional analysis (Buckingham Pi theorem). There are five outside diameter, 250 mm in length, and 0.25 mm in thickness.
variables, which are displacement (y), pile diameter (D), area The model piles made of stainless steel (Ep=210 GPa) were
(A), force (F), and pile lateral stiffness (EI). In order to satisfy the instrumented at two different levels of strain gauges as shown in
similarity between the model testing and the prototype, the Fig. 3. Each model pile was calibrated to measure the bending
scaling factors are 1/N, 1/N, 1/N2, 1/N3, and 1/N4. This study was moment at a pile head, and coated with 0.5 mm thick layer of
to investigate the effects of group size, soil density, and pile epoxy for protection against interface friction with sand during
spacing in pile groups. The soil modulus was not scaled because the test. Experimental works were conducted at 1g with a scale
the stress of soil material scaled to 1/N and the strain of soil 1:34, and modeled an equivalent prototype pile with outside
material same to prototype. diameter, d, of 408 mm, bending stiffness, EIp, of 45 GNm2, and
an equivalent diameter-to-thickness ratio, d/t, of 48. Pile embed-
ment was 17.5 diameters, which is 210 mm at model scale.

2.2 Testing Procedure and Equipment

The lateral loading was applied to the pile group using 1000 N
capacity electric motors at an elevation of 49 mm above the
ground surface. The reaction frame was used to distribute the
motor forces to the strongbox container. A schematic diagram of
the three rows of piles and the connection of the 9 piles to the cap
is shown in Fig. 3. The pile arrangement of the group pile used in
the model tests is shown in Fig. 4. The loading frames consisted
of rectangular plate and a pile cap that was essentially rigid
compared with the lateral pile stiffness. Applied load was
transferred from the loading frame to each pile along a guide-
frame using a fixed connection (pile cap) attached 30 mm above
the ground surface. Transferred lateral load to each pile head is
derived from the measured bending moment at a pile head.
A lubricant was used on the surface of the guide-frame for a
low-friction mechanism in load transfer between pile and applied
load during the test. A fixed-head load test was also performed
for comparison on an isolated single pile with pile group
behavior using the same strongbox container. A load cell was
placed immediately behind the electric motor to measure the
total loads applied to the loading frame. The displacement of the
pile group during the test was measured using two Linear Variable
Differential Transducers (LVDT), which were attached to the
corner piles at the center line of the loading frame. Measured
stress and displacement data from the load cell, LVDT, and strain
gauges were simultaneously stored on a computer data acquisi-
tion system.

2.3 Descriptions of Sand Preparation

The sands for model test were sampled at the Nak-Dong River
in Daegu, Korea. Sampled sand was prepared in a dry state for
model test. Sands passing the US-sieve size of 40 were used to
minimize the particle size effects for the lateral behavior of the
pile. The grain size distribution for the Nak-Dong river sand is
shown in Fig. 5 and key characteristics of sand used are given in
Table 1. Also, X–ray diffraction test results on the Nak-Dong
River sands indicate that quartz is the most prevalent mineral in
Fig. 4. Pile Group Array Type and Pile Spacing Ratio for Model sand. To compute the relative density of sand, the limiting
Pile Group Test: (a) Pile Array Type for Model Group Pile density test (ASTM D-4253, 4254) was carried out to determine
Test, (b) Pile Spacing Ratio (PSR) and Loading Direction of both the maximum void ratio and minimum void ratio. The sand
Pile Groups for model tests was prepared at three different relative densities

