You are on page 1of 18

Compuwrs & Srrurrures Vol. 57. No. 4. pp. 605-622.

1995
Copyright ‘(‘8 1995 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0045-7949(95)00068-2 oo‘ls-7949195 $9.50 + 0.00

NONLINEAR FINITE ELEMENT ANALYSIS OF


REINFORCED CONCRETE FOUR-PILE CAPS

C. Sam? and P. Krishna IyerJ


TDepartment of Civil Engineering, College of Engineering, Trivandrum 695 016, India
IDepartment of Applied Mechanics, Indian Institute of Technology, Madras 600 036, India

(Received II May 1994)

Abstract-An investigation to study the behaviour of reinforced concrete four-pile caps by three-dimen-
sional nonlinear finite element analysis is presented. Four-pile caps having different reinforcement layouts
were considered. Eight-noded isoparametric brick elements with incompatible modes and two-noded
three-dimensional truss elements were used to model the concrete and steel, respectively, of the pile caps.
The steel elements were assumed to he on the edges of the brick elements. The smeared cracking model
was employed for modelling cracks during the analysis. The nonlinear components, such as multiaxial
compressive behaviour of concrete, including strain softening, cracking of concrete and yielding of steel
reinforcement, were considered. The load deflection characteristics, the strains in the reinforcement, the
crack patterns and the ultimate loads are presented herein. Some of the theoretical results are compared
with those obtained from experiments carried out on reinforced concrete pile cap specimens.

1. INTRODUCTION same quantity of reinforcement but with all the


reinforcement concentrated over the piles.
A pile cap, a thick reinforced concrete block, is an
Clarke [IO] conducted experiments on four-pile
important structural element in pile foundation, de-
caps to investigate the influence of different arrange-
signed to transfer loads from a column to a group of
ments of reinforcement. He found that distributing
piles. From the literature available for the last four
the reinforcement uniformly reduced the failure load
decades, it was found that the research into pile caps
by 14%, when compared to pile caps with the same
has been concentrated along two main directions:
quantity of reinforcement concentrated over the piles.
(i) linear elastic analysis; and
Gogate and Sabnis [l I] suggested a procedure for
(ii) tests on reinforced concrete pile cap models. designing thick pile caps. From the experiments
The two approximate methods mainly available for conducted on four-pile caps with varying amounts of
the linear analysis of pile caps are the truss analogy uniformly placed reinforcements, they [ 121 concluded
method and the beam method. Yan [l], and Blevot that reinforcement above 0.2% has little or no effect
and Fremy [2] suggested some modifications to the on the failure loads and the failure was identical to
existing truss analogy method to include the effects of a conical failure under the column.
column size. Banerjee [3], Whittle and Beattie [4], and Adebar et al. [13] conducted experiments on six
Sharma [5] considered the pile cap as a beam span- large models of four-pile caps to examine the suit-
ning the piles. The American [6], British [7] and In- ability of three-dimensional “strut and tie model” for
dian [8] standards assume that a pile cap acts as a the pile cap design. They concluded that the strut and
beam spanning between piles and can be analysed by tie model can accurately represent the behaviour and
the simple beam theory. These methods of analyses failure loads of pile caps.
evaluate the elastic stresses by making simplifying As the pile caps are basic to the safety of structures
assumptions regarding geometry and the material and not open to visual inspection under service, a
properties. Recently a general three-dimensional the- sound knowledge of their exact behaviour is essential.
ory of elasticity solution of pile caps with any number A complete and satisfactory solution can be achieved
of piles has been obtained by the authors [9]. only by performing a nonlinear analysis of the pile
The behaviour of the reinforced concrete pile cap caps throughout the entire load range, taking into
was studied experimentally by many researchers, account the loading history and several nonlinear
among them the following may be cited. effects.
Blevot and Fremy [2] conducted experiments on With this view in mind a detailed investigation of
reinforced concrete pile cap models to investigate the four-pile cap specimens using non-linear finite el-
behaviour due to differences in the reinforcement ement method was undertaken. The nonlinear effects
layout. For four-pile caps they found that spreading considered in the analysis were: (i) multiaxial com-
out the reinforcement uniformly reduced the failure pressive behaviour of concrete including strain soft-
load by 20% when compared to pile caps with the ening, (ii) cracking of concrete, (iii) yielding of steel

605
606 C. Sam and P Krishna Iyer

reinforcement, (iv) tension stiffening and (v) change truss elements were used to model concrete and steel,
in Poisson’s ratio. The other inelastic phenomena respectively. The two-noded three-dimensional truss
such as creep of concrete and bond slip between steel elements used to represent reinforcements were as-
and concrete were not considered. The object of this sumed to lie on the edges of the brick elements [14].
study was to understand the structural behaviour of The steel bars are marked by dotted lines superim-
pile caps, namely (i) the load deflection character- posed over the relevant common edge between
istics, (ii) the strains in the reinforcement, (iii) the concrete elements, as shown in Figs 7, 12 and 15. The
crack patterns and (iv) the ultimate load by perform- concrete material below the reinforcement was
ing a nonlinear analysis. These results were also neglected in the finite element model and hence only
compared with those obtained from tests conducted the effective depth of the pile cap was considered.
on reinforced concrete pile cap specimens. Also the weight of the pile cap was not included. The
This paper describes the results of the nonlinear shape function defining the geometry and the vari-
finite element analysis of four-pile caps having differ- ation of displacement of a three-dimensional eight-
ent reinforcement layouts keeping the geometry, per- noded isoparametric element shown in Fig. la is
centage of reinforcement and concrete grade etc. the given by
same.
N,=1/8(1+rr,)(l+ssi)(l+tt,) i=l,2 ,..., 8
2. FINITE ELEMENT MODELLING OF REINFORCED
CONCRETE AND NONLINEAR ANALYSIS (1)

Eight-noded isoparametric brick elements with in- in which r, S, f are natural coordinates and r,, s,, t, are
compatible modes and two-noded three-dimensional the natural coordinates of the node i.

