International Conference on Deep Foundations – CPRF and Energy Piles,

Frankfurt am Main, 2009

Optimised Design of Combined Pile Raft Foundations

Oliver Reul, Mark F. Randolph

Proc. International Conference on Deep Foundations – CPRF and Energy Piles, Darm-
stadt Geotechnics No. 18, 149 - 169
 Optimised Design of Combined Pile Raft Foundations

Prof. Dr. Oliver Reul

University of Applied Sciences Darmstadt / CDM Consult GmbH, Alsbach

Prof. Dr. Mark F. Randolph

Centre for Offshore Foundation Systems, The University of Western Australia, Perth

1 Introduction

An increasing number of structures, especially tall buildings, are founded on Com-

bined Pile Raft Foundations (CPRF), referred to simply as piled rafts in this paper.
For this reason, design strategies that enable an optimised design have a major im-
portance for an economic construction to be achieved.

Serviceability is the decisive criterion for the design of vertically loaded piled rafts
of tall buildings under practical relevant loads. The magnitude of settlements and
of differential settlements and the resulting tilting of the building, which can lead
to impairments of technical installations such as elevators, are therefore of special
importance. Moreover, the objective of an economic optimised foundation design
has to include minimisation of the bending moments acting on the raft.

An optimised design of a piled raft can therefore be defined as a design achieving

maximum economy of the solution, that is minimum cost for the installation of the
foundation, while maintaining satisfactory performance.

This paper presents the results of a parametric study by means of three-dimensional

elastoplastic finite element analyses. In the study the pile positions, the pile num-
ber, the pile length and the raft-soil stiffness ratio as well as the load distribution
on the raft has been varied. Based on the parametric study design strategies for an
 -2-

optimised design of piled rafts subjected to nonuniform vertical loading are dis-
cussed.

Additionally, for the recently finished high rise building WestendDuo in Frankfurt
the stages of the design process are presented and the settlement performance as
well as the building costs are compared for different foundation layouts.

2 Parametric Study

2.1 Numerical Model

The parametric study presented in this paper was carried out with a structural model
based on the finite element method modelling a subsoil condition based on that in
Frankfurt am Main, Germany, which is characterised mainly by tertiary soils and
rock. These comprise Frankfurt clay at the top underlain by the rocky Frankfurt
limestone. The Frankfurt clay is a stiff, overconsolidated clay with liquid limit,
plastic index and natural moisture content very similar to the London clay (Butler
1975), and typical of soil types where piled raft foundations are suitable. Sand and
limestone bands of varying thickness are embedded in the Frankfurt clay, which
results in a nonhomogenous appearance of the layer as a whole. Since the com-
pressibility of the Frankfurt limestone is small compared to the Frankfurt clay, the
boundary between Frankfurt clay and Frankfurt limestone is assumed to be a rigid
base in the finite element model.

The nonlinear material behaviour of the soil was simulated with an elastoplastic
cap-model. The material parameters used in the threedimensional finite element
analyses are documented in Reul and Randolph (2004).

Verification and calibration of the structural model is based on the back analysis of
static load tests and the measured bearing behaviour of foundations. Reul and Ran-
dolph (2003) show the results of the detailed back analyses of three piled raft foun-
dations in Frankfurt.
 -3-

Fig. 1 Model conditions in the parametric study and finite ele-

ment mesh
Fig. 1 shows the model conditions and finite element mesh applied in the paramet-
ric study. Because of the threefold symmetry, only one eighth of the system had to
be modelled with finite elements. The foundation level was set 14 m below ground
level. Only the soil below the foundation level is modelled with finite elements,
with the soil above the foundation level being considered as applying a surcharge
only. The base of the compressible soil layer, i.e. the Frankfurt clay, was set 83 m
below ground level. In the horizontal direction the model extends 190 m, i.e. 171 m
beyond the edge of the raft, and infinite elements are used at the edge of the mesh
in order to minimise any boundary effects.