Vol. 15, No. 1 / January 2011 − 67 −

 Byung Tak Kim and Gil Lim Yoon

3. Pile-Soil Interaction Models

The p-y method was used in this paper to describe single pile
behavior for predicting the laterally loaded pile group response.
It is well recognized that the p-y method models pile-soil
interaction well beyond the elastic range into nonlinear stress and
failure stress states. The p-y curves in this paper for the Nak-
Dong River sands are composite curves, which are composed of
an initial linear zone and an ultimate soil resistance. The p-y
curves in the pile groups were estimated using the p-multiplier
concept from back-analysis of the results measured in laterally
loaded pile group tests. The p-multiplier Pm was determined by
comparing with the p-y curves obtained in laterally loaded single
Fig. 5. Grain Size Distribution Curve for the Nak-Dong River Sand pile tests as shown in Fig. 6.
The p-y responses for the soil-pile interaction were obtained by
Table 1. Representative Soil Properties for the Nak-Dong River
double differentiation of the bending moment profile M(z) to
Sand
obtain the force per unit length p and by double integration of the
Soil Properties Values curvature M/EI to obtain the lateral displacement y. The
Specific gravity, Gs 2.67 nonlinear characteristics of the soil springs from single pile
Mean particle size, d50 (mm) 0.26 model tests were obtained by combining the initial stiffness of
Uniform coefficient, Cu 2.0
the p-y curve, kini, and ultimate soil-pile reaction, pu, to produce
p-y curves with depth, based on the following hyperbolic func-
Curvature coefficient, Cc 1.2
tion:
Maximum void ratio, emax 1.19
y
Minimum void ratio, emin 0.81 p = ----------------- (1)
1 y
------ + ----
40 (Dr=73%) kini pu
Peak friction angle, φpeak (deg) 35 (Dr=50%)
27 (Dr=27%) Eq. (1) was found to fit remarkably well the measured data
points from single pile model tests. In this paper, the initial
stiffness of the p-y curve, kini, based on the measured data points
of 27% (loose sand), 50% (medium sand), and 73% (medium was obtained by combining the coefficient of subgrade modulus
dense sand) to investigate the soil density effects on model pile with depth, and the ultimate soil-pile reactions, pu, are computed
group behavior. To obtain uniform model sand ground condi- from p-y curves at each depth by fitting the measured data points.
tions, a pluviation technique using a traveling spreader was The initial stiffness and ultimate soil-pile reaction for the single
employed in the model box. pile were used to calculate the p-multiplier Pm for pile group.
The model piles were installed at the center of the strongbox Detailed descriptions for the initial stiffness and ultimate soil-pile
container with several types of group pile layouts and then the reaction of the Nak-Dong River sand can be found in Kim et al.
piles were kept in a vertical state using a supporting guide frame. (2004). To predict the pile response to lateral loading, a computer
In the preparation of the model sand, the spreader container box
filled with dry sands was kept 200 mm above the current surface
of the model ground in order to obtain a uniform soil density.
To check the relative density of sand, 225 small boxes
(50×50×50 mm) having a volume of 125 cm3 boxes were placed
on the bottom of the model box. After the traveling spreader was
used, the weight of each small box was measured to compare
with the required relative density. The average error between the
measured and the required relative density was found to be ±3%
in the model tests. Using the pluviation technique, consistent unit
weights were obtained for loose sand (Dr=27%), medium sand
(Dr=50%), and medium dense sand (Dr=73%). The unit weights
were 12.54 kN/m3, 13.13 kN/m3, and 13.72 kN/m3 (average
values), respectively.

Fig. 6. Concept of p-Multiplier for Pile Group

− 68 − KSCE Journal of Civil Engineering

 Laboratory Modeling of Laterally Loaded Pile Groups in Sand

program was developed which utilizes the finite different

method and treats the pile as an elastic beam on nonlinear soil
springs. Detailed description of the computer program is given in
Kim (2000).

4. Experimental Results

A series of 50 model pile group tests was performed to

investigate the effects of soil density, type of pile group array and
pile spacing on the average lateral resistance of the group pile as
well as the individual pile row responses. The load was applied
to the single pile and group piles in an incremental manner, so
that the nonlinear load versus lateral deflection curve could be
adequately defined. A large amount of data were collected on
lateral load, lateral deflection of the pile cap, and shears in the
individual piles. This paper analyzes typical results with an
emphasis on group effects and pile group behaviors.
The ratio between the average lateral load carried by a pile in
each row and the total loads carried by a pile group is shown for
the three row piles in Fig. 7. Although the ratios are somewhat
dependent on the pile deflection, they show relatively constant
for the deflections greater than about 2 mm. For the case of
higher deflections, the front, middle, and rear piles for the center
column typically carry only 13%, 10%, and 8%, respectively, of
the total pile group loads.

