Professional Documents
Culture Documents
Computers and Geotechnics: A Pile Group Foundation by A Three-Dimensional Nonlinear
Computers and Geotechnics: A Pile Group Foundation by A Three-Dimensional Nonlinear
ABSTRACT
INTRODUCTION
various components (namely, the pile-cap, the piles, and the soil medium) [1-14]. Most of
the previous studies have been concerned with vertical piles and relatively few investigations
have involved battered piles [7]. For three-dimensional finite element analysis, inclined piles
pose difficult problems, especially from the point view of mesh generation.
For a structure having potential for high lateral loads it is desirable to consider in the
analysis the battered pile geometry and nonlinear behavior of soil medium, as well as the
various components of the pile group foundation together as a three-dimensional (3-D)
system, so that a realistic response of the system can be obtained.
In this paper, the behavior of a pile group foundation subjected to arbitrary 3-D
loading is analyzed by using an incremental-iterative finite element procedure. The material
nonlinearity of the soil medium is represented by a generalized plasticity model [7,15,16].
An algorithm based on the interpolation of nodal displacements and finite difference
approximation of the displacement field is used to evaluate the bending moments and the
shear forces in the piles and the pile cap. An analytical parametric study is conducted to
investigate the effects of pile inclination and pile-cap thickness on the overall behavior of the
system, particularly on the distribution of forces in individual piles, and on the distribution
of bending moments and shear forces in each pile.
Approximating the soil surrounding the pile by a Winkler medium [17] and using the
matrix approach to obtain the individual pile forces in a pile group foundation was first
proposed by Hrennikoff [6]. Aschenbrenner [1] extended this approach for three-dimensional
analysis of pile group foundations where piles were assumed to be hinged at the pile cap.
A modified Hrennikoff's method was employed by Reese and O'Neill [14] to analyze the
behavior of a three-dimensional battered pile-foundation system. The torsional response was
taken into consideration in this analysis, but the axial, lateral, and torsional behaviors of the
pile group were assumed to be linearly independent. The principle of superposition was used
to obtain the total pile head forces. Accordingly, this method could not be used for problems
involving nonlinearity.
The procedure proposed by Desai, et al. [5] utilizes the beam-column, plate and
nonlinear spring elements to idealize the behavior of piles, cap, and foundation, respectively.
The pile cap was treated as a thin plate. Wittke [14] used the finite element method (FEM)
to study the behavior of a laterally loaded single pile in a homogeneous and linearly elastic
soil by using a three-dimensional idealization. Ottaviani [11] examined the behavior of a
vertically loaded single pile as well as a pile group by invoking similar assumptions. In these
studies no relative displacement was allowed at the interfaces between the piles and the soil.
Poulos [19] and Butterfield and Banerjee [3] assumed a symmetrical distribution of shear
stress around the axially loaded pile groups which contradicts the findings of Ottaviani [11].
67
The load transfer mechanism was also studied considering the presence of a pile cap.
Faruque and Desai [20] performed a three-dimensional analysis of a single pile including both
material and geometric nonlinearities. Muqtadir and Desai [9] have reported results of a fully
three-dimensional finite element analysis of a pile group where both the material and the
boundary nonlinearities were taken into consideration. The results of this study were
compared with experimental observations reported by Frucco and Associates [21 ] for a model
pile group.
The finite element (FE) method is well documented in the literature [22-26]. Hence,
it is not detailed herein. Only a brief description is presented.
and
Here, [k] is the element stiffness matrix, {Q} is the element load vector, { x"} is the body
force vector, { T } is the surface traction vector, and {a0} is the initial stress vector. Also,
[13] represents the strain-displacement transformation matrix and [C] e+ is the elasto-plastic
constitutive relation matrix.
The stiffness matrices for all elements are evaluated and assembled and the load vector
is assembled to obtain the global equations of equilibrium. A 2x2x2 Gaussian quadrature
68
[26] integration scheme is adopted to evaluate the element stiffness matrices and the load
vectors. The frontal solution technique developed by Irons [24] and described by Owen and
Hinton [25] is used for the solution of the nonlinear equilibrium equations for the system.
