You are on page 1of 16

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

net/publication/248877239

Pile Group Analysis: A Study of Two Methods

Article  in  Journal of Geotechnical Engineering · March 1983


DOI: 10.1061/(ASCE)0733-9410(1983)109:3(355)

CITATIONS READS

55 4,526

2 authors, including:

Harry G. Poulos
The University of Sydney
307 PUBLICATIONS   14,618 CITATIONS   

SEE PROFILE

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

Analysis of piled raft load sharing under vertical and horizontal loading View project

Foundations View project

All content following this page was uploaded by Harry G. Poulos on 17 September 2014.

The user has requested enhancement of the downloaded file.


THREE-DIMENSIONAL ANALYSIS OF PILE GROUPS
By Y. K. Chow, 1 A. M . ASCE

ABSTRACT: A numerical method is presented for the analysis of general three-


dimensional pile groups. The features of the pile-group model may include bat-
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

tered piles, different pile sizes, non-uniform pile sections, soil nonlinearity, soil
inhomogeneity, and pile-soil-pile interaction. A typical six-pile group is ana-
lyzed and compared with results from three other computer programs for pile-
group analysis that are based on different approaches. This method is then
used to analyze field and laboratory tests on groups of battered and vertical
piles. The computed solutions are shown to be in good general agreement with
the measured data.

INTRODUCTION

A number of theoretical approaches are now available for the analysis


of general pile groups. The influence of pile batter and soil yielding, as
well as pile-soil-pile interaction, can be studied with different degrees
of rigor, depending on the approximations involved in the analysis.
The methods of analysis based on the theory of elasticity (or some
simplifications of it) are largely derived from the works of Banerjee and
Driscoll (1976), Poulos (1980), and Randolph (1980). Various computer
programs for pile-group analysis have been developed based on these
methods.
Banerjee and Driscoll (1976) presented a boundary element method for
pile-group analysis. The soil is modeled as a homogeneous, linear elastic
material. PGROUP, the computer program developed from this work,
has since been upgraded to include a soil model with a linearly increas-
ing modulus following the work of Banerjee and Davies (1980). Note
that Banerjee and Davies have also described a nonlinear method of
analysis (1980) in which volume cells were introduced into the soil do-
main to handle soil yielding.
DEFPIG, written by Poulos (1980) is based on a simplified boundary
element approach for the single pile analysis and the calculation of the
interaction factors for two equally loaded identical piles. Soil nonline-
arity is modeled by limiting the stresses at the pile-soil interface, while
soil inhomogeneity is approximated with an averaging procedure using
the point-load solutions of Mindlin (1936).
The PIGLET program (Randolf 1980) is based on analytical solutions
that are either derived theoretically or fitted to finite element results to
give the response of single piles. Pile-soil-pile interaction is based on
interaction factors determined from expressions fitted to the results of
finite element analyses. Linear elastic soil behavior is assumed.
The interaction-factor approach used in DEFPIG and PIGLET for pile-
soil-pile interaction only gives the loads and bending moments at the
x
Lect., Dept. of Civ. Engrg., National Univ. of Singapore, 10 Kent Ridge Cres-
cent, Singapore 0511.
Note.—Discussion open until November 1, 1987. 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 August 1, 1986. This paper is part of the Journal of Geotechnical Engineering,
Vol. 113, No. 6, June, 1987. ©ASCE, ISSN 0733-9410/87/0006-0637/$01.00. Paper
No. 21591.
637

J. Geotech. Engrg. 1987.113:637-651.


pile heads, but not their distributions along the piles. The loads and
bending moments along the pile can only be approximated, for instance,
utilizing the single pile solutions with the corresponding pile head loads
and bending moments.
A popular method of single pile analysis models the pile using beam-
column elements, while the soil response is based on the modulus of
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