In the parametric study square unpiled rafts and piled rafts with an edge length of
B = 38 m have been considered. Three basic pile configurations were investigated
(Fig. 2a). Pile configuration p1 has the piles uniformly distributed under the whole
raft area. In pile configuration p2 the piles are placed only in the central area of the
raft (beneath the core loading as detailed below). Pile configuration p3 has piles
under the central area of the raft as well as under the edges of the raft. The number
of piles was varied between n = 9 and n = 169 and the pile length was varied be-
tween Lp = 10 m and Lp = 50 m. The pile diameter was held constant at dp = 1.0 m.
Pile spacing was varied between s = 3 m and s = 6 m.
 -4-

Fig. 2 System configurations and load configurations for the

parametric study
The raft-soil stiffness ratio was varied between Krs  0.008 (approximately fully
flexible) and Krs  54 (approximately rigid), where Krs is defined according to Hor-
ikoshi and Randolph (1997):
0.5 3
E 1   s2  B   t r 
K rs  5.7 r     (1)
E s ,eq 1   2r  L   L 
with Er = Young’s modulus of the raft; Es,eq = equivalent Young’s modulus of the
soil; r = Poisson’s ratio of the raft; s = Poisson’s ratio of the soil and tr = thickness
of the raft. In the scope of this work the equivalent Young’s modulus for the soil,
Es,eq, is defined to take the nonlinear soil behaviour in to account (Reul and Ran-
dolph 2004). Regardless of the varying raft-soil stiffness ratios, which were
achieved by varying the raft thickness, the weight of the raft was the same in all
the analyses.

As shown in Fig. 2b four different load configurations have been studied. For Load
type I half of the load has been applied in a core area at the centre of the raft and
the other half at the edge of the raft. The core area is 25 % of the total area of the
raft. This is a typical configuration for tall buildings with stiff structural elements
such as elevator shafts and stairways at the centre of the raft and columns at the
edge to support the facade. Load type II and III have loads applied only in the core
 -5-

area or the edge area, respectively. Load type IV describes a uniform load over the
whole raft area, a load configuration frequently applied in parametric studies on the
behaviour of (piled) rafts (e.g. Fraser and Wardle 1976, Horikoshi and Randolph
1998, Katzenbach and Reul 1997). The total applied load in the four different load
configurations is the same.

The aim of the application of the four different load configurations is mainly to
investigate the global response of the piled raft under nonuniform loading. The in-
vestigation of the necessity to install piles, if column loadings are present as dis-
cussed by Poulos (2001), is outside the scope of what is considered here.

The construction process and the loading has been modelled in a step-by-step anal-
ysis (Reul and Randolph 2004). The maximum load including the weight of the raft
and the uplift due to pore pressures amounts to Peff = 721.7 MN, which is approxi-
mately 20 % of the ultimate capacity of an equivalent unpiled raft under drained
conditions. The load is within the range of values known for piled rafts in Germany
(Reul and Randolph 2004). All results presented in this paper are related to the
situation after the installation of the raft, i.e. deformations due to the weight of the
raft are not considered.

In the scope of this research work a total of 259 system configurations has been
analysed (Reul 2001).

2.2 Results of the parametric study

Tab. 1 summarises the results of the parametric study (Krs  3.6) in terms of foun-
dation configurations that give the smallest average settlement, saverage, the smallest
differential settlement s, and the smallest maximum positive and maximum neg-
ative bending moments, for two different load levels. Unless the aim of a founda-
tion design process is solely to minimise a certain parameter such as the differential
settlements, these configurations are obviously not optimised design solutions. The
results show that not only minimum values for different investigated parameters
are achieved with different foundation configurations, but also that the foundation
configurations giving the minimum value change with increasing load levels. This
is an important fact when targeting the optimised design of foundations.
 -6-