X Cartesian coordinate system Natural coordinate system

c _ Zi -Zj
z--
L
VZ

Fig. 1. (a) Eight-noded isoparametric brick elements. (b) Two-noded three-dimensional truss elements.
Reinforced concrete four-pile caps 607

The geometry of the element is described as investigated and an improvement has been suggested
by Taylor et al. [16]. The strain vector is given by

‘, (2)

where xi, y, and z, are the global coordinates of node = i$, M4 1+ [P’l{4 (6)

The variation of the displacement inside the el- where [P'] represents the differentiation of [P]with
ement can be expressed using the same shape function respect to the global coordinates x, y and z. Applying
as minimum potential energy, the element equilibrium is
given by
(3)
KWI = h2). (7)

where u,, vi and w, are the nodal displacements of the Since there are no external forces associated with as,
node i in the Cartesian/global coordinate system. the above equation can be written as
One of the main causes of the inaccuracies in the
general eight-noded solid isoparametric element is its
inability to represent certain simple stress gradients
and hence it is not adequate in simulating flexural
response. To simulate the flexural response, Wilson et where
al. and Taylor et al. [15, 161 introduced additional
displacement modes in the general eight-noded solid
element. These additional modes are represented by
PLI = PITK’lPI do

P, = (1 - r2), PZ = (1 - s2), P, = (1 - t2) (4)


K,l = PITKI[f”l dv
in X, Y and 2 directions, respectively, where r, s and
t are natural coordinates and are given by
[K,,] = [P']'[C][l$
do

r = x/a, s = y/b, t= z/c.

These additional modes are called incompatible


K,l = [P’l’KI[f”l dv,
f
modes, as they are not activated at nodes at the
surface. So the displacement approximations for a {Q} represents nodal force vector acting on the
general eight-noded solid element with incompatible elements, jd} represents the nodal displacements,
modes are given by [&I is the stiffness matrix associated with displace-
ment vector {d}, [K,,] is the stiffness matrix associ-
ated with the unconforming generalized
+ [Pl{aj (5) displacements {a}.
Out of the three criteria to be satisfied for conver-
gence it may be noted that the condition pertaining
in which
to the constant-strain state alone is not satisfied.

1
Let {di} be the displacement vector for a constant-
p, p, p, 0 0 0 0 0 0
strain deformation state. Under this condition the
generalized displacement vector (a} is to be zero in
0 0 0 0 0 0 P, P, P, order that the incompatible modes are not activated.
[PI= 0 0 0 P, P, P, 0 0 0
The later part of eqn (8) can be expressed as

and {a}‘= [x, a2 a3 a4 a5 a6 a, a8 a,]. KJ~‘,J + K,,l{~~ = 0. (9)


The eight-noded solid element with incompatible
modes is an improvement over that without incom- For the equation to be valid, {a} = {0}:
patible modes. However its accuracy is found to be
dependent on the discretization of the structure. It K,l{4 1 = 0 (10)
yields excellent results for flexure problems if the
elements are regular hexahedrons, but if they are not i.e.
then the stresses calculated are found to be erro-
neous [ 15, 161. The inconsistencies in the general form
[P']T[C][B]{d;}
dv = {O}.
of elements with incompatible modes have been
fI
608 C. Sam and P. Krishna Iyer

Now [C] is a constant, The stiffness matrix is of size 33 x 33 and includes


coefficients pertaining to incompatible modes. How-
ever these terms can be eliminated by the static
condensation procedure and the condensed stiffness
is a constant because of constant strain, i.e. matrix will be of the order of 24 x 24 pertaining to
the nodal degrees of freedom. For details one may
refer to Ref. [17].
[PIT du = [O]. (11) Figure lb shows a two-noded three-dimensional
truss element arbitrarily oriented. The direction cosi-
For isoparametric formulation, eqn (11) becomes nes of the element are given by

x, - xi
[P’](J] dr ds dt = [0], (12) Q,, qyq, c+$ (14)
I.