subgrade reaction or load-transfer curves. Hrennikoff (1949) utilized this


approach for the analysis of pile groups by considering the relative po-
sitions of the individual pile heads, and applying the appropriate bound-
ary conditions as imposed by a rigid pile cap. Reese, et al. (1970) ex-
tended the linear analysis of Hrennikoff (1949) to include nonlinear soil
behavior utilizing load-transfer curves in a general three-dimensional pile-
group analysis. A feature of this approach absent in the elasticity-based
methods is that the nonlinear soil response is modeled using (static or
cyclic) load-transfer curves, which are widely used in the offshore in-
dustry. However, the use of the modulus of subgrade reaction or load-
transfer curves disregards soil continuity, and hence precludes pile-soil-
pile interaction.
The numerical procedure of Reese, et al. (1970) was improved by
O'Neill, et al. (1977) through the incorporation of pile-soil-pile interac-
tion effects utilizing Mindlin's solutions. Pile-soil-pile interaction is ap-
proximated with an iterative procedure by "softening" the single pile
load-transfer curves with a multiplier, involving the displacements of
the pile nodes and additional induced displacements at the nodes caused
by soil reactions generated at the other nodes. The definition of this
multiplier, used to "soften" the load-transfer curves, appears to be
somewhat arbitrary (Ha 1976).
This paper describes a method of analysis that incorporates the salient
features in the general three-dimensional pile group model of O'Neill,
et al. (1977), but considers pile-soil-pile interaction directly. This pro-
cedure may be considered a refinement of that of O'Neill, et al. (1977),
and there are significant differences in its implementation. The numer-
ical procedure for three-dimensional pile-group analysis is a generaliza-
tion of the analytical methods for vertical pile groups subjected to axial
and lateral loads (Chow 1986; Leung and Chow 1986). The accuracy of
this proposed approach is evaluated by comparing the approach with
some computer programs available for pile-group analysis for the per-
formance of a "typical" battered pile group. Finally, the analyses of a
series of full-scale field tests on groups of vertical and battered piles,
and a model pile-group test are reported.

METHOD OF ANALYSIS

In the numerical approach, the individual group piles are modeled


using general beam-column elements. The soil at the individual piles is
represented by the modulus of subgrade reaction or load-transfer curves,
while the interaction between piles is based on the flexibility approach
using Mindlin's solutions. The pile cap is assumed not to be in active
contact with the ground.
Pile Stiffness Matrix.—The individual piles in the group are idealized
by general beam-column elements connected at the nodes. A typical ele-

638

J. Geotech. Engrg. 1987.113:637-651.


Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1.—Global and Local Axes Systems for Pile Element

ment is shown in Fig. 1. Associated with each element is a set of local


axes: x, y, and z. Each end or node of the element has six degrees of
freedom, namely, displacements in the three orthogonal directions—u ± ,
Uy and ut—and rotations about the three axes—6*, %g and Qt. These are
shown in Fig. 1.
The pile element load-deformation relationship or stiffness matrix may
be derived using conventional finite element approaches. The explicit
form of this stiffness matrix [Ke] is well known, and may be obtained,
for instance, from standard texts on structural analysis (see, for example,
Coates, et al. 1972). This matrix [Ke] is of the order 12 x 12, representing
the six degrees of freedom at each of the two nodes of the element.
In a three-dimensional pile group, the set of local axes for the element
will be common for all elements within the pile (assuming the piles are
straight), but this is not necessarily the same for different piles in the
group, unless they are all oriented in the same manner; for instance,
vertical pile groups. Hence, it is necessary to relate the forces and de-
formations of the group piles to one single system of axes common to
all—the global axes x, y, and z, as shown in Fig. 1.
In the next stage of the analysis, the stiffness matrix [Ke] is trans-
formed from the local to the global axes. This can be achieved through
the transformation matrix [T], which can be derived, for instance, fol-
lowing the procedure described by Coates, et al. (1972). The element
stiffness matrix with respect to the global axes my be shown to be giv-
en by
[Ke] = [T]T[Ke][T] (1)
T
in which [T] = transpose of the transformation matrix. The element
639

J. Geotech. Engrg. 1987.113:637-651.


stiffness matrices of the group piles are assembled to yield the stiffness
matrix of the piles [Kp], defined with respect to the global axes.
Soil Stiffness Matrix.—The soil response at the individual piles is
modeled by the modulus of subgrade reaction or load-transfer curves,
which are essentially nonlinear moduli of subgrade reaction. The linear
moduli of subgrade reaction of the soil for the different deformation modes
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

