2678
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
for single piles is unnecessary because this proof is inconsistent
with the concept of piled rafts. Within the same numerical load
test, proof of the serviceability limit state (SLS) for the piled
raft can be performed and, under the total load of 25 MN, a
settlement of 20mm is calculated, i.e. below the allowable value
of 25mm. It may be observed that at this load level the raft
carries 39% of the total load. Finally, it should be emphasized
that the piled raft solution leads to a significant reduction in the
required number and length (
L
) of the piles as compared to the
conventional pile group, resulting in a saving of 63% in total
pile length, i.e. from 488m for the 4x4 pile group (
L
= 30.5m) to
180m for the 3x3 piled raft (
L
= 20m).
4 CASE HISTORY
The case history for the Messe-Torhaus building in Frankfurt is
presented (Sommer et al 1985). The building is supported by
two separate piled rafts, each with 42 bored piles with a length
of 20m and a diameter of 0.9m. The piles under each raft are
arranged in a 6x7 rectangular configuration with a centre-to-
centre spacing of 2.9m and 3.5m along the shorter and the larger
side of the raft, respectively. Each raft is 17.5m x 24.5m in plan,
2.5m thick and is founded at 3m below ground surface.
The piled raft is embedded in the Frankfurt clay and, within
PGROUPN, it is assumed that
C
u
increases linearly with depth
from 100 kPa at the foundation level to 200 kPa at the pile base,
with a correlation
E
s
/C
u
= 600 and
ν
s
= 0.5. The same soil
parameters were adopted in the variational approach by Chow et
al (2001) so that a direct comparison between analyses may be
made. For consistency with the non-linear Chow analysis, an
elastic-perfectly plastic soil model has been adopted, while a
total load of 181 MN is assumed to act on the piled raft (as only
approximately 75% of the total structural load of 241 MN was
applied at the time of the measurements reported herein). In
addition, the following parameters have been assumed (as these
were not reported by Chow): an adhesion factor (
α
) of 0.7 (in
order to achieve an ultimate pile load of about 7 MN, given that
the measurements showed that piles were carrying at least this
amount of load), and a Young's modulus of 23.5 GPa for the
piles and of 34 GPa for the raft. The latter value results in
K
rs
=
2.2 and hence the PGROUPN assumption of rigid raft is valid,
as confirmed by the field measurements which showed that the
raft actually behaved as fully rigid.
The settlement of the piled raft and the proportion of load
carried by the raft are reported in Table 1 showing a good
agreement between analyses and measurements. In this case,
soil nonlinearity appears to have only a relatively small effect
on the computed response (at least in terms of settlement and
load carried by the raft). The rather low value of the measured
load carried by the raft (20%) suggests that the effect normally
intended by a piled raft was not realised, thereby indicating a
quite conservative design. Indeed, the contact pressures between
raft and soil are scarcely larger than those due to the dead
weight of the raft (i.e. about 25 MN, resulting in a load
proportion of 14%), so that almost the complete load of the
superstructure is carried by the piles. It is also noted that, while
the aim of reducing settlements of the foundation in comparison
to a shallow foundation has been reached (resulting in a
reduction of about 50%), a more efficient design could have
been achieved using fewer piles of greater length. Indeed,
PGROUPN shows that an identical value of settlement can be
Table 1. Settlement and load proportion carried by raft
attained with a significantly smaller total pile length,
specifically with 25.5m long piles in a 4x5 group configuration
(at a spacing of 5.0m and 5.5m along the shorter and the larger
side of the raft, respectively). In this case, a better ratio of the
raft-pile load sharing could have been achieved (i.e. 23%) with
a saving of 39% in total pile length, i.e. from 840m for the
original 6x7 group (
L
= 20m) to 510m for the 4x5 group (
L
=
25.5m). Finally, it is noted that PGROUPN non-linear analyses
for the 6x7 and 4x5 group configurations run in 3 and 1 min,
respectively, on an ordinary computer (Intel Core i7 2.7 GHz),
thereby resulting in negligible computing costs for design.
5 CONCLUSIONS
The paper has described a practical analysis method, based on a
complete BEM solution and implemented in the code
PGROUPN, for determining the non-linear response of piled
rafts. The method has been successfully validated against
alternative numerical analyses and field measurements.
It has been shown that the concept of piled raft, generally
adopted for "large" flexible piled rafts, can also be applied
effectively to "small" rigid piled rafts (and to any larger piled
raft in which the assumption of rigid raft is valid), making
PGROUPN suitable to a wide range of foundations such as
bridges, viaducts, wind turbines and ordinary buildings. In such
cases, if the raft can be founded in reasonable competent ground
(which can provide reliable long-term resistance), then the extra
raft component of capacity can be used to significantly reduce
the piling requirements which are necessary to achieve the
design criteria (e.g. ultimate bearing capacity, settlement).
Given the relatively high load level at which the piles
operate within a pile-raft system, the influence of soil
nonlinearity can be significant, and ignoring this aspect can lead
to inaccurate predictions of the deformations and the load
sharing between the raft and the piles. Consideration of soil
nonlinearity would also be required if PGROUPN is used to
perform a numerical load test following the methodology
outlined in the International CPRF Guideline. Due to the
negligible costs (both in terms of data preparation and computer
execution times), a large number of cases can be analysed
efficiently, enabling parametric studies to be readily performed.
This offers the prospect of more effective design techniques and
worthwhile savings in construction costs.
6 REFERENCES
Basile F. 2003. Analysis and design of pile groups. In
Num. Analysis
and Modelling in Geomech.
(ed. J.W. Bull), Spon Press, 278-315.
Bond A.J. and Basile F. 2010. Repute 2.0, Software for pile design
and analysis.
Reference Manual
, Geocentrix Ltd, UK, 49p.
Chow Y.K., Yong K.Y. and Shen W.Y. 2001. Analysis of piled raft
foundations using a variational approach.
Int. J. Geomech
. 1 (2),
129-147.
Horikoshi K. and Randolph M.F. 1997. On the definition of raft-soil
stiffness ratio.
Géotechnique
47 (5), 1055-1061.
Katzenbach R. 2012.
Combined Pile-Raft Foundations
. International
CPRF Guideline.
Kuwabara F. 1989. An elastic analysis for piled raft foundations in
homogeneous soil.
Soil and Foundations
29 (1), 82-92.
Poulos H.G. 2001. Piled-raft foundation: design and applications.
Géotechnique
51 (2), 95-113.
Randolph M.F. 2003. 43rd Rankine Lecture: Science and empiricism in
pile foundation design.
Géotechnique
53 (10), 847-875.
Shen W.Y., Chow Y.K. and Yong K.Y. 2000. A variational approach
for the analysis of pile group-pile cap interaction.
Géotechnique
50
(4), 349-357.
Sommer H., Wittmann P. and Ripper P. 1985. Piled raft foundation of a
tall building in Frankfurt clay.
Proc. XI ICSMFE
, 2253-2257.
Viggiani C., Mandolini A. and Russo G. 2012.
Piles and pile
foundations
. Spon Press, 278p.
Settlement
(mm)
Load carried
by raft (%)
Measured (Sommer et al 1985)
45
20
Chow et al (2001)
45
26
PGROUPN
44
21
PGROUPN (linear elastic)
43
21
PGROUPN (4x5 group, L= 25.5m)
44
23
