Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

A DESIGN METHOD FOR PILED RAFT FOUNDATIONS

V.Balakumar
Senior consultant, Simplex Infrastructures Limited, Chennai, Tamil Nadu, India.
Erwin Oh
Lecturer, GriffithUniversity Gold Coast Campus, Brisbane, Australia.
Mark Bolton
Lecturer, GriffithUniversity Gold Coast Campus, Brisbane, Australia.
A.S.Balasubramaniam
Professor of Civil Engineering, GriffithUniversity, Brisbane, Australia.

ABSTRACT: The design of piled raft foundation involves two stages namely a preliminary stage and the final stage. The
preliminary design stage involves the identification of the essential parameters namely the number of piles, their diameter and the
length along with Es value of the subsoil strata for an optimum design which can produce the required settlement reduction. The
existing design methods, whose accuracy depends upon the accuracy of the evaluation of in-situ parameters like Es although can
produce satisfactory results, the computational efforts involved and the time does not justify the use of them for the preliminary
analyses which will involve repetitions. This paper outlines a simple analytical procedure and the effectiveness of the pressuremeter
tests in predicting the piled raft behaviour, so that the results can be used in the detailed design.

RSUM : La conception dune fondation avec radiers sur pieuxcomporte deux tapes : une tape prliminaire et une tape finale.
Ltape prliminaire comprend lidentification des paramtres fondamentaux, savoir le nombre de piliers, leur diamtre et leur
longueur ainsi que la valeur Es de la strate souterraine pour une conception optimale capable de crer la rduction daffaissements
sollicite. Les mthodes de conception existantes, dont lexactitude dpend de celle des paramtres in situ tels quEs, bien que
passibles de donner des rsultats satisfaisants, ne justifient pas leur emploi pour les analyses prliminaires sujettes rptitions, en
raison des efforts de calcul et de temps quelles impliquent. Cet article expose une procdure analytique simple et lefficacit des
essais pressiomtriques dans la prvision du comportement du radier sur pieux, de manire permettre lutilisation des rsultats
pour la conception dtaille.
KEYWORDS: Piled raft, tri-linear pressuremeter.

INTRODUCTION.

the interaction process between the observational methods

(Katzenbach etal., 2000 a; Balakumar and Ilamparuthy, 2007),
small scale model studies (Horikoshi1995; Balakumar
etal.,2005) and numerical and analytical simulations leading to
the improvement of the design process. Consequent to this
number of tall and heavily loaded structures have been
supported on piled raft and the performance of some of these
piled rafts have been monitored and the results may be used to
refine the design in the future. (Poulos, 2008; Yamashita etal.,
2010).

The combined piled raft foundation system provides a skilful

geotechnical concept wherein the applied load is transferred by
means of a load sharing mechanism which is generated through
a process of interaction between the pile soil and the raft. The
piled raft foundation system differs from the traditionally
designed pile supported raft in the fact that in the case of piled
raft the presence of the raft and its contribution in sharing the
load with the pile group is recognized. The piled raft foundation
utilises the pile group for the control of settlement, with the
piles providing most of the stiffness at the service loads while
the raft elements provide the additional capacity at the ultimate
load levels. The concept of piled raft system was born out of the
fact that any structure has a certain magnitude of permissible
settlement and the foundation system has to aim at reducing the
settlement as close to the permissible value as possible rather
than eliminating the settlement completely. In the last two
decades researchers like Cooke (1986), Burland(1995), and
Poulos(2001) have provided considerable insight into the
behaviour of piled raft .

THE DESIGN PROCESS

In the design process of piled raft the initial stages of design
involves the determination of the optimum number of piles, pile
length and diameter required to be placed in a strategic manner
to produce the required settlement reduction along with the load
shared by the pile group. This process may require a large
number of trials depending on the nature and requirement.
Hence the analytical procedure has to be computationally
simple so that the efforts and the time will be less. The existing
methods although produce satisfactory results, involve more
complicated computational efforts .Further in solving the
complicated three dimensional problems such as piled raft,

The development of sophisticated computational facilities

and FEA codes like Defpig,Napra,and HyPr etc have enhanced

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

many simplified assumptions are to be made and the

rigourousness of the method may have to be diluted to make the
problem computationally viable. Therefore there is a need for a
simple method that can solved by treating the problem as
axisymmetric or plane strain problem in the case of preliminary
design to establish parameters like pile length, numbers
diameter and the layout to be used in the final design.