of the pile can be related to the elastic soil properties. The axial and
torsional subgrade reaction moduli may be determined from the work
of Randolph and Wroth (1978) and Randolph (1981), respectively. The
lateral subgrade reaction modulus may be selected so that the response
of the pile head is similar to that of a more rigorous method of analysis i
like the integral equation method (Poulos and Davis 1980). I
The nonlinear soil response may be modeled using load-transfer curves. f
This approach is widely used in the offshore industry, and the load-
transfer curves are generally referred to as "t-z" and "p-y" curves for
the axial and lateral responses, respectively. The "t-z" and "p-y" curves
may be determined following the recommendations of the American Pe-
troleum Institute (1982). Some guidelines for the construction of the tor- I
sional load-transfer curves have been presented by O'Neill (1964) and |
Dutt (1976) for clay and sand, respectively.
In the case of battered piles, the soil subgrade reaction moduli and
load-transfer curves for the axial, torsional and lateral responses are as-
sumed to be independent of the batter angle. Evidence to support this
was reported by Poulos and Madhav (1971), who considered a practical
range of batter angles.
In essence, the soil subgrade reaction moduli and load-transfer curves
represent the soil responses with respect to the local axes of the pile (see
Fig. 1). In a general pile-group analysis, it is thus necessary to transform
the soil behavior from the local axes to the global axes. This can be
achieved in a similar manner as the pile element stiffness matrix. The
soil stiffness matrix [Kf] at the node is of the order 6 x 6 , corresponding
to the six degrees of freedom at the node, as opposed to the pile element
stiffness matrix [Ke], which is of the order 12 x 12. This is because of
the assumption that the soil stiffnesses are "lumped" at the nodes, thus
uncoupling the soil stiffnesses at the two ends or nodes of the element.
Note that the soil stiffness matrix [K* ] is a diagonal matrix with respect
to the local axes. This is due to the assumption that the soil stiffnesses
in the axial, torsional and lateral modes of deformation at the node are
uncoupled. The rotational stiffness of the soil is ignored and, numeri-
cally, this can be done by placing a very small number (relative to the
other soil stiffnesses) in the corresponding diagonal elements. The trans-
formation of the soil stiffness matrix [Kf ] at the node from the local to
the global axes may be achieved using the following equation:
[Kf] = [T*]T[Kf][T*] , (2)
in which [Kf] = soil stiffness matrix with respect to the global axes; [T*]
= transformation matrix, which is a submatrix of [T] in Eq. 1; and [T*]T
= transpose of the transformation matrix.
The procedure presented thus far only describes the soil response at
the individual piles, and does not consider pile-soil-pile interaction. The
loading of a pile in the group will generate soil reactions at the nodal
640

J. Geotech. Engrg. 1987.113:637-651.


points of the pile. Since the soil is a continuum, the generated soil re-
actions will induce additional displacements of other piles in the group.
The pile-soil-pile interaction problem in the present approach is based
on the superposition of the influence of adjacent piles utilizing the flex-
ibility concept. The overall deformation of any degree of freedom, «,,
because of its own loading plus the loadings at the other degrees of
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

freedom of the piles in the group may be expressed in the following


discrete form:

«< = 2 / # p / (3)
in which fa = flexibility coefficient denoting the deformation at degree
of freedom i because of a unit load acting at degree of freedom ;'; Pj —
soil reaction load at degree of freedom /; and n = total number of degrees
of freedom in pile group. Eq. 3 may be written for each degree of free-
dom, giving the following matrix equation;
{«} = [F.KP} (4)
in which {«} = deformation vector for the n degrees of freedom; [Fs] =
soil flexibility matrix; and {P} = soil reaction load vector.
The flexibility coefficients fa in [Fs] corresponding to the degrees of
freedom at each nodal point in the pile group may be obtained by in-
verting the soil nodal stiffness matrix [Kf ] in Eq. 2 to give the nodal
flexibility matrix [F*] (= [K*] -1 ) with respect to the global axes. In this
manner, piles that are battered at any arbitrary angle can be dealt with
readily. The off-diagonal nonzero elements in [Ff ], a result of the trans-
formation from the local to the global axes, represent the coupling of
the degrees of freedom at the nodal point because of the pile batter. The
other off-diagonal elements in the flexibility matrix [Fs] represent pile-
soil-pile interaction. In determining these values, the influence of the
continuously distributed loads at the pile shaft and base is assumed to
be similar to that of equivalent point loads applied at the nodal points
along the centerline of the piles. The accuracy of this assumption was
verified for vertical pile groups (Chow 1986; Leung and Chow 1986).
Note that the use of the point-load solutions is considerably more effi-
cient than the distributed load solutions used in PGROUP and DEFPIG.
The flexibility influence coefficients fa that result from pile-soil-pile in-
teraction are determined from the solutions by Mindlin (1936) (also see
Poulos and Davis 1973) for the influence of point loads in the interior
of an elastic half-space (see Fig. 2), and are defined with respect to the
global axes. These solutions are strictly valid for a homogeneous soil
medium, but interaction effects in a nonhomogeneous soil may be ap-
proximated by using the mean soil shear modulus at locations i and;' in
Mindlin's solutions.
Note the following points about the flexibility matrix [Fs]:

1. The off-diagonal elements, fa = 0, for interaction between degrees


of freedom not at the same node but associated with the same pile form
the inherent assumption in the load-transfer method for modeling soil
behavior in single piles. The coupling between some of the degrees of
641

J. Geotech. Engrg. 1987.113:637-651.


Surface
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

2 = (r 2 >(z*c) 2 ) 1 / 2

=V 2 *(z-c) 2 ) 1 / 2

r =(x2*y2) 1 / 2

FIG. 2.—Influence of Point Loads in Interior of Elastic Half-Space

freedom at the same node arises because of the pile batter.


2. The pile-soil-pile interaction for the rotational/torsional degrees of
freedom is not considered; the flexibility matrix [Fs] is augmented with
zeros in the corresponding rows and columns.

The flexibility relationship in Eq. 4 is inverted to give the stiffness ma-


trix of the soil (with respect to the global axes) as follows:
{P} = [Ks]{u} = [F.rHu] (5)
Stiffness Matrix of Pile-Group System.—The load-deformation rela-
tionships of the group piles and the soil are first considered separately.
The stiffness relationship of the pile-group system is then derived by
considering the equilibrium of the pile-soil interaction forces, and the
compatibility of displacements of the pile and the soil. This relationship
may be written as
{Q}=[K]{u} (6)
in which {Q} = external applied loads; and [K] = stiffness matrix of pile
group system (= [Kp] + [Ks]).
The solution procedure for the analysis of a three-dimensional pile
group may be summarized as follows:

1. Determine the pile element stiffness matrices in the local axes, and
transform these matrices from local to global axes using Eq. 1.
2. Assemble the element matrices to give the stiffness matrix of the
group piles [Kp].
3. Determine the nodal soil stiffness matrices in the local axes, and
642

J. Geotech. Engrg. 1987.113:637-651.


transform these matrices from local to global axes using Eq. 2.
4. Invert the nodal soil stiffness matrices to give the nodal soil flexi-
bility matrices.
5. Assemble the nodal soil flexibility matrices into the global soil flex-
ibility matrix [Fs],
6. Evaluate the off-diagonal flexibility coefficients of [Fs] using Mind-
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

lin's solutions.
7. Invert the soil flexibility matrix [Fs] to give the soil stiffness matrix

8. Assemble the pile and soil stiffness matrices to give the stiffness
matrix of the pile-group system [K] (Eq. 6).
9. Solve the stiffness relationship in Eq. 6 to give the deformations of
the pile group with respect to the global axes.

The nonlinear behavior of the pile group is assumed to be due to the


nonlinear soil response at the individual piles, while pile-soil-pile inter-
action is assumed to be linear elastic. Thus, the incremental forms of
Eqs. 2, 3, 4, and 5 will be required, with the tangent stiffnesses of the
soil evaluated from the appropriate load-transfer curves at the relevant
load level. The foregoing solution procedure is then executed at each
load increment.
The numerical procedure presented is completely general, and is able
to consider three-dimensional pile groups with piles raked in any direc-
tion. There is, however, a certain limitation insofar as computer re-
sources are concerned. Although the stiffness matrix of the group piles
[Kp] is banded (if the pile nodes are properly numbered), the soil flex-
ibility matrix [Fs] and hence the soil stiffness matrix [Ks] are generally
fully populated. With six degrees of freedom at each nodal point for a
general pile group, the computer resources required for core storage as
well as solution time would be significantly large for practical pile groups.
Accordingly, a computer program has been developed in which only
piles inclined in the direction of horizontal loading can be considered.
In this case, the degrees of freedom at the nodes have been reduced to
three (two translation and one rotation). Even with this assumption, a
wide range of practical pile-group problems can be solved. Note that this
limitation also exists in two widely used pile group computer programs,
namely PGROUP and DEFPIG.
Two features can be exploited to reduce the size of the problem: (1)
The symmetry of the loading conditions and the arrangement of the group
piles; and (2) a graded discretization of the piles, with smaller elements
at the top; this is particularly important when the flexural responses of
the group piles are dominant.