4.1 Lateral Load – Deflection Responses

Fig. 8. Average Lateral Load-Deflection Response for Group Tests
The effect of soil density on pile group behavior was investi- with Varying Relative Density and Single Pile Tests (PAT-1;
gated for both square pile groups and parallel pile groups. For 3×3 pile group): (a) 3d Pile Spacing, (b) 6d Pile Spacing
medium dense sand (Dr=73%), medium sand (Dr=50%), and
loose sand (Dr=27%). The average lateral load versus average
lateral deflection curves in group piles were compared with those The difference of lateral deflection between the dense and
of single piles for the same soil condition. Fig. 8 shows the soil medium sands is larger than that between the dense and loose
density effects on lateral load–deflection response for the 3×3 sands under the same average lateral loading. At the average
pile group with a square arrangement. The higher soil density lateral load of 20 N, the deflection of the pile group is about 7.0
provided a stiffer response and a higher average lateral resistance times higher than that of the single pile for medium dense sand,
(> 80%) compared with those of loose sand. 2.1 times higher for medium sand, and 1.5 times higher for loose
sand. In the case of 6-diameter pile spacing (s/d=6.0), the
average lateral load–deflection curves for both medium sand and
loose sand are close to the lateral–deflection curve for the single
pile. For a given lateral load, heaving of the sand at ground
surface occurs around the sand of the lead row of piles since the
lead row piles carry the highest lateral load. These trends are
greater in medium dense sand than in medium and loose sand
due to the greater shadowing effect at higher density sand.
It is believed that the layout of pile groups used in the field
varies with soil condition, loading types, and pile installation
methods. In this study, six types of pile group layouts are
considered. However, due to space limitations, only three types
of pile layout are considered in the paper; and these are square,
series, and parallel arrangements that are reviewed for
Fig. 7. Variations of Load Distribution Ratio with Lateral Deflection
comparative studies. Fig. 9 shows the average lateral load –
in Medium Dense Sand (PAT-1; 3×3 Pile Group, PSR=1.0, deflection response for both group pile and single pile tests in
s/d =3.0) medium sand. For the same lateral deflection, the average lateral

Vol. 15, No. 1 / January 2011 − 69 −

 Byung Tak Kim and Gil Lim Yoon

Fig. 9. Average Lateral Load-Deflection Response for Group and Fig. 10. Group Efficiency-Lateral Deflection Relationship in Medium
Single Pile Tests with Varying Pile Group Arrays in Medium dense Sand and PSR=1.0
Sand (s/d=3.0)

resistance of a 1×3 pile group is about 25% less than that of the
single pile. Under the same lateral deflection, the average lateral
resistance for the 2×3 pile group and the 3×3 pile group is about
27% and 39% less, respectively, than that of the 3×1 pile group.

4.2 Group Efficiency

Group efficiency is important for estimating the pile group
resistance from the result of single pile resistance. Group effici-
ency, h, at a given deflection is expressed here in terms of ultimate
lateral resistance of a pile group to that of a single pile as :
Q
η = -----------------
LG
- (2)
n1 n2 QLS
Fig. 11. Group Efficiency-Pile Spacing Relationship for PAT-1: 3×3
where, QLG = lateral resistance of the pile group; QLS = lateral Pile Group and PSR=1.0
resistance of a single pile; n1 = number of rows in a pile group;
and n2 = number of columns in a pile group.
McVay et al. (1994) conducted centrifuge tests on a free-head
single pile and a 3×3 pile group with 3d and 5d pile spacing.
They showed that group efficiency is not greatly dependent upon
soil density. Recently, Wakai et al. (1999) carried out the 1-g
model tests on free and fixed-head pile groups (3×3 pile group)
consisting of a nine pile group (s/d=2.5) subjected to lateral
loading, which showed that group efficiency was 0.45-0.7 when
the deflection reached 0.1d.
It may be noted that group efficiency based on the ultimate
lateral resistance will be higher than that based on the lateral
resistance at a given deflection. This is because group efficiency
is dependent upon the deflection of the pile. Therefore, the larger
the deflection of the pile the higher the group efficiency. Fig. 10
Fig. 12. Group Efficiency-Pile Spacing Relationship for Series Pile
shows the group efficiency–lateral deflection relationship
Group in Medium Dense Sand and PSR=1.0
developed in this paper for the conditions of different pile
spacing and different pile group arrangement.
Group efficiency, in this paper, was determined when the pile group, the group efficiencies are about 0.4-0.7 for the
lateral deflection reached the reference deflection of 2.4 mm medium dense sand, 0.5-1.04 for the medium sand, 0.66-1.08 for
(0.2d). Figs. 11-13 show that the measured group efficiency the loose sand, and for the case of the medium dense sand and
increases with an increase in pile spacing. For the case of a 3×3 2×2 pile group, the group efficiency is about 0.52-1.00; the lower