where ]J) is the determinant of the Jacobian matrix


where xi, y,, zi and x,, y,, zj are the coordinates of the
[.I]. Equation (12) is automatically satisfied if the
nodes i and j, respectively.
element is a parallelepiped in which ]j] is a constant.
The stiffness matrix of the element is given by
For arbitrarily shaped elements, the Jacobian is not
constant, since the terms in [P’] are first powers of r,
s, t, and eqn (12) will be satisfied if ]JI is made [k]=z
L
constant. This can be done by evaluating IJI from a
Jacobian matrix [.I] computed at r = s = t = 0, re-
gardless of the actual coordinates of the integration 1; lJ, lX12 - 1f - 1, IF - 1, 1, 1
points. The Jacobian matrix [.I] used for evaluating l,l, 1; - - 1.; -
[II,] is not modified. In essence the Jacobian matrix
l,r; IJ l,l:
1; - l,J,
1.J: - !, 1, - I$:
1; . (15)
involved in the formulation of [B,] in eqn (6) is
evaluated at the integration points, while the Jaco- -1: -l,l,. -&I, 1: lJ, 4
bian involved in [P’] in eqn (6) is evaluated at - l,l, -If -I$; l,JJ 1; r,.1:
r=s=t=O.
-1j: -& -1; 1.J; I, 1: 1;
J
Element st@ness matrix
The well-known initial stiffness iterative method
The element stiffness matrix is evaluated using the was chosen for the present nonlinear analysis [ 181.
relation The incremental process was stopped when the diver-
gence of the residual force occurred (which was taken
[BTIC][B]IJI dr ds dt. (13) to have occurred if the maximum residual force in the
current iteration was greater than that in the previous
The numerical integration is performed by using iteration for the same load step). A convergence
2 x 2 x 2 Gauss quadrature scheme. tolerance of 1.5% was found to be adequate.

(b)
ty (compression 1

COMPRESSION
I I -
EC ELI compression
STRAIN
STRAIN

Fig. 2. (a) Uniaxial stress-strain curve of concrete. (b) Increase of strength parameter under multiaxial
conditions.
Reinforced concrete four-pile caps 609

3. THE ANALYTICAL MODEL USED FOR THE where


PRESENT STUDY
d = $/E” + ($3 - 2p2)EJ& - (2p3 - 3p2 + I)
The analytical model used to represent the be- ..
tp*-2p + 1)P
haviour of concrete for the present study is mainly
based on the work done by Bathe et al. [19-211 and B = (2&,/_17 - 3) -2A
is described briefly in the following.
The material is modelled as a hypoelastic one based C=(2-l?,,/&)+A
on a uniaxial stress-strain relation that is generalized
to take biaxial and triaxial stress conditions into
p=E,/C c
account. The concrete material is considered as or-
thotropic with the directions of orthotrophy defined
by the principal stresses.
and
3.1. Stress-strain relations
The general stress-strain relation for a multiaxial
case is derived based on the uniaxial stress-strain
relation. All uniaxial parameters obtained from The strength parameters occurring in the above
Fig. 2a carry a bar. equations, p, $, Es,. Eu, Cc, Cc, cf., C,, are obtained
from uniaxial tests. The stress-strain relation in eqn
3.1.1. Uniaxial conditions. A typical uniaxial stress, (17) assumes monotonic loading conditions. For
8, to uniaxial strain, .C,relation (assuming continuous strain states beyond C, in compression, it is usually
loading of the material) is shown in Fig. 2a. This assumed that the material cannot resist any load and
stress-strain relation shows that there are basically the stresses are gradually reduced to zero.
three strain phases; namely, corresponding to C, > 0,
3.1.2. Multiaxial conditions. The behaviour of con-
0 2 C > CCand C, > C > F,, where C; is the strain corre-
sponding to the uniaxial cut-off tensile stress 6,. Z, is crete under multiaxial stress condition is very com-
the strain corresponding to the minimum crushing plex and has not been assessed experimentally in a
stress, CC. that can be reached, and F, is the ultimate complete manner. The confining compressive stress
compressive strain. If C, > 0. i.e. the material is in gives a modified stress-strain curve and is given in
tension, the stress-strain relation is linear and a Fig. 2b. For the present study, the material is con-
constant Young’s modulus E0 is employed: sidered as orthotropic with the directions of or-
thotropy defined by the principal stresses. To obtain
the stress-strain relations in loading conditions, the
(16)
principal stresses are calculated and for each principal
stress direction a uniaxial tangent Young’s Modulus,
/?rz. corresponding to the strain in the principal stress
direction or, is evaluated using eqns (16) and (I 8).
For C < 0, the following relations are used: When using eqn (18) the current strain tp, is em-
ployed and to account for multiaxial stress conditions
the material variables CC, CU, C, are replaced by the
6 (G/E;)(~/C) variables a:, c?:, Cf and EL defined in eqns (21)-(24).
(17)
E = 1+ A(i/t,) + B(c/t,)* + C(C/EJ3 Let or,. cp2and or,, be the principal stresses, with
crp3< up2 < or, and E,, , & and i?r3 the corresponding
and hence uniaxial Young’s moduli.
The stress-strain matrix, considering the three-di-
$[l - B(C/E,)‘- 2C(C/t,)3]
E= (18) mensional stress conditons. corresponding to these
[I + A (C/C,) + B(C/S,)Z + C(E/CC)3]*’ directions is given by

7
(1 - v)&, vE,? v-43 0 0 0
(1 -v& vEZ3 0 0 0
1 (1 -v,Er,, 0 0 0
(1 - 2v) E (19)
Ic]=(* +v)(l-2v) 2 I2 0 0

(‘--%2 23 0

where v is the Poisson’s


1 ratio.
(1 -WE
2
I3
610 C. Sam and P. Krishna Iyer

Fig. 3. Triaxial compression failure envelope.