COMPARISON WITH THEORETICAL RESULTS

The availability of theoretical results for the performance of a typical


3 x 2 pile group obtained using three widely used pile-group-analysis
computer programs, namely PIGLET, DEFPIG and PGROUP (Poulos and
Randolph 1983; Randolph 1985), provides an excellent opportunity to
compare the accuracy of the proposed approach described in this paper.
The pile group examined is illustrated in Fig. 3. Analyses were per-
643

J. Geotech. Engrg. 1987.113:637-651.


S
s
1" '1
°5
PLAN
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

o2 3o=
l/d = 25
Jj/^M s/d = 3
Ep/Es = 103
Vc = 0-5

ELEVATION

—d

FIG. 3.—Configuration of Six-Pile Group

Present Approach
— DEFPIO

FIG. 4.—Comparison of Group Deflection Influence Factors

644

J. Geotech. Engrg. 1987.113:637-651.


0 7-5 15 7-5
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

1
fl-7 - 1
•s^T'"' •
4

kn
01
C|M
-0-01 x^
^ ^ 1
0

. Present Approach PIGLET


• DEFPIC PGROUP

FIG. 5.—Comparison of Load Influence Factors

formed for cases where the corner piles were raked in the same plane
at angles of 0°, 7.5°, and 15°.
The vertical (v) and horizontal (u) deflections, a n d rotation (G) of a pile
group u n d e r vertical load, V, horizontal load, H, a n d m o m e n t , M, m a y
be written as
V H M
(7a)

v_luv + _H_ M
~^~~, I«H + TTTi I»M (7b)
' Esd E.i" Esd2

V H 1 M
2 *ev + TTTihH
*H + hi (7c)
Esd Esd* E5d3
in which E s = Young's m o d u l u s of the soil; d = pile diameter; a n d IvV,
UH, IVMI hv, l«H, IUM, hv, UH, a n d lm = deflection influence factors.
Fig. 4 shows a comparison of the deflection influence factors com-
puted using the four methods. There is a good general agreement among
the four solutions. The present approach is in closest agreement with
PGROUP—the most rigorous of the four computer programs. N o t e that
UH = IUM from the reciprocal theorem, a n d IvH, IvM, IuV, a n d Iw are zero
for symmetrical pile groups.

645

J. Geotech. Engrg. 1987.113:637-651.


The axial (A) and lateral (L) loads and the bending moment (B) at the
heads of the individual piles in the group may be written as

A, = VC,v + HCaH + -CaM (8a)


Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

Li=VClv + HCm + -Cm (8b)


a
B, = VCwd + HCbHd + MCm (8c)
Fig, 5 shows the computed load influence factors for piles 3 and 6. There
is a good general agreement in the influence factors obtained using the
four approaches with the exception of CIV and Cw, where PIGLET, in
particular, gave significantly different values. There is, again, an excel-
lent agreement between the present approach and that of PGROUP. Since
PGROUP is the most rigorous of the four programs, it appears that PIG-
LET overpredicts the lateral loads and bending moments in piles 3 and
6 under vertical group loads when the angle of rake is significant.

COMPARISON WITH TEST RESULTS

There are few comprehensive test results on battered pile groups. Apart
from the instrumentation difficulties, the costs involved in such a ven-
ture (especially of full-scale field tests) can be astronomical. In this sec-
tion, full-scale tests and a laboratory model test reported by Feagin (1953)
and Davisson and Salley (1970), respectively, are analyzed. Results from
other approaches are included for comparison where possible.
Field Tests of Feagin (1953).—Feagin (1953) reported a series of full-
scale field tests on groups of vertical and battered timber piles with heads
fixed in concrete monoliths. The embedded lengths of the piles were
9.14 m (30 ft). The mean diameters at the pile head and pile base were
330 mm (13 in.) and 229 mm (9 in.), respectively. The piles were driven
at about 914 mm (3 ft) centers by a Vulcan No. 1 hammer into fine to
coarse sand. The battered piles were inclined 20° with the vertical.
In the present analysis, the soil Young's modulus was assumed to
vary linearly with depth, and the appropriate value was obtained by
fitting the theoretical solution to the measured field value for a group
of eight vertical piles. The linearly varying modulus of subgrade reaction
for the flexural response of the individual piles was deduced through
comparisons with the solution by Randolph (1980). Since the pile heads
were cast into concrete monoliths 1.52 m (5 ft) thick, calculations were
carried out assuming two different boundary conditions for the concrete
blocks: (1) No rotation of the pile cap permitted; and (2) rotation al-
lowed. The linearly varying soil Young's moduli deduced for these two
conditions were different: 15,750 kPa/m (58 Ib/cu in.) and 19,000 kPa/
m (70 lb/cu in.) for conditions 1 and 2, respectively. The soil Poisson
ratio was assumed to be 0.3.
The computed responses of the pile groups are presented in Fig. 6
with the measured values and solutions from PGROUP, PIGLET, and
DEFPIG (Banerjee and Davies 1980; Poulos and Randolph 1983). Note