200mm is far higher than the other lengths namely

75mm,100mm and 120mm. The typical characterisation curve
of the piled raft shown with various pile lengths are given in
Figure 2 for a pile diameter of 10 mm, which shows that
irrespective of the pile length, the behaviour has three phases.
Although the settlement up to which the linear elastic stage
(portion OA of the curve) remains same as 1mm, the load
corresponding to this varies. As can be seen at higher length the
linear behaviour extends nearly upto 30% of the load taken by
the piled raft corresponding to settlement equal to 10% of the
pile length. The second stage of the curve AB is the stage where
the behaviour tends to become elasto- plastic, which extends up
to a settlement level 9 mm for 200 mm long pile, 7.5 mm for
120 mm long pile and 4.5 mm for 75 mm long pile.

It is always recognized in geotechnical engineering that the

most difficult part is the evaluation of the in-situ parameters
particularly the elastic modulus. In most of the cases, such
parameters are obtained either from laboratory tests or from
standard correlations between tests like SPT and Es values,
which can affect the accuracy of results. However over the past
few years there is a considerable shift from the laboratory
testing to in-situ testing and this has led to the use of the results
from in situ tests such as CPT and pressuremeter tests
extensively to determine the stress strain characteristics and
essential parameters like the in-situ elastic modulus of the soil
over the length of the pile. A well tried procedure for predicting
such parameters along with the shaft friction development has
been published by Roger Frank et al (1991) using pressuremeter
tests. Therefore it was felt necessary to study whether such
predictions can be used to evaluate the numerical details such as
the number of piles, length, diameter, and layout required for
the design of the piled raft.

8MM RAFT AND 10MM DIA. PILE - VARIATION IN LENGTH

LOAD, kN
0

10

0
d = 10mm
t = 8mm
N = 21
R.A. = 36
D = 200mm
Bed = MD

2
4

SETTLEMENT, mm

6
8
10
12
14
16

PLAIN RAFT
200 mm
120 mm

SCOPE OF WORK
With the above in mind it was decided to study the various
options available to idealise the piled raft model which would
be amenable for a simple numerical procedure and will give the
load settlement, settlement reduction and load sharing
behaviour of the piled raft. Further in order to evaluate the
elastic modulus and other parameters over the pile depth various
in-situ test options were also studied. It was found that the
equivalent pier approach would be the most suitable approach
for modelling the piled raft. The paper presents the details of the
study and the conclusions of the study.

18

100 mm

20

75 mm

Piled raft
Pile length

22

Fig. 1: Load-Settlement Response of Circular Piled Raft with

Various Pile Lengths

THE STUDY
In the design of piled raft the requirement is the settlement
reduction and the data for the design is the load shared by the
raft and the pile group. It is only from the group capacity
required, the number of piles required, diameter and the length
can be evaluated. In order to study the load settlement response,
a series of 1g small scale model tests were carried out on
circular and square shaped piled raft placed on sand bed. Poorly
graded sand was rained in pre-calibrated manner so that the
required densities could be achieved; the tests were carried out
on unpiled raft, free standing pile group and piled raft.
Extensive parametric studies were also carried out but the
presentation is restricted to the load settlement and load sharing
response typically for circular piled raft under medium dense
bed condition. The studies showed that the performance of the
piled raft was identical in all the other cases. Details of the
models test set up and other details are presented elsewhere
(Balakumar etal., 2005).

Fig. 2: Characteristic Response of Piled Raft

for Various Pile Lengths

T
o have better understanding on load sharing between the raft
and pile group of piled raft, three dimensional nonlinear
analysis was carried out using ANSYS code. Only quarter
model of piled raft was analysed taking advantage of the
symmetry (Figure 3). The bed density was kept as medium
dense with = 37.5 and dry unit weight = 15.5 kN/m3. MISO
material model was used for the soil. The continuum was
modelled using solid 45 elements with three degrees of freedom
at each node. In the analysis the bed dimensions were kept
same as that of the model tested in the laboratory. The raft and
piles were also modelled as solid 45 elements in order to
maintain the elements compatibility. The load was applied as
pressure in small increments till the load on the raft equal to the
final test load. Figure 3 shows the quarter model including finite
element meshing adopted in the analysis.

LOAD SETTLEMENT AND LOAD SHARING RESPONSE.