The shear modulus in a coordinate plane is calcu- +/CC (i = 1 , . , 6;j = 1,2,3) that correspond to
lated from the weighted Young’s modulus corre- the stress magnitudes crp2= upI, ap2 = j&r,, (where fl is
sponding to that plane: a constant) and err = up3.
To identify the compression failure, the largest
E 1 principal stress bp, is employed to establish from Fig.
=p IepilEpi + lop, I&,
(20) 3 by interpolation, the biaxial failure envelope opz and
GCJ=2(1 2(1+v) ]a,,]+]or,,] .
up3. The material has been assumed to be crushed if
the stress state corresponding to cp2 and up3 lies on or
Er, is evaluated using the uniaxial stress-strain re- outside the biaxial failure envelope. To include the
lationship in eqn (18). effect of change in Poisson’s ratio, the expressions
The above stress-strain relations are employed in reported by Ottosen [23] are used.
the calculation of the stress increment before tension The compression failure envelope is also employed
or compression failure. to establish the uniaxial stress-strain law accounting
for multiaxial stress conditions. Having calculated
3.2. Material failure envelope (compression)
the current principal stresses, to establish the
To model the failure of the material in compression uniaxial stress-strain law it is assumed that or,, and
in two- and three-dimensional analyses and to ac- up2 are held constant and the minimum compressive
count for multiaxial stress conditions in the uniaxial stress that would have to be reached in the
stress-strain law of Fig. 3, the compressive ultimate third principal stress direction to cause crushing
strength envelope suggested by Bathe et al. [19-211 is of the material is calculated using the strength
employed. The shape of the compressive strength envelope Fig. 3. Let the stress be 5: and then
used is largely based on the experimental results
reported by Kupfer et al. [22].
The compressive failure envelope is incorporated y, = 0; J& (21)
using 24 discrete stress values and applied as input.
Firstly, the values a,,/~?, are input. These values
define at what stress magnitudes, up,, the discrete CT:=y,5, (22)
two-dimensional failure envelopes for additional
stresses Orz and ep3 are input. These failure
envelopes are defined by the failure stress values t; = (C,yi’ + C,y,)i, (23)
Reinforced concrete four-pile caps 611

and tions for the stress calculation is

1
r:=(C,y:+C*y,)c;. (24)
icl = (1 - v)*

The constants C, and C, are defined by the user;


normally assumed as 1.4 and - 0.4, respectively. The E’O 0 0 0 0
values di, ah, ZL and E: are used instead of the
unprimed variables in order to establish the uniaxial
& ~4, 0 0 0
stress-strain law under multiaxial stress conditions, &b, 0 0 0 > (25)
Fig. 2(b). G:2 0 0
(1 -v2)G2, 0
3.3. Tension failure Symmetric G:,
To identify whether the material has failed in
tension, the principal stresses are calculated first. where EC, Gi2 and Gi, are evaluated as shown in Fig.
Tensile failure occurs if the value of the maximum 4a and b. In these figures Z,,,is a variable input by the
principal stress exceeds the tensile failure stress. In user, which takes into account the tension stiffening
this case it is assumed that a plane of failure develops and shear retention. Figure 4(b) shows that CL is a
perpendicular to the principal stress direction. The fraction of the initial shear modulus, E,,/2(1 + v),
effect of this material failure is that the normal and corresponding to the total strain normal to the tensile
shear stiffnesses across the plane of failure and the failure plane. l?ri is the uniaxial Young’s modulus
corresponding normal stress and shear stresses are evaluated using eqn (18) and E,, and G*, are evalu-
reduced. ated using eqn (20).
Prior to tensile failure the stress-strain material law
3.4. Post-tensile cracking behaviour
is given by eqns (19). Assuming that crp, is larger than
the tensile failure stress, the new material stress-strain Once a tensile failure has formed at a Gauss point,
matrix corresponding to the principal stress direc- it is checked for subsequent failure. A subsequent

STRAIN NORMAL
TO TENSILE
FAILURE PLANE

(a)

I I
I I
I I I STRAIN NORMAL
I t
= TO TENSILE
5 Ee % FAILURE PLANE

(b)

Fig. 4. Material moduli for stress calculation under tensile failure. (a) Calculation of Young’s modulus
E normal to tensile failure plane. (b) Calculation of shear modulus in tensile failure plane.
612 C. Sam and P. Krishna Iyer