646

J. Geotech. Engrg. 1987.113:637-651.


Lateral Load (kN/Pile)
Schematic Diagram for 8-35 mm Deflection
(3

OJ .2
a> at

(T)
£=
a> o
«l

ll Li_l

a.
C3
Q_
UJ

it)
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

111- (2) (3) (5) (6) (7) (8)


0 O 1

1 t2 o ol 427 42-7 427 42-7 42 7 42-7


^
o o|

2
m

H
O
o o o

'
0

3-66
O

:h -tinu nu iiu n*u


•'
f?|| n n -
51 6 400 44-5 47-2 43-6 44-5

sir
ft O O

3 9 O O In Sjll r^il ft n n n 62-3 596 605 650 64-1 70 3


1
Ml ' ^ UU u

i
O O 0
9 9 9
4 O O 0
""• f? St If 63 2 63-2 62-3 60-5 66 7 75-6
' * 3.05 "
MM£
» o o
all
5 sir i-"
9 O O

2-71 - 1 tf U U
650 72-1 71-2 72-1 75-6 77-4

6 fl '•
B O O

3-66
:l-
•' TTrrtr
801 694 62-3 74-7 66-7 72-1

7 i 0
9
9
9

3-66
9

all
i—i 1 n i

ii u tf if
n n 801 82.8 47-2 730 48-9 49-8

B
mf 10 9 ©
•"S 1 O 0 9 :l- gii n n ~
140-6 1041 91-6 104-1 101-4 113-9
^ l t ^ TuX%
O O 9

9
h"
0 O

3-66
9 9

H
~\U ^wT
1) ^ W
97 9 69-4 62-3 - 66 7 72-1

o Vertical Pile • Battered Pile


a
No rotation of pile cap
b Rotation of pile cap permitted
Note: 1 ton = 8 9 0 kN, 1 in. = 25-4 mm
All dimensions in m.

FIG. 6 . — C o m p a r i s o n with Field Tests of Feagin (1953)

that regardless of the boundary conditions assumed for the pile cap,
both solutions generally agree reasonably well with the test results. In
the case where pile cap rotations were not permitted, the result for test
2 appears to be somewhat low. When rotations of the pile cap were
permitted, the present results are consistent with those obtained using
PIGLET and DEFPIG, in which the same assumption was made.
Note that the back-figured values for the linearly increasing Young's
647

J. Geotech. Engrg. 1987.113:637-651.


TABLE 1.—Comparison with Model Test

Pile Moment at Pile Head (Nm)


number Measured Present approach PGROUP PIGLET Measured
(1) (2) (3) (4) (5) (6)
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

1 1.30 1.55 1.06 1.18 22.2


2 1.27 1.55 1.06 1.18 23.1
3 1.20 1.36 1.13 1.17 16.9
4 1.18 1.36 1.13 1.17 16.0
5 1.49 1.53 1.23 1.55 23.1
6 1.22 1.53 .1.23 1.55 16.9
Note: 1 lb-force-in. = 0.113 Nm; 1 lb-force = 4.45 N.

moduli in PGROUP, PIGLET, and DEFPIG were 10,860 kPa/m (40 lb/
cu in.), 12,490 kPa/m (46 lb/cu in.), and 21,990 kPa/m (81 lb/cu in.),
respectively. It appears that these differences arise from the different
approximations made in determining pile-soil-pile interaction in non-
homogeneous soils.

U
381 mm
m

e 5
o 3
o 1
o
E
en
Csl
o
6
0
4
o
2
]
\aSB-
-a 127 mm
127 mm

i 222 N

138 N

UJ
FIG. 7.—Details of Pile Group Model

648

J. Geotech. Engrg. 1987.113:637-651.


Results of Davisson and Salley (1970)
Shear at Pile Head (N) Axial Load at Pile Head (N)
Present Present
approach PGROUP PIGLET Measured approach PGROUP PIGLET
(7) (8) (9) (10) (11) (12) (13)
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

20.0 16.5 14.4 80.1 63.5 68.1 73.8


20.0 16.5 14.4 56.9 63.5 68.1 73.8
16.1 17.3 14.1 28.9 38.2 36.5 34.6
16.1 17.3 14.1 24.0 38.2 36.5 34.6
19.0 18.7 19.6 12.2 13.7 10.0 5.4
19.0 18.7 19.6 13.6 13.7 10.0 5.4