Loadtype Loadlevel saverage s m+  m

Vult/Peff (s  0)
20 p1, n = 169, p2, n = 16, p2, n = 49, p3, n = 33,
Lp = 50 m, Lp = 30 m, Lp = 50 m, Lp = 50 m,
tr = 3 m tr = 1 m tr = 3 m tr = 3 m
7 mm 0 mm 0.199 MNm/m 0.121 MNm/m
I
5 p1, n = 169, p2, n = 16, p2, n = 49, p3, n = 33,
Lp = 50 m, Lp = 30 m, Lp = 50 m, Lp = 30 m,
tr = 3 m tr = 3 m tr = 3 m tr = 3 m
38 mm 2 mm 0.760 MNm/m 0.259 MNm/m
20 p1, n = 169, p2, n = 49, p3, n = 73,for unpiled rafts
Lp = 50 m, Lp = 50 m, Lp = 50 m, and pile config-
tr = 3 m tr = 3 m tr = 1 m uration 2 with
tr  3 m:  m =
9 mm 7 mm 0.640 MNm/m 0
II
5 p1, n = 169, p2, n = 49, p2, n = 49,for unpiled rafts
Lp = 50 m, Lp = 50 m, Lp = 50 m, and pile config-
tr = 3 m tr = 3 m tr = 1 m uration 2 with
tr  3 m:  m =
46 mm 25 mm 2.828 MNm/m 0
20 p1, n = 169,  p3, n = 33, p3, n = 40,
Lp = 50 m, Lp = 30 m, Lp = 50 m,
tr = 1 m tr = 3 m tr = 1 m
5 mm 0.183 MNm/m 0.379 MNm/m
III
5 p1, n = 169,  p3, n = 33, p3, n = 33,
Lp = 50 m, Lp = 50 m, Lp = 50 m,
tr = 1 m tr = 3 m tr = 1 m
29 mm 0.488 MNm/m 1.203 MNm/m
20 p1, n = 169, p3, n = 73, p1, n = 169, for all unpiled
Lp = 50 m, Lp = 50 m, Lp = 50 m, rafts:
tr = 3 m tr = 1 m tr = 1 m
7 mm 0 mm 0.065 MNm/m  m = 0
IV
5 p1, n = 169, p2, n = 49, p2, n = 49, for all unpiled
Lp = 50 m, Lp = 10 m, Lp = 50 m, rafts:
tr = 3 m tr = 1 m tr = 3 m
37 mm 1 mm 0.273 MNm/m  m = 0
Tab. 1 Summary of the parametric study: Foundation configura-
tions that give the smallest average settlement, the small-
est differential settlement, and the smallest maximum
positive and maximum negative bending moments
 -7-

Based on the parametric study a set of charts has been developed for the system
and load configurations shown in Fig. 2, where the coefficient for the average set-
tlement, the coefficient for the differential settlement and the coefficient for the
maximum positive bending moment are plotted as a function of the total pile length
(Reul 2001). The coefficients are defined as follows:

s pr  s  s r (2)
s pr  s  s r (3)
m pr    m  m r  (4)
with s pr = average settlement of the piled raft; s r = average settlement of the
equivalent unpiled raft;  s = coefficient for the average settlement; s pr = differ-
ential settlement of the piled raft; s r = differential settlement of the equivalent
unpiled raft; s = coefficient for the differential settlement; m pr = maximum
positive bending moment of the piled raft; mr = maximum positive bending mo-
ment of the equivalent unpiled raft and  m = coefficient for the maximum positive
bending moment.

Eqs. (2) to (4) given above define constraints, which ensure the satisfactory bearing
behaviour of the piled raft. Generally, the majority of these coefficients can be as-
sumed to be smaller than 1. Otherwise the performance of the unpiled raft would
not need to be improved by the installation of piles. In establishing these coeffi-
cients quantitatively, local standards and practice give the benchmarks and the lim-
iting criteria (Cunha et al. 2001).

Fig. 3 shows as an example the charts for a foundation under core-edge loading
with n = 33 piles. It is worth mentioning that there are in fact system configurations
that yield increased differential settlements and maximum positive bending mo-
ments of the piled raft compared to the unpiled raft (s > 1, m+ > 1).

The normalised bending moments per unit length acting on cross-sections perpen-
dicular to the x-axis are plotted in Fig. 4 for the cross-section A-A for uniform and
core-edge loading, respectively. Three different pile configurations (Krs  1.5,
Lp/dp = 30, s = 6dp) are compared with unpiled rafts with the same raft-soil stiff-
ness ratio subjected to the same loading. For the unpiled rafts as well as for the
piled rafts the normalised bending moment distribution depends on the type and
level of load.
 -8-

Fig. 3 Example for the variation of the coefficients for the aver-
age settlement, for the differential settlements and the
maximum positive bending moments with the total pile
length
From Fig. 4 it can be concluded that for the configurations investigated, the instal-
lation of piles yields no benefits in terms of the reduction of bending moments
compared to the unpiled rafts in the case of uniform loading (Loadtype IV). Pile
configuration p2 with piles placed only under the central area of the raft changes
the deformation mode of the raft from sagging to the structurally less favourable
hogging, although the absolute magnitudes are reduced.
 -9-

Fig. 4 Normalised bending moments per unit length in cross

section A-A (Krs  1.5, Lp/dp = 30)
For core-edge loading (Loadtype I) pile configuration p3 most efficiently reduces
the negative bending moments at the expanse of slightly increased maximum mo-
ments in the centre of the raft for Vult/Peff = 5 while pile configuration p2 shows
increased negative bending moments compared to the unpiled raft (Fig. 4).