failure is assumed to form perpendicular to the y2 is the ratio of the actual stress to the corresponding
direction of the one that developed first, once a value of that stress at failure (cr,,/e,).
normal stress perpendicular to the original failure
plane has reached the tensile failure stress. It follows
that, at any integration point, the direction of the 4. CHoIcEoF CONSTITUTIVE RELATION FOR
third tensile failure plane is fixed once the failure has REINFORCING STEEL
occurred in two directions. The modified stress-strain
In the present study the steel reinforcement is
relations similar to eqn (25) are used for further stress
considered to be elastoplastic with yield stress _+a,
calculations.
and elastic modulus ES,. It is further assumed that
It may also happen that after tensile failure of the
reinforcing bars carry only uniaxial stresses. The
material (in one or two directions), the material fails
same relationship is used in tension as well as in
in compression crushing, which is identified as usual
compression.
by entering the compression crushing envelope in Fig.
3 with the principal stress(es) that act(s) along the
tensile failure plane(s). If the material has crushed in 5. EXPERIMENTAL ANALYSISOFPILECAPS
compression, it is assumed that the material strain
softens in all the directions. For the purpose of the experimental study, the pile
caps were modelled as rectangular blocks subjected to
3.5. Post-compression crushing and strain softening patches of loading on the top and bottom surfaces (at
behauiour of concrete the column and pile positions). The column and piles
As shown in Fig. 2a, for a uniaxial strain smaller were assumed as square and the column dimensions
than E,, the material has crushed and softens with the were so chosen that the intensity of the column stress
increasing compressive strain, i.e. E is negative. was twice that of the pile stress (as a typical example).
Under multiaxial stress conditions the compression The centre-to-centre distances between the piles were
crushing is identified using the multiaxial failure kept equal to three times the side dimension of the
envelope. Once the material has crushed, isotropic square pile, as is the usual practice. The column stress
conditions are assumed using the stress-strain law and pile stress were considered uniform. The clear
with the constants 5: and FAetc., corresponding to the overhang of the pile cap beyond the external face of
multiaxial conditions at crushing. The Young’s pile was assumed as 3/8-times the side dimension of
modulus for the isotropic material is evaluated using the pile. The depth-to-sparr’ratios of all the models
were kept as 0.75.
Based on the availability of the testing machine
base to accommodate the specimens, the dimension
of the square pile was chosen as 10 cm. The details of
where cp3 and AC,, are the strain and the incremental the specimens are given in Fig. 5.
strain measured in the direction of the principal stress Aggregates of 10 mm maximum size were used for
up3. When tp3 becomes equal to or less than F,, the casting the specimens. Mild steel bars having yield
stresses are set to zero. strength of 300 N mmV2 were used as reinforcement.
The average compressive strength of concrete ob-
3.6. Change in Poisson’s ratio tained from the cylinder test was 19Nmm-‘. The
reinforcements were fully anchored to avoid bond
From experiments, it has been observed that the
failure. The pile caps and the cylinders were cast
ratio of lateral strain to principal compressive strain
together and cured for 28 days.
remains constant until approximately 80% of the
In all the pile cap specimens reinforcements (flex-
maximum compressive stress is reached and at which
Ural) were provided only in the tension surface. No
stress the apparent Poisson’s ratio begins to increase.
compression and shear reinforcements were provided.
The expressions suggested by Ottosen [23] were used
In the present investigation minimum reinforcements
here to calculate the current Poisson’s ratio.
were only provided as per Indian Standard Specifica-
tion [24], which is slightly higher than 0.2%.
v, = v when y2 <y, (27) The percentage of reinforcement provided for all
the specimens tested is same. However minor differ-
ences in the percentage of reinforcement among the
three layouts could not be avoided due to the
difficulty in the choice of the bars.
A 300 ton compression testing machine was used to
in which v, is the initial Poisson’s ratio (0.15). apply concentric compressive load onto the column.
The load on the specimen was increased monotoni-
vi = 0.36 cally to failure. Each test was represented by two
identical models for the purpose of confirmation of
y, = 0.8 the test results.
Reinforced concrete four-pile caps 613

ELEVATION

L
37.5
-k--
SLAB TYPE 100 15 NOS OF 6mm DIA RODS
IN BOTH DIRECTIONS
REINFORCEMENT t
LAYOUT
200 X
o-
35.36
ioo.oo70.7’
t -135.00
100 165.00
_f- ’ ’
3 7.5
r
(al
t i!_
3_-
37.5
-r
BUNCHED SQUARE100
TYPE
4 NOS OF 8mm DIA RODS
REINFORCEMENT t
A
LAYOUT 200

100.00
135.00
1 165.00
-200.00
37.5 237.50

5 NOS OF 10 mm DIA RODS


IN BOTH DIR ECTI ONS

LAYOUT

Fig. 5. Details of four-pile caps having different reinforcement layout.

The material properties used (obtained from the 6. RESULTS AND DISCUSSIONS
experiments) in the numerical analysis were
Three reinforced concrete pile caps with four piles,
havmg the same geometry, dimensions, material
O,= 19Nmm-’ ~~~=300Nmrn~~
properties and percentage of reinforcements, but with
different reinforcement layouts, were considered for
I?,= lSNmm_? Es, = 200,000 N mm-’
the numerical and experimental analyses. The three
or = 1.9 N mm-’ different layouts of reinforcement considered here
are:
c, = -0.0020 (a) distributing the reinforcement within the width
of the pile caps in both the directions, slab type
Z” = -0.0035 reinforcement layout, as shown in Fig. 5a;
(b) bunching the reinforcement over the piles
t, = - 0.0020 within the width of pile between the adiacent
piles, bunched square-type reinforcement layoui (Fig.
$ = 22077 N mm-?. 5b);
614 C. Sam and P. Krishna Iyer

I
t-u I
r
I
) I
1
(c) bunching the reinforcement over the piles
within the width of pile between the diagonally
opposite piles, bunched diagonal-type
layout (Fig. 5~).
reinforcement

(a) Four -pile caps with slab -type reinforcement layout


Experiment. The details of the pile cap having
slab-type reinforcement layout are given in Fig. 5a.
Figure 6 shows the experimental crack pattern for
this case at the time of failure. Initially a few visible
cracks developed at the bottom surface closer to the
internal faces of each pile at a load of 450 kN. These
cracks (developed near the internal faces of each pile)
extended upwards at an angle as the load was in-
creased. This type of crack propagation could be
noticed on all four vertical faces. Prior to failure,
cracks developed surrounding the faces of column at
the top surface of the pile cap. From the crack pattern
it may be concluded that the failure was due to
TOP SURFACES AND SIDES
punching of column or piles. The average failure load
for this case was 690 kN.