Constitutive Models
The code used in this study has a provision to account for various constitutive models
in the analysis. The models implemented in this code include linear elastic model, nonlinear
elastic (Hyperbolic) model, variable moduli model, Drucker-Prager model, critical state
model, cap model, and the generalized plasticity model developed by Desai and his
co-workers [15,20,27]. In this study, the pile cap and the piles are assumed to be linearly
elastic. Nonlinearity of the soil medium is represented by the generalized plasticity model
in which both yielding and failure surfaces are described by a single mathematical function
to avoid computational difficulties at the junction of yield and failure surfaces which may
arise in two-surface models such as cap model [27].
In the generalized plasticity model used here, the yield function is represented in the
form [15,16]
where J2D = second invariant of the deviatoric stress tensor, Jl, J3 = first and third
invariants of the total stress tensor, respectively, k = measure of the cohesive strength of the
material, and a, $, y = material response functions. The response function B is made
dependent on the history of deformation and is called the growth or evolution function,
while a, y and k are treated as material constants. The explicit form of B can be expressed
as follows
where ~ is the trajectory of plastic strain, formed by the components of the plastic strain
tensor, ~ , according to the following expression
I
- f (6)
¥ {aF}
ao ~o ~ a~
where [C] = constitutive relation matrix for linearly elastic behavior, {0F/0o} = gradient
of F with respect to stress tensor aij and YF = length of the gradient vector in the stress
space.
Problem Description
The problem under consideration is a pile group foundation as shown in Fig. 1. The
group consists of four vertical piles having uniform cross-section (0.45 m x 0.45 m) and
equal length (14.0 m), fully embedded in the soil medium. The pile cap is 3.0 m x 3.0 m
in plan dimensions. The piles are arranged symmetrically in both x and y directions. The
cap is subjected to a vertical load (P) and a horizontal load (Q). Considering the symmetry
of the problem with respect to the X-axis, only half of the pile cap is analyzed here.
Therefore, due to this symmetry, only half of the vertical load (P/2) or/and the horizontal
load (Q/2) need to be applied to this system. Also, note that results applicable to pile #1 are
also applicable to pile #3, and results applicable to pile #2 are also applicable to pile #4. For
brevity, only results pertaining to pile #1 and Pile//2 will be discussed in this paper.
70
P
PileCap
7z //I
//': I
:
,
z ¥ I
Sle #
~ Pile01
_4
T
t
[
-f
I 3.Ore
IIIII!1 .....
l llllll
IIIIII ,
25.0
IIIIII
!111111
A
S
SS
•
"Iljl/
i
I
B q>
I__
! - 20.0 m -~/'//~"
these boundary conditions, makes the total number of active DOF to be 1564.
Material Properties
The piles and the cap are assumed to behave as linearly elastic during the loading
process. The following material properties are used for the generalized plasticity model to
represent the nonlinearity of the soil medium: Young's modulus, E s = 1.12 x l05 kN/m 2,
Poisson's ratio v s = 0.36, a = 0.212, V = 25.28 k N / m 2, k = 0.0 kN/m 2, B = 0.005, and
n = 0.74. While the elastic properties used for piles and pile cap are: Young's modulus, Ep
= 3.0 x 107 kN/m 2 and Poisson's ratio, vp = 0.17.
Calculation of Flexural Moment Along the Central Line of the Pile Cap
To calculate the flexural moments along the central line, a finer and uniform (square)
grid of the pile cap is constructed. For the numerical example considered herein, each
73
element in the original FE grid has the dimensions of 0.45 m x 0.45 m (Fig. 2). A refined
FE grid is constructed from the original grid. Each element in the refined grid has the
dimension of 0.15 m x 0.15 m. The total number of nodes used to construct the finer,
uniform grid is 231. Nodal displacements in the vertical (Z) direction for all nodes of the
uniform grid are obtained from the nodal displacements of the original FE mesh of the pile
cap (see Fig. 3). The central finite difference technique is used to calculate the bending
moment along the central line of the pile cap using the nodal displacements of the uniform
grid. The bending moments at the free edges of the pile cap are assumed to be zero. Note
that, the bending or the flexural moments reported here for the pile cap or the piles are the
moments about the Y-axis only.