Figure 1 presents the load settlement response of circular piled
raft with varying pile lengths and and figure presents the
charecterised form of the load settlement response.It is clearly
seen that at any given settlement the load taken by the piled raft
is more than the unpiled raft for the corresponding settlement.It
is seen that the load taken by the piled raft with pile length of

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

Fig. 5: Characterisation Curves of Experimental and

Numerical Analysis

Fig. 3: Finite Element Mesh of a Circular Piled Raft

The probability of wide variation while correlating the field

data with the laboratory test results cannot be ruled out.
Therefore the evaluation of parameters from the in-situ tests
gain considerable importance.A well tried procedure for
predicting the shaft friction development has been published by
Roger Frank et al., (1991) using pressuremeter is discussed
below.

LOAD-SETTLEMENT BEHAVIOUR
Figure 4 presents the load settlement curves of circular piled raft
obtained from 1g model test and the numerical model. Figure 5
presents comparison of characteristic load-settlement response
of circular piled raft between experiment and numerical
analysis. The results obtained from the 1g model test and
numerical model agree very closely, till the settlement level of 4
mm.
CIRCULAR RAFT ANSYS Vs LAB
0

LOAD, kN
5
6

PREDICTION OF PILE BEHAVIOUR. PRESSSUREMETER

AND ITS APPLICABILITY
The pressuremeter is an effective tool that has been extensively
used to obtain the in-situ parameters and for the last three
decades foundations have been designed
based on the
parameters obtained from the in-situ tests. The large volume of
data collected over a period of time particularly the French
Highway authorities has enhanced the confidence level of the
designers in using them for the design of deep foundations.
Frank etal., (1991) have studied the load settlement response of
two piles forming a part of a bridge foundation. and had
established that their behaviour can be predicted by conducting
the pressuremeter test.Their prediction of pile behaviour is
based on a tri-linear relationship for the skin friction
mobilisation based on the pressuremeter tests The model they
had used is given in Figure 6.

10

SETTLEMENT, mm

4
8
12
16
EXPERIMENTAL
20

NUMERICAL - MISO

24

Fig. 4: Comparison of Load-settlement

As the load increases, the difference in the settlement

between the 1 g model and the numerical model results
increased marginally. However the maximum variation in the
load between the numerical model and the small scale model
results was less than of 5%. This comparison indicates a close
agreement between the numerical model and 1g model test.
Similar observation is made in the analyses of piled raft in loose
and dense sand. Thus the nonlinear analysis using MISO model
idealization for the soil predicts the performance of piled raft
reasonably well. Although it has been established that the
characterised load settlement response predicted by analytical
study and the 1g model tests have agreed very closely, the basic
parameter namely the Es value of the soil was obtained from the
direct shear tests and the standard correlations available from
the literature The agreement in the results of 1g model and the
numerical analyses can be attributed to the fact that the
supporting medium was prepared under laboratory conditions.
However in the case of field samples the accuracy of the
parameters obtained largely depends upon the sampling
efficiency and the care taken in the extrusion of the specimen
and the preparation of test conditions.

Figure 6: Tri-linear Model, Frank et al (1991)

The first segment has a constant slope .The slope of the

second line as more flat and third segment represents the

Proceedings of the 18th International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013

mobilisation of total skin friction. A typical tri-linear model of

shaft friction mobilisation. The end of the second part is the
limiting value of the friction .The evaluation of the friction is
done in their case based on LCPC-SETRA (1985) RULES. The
slopes of the lines depend upon the pressuremeter modulus and
radius of the pile. The prediction experimental load distribution
given by them has a similar trend as predicted by the numerical
analyses of the 1g model tests.

equivalent pier has the same plan area of the pile group and the
length of the pile can be taken as the length of the pier.The load
settlement response predicted by the pressuremeter with the
shaft friction mobilisation can be compared with the equivalent
pier analyses to validate the in-situ Es value over the length of
the pile (pile group) and the shaft friction over the length of the
pile group. This will also establish the shaft stress distribution at
any given settlement level and the in-situ Es value which can be
used in the detailed analyses.