2.3 Example Design Optimization

In the following example, an optimised design process is investigated for a foun-

dation (B = 38 m, Peff = 721.7 MN, Vult/Peff = 5) subjected to core–edge loading
(Loadtype I). With tr = 1 m (Krs  0.06) and tr = 3 m (Krs  1.5) two different raft
thicknesses have been considered. The constraints that ensure the satisfactory bear-
ing behaviour of the piled raft will be expressed in terms of the behaviour of the
corresponding unpiled raft as defined in Section 2.2. In the example the constraints
 - 10 -

are assumed to be the average settlement, the differential settlement, and the max-
imum positive bending moment per unit length, which are summarised in Tab. 2
for the unpiled raft.

The aim of the optimised design process is to find a foundation configuration that
gives the minimum total pile length nLp under the assumption of the following
constraints:

Average settlement of the piled raft, saverage  100 mm:

s  0.39 t r  1 m 
 (5)
s  0.45 t r  3 m 

Differential settlement (deflection ratio), s/B  1/1000:

 s  0.26 t r  1 m 
 (6)
 s  2.92 t r  3 m 

No hogging of the raft:

  s  0 t r  1 m and t r  3 m  (7)

Maximum positive bending moment of the pile raft reduced compared to the
unpiled raft, mpr+  mr+:
  m  1 t r  1 m and t r  3 m  (8)

The pile diameter is fixed to dp = 1 m and the pile length is limited to Lp  50 m.

tr saverage s s/B m+
[m] [mm] [mm] [-] [MNm/m]
1 255 148 1/257 7.8
3 221 13 1/2923 14.6
Tab. 2 Example design optimization: Results for the unpiled
rafts
 - 11 -

tr Pile n nLp nLp for feasible

configuration s-constraint s-constraint m+-con- configurations
[m] [-] [m] [m] straint [m]
[m]
1 49 1840 X1 490 X3
1 169 4182 4823 1690 4823
2 9 X1 260 90 X3
2 16 X1 426 160 X3
1
2 49 2232 (X2) 632 490 X3
3 33 1510 X1 330 X3
3 40 1614 X1 400 X3
3 73 2050 1916 730 2050
1 49 1387 490 2166 2166
1 169 3183 1690 4945 4945
2 9 X1 90 90 X3
2 16 X1 160 160 X3
3
2 49 1893 (X2) 490 490 X3
3 33 1160 330 1626 1626
3 40 1297 400 X1 X3
3 73 1936 730 730 1936
X1 Constraint not fulfilled
X2 Hogging of the raft
X3 No feasible configuration
Tab. 3 Example design optimization: Feasible foundation con-
figurations
Application of the charts developed by Reul (2001) for core-edge loading leads to
a number of feasible foundation configurations, which are summarised in Tab. 3.
The feasible total pile lengths vary significantly for the different investigated pile
configurations. The minimum feasible total pile lengths are nLp = 2050 m
(tr = 1 m) and nLp = 1626 m (tr = 3 m), respectively. For the given load configura-
tion and constraints, pile configuration p3 with piles installed under the core and
the edge of the raft is the favourable solution for both investigated raft thicknesses.
 - 12 -

Fig. 5 Example design optimization: System configurations

tr Pile n Lp,core Lp,edge s s m+ nLp
[m] config. [-] [m] [m] [-] [-] [-] [m]
1 3 73 30 22 0.39 0.08 0.18 1998
3 3 33 44 30 0.45 1.12 0.84 1116
Tab. 4 Example design optimization: Results achieved with
nonuniform pile lengths
A further reduction of the total pile length can be achieved with nonuniform pile
lengths. Based on the pile configurations p3 with n = 73 (tr = 1 m) and n = 33
(tr = 3 m) the length of the piles under the core and under the edge of the raft has
been varied. Tab. 4 and Fig. 5 summarise the optimised design, which has been
achieved with two additional analyses for n = 73 (tr = 1 m) and three additional
analyses for n = 33 (tr = 3 m). For both pile configurations, the average settlement
turns out to be the critical parameter in the design process. Even with this limited
number of additional analyses, the total pile length can be reduced further by 3 %
(n = 73, tr = 1 m) and 31 % (n = 33, tr = 3 m), respectively.