− 70 − KSCE Journal of Civil Engineering

 Laboratory Modeling of Laterally Loaded Pile Groups in Sand

Fig. 13.Group Efficiency-Pile Spacing Relationship with Various Fig. 14.Effects of the Number of Pile in Group on Group Efficiency
Group Pile Arrangements in Medium Sand and PSR=1.0 for PSR=1.0

values are associated with 3d spacing and the higher values with
a 6d spacing.
Oteo (1972) reported on aluminum pile (3×3 pile groups) test
results in medium sand under 1-g model tests. The load was
applied at a height of 50 mm (0.25 L) above the ground line. The
measured group efficiency in this study for 3×3 pile groups at 6d
spacing is about 48% higher than the reported efficiency, which
is determined at deflection of 0.2d, by Oteo. However, the
measured group efficiency at 3d spacing is in good agreement
with that of Oteo.
For the rectangular pile groups, the group efficiencies are
about 0.53-0.87 for the 3×1 pile group, 0.40-0.73 for the 3×2 pile
group, 0.75-1.10 for the 1×3 pile group, and 0.58-0.96 for the
2×3 pile group; the lower values are associated with 3d spacing Fig. 15. Effects of the Pile Spacing Ratio on Group Efficiency in
and the higher values with 6d spacing. The group efficiencies for Medium Sand and PAT-1: 3×3 Pile Group
the parallel type of pile group are about 26%-45% higher than
those for the series type of pile group. This is possibly due to the shows that the group efficiency of most pile group arrangements
fact that the shadowing effect in the direction of the loading axis at 6d spacing is close to 1.0 or beyond. As expected in Fig. 14,
is higher than that in perpendicular direction of loading axis. the group efficiency decreases with an increase in the number of
Williams (1979) reported the results of model tests on pile piles in the group, and the trend is dependent on the pile spacing
groups of two piles and three piles in a line along the direction of and number of piles in the group. The decrease in the group
loading for pile spacing from 6d to 12d spacing. The measured efficiency with an increase in the number of piles in the group for
group efficiency in this study for 3×1 pile groups at 6d spacing is medium dense sand is higher than that for the medium sand. In
about 18% higher than the reported efficiency, which is particular, group efficiency for both medium sand and 6d spacing
determined to ultimate lateral capacity, by Williams. This may be is independent of the number of piles in the group.
due to the free rotation allowed to the pile heads for the line pile The variation of group efficiency with varying pile spacing
groups by Williams. ratios in medium sand and for a 3×3 pile group is shown in Fig.
Very recently, Pise and Patra (2001) reported the group 15. Group efficiency, in which the pile spacing, s2, increases with
efficiency for 2×1, 3×1, 2×2, 3×2 pile groups with L/d=12. They constant s1 pile spacing, i.e. PSR (=s2/s1) ≥ 1.0, tends to linearly
showed that, for L/d=12, the group efficiency is about 0.9-1.2 for increase. However, group efficiency, in which the pile spacing,
the 3×1 pile group and 0.75-1.0 for the 3×2 pile group. The s1, increases with constant s2 pile spacing , i.e. PSR≤1.0, tends to
group efficiencies, which is determined to ultimate lateral nonlinearly increase and becomes constant at about 0.6 below
capacity, obtained from their tests are higher (about 37%-88%) PSR of 0.75.
in comparison to those determined in this study for the pile
spacings from 3d to 6d. 5. P-multipliers from Model Tests
Fig. 13 shows the group efficiency–pile spacing relationship
for all the pile groups used in this study in medium sand, which It is known that for a given deflection in the pile group, the lead