PARAMETRIC STUDY
R, E, t 3
(8)
12(1 - v2p)
where Ep and vp are the elastic properties of the pile cap material, and (t) represents the
thickness of the pile cap. Assuming the rigidity of the pile cap (Rp) is 1.0 at t = 0.25 m,
pile cap rigidities at other thicknesses are calculated, relative to the pile cap rigidity at t =
0.25 m.
TABLE 1: Pile Head Forces Due to Variation in Pile Cap Thickness (P=400 kN, Q = 0 kN)
Pile #1 or #3 Pile #2 or #4
(cm) Fx Fy Fz Fx Fy Fz
TABLE 2: Pile Head Forces Due to Variation in Pile Cap Thickness ( P = Q = 4 0 0 kN)
Pile #1 or #3 Pile #2 or #4
(cm) Fx Fz Fx Fz
Distribution of axial and shear forces in pile #1 for different pile cap thicknesses are
75
shown in Figs. 4 and 5, respectively. The difference in the axial force distribution in pile
#1 due to change in pile cap thickness seems to be predominant at shallow depth and
maximum at the pile head, where it is connected to the pile cap. Also, it can be observed
that as the thickness of the pile cap increases the axial force in pile #1 increases in
magnitude. An opposite trend is observed for the axial force distribution in pile//2 [10].
Here the axial force decreases in magnitude with the increase in pile cap thickness.
Fig. 5 shows that the difference in the shear force distribution for different pile cap
thicknesses along pile #1 is negligible at depths more than 6.0 m, and noticeable only at
shallow depth, especially at the pile head. Also, it is observed that as the thickness of the
pile cap increases, the shear force distribution along pile #1 decreases especially in the upper
part of the pile. However, the change appears to be insignificant.
Further, from Figs. 4 and 5, it is noted that the axial and shear forces in both piles
are maximum at the surface and minimum at the tip, as expected. A part of the axial load
carried by the pile is transferred to the soil medium through skin friction. The shear force
is carded by the pile only at the surface, while it is carried by both the pile and the
surrounding soi~ along the embedded length of the pile, As a result, the axial and shear
forces in both piles reduce with depth. The axial force distribution is more sensitive to the
variation of the pile cap thickness than the shear force distribution. As the cap becomes more
rigid, it induces a pulling effect in pile//2 due to application of lateral load. Thus, the
soil-structure interaction effects not only depend on the characteristics of the piles and the soil
but also on the pile cap and the superstructure (not included here).
The distribution of flexural moments (about the y-axis) along the central line of the
pile cap for different pile cap thicknesses is shown in Fig. 6. It is observed that the variation
of the pile cap thickness has a significant effect on the moment distribution in the pile cap.
The rate at which the flexural rigidity of the pile cap increases with the increase in pile cap
thickness is more than the rate at which the flexural displacements of the cap decreases. As
a result, the flexural moment in the pile cap increases with increasing pile cap thickness.
The distribution of flexural moments (about the y-axis) in pile #1 for different pile
cap thicknesses is shown in Fig. 7. It is observed that the change in pile cap thickness does
not cause a significant change in the flexural moment, except near the pile head region where
the change in shear force is maximum (see Fig. 5). The difference in the flexural moments
reaches maximum at the pile head. A cap with relatively large thickness behaves
approximately as a rigid body, compared to a cap of smaller thickness. This varying
interaction between the pile cap and the piles affects the mechanism of load transfer between
different piles in a group. Simplified analyses do not account for such interaction effects.
Pile head moments of pile #1 and pile//2 for different pile cap thicknesses are shown
in Table (3). The maximum negative moments induced in both piles for different cap
thicknesses are shown in Table (4). It can be observed from Table (3) that the pile head
moment in pile #1 is maximum at t = 0.15 m, while minimum in pile//2. Note that this is
76
?(
8.
1,
II.
13.
Figure 4 Distribution of Axial Force Along Pile #1 for Different Pile Cap Thicknesses
(P = 400 kN, Q = 400 kN)
0 i
: ,. , t--0. 6 0 m
L~ t=0.15 m ~/ ~ t=0.45 m
)i t=0.35 m
I,
i,
?,
|,
eL
lb
L3,
t $.