3 Case History: WestendDuo, Frankfurt

3.1 Situation

The building complex WestendDuo in Frankfurt, a reinforced-concrete skeleton

construction consisting of two connected, 96 m high office towers (Fig. 6) and a
low rise section with a maximum height of approximately 20 m, was constructed
from November 2004 to November 2006. The WestendDuo has replaced the office
tower of the HOCHTIEF AG. The old office tower, which was founded on a raft,
was built during two construction phases from 1966 to 1968 and from 1973 to 1974.
The demolition works included the 83 m high tower and the 3-storey basement.
 - 13 -

Fig. 6 WestendDuo, Frankfurt

The basement area of the new building complex with a size of approximately
4100 m² is equivalent to the basement area of the demolished building. At the
north-western corner of the site, the heritage-protected historic Miquel-Villa is sit-
uated in the immediate vicinity of the 4-storey basement of the WestendDuo (Fig.
7).

The load of the superstructure of the WestendDuo amounts to approximately

G+Q = 695 MN + 220 MN = 915 MN (G: dead loads; Q: live loads) and is trans-
ferred to the raft mainly in the approximately 17 m  30 m core area of the two
towers via walls and at the edge of the building via columns (maximum column
load G+Q = 23.3 MN).

For the construction of the 4-storey basement of the WestendDuo a 15 m deep ex-
cavation pit was required after the demolition of the existing basement. The retain-
ing wall comprised bored piles with a diameter of 0.9 m supported by four rows of
 - 14 -

pre-stressed anchors. The spacing between the bored piles was covered with shot-
crete. The groundwater was lowered by means of wells equipped with pumps lo-
cated outside the excavation pit. To relax the groundwater circulating in tertiary
sand and limestone bands and to ensure the stability of the excavation pit against
uplift additional borings were placed inside the excavation pit.

The heating and cooling power of 300 kW required for the building is extracted
from the groundwater. Groundwater is pumped with a rate of approximately
43 m³/h from two 140 m deep production wells, sent through a heat pump/cooling
unit and re-infiltrated via three 140 m deep injection wells. The groundwater is
pumped with a temperature of 18°C and re-infiltrated with a temperature of 10°C
in winter (heating cycle) and 26°C in summer (cooling cycle), respectively.

The subsoil condition on the project site is characterised mainly by tertiary soils
and rock with artificially filled soils and quaternary sand and gravel with a thick-
ness of approximately 6 m just below the ground surface. The tertiary soils consist
of Frankfurt clay with a thickness of approximately 79 m at the top underlain by
the Frankfurt limestone.

Fig. 7 WestendDuo: Ground plan of the site

 - 15 -

The groundwater circulates in the quaternary sand and gravel as well as in the ter-
tiary sand and limestone bands while the tertiary clay is practically impermeable.
The quaternary and tertiary groundwater layers are connected. A groundwater
drawdown in the tertiary layers may result in a reduction of the hydraulic head
within an area with a radius of several hundred metres. Measurements showed the
groundwater level at a depth of approximately 10.4 m due to neighbouring ground-
water drawdowns. The natural groundwater level is situated approximately 6.8 m
below the ground surface.

3.2 Foundation design

In the course of the technical and economic design process for the foundation five
design alternatives were investigated by means of three-dimensional elastoplastic
finite element analysis. In a pre-design study the unpiled raft F1 with a constant
raft thickness of tr = 2.4 m had been identified as the starting configuration of the
design process. Furthermore, an unpiled raft with a constant raft thickness of tr =
4.0 m (F2) and piled rafts K1, K2 and K3 where the number of piles, the pile length,
the pile diameter and the raft thickness were varied, were investigated. A detailed
documentation of the numerical analyses carried out for the foundation design of
the WestendDuo is given by Reul et al. (2006).

Tab. 5 compares the main results, i.e. the maximum settlement, the deflection ratio
of the raft as defined by Burland et al. (1977) and the piled raft coefficient. The
piled raft coefficient, pr, describes the ratio of the sum of all pile loads, Ppile, to
the total load on the foundation, Ptot. A piled raft coefficient of unity indicates a
freestanding pile group whereas a piled raft coefficient of zero describes an unpiled
raft.

For the unpiled rafts maximum settlements of smax = 13.7 cm (F1) and smax
= 13.5 cm (F2) have been calculated. For configuration F1 the deflection ratio /L
 1/300 is critical because the resulting cracks in the concrete structure might cause
problems for the impermeable basement. For the piled rafts the maximum settle-
ments and the deflection ratios are significantly reduced and range between smax
= 6.2 cm (K1) and smax = 7.2 cm (K3) and /L  1/900 (K1) and /L  1/500 (K3),
respectively.