Finite element analysis. The mesh used for the finite


element analysis is shown in Fig. 7. Owing to the
existence of the two planes of symmetry, only one-
quarter of the pile cap was considered.
At a load of 200 kN, cracks developed at the
bottom Gauss points of the bottom level elements in
the central region of the pile cap and in the portion
closer to the internal faces of all the piles. As the load
was increased the cracks developed near the internal
BOTTOM SURFACE
faces of piles extended upwards faster than those at
Fig. 6. Final experimental crack pattern of four-pile caps
with slab-type reinforcement layout. the central region. The propagation of cracks and

ELEVATION

r----7 r---I
I
I I I
I I I
I
i___l i___l PLAN

r--1
r---1 I

I
I
I
I I I
I
I
i-__._ J I-___;

‘z
Fig. 7. Finite element mesh for one-quarter of four-pile caps with slab-type reinforcement layout.
Reinforced concrete four-pile caps 615

their corresponding loads are given in Fig. 8. A large The strains along the reinforcement at various
number of cracks appeared at the top surface sur- stages of loading are plotted in Fig. 9. In the initial
rounding the column faces before failure, and failure stages of loading the strain at the centre was larger
took place at a load of 600 kN. than that at the support. However, as the load
The smeared cracking model adopted here shows increased this trend changed and the strains at the
that a portion of concrete below the column similar centre and the support were more or less equal
to the frustrum of a pyramid extending from the towards failure. This indicates that all the forces
outside faces of column (at top) to the inside faces of carried by the concrete were transferred to steel at the
piles (at bottom) has lost its strength for further final stages of loading. The load deflection diagram
resistance to load. for this case is shown in Fig. 10.

240 kN 280 kN 320 kN

Fig. 8(a). Finite element crack-pattern of four-pile caps with slab-type reinforcement layout at different
Gauss point levels for different loads.
C. Sam and P. Krishna Iyer

(e)

(d)

(b)

(a)

360kN 400 kN 440 kN 480 kN


Fig. 8(b). Finite element crack-pattern of four pile caps with slab-type reinforcement layout at different
Gauss point levels for different loads.

(b) Four-pile caps with bunched square-type re- inner faces of the piles extended upwards at an angle
inforcement layout as the load was increased. Further cracks developed
Experiment. The dimensions of the pile cap and the surrounding the faces of column (at the top surface
reinforcement details are outlined in Fig. 5b. The of pile cap) as the load was increased to the ultimate
experimental crack pattern for this case at the time of load. From the crack pattern at the time of failure,
failure is given in Fig. 11. It was noticed that the one can conclude that, though the diagonal cracks
diagonal cracks developed at the central region of the were developed initially at the central portion of the
pile cap (at the bottom surface) at a load of 350 kN. bottom surface, possibly due to absence of reinforce-
This was followed by development of cracks at the ment, the final failure was due to punching of column
inner faces of piles. The cracks developed near the or piles. The average failure load was 630 kN.
Reinforced concrete four-pile caps 617

Finite element analysis. The finite element dis- The smeared cracking model adopted here conveys
cretization of the pile caps and the position of the behaviour that a portion of concrete similar to
reinforcement is shown in Fig. 12. Only one-quarter frustrum of a pyramid extending from outside faces
of the pile cap was considered because of symmetry. of the column to inside faces of pile (at bottom) has
Cracks were found to develop first at the bottom lost its strength to resist further load, as in the
Gauss points of the bottom level elements in the previous case. The failure load was 560 kN.
central region of the pile cap surface and the internal The strains along the reinforcement at different
faces of the piles at a load of 200 kN. As the load was stages of loading are given in Fig. 13. At the initial
increased the cracks developed surrounding the faces stages of loading, the strains in the reinforcement at
of piles extended upwards at a faster rate than those the centre and near the support were considerably
at the central region. The propagation of cracks and different. But as the load was increased the difference
the loads corresponding to them are similar to Fig. 8. between the strains in the reinforcement at the centre
Prior to failure, a large number of cracks appeared and at the support reduced. At failure, the strains in
surrounding the faces of the column. the reinforcement between the internal faces of piles

(4

0 0006
P 200kN

0 280

5 , 360

2 0 0001 * 440
L
0 520
0 0002
4 600

- - 0 0000

(b) 0 000.4

0 0006

P 100 kN

0 0006 q 280

* 360
* UO
0 000‘
0 520

< 600

0 0002

0 000

Fig. 9a. Strain along reinforcement at various column loads four-pile caps slab-type reinforcement layout.
618 C. Sam and P. Krishna lyer

between the opposite piles are given in Fig. SC. The


crack pattern at the final stages of loading for this
case is shown in Fig. 14. In this case the cracks were
found to develop initially near the internal faces of
piles and the central portion of the pile cap at the
bottom surface at a load of 400 kN. These cracks near
the internal faces of piles extended upwards at an
angle towards the top surface as the load was in-
creased. These can be noticed from the crack pattern
at the bottom surface and the sides. As the load was
increased further, several cracks developed at the
bottom surface in the pile-free area. In addition
cracks were found to develop at the top surface
surrounding the column faces. The maximum average
load sustained by this pile cap model was 680 kN.