Shear Force, kN
Figure 5 Distribution of Shear Forces Along Pile #1 for Different PLle Cap Thicknesses
(p = 400 kN, Q = 400 kN)
77
I
+aaI f,\ .t--o'6° m
! :~ t=0.45 m
| I1\ \\
| II ,, \
~ooJ II \,, ,, t=0.35 m
] I1,', \ ~\/
I It: ". ~
| ~l/ ",. \ , / ~ t=o. 25 m
.........
:,r ~ .... ,k : ":-- ,
i'
!
I ~% /I
- ~00, i I
-SOIl.
_ • .t +_. . .
o.o o.o ~.o ~.s ~.o m[s i;o
Distance, m
Figure 6 Distribution of Moment Along the Central Line of the Pile Cap for Different
Pile Cap Thicknesses (P -- 400 kN, Q = 400 kN)
0 ,
.t=O. 15 m
z~ f/ t=0.25 m
3, ~,~jt--O. 45 m
~. ~ t=0.35 m
1.
I.
?,
I0.
I1,
Ii.
LJ,
LI4,
11,
Moment, kN.m
Figure 7 Distribution of Moment Along Pile #1 for Different Pile Cap Thicknesses
(P = 400 kN, Q = 400 kN)
78
in total agreement with results reported in Table (2). In view of Table 4, it is observed that
the negative moments induced in both piles are maximum at t = 0.15 m, while minimum at
t = 0.60 m.
TABLE 3: Pile Head Moments Due to Variation in Pile Cap Thickness ( P = Q = 4 0 0 kN)
The finite element (FE) mesh used for the vertical piles is also used for the battered
piles with the same number of nodes and elements except with some modification in the mesh
configurations. The FE mesh for the battered cases is achieved by rotating (in the region
where the piles are located) the two sides of the finite element grid for the vertical case in
the xz plane, to obtain the required slope for each battered case. The regions away from the
piles are kept unchanged.
TABLE 5: Pile Head Forces Due to Variation in Pile Inclination (P=400 kN, Q = 0 kN)
The distribution of axial forces in pile #1 for different inclination of piles is shown
in Fig. 9. The corresponding distribution of shear forces is also shown in Fig. 10. Fig. 9
depicts that the axial force in pile #1 at and near the pile head increases as the angle of
inclination of the pile increases. An opposite trend is observed in the middle portion. The
axial forces at the tip portions of the pile do not change appreciably due to variation in pile
inclination. From Fig. 10, it is observed that as the angle of inclination of piles increases,
the positive shear forces increase whereas the negative shear forces decrease in pile #1. The
change in the angle of inclination of piles has significant effects near and at the surface
(where piles are connected to the cap).
The distribution of moments along pile #I for different inclination of piles is presented
in Fig. 11. While, pile head moments and maximum negative moments for both piles are
shown in Table 7 and 8. In view of Fig. 11 and Tables 7 and 8, it is noted that as the angle
of inclination of piles increases, the negative moments increase whereas the positive moments
decrease in pile #1. An opposite trend is seen for the moment distribution in pile #2 [10]
(i.e., the negative moments decrease and the positive moments increase with increasing
inclination). As the angle of inclination increases the normal component of displacement
decreases in pile #1 and increases in pile #2. This causes the flexural moments along the
length of the pile to increase and the pile head moments to decrease for pile #1. An opposite
trend is observed for pile #2. The difference in the magnitude of moments appears to be
large when the pile geometry is changed from vertical to battered (5:1). Due to the change
in the angle of inclination, the axial component of displacements increases in pile #1 while
decreases in pile #2. This causes the axial forces to increase in pile #1 and to decrease in pile
#2. This observation supports the results in Table 6. An opposite trend is noticed for the
distribution of shear forces (i.e., shear forces decrease in pile #1 and increase in pile #2,
81
_
P+tO #I
I_
Pile / ~
$ %.!
a) Case #I b) Case ~ c) Case
L.
3. Battered 4:1
¼,
$.
I.
Battered 5:1
7,
I.
t. Ver
L0.
II.
|Z'
L3,
Lg,
t|, ',, , -, ,
Figure 9 Distribution of Axial Force Along Pile #1 for Different Inclination of Piles
(P - 400 kN, Q = 400 kN)
82
because the normal component of displacements decreases in pile #1 and increases in pile
#2).