The share of the piles in the load transfer to the subsoil, as described by the piled
raft coefficient pr, is only 8 % higher for configuration K1 than for configurations
 - 16 -

K2 and K3. This is remarkable since the total pile length nLp is approximately
44 % higher for configuration K1 than for configurations K2 and K3.

For the case of a raft that is sagging, i.e. the settlements at the centre of the raft are
larger than at the edge, a maximum deflection ratio of /L = 1/500 is generally
considered to be the limit to prevent significant cracks, which could endanger the
impermeability of the structure (Schultze & Horn 1995). For the piled rafts (K1,
K2, K3) and the unpiled raft with the thicker raft (F2; tr = 4.0 m) this constraint is
fulfilled.

As a result of the optimisation process foundation configuration K3 (Fig. 8) was

selected for construction, yielding tolerable deformations and the most efficient
solution economically. The costs for the different foundation configurations are
shown in Tab. 6 where the costs of the single components have scaled to the total
cost of foundation configuration K3. Please note that only the components that are
directly affected by the foundation design have been considered.

Foundation configuration K3 has approximately 9 % to 11 % lower costs than the

other piled rafts K1 and K2 and approximately 23 % to 26 % lower costs than the
unpiled rafts F1 and F2. Moreover, for foundation configuration F1 the servicea-
bility can not be ensured due to the high deflection ratio.

Foundation tr n Lp dp nLp smax /L αpr

configuration hr / lr
[m] [-] [m] [m] [m] [cm] [-] [-]
F1 unpiled raft 2.4/2.4     13.7 ~1/300 
F2 unpiled raft 4.0/4.0     13.5 ~1/500 
K1 piled raft 2.4/2.4 28 30 1.3 840 6.2 ~1/900 0.460
K2 piled raft 2.4/2.4 26 25/20 1.2 585 7.0 ~1/700 0.377
K3 piled raft 1.8/1.2 26 25/20 1.2 585 7.2 ~1/500 0.380
tr thickness of the raft (hr: high rise nLp total pile length
section; lr: low rise section) smax maximum settlement
n number of piles /L deflection ratio of the raft
Lp pile length pr piled raft coefficient
dp pile diameter

Tab. 5 WestendDuo: Results of the foundation analyses for the

design process
 - 17 -

Fig. 8 WestendDuo: Layout of the foundation

 - 18 -

Component Unpiled raft Piled raft

F1 F2 K1 K2 K3
Retaining structure for the 45.5 % 50.0 % 45.5 % 45.5 % 42.5 %
excavation pit
Excavation and disposal of 17.9 % 22.5 % 17.9 % 17.9 % 15.5 %
soil
Dewatering system 8.1 % 8.3 % 8.1 % 8.1 % 8.1 %
Foundation piles 0.0 % 0.0 % 9.9 % 6.1 % 6.1 %
Raft 50.6 % 54.2 % 28.3 % 29.7 % 26.4 %
Measurement devices (foun- 0.7 % 0.7 % 1.4 % 1.4 % 1.4 %
dation, dewatering, retaining
structure)
 122.9 % 135.8 % 111.2 % 108.8 % 100.0 %

Tab. 6 WestendDuo: Costs for the different foundation configu-

rations
The comparison shows that with the selected foundation configuration K3 a signif-
icant cost reduction, especially for the raft, could be achieved. Compared to foun-
dation configuration F2 the selected foundation configuration K3 reduced the re-
quired reinforcement steel from approximately 2000 tons to approximately
1150 tons.

3.3 In-situ-Measurements

According to design codes such as Eurocode EC 7 the foundations of high rise

buildings are usually classed with geotechnical category GK3. Therefore in-situ
measurements are an essential aspect of the safety concept. Furthermore they are
used for quality control and documentation of deformations of the foundation and
of neighbouring structures. In the present case, three foundation piles were
equipped with a load cell at the pile head, and five contact pressure cells and five
pore pressure cells were placed beneath the raft to establish the load transfer to the
subsoil. Fig. 8 shows the location of the measurement devices in the ground plan.
 - 19 -

Fig. 9 WestendDuo: Measured settlements in November 2006

after the building had been finished
The deformations of the foundation are monitored with 23 geodetic survey points
located in the basement. Fig. 9 shows the maximum settlement measured in No-
vember 2006 when the building had just been finished amounting to s = 4.7 cm. It
can be concluded that even allowing for small time dependent settlements due to
consolidation and creep, the maximum predicted settlements (Tab. 5) will not be
exceeded.