Vol. 15, No. 1 / January 2011 − 71 −

 Byung Tak Kim and Gil Lim Yoon

row piles carry the greatest load whereas the trailing row piles
carry lower load. The tendency for a pile in a trailing row to exhibit
less lateral resistance is due to “shadowing”. This shadowing
effect becomes less significant as the spacing between piles
increases and is relatively unimportant for a spacing greater than
about 6d spacing based on model tests (Cox et al., 1984).
The effect of soil density and pile spacing on the ratio of total
lateral load taken by the lead row piles is summarized in Table 2
for PAT-1 (3×3) and PAT-2P (2×3) pile groups with PSR=1.0. In
addition, the ratio of total lateral load taken by the lead row piles
for varying PSR is summarized in Table 3 for dense and medium
sands. McVay et al. (1994) determined that the magnitude of
shear force carried by an individual row pile is a function of soil
density and pile spacing. However, with regard to Nak-Dong Fig. 16. Variations of Load Distribution Ratio of Each Pile Row with
River sand, the ratio of total lateral load taken by the lead row PSR in the Center Column Pile in Dense Sand and PAT-1:
piles is independent of the pile spacing at the higher density. 3×3 Pile Group
Table 2 indicates that the higher the soil density, the lower the
ratio of total lateral load taken by the lead row piles at the same clays, stiff clays, and sands, and widely incorporated in computer
pile spacing. In addition, the closer the pile spacing, the higher models (Matlock, 1970; Reese et al., 1974, 1975). For closely
the ratio of the total lateral load taken by the lead row piles at the spaced piles, Brown et al. (1988) proposed that the p-y curves for
same soil density. individual piles in a group are obtained using p-multipliers (Pm)
Fig. 16 presents the effects of the pile spacing ratio (PSR=s2/s1) to reduce all the p values on a single pile p-y curve, as shown in
on the percentage of total lateral load taken by the center column Fig. 6. With this approach, it is possible to reduce the computed
pile for each row pile. The effect of PSR on the lead row pile is load-carrying capacity of each pile in the group relative to the
more uniform than on the middle and trail row piles. However, load-carrying capacity of the single pile as observed in pile load
the pattern of percentage variation of total lateral loads taken by test results. In this study, p-multiplier values for a pile group
the middle or trailing row piles for PSR >1.0 is different from were evaluated from the back-analysis based on the measured
that of the lead row pile. The percentage of these rows increases data using the p-y model suggested by Kim et al. (2004) for
with PSR (s1) presumably due to group effects becoming less single pile. The parameters of p-y model, kini and pu, for fixed
important as pile spacing s2 increases and the overlap of the head single pile were obtained from model tests (Kim et al.,
failure zone decreases. In addition, the minimum percentage of 2004) and these values are used to evaluate the p-multipliers.
total lateral loads taken occurred at a PSR value of unity. Based on the foregoing observations, sets of multiplier factors
Laterally loaded pile response is typically analyzed by the for 3×1~3 and 1×3 pile groups are presented in Table 4. The
finite difference method of the pile along with a nonlinear spring multiplier factors presented in Table 4 are the values for five
model to represent the resistance provided by the surrounding individual piles in a pile group. The factor for the lead row is
soil. The load-deflection responses for the soil are known as p-y independent of the type of pile group layout and pile spacing.
curves, where p is the lateral soil resistance and y is the lateral The factors for the middle and the trail rows of the pile groups
deflection. Generic p-y curves have been developed for soft are, however, highly dependent on the type of layout and pile
spacing. This trend was observed in medium sand. In this study,
Table 2. Load Distribution Ratio of the Lead Row Pile for PSR=1.0 the proposed Pm values for the other conditions are average
PAT-1 (3×3 pile group) PAT-2P (2×3 pile group) values for the pile groups because the shear distribution for each
Dr (%) Pile Spacing row of piles was measured in model testing.
The effect of pile group spacing and number of piles in the group
3.0d 4.0 d 6.0 d 3.0 d 4.0 d 6.0 d
on the average Pm value for the medium dense and medium sands
73 0.42 0.42 0.40 0.62 0.60 0.57 is shown in Figs. 17 and 18 for the different pile group layouts. Fig.
50 0.47 0.42 0.35 0.65 0.59 0.52 17 highlights that the Pm value increases remarkably increasing
27 0.52 0.44 0.34 - - - pile spacing from 3-diameter to 6-diameter, and then approaches
unity for larger than 6-diameter pile spacing. The Pm values for
Table 3. Load Distribution Ratio of the Lead Row Pile for Varying different pile groups under 3-diameter spacing were estimated
PSR from 0.3 to 0.55 in medium dense sands and 0.6 to 0.9 in medium
PSR 0.50 0.75 1.0 1.33 2.0 sands, respectively. These results are very close to the average Pm
values of about 0.5, proposed by McVay et al. (1998), for 3×3 pile
Medium dense sand 0.45 0.43 0.42 0.41 0.42
group in medium dense sand.
Medium sand 0.48 0.47 0.45 0.48 0.49 As shown in Fig. 18, the Pm value for medium dense sand