CONCLUDING REMARKS
/ .i+
I //
4,
2 Battered 4:1 ~ i'+'
7q
$,
Battered 5 : i ~ (,"
!.
ta,
Vertical / ~
tl,
1
t2, I
I
I
I |,
ill,
I
tl'
-rio -,&, .N -~o -,~o -~,, ~ +o"
Shear Force, kN
Figure I0 Distribution of Shear Force Along Pile #I for Different Inclinationof Piles
(P = 400 kN, Q = 400 k.N)
-, ".,
• , ".
,, ',
Et IVertical
Battered 4 :1 ',,'..
~u i .
+,'.
tO.
II,
Ii,
%
I |.
I|.
Moment:, kN.m
ACKNOWLEDGEMENTS
The authors wish to express their thanks and gratitude to Professor C.S. Desai,
University of Arizona, Tucson for providing the original computer code.
REFERENCES
. Butterfield, R. and Banerjee, P.K., The Problem of Pile Group-Pile Cap Interaction.
Geotechnia_ue, 21 (1971) 135-142.
4. Desai, C.S. and Appel, G.C., 3-D Analysis of Laterally Loaded Structures.
Proceedine. 2nd International Conference on Numerical Methods in Geom~hanics,
Blacksburg, Virginia, 1 (1976) 405-418.
. Desai, C.S., Kuppusamy, T. and Alameddine, A.R., Pile Cap-Pile Group Soil
Interaction. Journal of Structural Division. ASCE, 107 (1981) 817-834.
I0. Najjar, Y.M., Three-Dimensional Nonlinear Finite Element Analysis of Pile Groups.
M.S. Thesis, Dept. of Civil Engineering and Environmental Science, University of
Oklahoma, Norman, OK (1986).
11. Ottaviani, M., Three-Dimensional Finite Element Analysis of Vertically Loaded Pile
Groups. Geotechnique, 25 (1975) 159-174.
12. Poulos, H.G., Behavior of Laterally Loaded Piles: I-Single Piles. Journal of S0il
Mechanics and Foundation Division. ASCE. 92 (1971) 711-731.
14. Reese, L.C. and O'Neill, M.W., Generalized Analysis of Pile Foundations. Journal
of Soil Mechanics and Foundation Division. ASCE. 96 (1970) 235-249.
15. Desai, C.S. and Faruque, M.O., Constitutive Model for (Geological) Materials. J.
Eng. Megh. Div., ASCE, ll0 (1984) 1391-1408.
17. Hrennikoff, A., Analysis of Pile Foundation with Batter Piles. Proceedings. ASCE,
79 (1949).
18. Wittke, W., Horizontal Belastetem Grossbobrepfahlen Noch Der Methods Finite
Element. Der Banirgenienr, 449 (1974).
19. Poulos, H. G., Analysis of the Settlement of Pile Groups. Geotechnio_ue, 18 (1968)
449-471.
20. Faruque, M.O. and Desai, C.S., Implementation of General Constitutive Model for
Geological Materials. InL J. for Num. Anal. Meth. in Geomechanics, 9 (1985)
415-436.
21. Frucco and Associate, Pile Driving and Loading Tests: Lock and Dam No. 4,
Arkansas River and Tributaries, Arkansas and Oklahoma. U.S. Army Coros of
En2ineers District, Little Rock, Arkansas (1974).
22. Bathe, K.J. and Wilson, E.L., Numerical Methods in Finite Element Analysis.
Prentice-Hall, Englewood Cliffs, New Jersey (1976).
86
23. Desai, C.S. and Abel, J.F., Introduction to the Finite Element Method. Van
Nostrand Reinhold Company, New York (1972).
24. Irons, B.M., Frontal Solution for Finite Element Analysis. Int. J. Num, Meth. of
En__~gg~.,Vol. 2, John Wiley and Sons, England (1970).
25. Owen, D.R.J. and Hinton, E., Finite Elements in Plasticity. Pineridge Press Limited,
Swansea, U.K (1980).
26. Zienkiewicz, O.C., The Finite Element Method in Enginosring Science. 3rd Edition,
McGraw-Hill Book Co., New York (1977).
27. Desai, C.S. and Siriwardane, H.J., Constitutive Laws for Engirmering M~um'ials with
Emohasis on Geological Materia~. Prentice-Hall, Inc., Englewood Cliffs, New
Jersey (1984).