APPLICABILITY TO PILED RAFT DESIGN

Figure 7 and figure 8 present the shaft stress distribution over
the length of the pile by pressuremeter test results and from the
numerical analyses of the 1g model tests. it is seen that the trend
of the shaft stress distribution obtained from both the cases
agree closely,indicating that the tri-linear model assumed in the
analyses of the pressuremeter results and the actual behaviour of
RAF T - LENGT H RAT IO Vs SHAF T
piled raft CIRCULAR
obtained from
the 1g model are identical.
ST RESS (20mm)

CONCLUSIONS
The design economy in the piled raft design depends upon the
optimum design of the pile group of piled raft. Therefore before
the final design number of trials has to be made and also the
evaluation of in situ parameters has a considerable influence.
Keeping the above in mind the study carried out has shown that
the tri-linear relationship for the shaft friction mobilisation
adopted by Frank etal.,(1991 )to study the performance of two
single piles was in agreement with the trend of the characterised
load settlement response of the piled raft. Also the trend of shaft
friction mobilisation over the length of the pile agreed with the
prediction made from the pressuremeter test results in the case
of single pile. Based on the above it is concluded that the
equivalent pier theory can be used in combination with the
pressuremeter test results to predict the load settlement and load
sharing behaviour of the piled raft adopting parameters
determined from in-situ tests which are more reliable.

SHAFT STRESS, N/m m 2

0.00

0.40

0.80

1.20

1.60

2.00

2.40

2.80

0.00

LENGTH RATIO 1

0.11
0.22
0.33
0.44
0.55
0.66
0.77

Central pile

0.88

Inner pile

0.99

Outer pile

1.10

REFERENCES

Figure 7: Variation of stress along the shaft of

typical piles along the centre line of raft for 8.10kN

Katzenbach R., Arslan V. and Moorman ch (2000a), Numerical

Stimulations of Combined Piled Raft Foundations for the New
High Rise Building, Max in Frankfurt am main, Proc. 2nd Int.
Conf. on Soil Structure Interaction in Urban Civil Engineering.
Poulos H.G. (2001), Piled Raft Foundation: Design and Application,
Geotechnique, Vol. 51, No. 2, pp 111-113
Horikoshi K. (1995), Optimum Design of Piled Raft Foundations,
Dissertation submitted for the degree of Doctor of Philosophy,
University of Western Australia.
Polous H.G. (2008), The Piled Raft Foundation for the Burj Dubai
Design & Performance. IGS Ferroco Terzaghi, Oration 2008.
Burland J.B. (1995), Piles as Settlement Reducer 18th Italian Congress
on Soil Mech., Pavia.

Figure 8: Comparison of Theoretical and Experimental Load

Distributions for Test Piles (Roger Frank et al.,) )1991)

Balakumar V. and Ilamparuthi K. (2007), Performance Monitoring of a

Piled Raft Foundation of Twelve Storied Building and Analytical
Validation, Indian geotechnical Journal, Vol. 37, No. 2, pp. 94115.

This establishes the fact that if the pile group of piled raft
can be idealised as a single large pier, then the procedure
adopted by frank et al(1991) can be used to predict the
behaviour of pile group of piled raft foundations. poulos (2001)
has shown that while studying the settlement behaviour of the
pile group, that if the pile group with the soil prism can be
considered as a single pier, then the procedure applied for a
single pile behaviour can be used for the prediction of the load
settlement response of the equivalent pier numerically using
axisymmetric analyses.the equivalent pier modulus eeq is given
by the expression
Eeq = ES+ (Ep - Es) At/Ag

Yamashita.K , Hamada.J, Yamada.T. (2010), Field Measurements On

Piled Rafts with Grid-Form Deep Mixing Walls on Soft Ground,
Geaotechnical Engineering-SEAGS Vol.42 No 2 June 2011
Cooke R.W. (1986), Piled Raft Foundation on Stiff Clays A
Contribution to Design Philosophy, Geotechnique, Vol. 36, No. 2,
pp. 169 -203
Roger Franke, Nicholos Kateziotis , Michel Bustamante, Stavaros
Christoulos (1991- Evaluation of Performance of Two piles Using
Pressuremeter Methods - Journal of Geotechnical Engineering, Vol
117,no.5 May 1991 ASCE pp 695-713

(1)

Balakumar V., Kalaiarasi V. and Ilamparuthi K. (2005). Experimental

and Analytical Study on The Behaviour of Circular Piled Raft on
Sand, Proc. 16th Intl Conf. on Soil Mechanics and Geotechnical
Engineering-2005, Osaka, Japan, pp. 1943-1947.

The value of the Es can be the pressuremeter modulu,Ep is

the pile material modulus, At is the total cross sectional area of
the piles and Ag is the gross plan area of the group. The