− 72 − KSCE Journal of Civil Engineering

 Laboratory Modeling of Laterally Loaded Pile Groups in Sand

Table 4. P-multiplier Factors for Laterally Loaded Pile Groups

Dr Group Average P-multiplier Factors

s/d
(%) Layout Pm P1 P2 P3 P4 P5
73 3.0 3×1 0.55 - 0.85 - 0.45 0.25
73 4.0 3×1 0.7 - 0.85 - 0.60 0.45
73 6.0 3×1 0.9 - 0.95 - 0.82 0.80
73 3.0 3×2 0.50 0.85 0.88 0.28 0.42 0.28
73 4.0 3×2 0.60 0.85 0.80 0.35 0.50 0.35
73 6.0 3×2 0.85 0.90 0.93 0.80 0.70 0.70
73 3.0 3×3 0.30 0.68 0.68 0.66 0.24 0.20
73 4.0 3×3 0.40 0.75 0.80 0.85 0.28 0.25
73 6.0 3×3 0.60 0.89 0.85 0.88 0.35 0.30
50 3.0 1×3 0.70 0.72 0.65 0.74 - -
50 4.0 1×3 0.80 0.84 0.73 0.84 - -
50 6.0 1×3 0.95 0.99 0.89 0.97 - -
50 3.0 3×3 0.50 0.70 0.67 0.71 0.35 0.30
50 4.0 3×3 0.70 0.80 0.80 0.84 0.45 0.32
50 6.0 3×3 0.93 0.92 0.91 0.90 0.73 0.65
50 PSR=1.33 3×3 0.55 0.70 0.72 0.75 0.40 0.33
50 PSR=2.0 3×3 0.60 0.70 0.80 0.80 0.50 0.30

Fig. 17.Pm – s/d Relationship for Each Relative Density (Case of

PSR = 1.0): (a) Medium dense Sand, (b) Medium Sand

decreases remarkably with an increase in the number of piles in

the group and the decreasing rate of Pm value becomes constant
for more than 9 piles. In a group for medium sand, however, the
Pm value is almost constant for the pile group with 6-diameter
pile spacing.
Empirical equations are proposed here to evaluate the average
Pm value based on extensive lab model test results. The proposed
formula as a function of the number of piles in the group n and
the pile group spacing s/d for PSR=1.0 are:
–0.4 –0.24 ( s ⁄ d )
0.7 Fig. 18.Pm – n Relationship for Both Different Soil Density and Pile
Pm(avg) = 2.0 ( s ⁄ d ) n for medium dense sand Spacing (PSR = 1.0)
(3)
0.36 0.57 – 0.56 ln ( s ⁄ d )
Pm(avg) = 0.70 ( s ⁄ d ) n for medium sand (4) 6. Conclusions

Fig. 19 shows the relationship between the average Pm value A series of laterally loaded laboratory model pile group tests
and the Pile Spacing Ratio (PSR) for a 3×3 pile group in a were carried out to investigate the pile group effects in from
medium sand. The Pm value with PSR=1.0 is the smallest, but the loose to medium dense sand. The following conclusions are
values slightly increase with increasing pile spacing s2 or s1 with drawn from the experimental works.
pile spacing s1 or s2 hold constant. • A pile spacing of more than six times the pile diameter in group
seems to be large enough to eliminate the group effects of the

