P I L E - C A P - P I L E - G R O U P INTERACTION IN

NONHOMOGENEOUS SOIL
By Y. K. Chow, 1 Member, ASCE, and C. I. Teh2

ABSTRACT: A numerical method of analysis is presented to study the behavior of

vertically loaded pile groups embedded in a nonhomogencous soil with the pile
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

caps in contact with the ground. The considered soil profiles consist of soil with
Young's moduli increasing linearly with depth. Parametric solutions are presented
to show the influence of the distribution of the soil Young's moduli on the behavior
of the groups. For the nonhomogeneous-scil profiles considered, the effect of the
cap in contact with the ground has a small influence on the stiffness of the group
compared to the case without a ground-contacting cap. The load carried by the
cap, as expected, is significantly affected by the distribution of the soil Young's
moduli. Case studies of field tests in clay show that the consideration of the nature
of the soil inhomogeneity at the sites better models the behavior of pile groups
with ground-contacting caps.

INTRODUCTION

The behavior of pile groups where the caps are not in contact with the
ground is reasonably well understood. Several numerical models are avail-
able for the analysis of this class of problem in homogeneous soil (e.g.,
Poulos 1968; Butterfield and Banerjee 1971a). Approximate methods that
deal with nonhomogeneous soil are also available (e.g., Randolph and Wroth
1979; Chow 1986a, b). Methods of analysis that treat nonhomogeneous soil
in a rigorous manner are reported by Banerjee and Davies (1977) and Chow
(1987a, 1989).
In practice, the pile cap is generally in contact with the ground, but the
work in this area is not as extensive. Reported methods of analysis have
dealt mainly with this interaction problem in a homogeneous soil medium
(Butterfield and Banerjee 1971b; Davis and Poulos 1972; Kuwabara 1989).
It can be expected that the consideration of soil inhomogeneity may be more
realistic in many instances. Although three-dimensional finite elements (which
may deal with soil inhomogeneity) have been used for the analysis of this
problem (e.g., Ottaviani 1975), the costs of such computations are rather
prohibitive.
In this paper, a rigorous method of analysis for the pile-cap-pile-group
interaction problem in nonhomogeneous soil is described. The results of a
parametric study are presented to show the influence of the various factors
on the behavior of this pile-group system. Case studies of field tests on pile
groups in clay with ground-contacting caps are also presented.

'Sr. Lect., Dept. of Civ. Engrg., Nat. Univ. of Singapore, 10 Kent Ridge Crescent,
Singapore 0511.
2
Lect., School of Civ. & Struct. Engrg., Nanyang Technol. Univ., Nanyang Av-
enue, Singapore 2263; formerly, CAD/CAM Specialist, Nat. Univ. of Singapore,
Singapore.
Note. Discussion open until April 1, 1992. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on March 5, 1990.
This paper is part of the Journal of Geotechnical Engineering, Vol. 117, No. 11,
November, 1991. ©ASCE, ISSN 0733-9410/91/0011-1655/$1.00 + $.15 per page.
Paper No. 26315.

1655

J. Geotech. Engrg. 1991.117:1655-1668.

 METHOD OF ANALYSIS

A method of analysis is presented for pile groups embedded in a linear

elastic isotropic soil with the pile cap in contact with the ground surface.
The problem considered is represented schematically in Fig. 1. The pile cap
is assumed to be rigid. The piles are assumed to be vertical and to have a
circular cross section with diameter d, and are of equal length L. The pile
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

material is assumed to be linear elastic with Young's modulus Ep. The center-
to-center pile spacing is s, and the overhang of the pile cap (measured from
the center of the edge pile to the edge of the cap) is b.
The soil layer is of a uniform thickness h overlying bedrock that is assumed
to be rigid. The soil profile considered in the present study has Young's
modulus increasing linearly with depth but with a constant Poisson's ratio
v. The soil Young's modulus is given by
E(z) = £(0) + Kz (1)
where E{z) = the soil Young's modulus at depth z; E(0) = the soil Young's
modulus at the ground surface; and \ = the rate of increase of the soil
Young's modulus with depth. Although this particular soil profile has been
considered in the present study, this approach is general enough to consider
arbitrary soil layering where the soil properties do not vary in the horizontal
direction.
The formulation of the problem considers pile-soil-pile interaction, cap-
soil-pile interaction, and the interaction among the cap subelements. This

i
rp :^w

j _

777777777777777777777777777-
Rigid rock
FIG. 1. Schematic Representation of Problem