4 Summary and Conclusions

The parametric study has shown that the optimised design of a foundation clearly
depends on the subsoil conditions, the load configuration and the load level. There-
fore, a generalization of the results presented in this paper to all possible cases to
be found in practice is not practicable. However, key results of the parametric study
can be summarised as follows (Reul and Randolph 2004).
 - 20 -

Generally, for the same total pile length smaller average settlements are achieved
with longer piles rather than with a higher number of piles. The average settlement
is the only parameter that is reduced compared to the unpiled raft, due to the instal-
lation of piles, for all configurations studied in the scope of this research work. The
overall stiffness of a piled raft decreases with increasing load level.

The differential settlements are much more sensitive to the raft-soil stiffness ratio
and the load configuration than the average settlements. For a raft under uniform
loading or core-edge loading, the differential settlements can be most efficiently
reduced by installation of piles only under the central area of the raft.

The calculation of bending moments is very sensitive to whether or not nonlinear

raft-soil interactions are considered. For a raft subjected to uniform loading the
installation of piles seems not to reduce the bending moments compared to the
unpiled raft.

The worked example for an optimised design process, as well as the case history
WestendDuo, shows the significance of a detailed parametric study based on a set
of initial design variables in the search for an optimised foundation design, as
pointed out by Cunha et al. (2001) in their investigation of design alternatives for
a piled raft case history. The examples suggest that the installation of piles under
the loaded areas, together with nonuniform pile lengths, yield the minimum total
pile length.

References

Burland, J.B., Broms, B.B., De Mello, V.F.B (1977)

Behaviour of foundations and structures. Proc. 9th Int. Conf. on Soil Mech.
and Found. Engrg., Tokyo, 2, 495-546. Rotterdam: Balkema

Butler, F.G. (1975)

Heavily over-consolidated clays - Review paper: Session III. Proc. Conf.
Settlements of Structures, 531-578

Cunha, R.P.; Poulos, H.G.; Small, J.C. (2001)

Investigation of design alternatives for a piled raft case history. J. of Ge-
otech. and Geoenv. Engrg., ASCE, Vol. 127, No. 8, 635-641.

Fraser, R.A.; Wardle, L. J. (1976)

 - 21 -

Numerical analysis of rectangular rafts on layered foundations. Géotech-

nique, 26, No. 4, 613-630.

Horikoshi, K.; Randolph, M.F. (1997)

On the definition of raft-soil stiffness ratio. Géotechnique, 47, No. 5, 1055-
1061.

Horikoshi, K.; Randolph, M. F. (1998)

A contribution to the optimum design of piled rafts. Géotechnique, 48, No.
2, 301-317.

Katzenbach, R., Reul, O. (1997)

Design and performance of piled rafts. Proc. XIVth International Conference
on Soil Mechanics and Foundation Engineering, Hamburg, Vol. 4, 2253-
2256, Rotterdam: Balkema

Poulos, H.G. (2001)

Piled-raft foundation: design and applications. Géotechnique, 51, No. 2, 95-
113.

Reul, O. (2001)
Numerical study on the bearing behaviour of piled rafts subjected to nonu-
niform vertical loading. Data Report, GEO:03294, The University of West-
ern Australia, Centre for Offshore Foundation Systems.

Reul, O.; Randolph, M.F. (2003)

Piled rafts in overconsolidated clay – Comparison of in-situ measurements
and numerical analyses. Géotechnique, 53, No. 3, 301-315.

Reul, O.; Randolph, M.F. (2004)

Design strategies for piled rafts subjected to nonuniform vertical loading. J.
of Geotech. and Geoenv. Engrg., ASCE, Vol. 130, No. 1, 1-13.

Reul, O., Ehrhardt, G., Rummel, B. (2006)

Entwurfsoptimierung einer Hochhausgründung im Frankfurter Ton. Vor-
träge des 5. Kolloquiums Bauen in Boden und Fels; Technische Akademie
Esslingen, 309-318

Schultze, E, Horn, A. (1995)

Setzungsberechnung. Grundbautaschenbuch, 5. Auflage, Teil 1, 225-254,
Ernst & Sohn, Berlin.