Vol. 15, No. 1 / January 2011 − 73 −

 Byung Tak Kim and Gil Lim Yoon

Davisson, M. T. (1970). Lateral load capacity of piles. HRR 333,

Transportation Research Board, National Research Council;
Washington DC, pp. 104-112.
Duncan, J. M., Evans, L. T., and Ooi, P. S. K. (1994). “Lateral load
analysis of single piles and drilled shafts.” Journal of Geotechnical
and Geoenvironmental Engineering, ASCE, Vol. 120, No. 6, pp.
1018-1033.
Focht, J. A. and Koch, K. J. (1973). “Rational analysis of the lateral
performance of offshore pile groups.” In Proceeding of the 5th
Offshore Technology Conference, Dallas, pp. 701-708.
Kim, B. T. (2000). Behavior of laterally loaded pile in Nonhomo-
geneous sand soil, PhD Thesis, Kyungpook National University,
Korea.
Kim, B. T., Kim, N. K., Lee, W. J., and Kim, Y. S. (2004). “Experi-
mental load-transfer curves of laterally loaded piles in Nak-Dong
Fig. 19.Pm – PSR Relationship for 3×3 Pile Group Array in Medium River sand.” Journal of Geotechnical and Geoenvironmental
Sand Engineering, ASCE, Vol. 130, No. 4, pp. 416-425.
Kong, L. G. and Zhang, L. M. (2007). “Centrifuge modeling of response
pile for both medium and loose sand, and in such cases, the of pile groups under torsion.” J. Geotech. Geoenviron. Eng., ASCE,
individual pile in pile groups behaves the same as a single pile. Vol. 133, No. 11, pp. 1374-1384.
For medium dense sand, however, the pile group behaves as a Matlock, H. (1970). “Correlations for design of laterally loaded piles in
soft clay.” In Proceedings of the 2nd Annual Offshore Technology
single pile at a pile spacing of about eight times the pile
Conference, OTC, Vol. 1, pp. 577-594.
diameter and beyond. The group efficiency obtained here for a McVay, M. C., Bloomquist, D., Vanderlinde, D., and Clausen, J. (1994).
pile spacing of 3 diameters is in good agreement with the “Centrifuge modeling of laterally loaded pile groups in sands.”
results of Oteo’s 1-g model tests. Geotechnical Testing Journal, ASCE, Vol. 17, No. 2, pp. 129-137.
• It is found that the p-multiplier approach used in this paper McVay, M. C., Casper, R., and Shang, T. (1995). “Lateral response of
provides a simple but accurate tool for characterizing the three row groups in loose to dense sands at 3D and 5D pile spacing.”
shadowing effects of laterally loaded vertical piles in a group. Journal of Geotechnical and Geoenvironmental Engineering,
The p-multiplier factors for the lead, the middle, and the trail ASCE, Vol. 121, No. 5, pp. 436-441.
McVay, M. C., Shang, T., and Casper, R. (1996). “Centrifuge testing of
row of 3×1, 3×2, 3×3 pile groups were estimated to be 0.84,
fixed-head laterally loaded battered and plumb pile groups in sand.”
0.58, and 0.4, respectively. Geotechnical Testing Journal, ASCE, Vol. 19, No. 1, pp. 41-50.
• With variable configurations of pile groups and pile spacing in McVay, M. C., Zhang, L. M., Molnit, T., and Lai, P. (1998). “Centrifuge
both medium and medium dense sands, the average p- testing of large laterally loaded pile groups in sands.” Journal of
multiplier factors for the middle and the trail rows change Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 124,
depending upon the type of pile group layout and the pile No. 10, pp. 1016-1026.
spacing. The average p-multiplier factor for a PSR >1.0 is Ng, C. W. W., Zhang, L. M., and Nip, C. N. (2001). “Nonlinear response
higher than that for a PSR <1.0. Empirical equations for the P of laterally loaded large-diameter bored pile groups.” J. Geotech.
Geoenviron. Eng., ASCE, Vol. 127, No. 8, pp. 658-669.
multiplier factor that considers the number of piles in a group
Ooi, P. S. K. and Duncan, J. M. (1994). “Lateral load analysis of groups of
and the pile group spacing for PSR = 1 in a medium to medium piles and drilled shafts.” Journal of Geotechnical and Geoenviron-
dense sand are presented as Eqs. (3) and (4). Finally, it is mental Engineering, ASCE, Vol. 120, No. 6, pp. 1034-1050.
concluded that, for a deflection ≥ 3 mm, an individual row’s Oteo, C. S. (1972). “Displacements of vertical pile group subjected to
contribution to a group’s lateral resistance is almost constant. lateral loads.” Proceedings of the 5th Europe Conference on Soil
Mechanics and Foundation Engineering, Madrid, Vol. 1, pp. 397-405.
References Peterson, K. T., Rollins, K. M., and Weaver, T. J. (1998). “Lateral load
behavior of full-scale pile group in clay.” Journal of Geotechnical
and Geoenvironmental Engineering, ASCE, Vol. 124, No. 6, pp.
ASTM D-4253 & 4254 (2002). Annual book of ASTM standards 2002,
468-478.
Baltimore, USA.
Pise, P. J. (1983). “Lateral load-deflection behavior of pile group.”
Ashour, M, Pilling, P., and Norris, M. (2004). “Lateral behavior of pile
Indian Geotechnical Journal, Vol. 13, No. 1, pp. 37-51.
groups in layered soils.” Journal of Geotechnical and Geoenviron-
Pise, P. J. and Patra, N. R. (2001). “Ultimate lateral resistance of pile
mental Engineering, ASCE, Vol. 130, No. 6, pp. 580-592.
groups in sand.” Journal of Geotechnical and Geoenvironmental
Brown, D. A., Morrison, C., and Reese, L. C. (1988). “Lateral load behavior
Engineering, ASCE, Vol. 127, No. 6, pp. 481-487.
of a pile group in sand.” Journal of Geotechnical and Geoenviron-
Poulos, H. G. (1971). “Behavior of laterally loaded piles. II: pile
mental Engineering, ASCE, Vol. 114, No. 11, pp. 1261-1276.
groups.” Journal Soil Mechanics and Foundation Engineering
Cox, W. R., Dixon, D. A., and Murphy, B. S. (1984). Lateral load tests of
Division, ASCE, Vol. 97, No. 5, pp. 733-751.
5.4 mm diameter piles in very soft clay in side-by-side and in-line
Randolph, M. F. (1981). “The response of flexible piles to lateral
groups, Laterally Loaded Deep Foundations: Analysis and Perfor-
loading.” Geotechnique, Vol. 31, No. 2, pp. 247-259.
mance, ASTM, West Conshohocken.