1656

J. Geotech. Engrg. 1991.117:1655-1668.

 approach is similar to that described for rigid surface foundations (Chow
1987b) and pile groups with caps not in contact with the ground (Chow
1987a), but it considers the interaction among the elements. Therefore, only
a brief outline of the principles is presented.
The problem is broken down into the following systems:

1. The cap acted on by an external applied load and interaction forces acting
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

on the cap.
2. The piles acted on by interaction forces acting on the piles.
3. A soil continuum acted on by a system of interaction forces acting on the
soil at the cap-soil and pile-soil interfaces.

The vertical soil deformations at the nodal points are determined using
the principle of superposition. They may be written in the following matrix
form:
w, = F,P, (2)
1
where w^ = (wj , wls2, pl p2 P T
. . . , w"", w , w , . . . , w ") = the vertical soil
displacement vector; Pv = (Pf, Pf, . . . , Pc/", Pp\ Pf, . . . , PP")T = the
vector of interaction forces acting on the soil; F, = the soil flexibility matrix;
the superscripts ci and pj correspond to the ith soil element at the cap and
the y'th soil element at the pile, respectively; and m and n are the total
number of soil elements at the cap and piles, respectively. The flexibility
coefficients in F, are evaluated numerically using a displacement-based ax-
isymmetric finite element formulation as described by Chow (1987a, b).
The pile cap is discretized into subelements (Chow 1987b) and the piles
are discretized into two-node elements with an axial mode of deformation.
By considering the equilibrium of the cap-pile-soil system interaction forces
and the compatibility of the cap, pile, and soil displacements, the load-
displacement relationship of a pile group with ground-contacting cap is given
by
(K„ + K > „ = P (3)
where Kp = the assembled stiffness matrix of the discrete elements of the
piles (this matrix is augmented with degrees of freedom associated with
the pile-cap subelements); K, = the soil stiffness matrix obtained from
the inversion of F,. from (2); w p = {wf, wf, . . . , wp"\ wpi, wp2, . . . ,
W
P")T = t n e vertical displacements vector of the cap sublements and pile
nodal displacements [these displacements are compatible with the soil dis-
placements given in (2)]; P = the vector of external applied loads; ci and
pj correspond to the ith cap subelement andy'th pile node, respectively; and
m and n = the total cap subelements and pile nodes, respectively.
The stiffness matrix of the pile elements is easily available in the literature
(e.g., Smith and Griffith 1988). The degrees of freedom in K p associated
with the pile-cap subelements are augmented with zeros; the rigid pile cap
is simulated numerically by enforcing equal displacements of the subele-
ments of the cap and the nodes at the pile heads using a prescribed dis-
placement technique as described by Chow (1986a).

COMPARISON WITH BOUNDARY ELEMENT SOLUTIONS

The present solutions are compared with the boundary element solutions
of Banerjee (1975), who used Mindlin's solution as the kernel function. The

1657

J. Geotech. Engrg. 1991.117:1655-1668.

 results are for two-pile groups and two-by-two-pile groups with Lid = 20,
sld = 2.5, and bid = 1.25, embedded in a linear elastic, homogeneous,
isotropic soil layer with ML = 2. Table 1 shows a comparison of the solutions
for two-pile groups obtained using the present method, the boundary ele-
ment method, and the three-dimensional finite element method (3-D FEM).
In the present method, 84 pile-cap subelements and 10 pile elements were
used to simulate one-half of the system, taking symmetry into account; the
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

solutions were not particularly sensitive when a slightly finer discretization

was used. The three-dimensional finite element solutions were obtained
using 1,800 eight-node brick elements with 2,226 nodes, and were computed
using the National University of Singapore's computer-assisted-design/com-
puter-assisted-manufacturing (CAD/CAM) system. The data preparation
has been eased considerably using this system, and even then the compu-
tations still consumed a significant amount of computer time. The stiffness
of the pile groups computed by the boundary element technique is the
lowest, whereas the three-dimensional finite elements give the highest stiff-
ness in the majority of the solutions. The loads carried by the caps are
generally highest in the boundary element solutions followed by the present
solutions with the majority of the finite element solutions generally giving
the lowest loads.
A comparison of the solutions for the two-by-two-pile groups is shown
in Table 2. Again, the boundary element solutions give lower stiffnesses
than the present method, but give higher loads carried by the caps. In this
method, a similar discretization as described was used, simulating a quarter
of the system, taking symmetry into account. Three-dimensional finite ele-
ment solutions were not attempted due to the high costs involved. The
differences seen here are mainly due to the inherent assumptions of the
different numerical models,