Finite element analysis. From the reinforcement


layout it can be seen that the axes of symmetry
coincide with the two diagonals. Hence the quadrant
considered in this case is different from that for the
other two models. The discretization and the position
of reinforcements are shown in Fig. 15.
At a load of 200 kN. a few cracks developed at the
Def lect~on (mm)
bottom Gauss points of the bottom layer elements
Fig. 10. Analytical load-deflection curves of four-pile caps near the internal faces of piles. As the load was
having different reinforcement layout. increased further to 240 kN. a large number of cracks
developed at the bottom Gauss points of the bottom
were almost the same. The steel stresses are small, level elements in the pile-free area and the cracks
even with minimum reinforcement, at the time of which developed near the internal faces of the piles
failure when compared to its yield strength. This extended upwards. Prior to failure, a large number of
confirms the conclusion arrived by Sabnis and cracks were found to develop at the top surface
Gogate [l2]. The load-deflection diagram for this surrounding the column faces. The theoretical failure
case is shown in Fig. IO. load was 600 kN.
In this case also the smeared cracking model
(c) Four-pile cap with bunched diagonal-type re- conveys the fact that the concrete at the bottom
inforcement la,vout surface in the pile-free area loses its strength due to
Experiment. The details of the pile cap having formation of cracks. These cracks extended upwards
bunched diagonal-type reinforcement over the piles to mid-depth of pile caps as the load was increased

TOP SURFACE AND SIDES BOTTOM SURFACE

Fig. Il. Final experimental crack pattern of four-pile caps with bunched square-type reinforcement
layout.
Reinforced concrete four-pile caps 619

_!l
e-

d-
ELEVATION
C-
ELEVATION
b-

a-
r---1 r----1
I I I
I I I
I
I I
L-_--i L-_-J

-X
PLAN

r---7 r--- 1
I 1 l I
I I
I
I
I I
L---1 L-_-1

PLAN
Fig. 12. Finite element mesh for one-quarter of four-pile caps having bunched square-type reinforcement
layout.

up to failure. Moreover, cracks also developed at the and support were very much different. At high load
top surface surrounding the faces of the column. The levels, however, this difference in the strain values
behaviour observed here is very similar to that found between the support and the centre decreased. At the
in the other two cases explained earlier. time of failure, the strains in the reinforcement be-
From the analysis of strains in the reinforcement, tween supports are almost the same. However the
given in Fig. 16. it was noticed that at low load steel stresses at the time of failure are small when
levels the strains in the reinforcement at centre compared to yield strength, even with minimum

Fig. 13. Strain along reinforcement at various column load four-pile caps with bunched square-type
reinforcement layout.
620 C. Sam and P. Krishna lyer

TOP SURFACE AND SIDES BOTTOM SURFACE

Fig. 14. Final experimental crack pattern of four-pile caps with bunched diagonal-type reinforcement
layout.

reinforcement. The loaddeflection diagram is given bunched diagonal-type reinforcement layout were

t-
in Fig. 10. 690, 630 and 680 kN, respectively; while the finite
element method predicted 600. 560 and 600 kN,
respectively, for the above models. It can be noticed
7. COMPARISON that the predicted failure loads for the above models
The maximum experimental load sustained by the are 87. 89 and 88% of their experimental failure
pile cap specimens with slab, bunched square and loads, respectively.

I
VI

Ii
e-
z
J

ELEVATION c
d-
z
B c-

r----l B
I
I I

I I
I
l____J
PLAN

Fig. 15. Finite element mesh for one-quarter of four-pile caps having bunched diagonal-type reinforce-
ment layout.
Reinforced concrete four-pile caps 621

Fig. 16. Strain along reinforcement at various column load four-pile caps with bunched diagonal-type
reinforcement layout.

Comparing the deflections of the three pile-cap noticed in all the specimens, irrespective of the re-
specimens at the centre of the bottom surface, it is inforcement layout.
found that the slab type reinforcement layout has The failure pattern exhibited in the test specimens
more stiffness when compared to the other two types was also of the punching type (punching of column
of reinforcement layouts. The pile cap with the or piles). The experimental observations thus confirm
bunched square-type reinforcement layout is the one the validity of the analytical procedure and the
having the least stiffness. This may possibly be due to assumptions used therein.
the absence of reinforcement in the central portion, At low load levels, the strains in the reinforcement
where the stresses are obviously high. between the centre of the pile cap and the points near
It is interesting to note that these observations are the internal faces of the piles were very different. This
at variance with the results obtained by the earlier is similar to the behaviour of a beam, where the
researchers in the field [2, IO]. However a firm con- stresses vary from a maximum at the centre to a
clusion cannot be drawn since the load carrying minimum at the supports, indicating that at the initial
capacity of a pile cap is dependant on various factors. stages of loading, the beam action is predominant in
such as spacing of piles. depth of pile cap, layout of the pile cap.
reinforcement, projection beyond the external face of At the time of failure, however, the strains (or
pile, grade of concrete. percentage reinforcement, stresses) in the reinforcement between the supports
column size. etc. were more or less the same. This is due to the transfer
The smeared cracking approach conveys the fact of the forces from concrete to steel after the develop-
that initially cracks developed at the bottom Gauss ment of cracks. This corresponds to a tied arch
points of the bottom level elements in the central behaviour, indicating thereby that at the final stages
region and the internal faces of the piles. As the load of loading the pile cap resists forces by strut action.
was increased further, new cracks developed and the This is similar to the observations made by Adebar
cracks near the faces of the piles extended upwards at et al. in their experiments [13].
a faster rate than those at the central region. Prior to
failure, a large number of cracks appeared at the top
surface surrounding the column faces. From the
8. CONCLUSIONS
crack pattern it can be visualized that a portion of
concrete, similar to the frustrum of a pyramid extend- An attempt has been made to analyse four-pile
ing from the outside faces of the column (at top) to caps having the same geometry, dimensions, percent-
inside faces of piles (at bottom), has lost its strength age of reinforcement and grade of concrete, but with
at the final stages of loading and leads to failure by different reinforcement layouts, by the nonlinear
punching of column or pile. The same behaviour was finite element method. From the theoretical and
622 C. Sam and P. Krishna Iyer