− 74 − KSCE Journal of Civil Engineering

 Laboratory Modeling of Laterally Loaded Pile Groups in Sand

Reese, K. T., Cox, W. R., and Koop, F. D. (1974). “Analysis of laterally Soils and Foundation, Tokyo, Vol. 39, No. 1, pp. 97-111.
loaded piles in sand.” In Proceedings of the 5th Annual Offshore Wang, S. T. and Reese, L. C. (1990). COM624P laterally loaded pile
Technology Conference, OTC. analysis program for microcomputer-version 2.0, Final Rep. FHWA-
Reese, K. T., Cox, W. R., and Koop, F. D. (1975). “Field testing and IP-90-005, Federal Highway Administration, Washington, D.C.
anlysis of laterally loaded piles in stiff clay.” In Proceedings of the Williams, D. J. (1979). The behavior of model piles in dense sand, PhD
7th Annual Offshore Technology Conference, OTC. thesis, University of Cambridge, Cambridge, U.K.
Townsend, F. C. and Ruesta, P. F. 1997. “Evaluation of laterally loaded Zhang, L. M., McVay, M. C., and Lai, P. (1999). “Numerical analysis of
pile group at Roosevelt bridge.” Journal of Geotechnical and laterally loaded 3×3 to 7×3 pile groups in sand.” Journal of Geo-
Geoenvironmental Engineering, Vol. 123, No. 12, pp. 1153-1161. technical and Geoenvironmental Engineering, ASCE, Vol. 125, No.
Wakai, A., Gose, S., and Ugai, K. (1999). “3-D elasto-plastic finite 11, pp. 936-946.
element analysis of pile foundations subjected to lateral loading.”

Vol. 15, No. 1 / January 2011 − 75 −

View publication stats