TABLE 1. Solutions For Two-Pile Groups

P/wEsd PJP (%)
Present Banerjee 3-D Present Banerjee 3-D
£„/£, method (1975) FEM method (1975) FEM
(1) (2) (3) (4) (5) (6) (7)
33 8.9 8.0 8.5 44.3 54 41.0
333 15.4 14.0 17.1 19.6 22 19.9
3,333 18.4 16.7 20.2 14.4 15 11.5
QC 18.9 17.3 22.3 13.7 14 10.2

TABLE 2. Solutions For Two-by-Two-Pile Groups

PlwEsd PJP (%)
Present Banerjee Present Banerjee
E„IES method (1975) method (1975)
(D (2) (3) (4) (5)
33 12.8 11.7 42.2 54
333 20.7 18.3 19.9 25
3,333 23.5 20.7 15.7 19
SO 23.9 21.0 15.2 18

1658

J. Geotech. Engrg. 1991.117:1655-1668.

 PARAMETRIC STUDY

Parametric solutions are presented to show the influence of some of the

factors on the behavior of pile groups embedded in a nonhomogeneous soil
with caps in contact with the ground. The soil was assumed to be undrained,
and therefore the Poisson's ratio of the soil, v, was taken to be 0.499. The
normalized depth of the soil considered was hIL = 2. Results are presented
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

for homogeneous soil profiles [E(0)IE(L) = 1] and compared to nonhom-

ogeneous soils with different degrees of inhomogeneity: E(0)/E(L) = 0.5
and E(0)IE(L) = 0. In all cases, the overhang of the pile cap considered
was bid = 1.5.
Fig. 2(a) shows the normalized stiffness of two-pile groups with caps
in contact with the ground for sld = 3 and EpIEs = 1,000 [where Es is
the average Young's modulus of the soil along the pile length, i.e.,
Es = E(L/2)] for a range of Lid ratios. The nondimensionalized stiffness
of the group is represented as PlwE,d, where P is the applied load on the
pile cap and w is the vertical displacement of the cap. When presented in
this form, the solutions give an appreciation of the effect of replacing a
nonhomogeneous soil profile by an "equivalent" homogeneous soil with the
same average soil Young's modulus along the pile length. The difference
between the solutions is more significant for shorter piles, becoming less
significant for longer piles. It is interesting that the effect of a cap in contact
with the ground, relative to one not in contact, is to increase the group
stiffness in homogeneous soil by about 13% for Lid = 10; the increase is
only about 7% for Lid = 20, and this figure is only about 3% when Lid
= 60. The results for pile groups where the caps are not in contact with
the ground are not shown to avoid cluttering the figures.
For E(0)/E(L) = 0.5, the increase in stiffness for Lid = 10, 20, and 60
are 11%, 5%, and 2%, respectively, whereas the corresponding figures for
E(0)/E(L) = 0 are 4%, 1%, and 0.4%. This smaller increase in stiffness in
the nonhomogeneous soil is to be expected due to the smaller value of the
Young's modulus of the soil in contact with the pile cap. The influence of
the nonhomogeneous soil is most significant on the load carried by the cap
[Fig. 2(b)]. For Lid = 10, the percentage of load carried by the cap is about
34% for the homogeneous soil [E(0)/E(L) = 1] reduces to 25% for £(0)/
E(L) = 0.5, and is only 7% for E(0)/E(L) = 0. For Lid = 60, the loads

15d 3d l-5d

\ "1 o o

r^rr —
10 20 30 (0 SO BO
L/d
Ibl

FIG. 2. Solutions for Two-Pile Groups (EpIE, = 1,000, sld = 3): (a) Stiffness of
Pile Group; and (to) Load Carried by Cap