Model Test of Davisson and Salley (1970).—A model test on a bat-


tered pile group in fine, fairly uniform sand subjected to a combined
vertical load of 222 N (50 lb) and a horizontal load of 138 N (31 lb-force)
was reported by Davisson and Salley (1970). The aluminum pipe piles
had an outer diameter of 12.7 mm (0.5 in.) with 0.8-m (0.03-in.) wall
thickness, and an embedded length of 533 mm (21 in.). Details of the
pile-group configuration are given in Fig. 7.
In the analysis, the pile cap was assumed to be rigid, and rotation of
it was permitted. The soil Young's modulus was assumed to vary lin-
early with depth, and was taken to be 19,000 kPa/m (70 lb/cu in.), which
is the same as in the previous example. The Poisson ratio of the soil was
assumed to be 0.3.
The computed horizontal deflection of 0.24 mm (0.0096 in.) agrees rea-
sonably well with the measured value of 0.23 mm (0.009 in.). Table 1
summarizes the measured and computed moments, shears, and axial
loads at the pile heads, as well as solutions from PGROUP (Banerjee
and Davies 1980) and PIGLET (Randolph 1980). There is a general agree-
ment among the three theoretical approaches and the measured results
in terms of relative magnitudes of the moments and loads, but the trends
of the moment/load distributions differ somewhat.
In the foregoing examples, linear moduli of the subgrade reaction were
used to model the soil response of the individual piles due to a lack of
experimental or field data on the influence of soil nonlinearity on the
distribution of loads or moments to the individual piles in the battered
pile group. There are, however, some very comprehensive data on full-
scale tests on vertical pile groups (which are special cases of the present
analysis) in which the influence of soil nonlinearity on the performance
of the group has been studied, the analyses of which were reported in
detail by Chow (1986), and Leung and Chow (1986).

CONCLUSION

A numerical method was described for the linear and nonlinear anal-
ysis of three-dimensional battered pile groups. The present approach is
shown to be at least as good as some of the computer programs cur-
rently available for pile-group analysis. Full-scale field tests and a lab-

649

J. Geotech. Engrg. 1987.113:637-651.


oratory model test were analyzed, a n d the results showed reasonably
good agreement with the measured values.

APPENDIX I.—REFERENCES

Banerjee, P. K., and Davies, T. G. (1980). "Analysis of some reported case his-
tories of laterally loaded pile groups." Proc. Int. Conf. Numer. Methods in Off-
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

shore Piling, London, U.K., 101-108.