experimental results the following conclusions are 9. P. Krishna Iyer and C. Sam, Three-dimensional analysis
of pile caps. Compur. Sfruct. 42, 39541 I (1992).
drawn:
IO. J. L. Clark. Behaviour and design of pile caps, with four
(1) The pile cap with bunched square-type re- piles, Technical report 42:489. Cement and Concrete
inforcement layout resists the least load when com- Association, London (1973).
pared to the pile caps with the other two types of 11. A. B. Gogate and G. M. Sabnis. Design of thick pile
caps. ACI J. 77, 18-22 (1980).
reinforcement layout.
12. G. M. Sabnis and A. B. Gogate, Investigation of thick
(2) At low load levels the beam action is predom- slab (pile cap) behaviour. ACI J. 81, 35-39 (1984).
inant, while at higher loads, the pile cap resists load 13. P. Adebar, D. Kuchma and M. P. Collins, Strut-and-tie
by strut action. irrespective of the reinforcement models for the design of pile caps: An experimental
layout. study. ACI strut. J. 81-92 (1990).
14. ASCE Committee on Concrete and Masonry Struc-
(3) A portion of concrete, below the column,
tures, Task Committee on Finite Element Analysis of
similar to the frustrum of a pyramid extending from Reinforced Concrete Structures: a state of the art report
the column faces (at top) to inside pile faces (at on finite element analysis of reinforced concrete struc-
bottom) weakens and leads to failure by punching of tures. ASCE Special Publication (1981).
column or piles, irrespective of the reinforcement 15. E. L. Wilson. R. L. Taylor. W. P. Doherty and T.
Ghabussi. Incomputible Displacement Models, Numeri-
layout. cul crud Computer Methods in Structural Mechanics
(4) The nonlinear analysis using the material (Edited by S. J. Fenves et ul.). pp. 43357. Academic
model and failure criteria adopted in this work is Press. London (1973).
capable of predicting the behaviour and ultimate load 16. R. L. Taylor. P. J. Beresford and E. L. Wilson, A
nonconforming element for stress analysis. In/. J. nu-
of pile caps fairly accurately.
mer. Meth. Engng IO, 121 l-1220 (1976).
17. C. S. Krishnamoorthy. Finite Element Anuiysis. Tata
REFERENCES McGraw-Hill, New Delhi (1987).
18. D. R. J. Owen and E. Hinton, Finile Elements in
1. H. T. Yan, Bloom base allowance in the design of pile Plasticity. Theory und Praclice. Pineridge Press,
caps, Parts I and II. Civil Engng Public Works Reo. 49, Swansea.
494495 and 6222633 (1954). 19. K. J. Bathe and Ramaswamy S.. On the three-dimen-
2. J. Blevot and R. Fremy, Semelles fur pieux. Atin. Inst. sional nonlinear analysis of concrete structures. Nuclear
Techn. Bet. Truraus Pub. 20, 223-295 (1967). /%gq Des. 52, 385409 (1979).
3. S. P. Banerjee. Notes on the design of pile caps. Civil 20. K. J. Bathe and J. A. Sundberg, A Concrete Material
Engng Public Works Rer. 51, 8722874 (1956). Model. Computaiional Modelling of‘ Reinforced Concrere
4. W. T. Wittle and D. Beattie, Standard pile caps. Parts Srrucrurrs (Edited by Hinton and R. Owen). Pineridge
I and II. Concrete 6, 29931 and 34-36 (1972). Press, Swansea (1986).
5. B. D. Sharma. Design of isolated pile caps. Ind. Con- 21. K. J. Bathe. J. Walczak, A. Welch and N. Mistry,
crete J. 53, 222-226(1979). . Nonlinear analysis of concrete structures. Compuf.
6. Building Code, Requirements of Reinforced Concrete Slruct. 32, 5633590 (1989).
(AC1 31811985). 22. H. Kupfer, H. K. Hilsdorf and Rusch, Behaviour of
7. British Standard for Structural Use of Concrete, Code concrete under biaxial stresses. J. Am. Concrete Ins. 66,
of Practice for Design and Construction. BS 8 I 10 Part 656-666 (I 969).
1 (1985). 23. N. S. Ottosen. Constitutive model for short-time load-
8. Indian Standard Code of Practice for Design and ing of concrete. ASCE Engng Mech. (1979).
Construction of Pile Foundation IS 291 I, Part l/Set. 3 24. Code of Practice for Reinforced Concrete, IS 456, New
(1979). Delhi (1978).

You might also like