1659

J. Geotech. Engrg. 1991.117:1655-1668.

 carried by the cap when E(0)IE(L) = 1, 0.5, and 0 are 12%, 9%, and 0.7%,
respectively.
The solutions for the pile group stiffness and load carried by the cap for
two-by-two- and three-by-three-pile groups with ground-contacting caps
(for sld = 3 and EpIEs = 1,000) are shown in Figs. 3 and 4, respectively.
As the group gets larger, the difference in the stiffness of the group in
nonhomogeneous and homogeneous soils becomes greater. The effect of a
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

ground-contacting cap increases the stiffness to a slightly smaller extent in

these larger pile groups compared to the two-pile groups. Again, it can be
seen that the nonhomogeneous soil has a significant influence on the load
carried by the cap. It is interesting that the load carried by the pile cap for
a given Lid ratio is quite similar for the two-, two-by-two-, and three-by-
three-pile groups considered.
The percentage of load carried by the piles in the three-by-three groups
is shown in Fig. 5. Due to the smaller load carried by the cap for groups in
nonhomogeneous soil, the loads carried by the piles are correspondingly
higher. The ratios of the load carried by the corner, edge, and center piles
With Lid = 30 for E(0)IE(L) = 1,0.5, and 0 are 4:2.5:1,3:2:1, and 2.2:1.6:1,
respectively. The corresponding ratios for Lid = 60 are 2.1:1.6:1,1.9:1.5:1,
and 1.6:1.3:1. The nonhomogeneous soil gives a more uniform distribution

l-5d 3d 15d

o o
o o

20 30 40 50
LAI
lb]

FIG. 3. Solutions for Two-by-Two-Pile Groups (Ep/E, = 1,000, sld = 3): (a) Stiff-
ness of Pile Group; and (b) Load Carried by Cap

l-Sd
01 02 Ol
\ 3d
02 03 02
N. 3d
Ol 02 Ol

30 10
L/d

FIG. 4. Solutions for Three-by-Three-Pile Groups (EPIE, = 1,000, sld = 3):

(a) Stiffness of Pile Group; and (ft) Load Carried by Cap

t660

J. Geotech. Engrg. 1991.117:1655-1668.

Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 5. Load Carried by Piles in Three-by-Three-Pile Groups (£,/£, 1,000,

sld = 3)

io2 io3 in1

Ep/Es
lb)

FIG. 6. Solutions for Three-by-Three-Pile Groups: Effect of EJES (Lid = 30,

sld = 3): (a) Stiffness of Pile Group; and (o) Load Carried by Cap

of the loads carried by the piles due to the smaller interaction among the
piles.
Fig. 6 shows the effect of EPIES on three-by-three groups with Lid = 30
and sld = 3. Generally, for higher EpIEs values, the difference in stiffness
of pile groups in the homogeneous and nonhomogeneous soils is more
significant, whereas the load carried by the cap reduces, implying that more
loads are transferred to the piles.
The effect of pile spacing sld on three-by-three groups with Lid = 30 and
EPIES = 1,000 is shown in Fig. 7. These pile spacings cover the range
normally used in practice. As can be expected, increasing the pile spacings
1661

J. Geotech. Engrg. 1991.117:1655-1668.

Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 7. Solutions for Three-by-Three-Pile Groups: Effect of sld (Lid = 30,

£,,/£, = 1,000): (a) Stiffness of Pile Group; and (b) Load Carried by Cap

EIOI/EIU
— 1

Pile
— — 05
-- 0
1

I-

2
~"3w«
\\\

3
_.
'

I0l 10s
s/d
lb]

FIG. 8. Load Carried by Piles in Three-by-Three-Pile Groups (Lid = 30): (a) Effect
of EPIES; and (b) Effect of sld.

increases the group stiffness as well as the load carried by the cap. Even
for sld = 8, the load carried by the cap when E(0)/E(L) = 1 is about 31%,
decreasing to 25% when E(0)/E(L) = 0.5, and decreasing rapidly to only
about 9% when E(0)/E(L) = 0.
The effects of EpIEs and sld on the distribution of loads in the piles for
three-by-three groups with Lid = 30 are shown in Figs. 8(a) and 8(b),
respectively. For increasing Ep/Es, the greater interaction among the piles
result in a more uneven distribution of loads. The increasing pile spacing
results in smaller interaction among the piles, giving rise to a more even
distribution of the loads.