Banerjee, P. K., and Driscoll, R. M. C. (1976). "Three-dimensional analysis of
raked pile groups." Proc, Institution Civ. Engrs., Part 2, 61, 653-671.
Chow, Y. K. (1986). "Analysis of vertically loaded pile groups." Int. } . Numer.
Anal. Methods Geomech. 10(1), 59-72.
Coates, R. C., Coutie, M. G., and Kong, F. K. (1972). Structural analysis. Nelson,
London, U.K.
Davisson, M. T., and Salley, J. R. (1970). "Model study of laterally loaded pile."
J. Soil Mech. Found. Div., ASCE, 96(5), 1605-1627.
Dutt, R. N. (1976). "Torsional response of piles in sand," dissertation presented
to the University of Houston, at Houston, Tex., in partial fulfillment of the
requirements for the degree of Doctor of Philosophy.
Feagin, L. B. (1953). "Lateral load tests on groups of battered and vertical piles."
STP No. 154. Symp. on Lateral Load Tests on Piles, ASTM, 12-29.
Ha, H. B. (1976). "Analysis of generally loaded non-linear three-dimensional pile
groups considering group effects," dissertation presented to the University of
Houston, at Houston, Tex., in partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
Hrennikoff, A. (1949). "Analysis of pile foundation with batter piles." Trans.,
ASCE, 79, 351-374.
Leung, C. F., and Chow, Y. K. (1986). "Response of pile groups subjected to
lateral loads." Int. J. Numer. Anal. Methods Geomech (in press).
Mindlin, R. D. (1936). "Force at a point in the interior of a semi-infinite solid."
Physics 7, 195-202.
O'Neill, M. W. (1964). "Determination of the pile head torque-twist relationship
for a circular pile embedded in a clay soil," thesis presented to the University
of Texas, at Austin, Tex., in partial fulfillment of the requirements for the de-
gree of Master of Science.
O'Neill, M. W., Ghazzaly, O. I., and Ha, H. B. (1977). "Analysis of three-di-
mensional pile groups with nonlinear soil response and pile-soil-pile interac-
tion." Proc. 9th Offshore Technology Conf. 2, 245-256.
Poulos, H. G. (1980). "An approach for the analysis of offshore pile groups."
Proc. Int. Conf. Numer. Methods in Offshore Piling. London, U.K. 119-126.
Poulos, H. G., and Davis, E. H. (1973). Elastic Solutions for soil and rock mechanics.
John Wiley and Sons Inc., New York, N.Y.
Poulos, H. G., and Davis, E. H. (1980). Pile foundation analysis and design. John
Wiley and Sons Inc., New York, N.Y.
Poulos, H. G., and Madhav, M. R. (1971). "Analysis of the movements of bat-
tered piles." Proc. 1st Australia-New Zealand Conf. on Geomech., Melbourne,
Australia, 1, 268-275.
Poulos, H. G., and Randolph, M. F. (1983). "Pile group analysis: a study of two
methods." /. Geotech. Engrg., ASCE, 109(3), 355-372.
Randolph, M. F. (1980). "PIGLET: A computer program for the analysis and de-
sign of pile groups under general loading conditions." Soil Report TR91, CUED/
D. Cambridge Univ., England.
Randolph, M. F. (1981). "Analysis of the behavior of piles subjected to torsion."
/. Geotech. Engrg. Div., ASCE, 107(8), 1095-1111.
Randolph, M. F. (1985). "Analysis and design of pile groups using a microcom-
puter." Proc. 5th Int. Conf. Numer. Methods in Geomech., Nagoya, Japan, 3,
1707-1714.
Randolph, M. F., and Wroth, C. P. (1978). "Analysis of deformation of vertically
loaded piles." /. Geotech. Engrg. Div., ASCE, 104(12), 1465-1488.

650

J. Geotech. Engrg. 1987.113:637-651.


"Recommended practice for planning, design a n d constructing fixed offshore
platforms." API RP2A. (1982). Amer. Petroleum Inst.
Reese, L. C , O'Neill, M. W . , a n d Smith, R. E. (1970). "Generalized analysis of
pile foundations." /. Soil Mech. Found. Div., ASCE, 96(1), 235-250.

APPENDIX If.—NOTATION
Downloaded from ascelibrary.org by Coffey Geotechnics P/L on 04/14/14. Copyright ASCE. For personal use only; all rights reserved.

The following symbols are used in this paper:


At = axial load in pile i;
Bj = bending m o m e n t at pile i;
Cav i Cati, CaM, Civ, Cm,
Q M / Cw, Cm, Cm = load influence factors;
d = pile diameter;
Ep = Young's modulus of pile material;
Es = Young's m o d u l u s of soil;
[Fs] = soil flexibility matrix;
[F*] = soil nodal flexibility matrix;
fij = flexibility influence coefficient;
H = horizontal load acting o n pile group;
IvV i IvH , IvM i IuV i hit i
IUM / hv, hti i UM — deflection influence factors;
[K] = stiffness matrix of pile-group system;
[K ] = pile element stiffness matrix;
[Ke] = pile element stifness matrix (local axes);
[Kp] = assembled stiffness matrix of group piles;
[Ks] = soil stiffness matrix;
[Kf] = soil nodal stiffness matrix;
[Kf ] = soil nodal stiffness matrix (local axes);
L, = lateral load at pile z;
I = pile length;
M = m o m e n t acting on pile group;
n = total n u m b e r of degrees of freedom;
{P} = soil reaction load vector;
Pj = soil reaction load at /;
{Q} = external applied load vector;
s = center-to-center pile spacing;
[T] = transformation matrix of pile element stiff-
ness matrix;
[T*] = transformation matrix of soil nodal stiff-
ness matrix;
u = horizontal deflection of pile group;
{w} = deformation vector;
U; = deformation at degree of freedom i;
= displacements in three orthogonal direc-
tions (local axes);
V = vertical load acting on pile group;
v = vertical deflection of pile group;
x,y,z = global axes;
x,y,z = local axes;
9 = rotation of pile group;
9* / % , 8z = rotations/twist about local axes; a n d
v, = Poisson ratio of soil.

651

View publication stats J. Geotech. Engrg. 1987.113:637-651.

You might also like