CASE STUDIES OF FIELD TESTS ON PILE GROUPS

Test of Koizumi and Ito (1967): Three-by-Three-Pile Group

Koizumi and Ito (1967) reported the results of a full-scale field test on a
three-by-three-pile group with cap in contact with the ground. The closed-
end steel pipe piles had an outside diameter of 300 mm, a wall thickness of
3.2 mm, and embedded lengths of 5.55 m. The pile spacing, s, was 900 mm
with a cap overhang, b, of 450 mm. The soil profile consists of slightly
organic silty clay, containing shells, which extends to a depth of about 13.5
1662

J. Geotech. Engrg. 1991.117:1655-1668.

 m. Below this depth gravel and dense sand layers extend to bedrock. The
shear strength of the clay increases linearly with depth, about 25 kN/m2 at
the foundation level and 40 kN/m2 at the pile toe.
Two analyses were carried out. The first was based on a nonhomogeneous
soil profile, and the other based on an equivalent homogeneous soil. The
soil Young's moduli were deduced by matching the computed displacement
with the measured displacement of the single pile at a test load of 100 kN
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

corresponding to the working load of the single pile with a factor of safety
of about 2.5. For the nonhomogeneous soil, the Young's modulus was
assumed to be proportional to the undrained shear strength of the clay as
reported. With this assumption, the back-analysis of the single pile as de-
scribed earlier gave £(0) = 12,500 kN/m2 and X = 1,352 kN/m2 per 1 m
depth for the nonhomogeneous soil profile, whereas the homogeneous soil
profile gave a uniform Young's modulus, Es, of 17,200 kN/m2; the soil
Poisson's ratio was taken to be 0.499.
The computed linear elastic load-displacement behavior of the pile group
is shown in Fig. 9, together with the field measurements. It can be seen
that at a factor of safety of about 2.5, the displacement of the group with
the nonhomogeneous soil assumption agrees closer with the measured dis-
placement; the homogeneous-soil assumption gives a slightly larger value.
A comparison of the computed and measured loads carried by the cap and
the piles is shown in Fig. 10. Again, the nonhomogeneous-soil assumption
agrees better with the field measurements. Fig. 11 shows that the computed
axial load distribution in the piles (obtained with the nonhomogeneous-soil
profile) at an applied load of 589 kN on the cap agrees closely with the
measured data.

Tests of Cooke et al. (1980): Three-Pile Group

Cooke et al. (1980) reported a comprehensive series of field tests on the
behavior of pile groups embedded in London clay. The tubular steel piles
had an external diameter of 168 mm, a wall thickness of 6.4 mm and embed-
ded lengths of about 4.6 m. Among the tests carried out was a test on a
group comprising a row of three piles installed at a center-to-center spacing
of three pile diameters with a cap of 0.5 m by 1.5 m in contact with the
ground.
The undrained shear-strength profile, based on laboratory triaxial tests
on 98-mm-diameter samples, increases linearly with depth with a value of
about 32 kN/m2 at the ground surface and 78 kN/m2 at the pile toe. The
soil Young's modulus for the nonhomogeneous case was assumed to be
proportional to the undrained shear strength as reported, and its profile
was deduced by matching the computed displacement with the measured
displacement of the single test pile at a working-load level of 20 kN; this
gave £(0) = 36,640 kN/m2 and X = 11,450 kN/m2 per 1 m depth. The
corresponding Young's modulus of the soil based on an equivalent homo-
geneous profile was Es = 64,800 kN/m2.
The computed load-displacement curves of the group are compared with
the field measurements in Fig. 12. The displacement of the group in the
nonhomogeneous soil is greater than the measured displacement; the dis-
placement in the homogeneous soil is even greater. The measured and
computed loads carried by the cap and the individual piles are shown in
Table 3. The solutions for the nonhomogeneous-soil profile are in closer
agreement with the measured values.
Fig. 13 shows the measured (Cooke et al. 1980) and computed distribution
1663

J. Geotech. Engrg. 1991.117:1655-1668.

 Load (kN)

200 400 1000 1200 K00

Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

—1

0-45 m
Ol 02 01
09 m
02 03 02
09m
01 02 01
JJ0-45m

Computed (nonhomogeneous) -—— Measured

(Koizumi and Ito 1967)
Computed! homogeneous)

FIG. 9. Load-Displacement Curves of Three-by-Three-Pile Group

2500
Measured (Koizumi and lto 1967)
— Computed (nonhomogeneous) /
— — — Computed (homogeneous) . /
2000

500 1000 1500 2000 2500

Applied load ( k N )

FIG. 10. Load Carried by Cap and Piles of Three-by-Three-Pile Group

1664

J. Geotech. Engrg. 1991.117:1655-1668.

 Load (kN) Load (kN) Load (kN)
0 50- 0 50

T
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

>
1 \
hi

Ii/
9

II
I lb)
I
Computed o---° Measured
(nonnomogeneous) (Koizumi and Ito 1967 ]

FIG. 11. Load-Transfer Behavior of Piles in Three-by-Three-Pile Group: (a) Pile

1; (b) Pile 2; and (c) Pile 3

of axial loads in the piles embedded in nonnomogeneous soil when they are
loaded singly, loaded as a group when the cap is not in contact with the
ground and when it is in contact; the loads are normalized with respect to
the loads at the pile head. The effect of the pile-soil-pile interaction (where
the cap is not in contact with the ground) is to inhibit load transfer near
the pile head with more load transferred to near the pile toe. The effect of
the cap is to inhibit the load transfer further near the pile head. These trends
are reproduced by the numerical model and the solutions are in reasonable
agreement with the field results.

CONCLUSIONS

A numerical model has been described for the linear elastic analysis of
pile groups embedded in a nonhomogeneous soil with the caps in contact
with the ground. In the majority of the cases analyzed in the parametric
study, the use of an equivalent homogeneous-soil profile tends to under-
estimate the stiffness of the groups. The effect of a cap does not increase
the stiffness of the group significantly and this effect is even less significant
in the nonhomogeneous-soil profiles studied. The nonhomogeneous-soil
profiles, as expected, result in a significant reduction of the load carried by
the pile cap with a corresponding increase in the loads carried by the in-
dividual piles.
Case studies of field tests on pile groups in clay with ground-contacting
1665

J. Geotech. Engrg. 1991.117:1655-1668.

 Load (kN)

20 10 60 80 100 120
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

¥o-i
E

E 02

13

_____ Computed (nonhomogeneous)

• Computed (homogeneous)
U
D Measured (Cooke et al.1980)

FIG. 12. Load-Displacement Behavior of Row of Three Piles

Computed Measured
O Loading on single pile
_,___ a Loading on group (cap not in contact with ground)
° Loading on group ( cap in contact with ground )

FIG. 13. Load-Transfer Behavior of Piles and Pile Groups: (a) Pile A; and (b) Pile
B

1666

J. Geotech. Engrg. 1991.117:1655-1668.

 TABLE 3. Load Carried by Pile Cap and Individual Piles in Row of Three Piles
D
ercentage of Load Carried
Cases Cap Pile A Pile B PileC
(D (2) (3) (4) (5)
Measured—Cooke et al. (1980) 9.2 27.2 ' 31.8 31.8
Computed—nonhomogeneous 14.3 24.7 30.5 30.5
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

Computed—homogeneous 21.2 22.0 28.4 28.4

caps show that the nonhomogeneous-soil profile better approximates the

effect of pile-cap-pile group interaction. The numerical results are shown
to be in reasonable agreement with the field measurements.

APPENDIX I. REFERENCES

Banerjee, P. K. (1975). "Effects of the pile cap on the load displacement behaviour
of pile groups when subjected to eccentric loading." Proc. 2nd Australia-New
Zealand Conf. on Geomech., Australian Geomechanics Society/New Zealand Geo-
mechanics Society, 179-184.
Banerjee, P. K., and Davies, T. G. (1977). "Analysis of pile groups embedded in
Gibson soil." Proc. 9th Int. Conf. Soil Mech. and Found. Engrg., International
Society for Soil Mechanics and Foundation Engineering, 381-386.
Butterfield, R., and Banerjee, P. K. (1971a). "The elastic analysis of compressible
piles and pile groups." Geotechnique, London, England, 21(1), 43-60.
Butterfield, R., and Banerjee, P. K. (1971b). "The problem of pile-group-pile cap
interaction." Geotechnique, London, England, 21(2), 135-142.
Chow, Y. K. (1986a). "Analysis of vertically loaded pile groups." Int. J. Numer.
Analytical Meth. Geomech., 10(1), 59-72.
Chow, Y. K. (1986b). "Discrete element analysis of settlement of pile groups."
Comput. and Struct., 24(1), 157-166.
Chow, Y. K. (1987a). "Axial and lateral response of pile groups embedded in non-
homogeneous soils." Int. J. Numer. Analytical Meth. Geomech., 11(6), 621-638.
Chow, Y. K. (1987b). "Vertical deformation of rigid foundations of arbitrary shape
on layered soil media." Int. J. Numer. Analytical Meth. Geomech., 11(1), 1-15.
Chow, Y. K. (1989). "Axially loaded piles and pile groups embedded in a cross-
anisotropic soil." Geotechnique, London, England, 39(2), 203-212.
Cooke, R. W., Price, G., and Tarr, K. (1980). "Jacked piles in London clay: In-
teraction and group behavior under working conditions." Geotechnique, London,
England, 30(2), 97-136.
Davis, E. H., and Poulos, H. G. (1972). "The analysis of piled-raft systems." Aust.
Geomech. J., Australia, G2(l), 21-27.
Koizumi, Y. and Ito, K. (1967). "Field tests with regard to pile driving and bearing
capacity of piled foundations." Soils and Found., 7(3), 30-53.
Kuwabara, F. (1989). "An elastic analysis for piled raft foundations in a homogeneous
soil." Soils and Found., 29(1), 82-92.
Ottaviani, M. (1975). "Three-dimensional analysis of vertically loaded pile groups."
Geotechnique, London, England, 25(2), 159-174.
Poulos, H. G. (1968). "Analysis of the settlement of pile groups." Geotechnique,
London, England, 18(4), 449-471.
Randolph, M. F., and Wroth, C. P. (1979). "An analysis of vertical deformation of
pile groups." Geotechnique, London, England, 29(4), 423-439.
Smith, 1. M., and Griffiths, D. V. (1988). Programming the finite element method.
John Wiley & Sons, Chichester, England.

1667

J. Geotech. Engrg. 1991.117:1655-1668.

 APPENDIX II. NOTATION

The following symbols are used in this paper:

b = overhang of pile cap;
d = pile diameter;
h = thickness of soil layer;
Downloaded from ascelibrary.org by Carleton University on 06/29/14. Copyright ASCE. For personal use only; all rights reserved.

£(0), E(z) = Young's modulus of soil at ground surface and at depth

z, respectively;
E(LI2), E(L) = Young's modulus of soil at depth L/2 and L, respectively;
Ep = Young's modulus of pile material;
Es = average Young's modulus of soil [= E(L/2)];
Fs = soil flexibility matrix;
Kp = assembled stiffness matrix of pile elements;
Ks = soil stiffness matrix;
L = pile length;
P = vertical load acting on pile group;
P = external applied load vector;
Pa = load on individual pile;
Pc = load carried by cap;
P5. = vector of interaction forces acting on soil;
Pf, Pps' = interaction force acting on /th and /th soil element at cap
and pile, respectively;
Pz = axial load in pile at depth 2;
s = center-to-center pile spacing;
w = vertical displcement of pile group;
y/p = vertical displacement vector of cap and pile nodes;
wP', wpp> = vertical displacement of /th cap subelement and y'th pile
node, respectively;
y/s = vertical displacement vector of soil;
wf, wpJ = vertical displacement of /th and /th soil element at cap
and pile, respectively;
2 = depth coordinate;
X = rate of increase of soil Young's modulus with depth; and
v = soil Poisson's ratio.

Superscripts
ci /th cap subelement;
m total number of cap elements;
n total number of pile nodes; and
PJ /th pile node.

Subscripts
p = cap and pile; and
s = soil.

1668

J. Geotech. Engrg. 1991.117:1655-1668.