Sei sulla pagina 1di 8

Optimization of a deep foundation with diaphragm

wall panels employing 3D FE analysis


F. Tschuchnigg1
Computational Geotechnics Group, Graz University of Technology, Austria

ABSTRACT
The project discussed in this paper is a tower located in Bucharest with a height of 137 m. To find the optimal layout of the
foundation elements with respect to minimising vertical displacements, conventional calculations are in general not sufficient
and advanced numerical modelling of the soil-structure interaction is essential. The paper shows results from numerical analyses
with the objective to assess the settlement behaviour of the tower. Due to the geometric layout 2D analyses proved to be too
conservative and therefore a number of 3D analyses have been performed. Different arrangements of diaphragm wall panels
have been investigated to find an economical and technical feasible solution for the layout of the foundation elements. For the
executed deep foundation concept a parametric study based on soil data recently published in the literature is presented. Finally
results from 3D finite element simulations of an in situ load test, performed to obtain additional information of the settlement
behaviour of the deep foundation, are presented. The test was conducted on a diaphragm wall element (barrette) using the
"Osterberg Method".

Keywords: finite element, deep foundation, soil-structure interaction, diaphragm wall

numerical analysis are required. 2D plane strain


1 INTRODUCTION models are simple and convenient for
preliminary studies, but due to the fact that in
If ground conditions are such that the load from general a two-dimensional representation of a
structures such as high-rise buildings cannot be group of deep foundation elements is not
supported by shallow foundations several options possible 3D analyses are preferred where, in
exist. Depending on the soil profile and the principle, no restrictions with respect to
corresponding soil properties a pile foundation, a geometric conditions and load distribution exist.
piled raft foundation or a diaphragm wall The project discussed, namely the Sky Tower,
foundation is the solution for most cases. For this is the most impressive part of the so-called
type of deep foundation systems assessment of Floreasca City Center in the northeastern part of
settlements and differential settlements are the Bucharest (Figure 1). The tower, with a height of
key issues. Thus ultimate limit state conditions 137 m will be the highest building in Bucharest.
are not considered in the proposed paper. In the first part of the paper, the results from
When dealing with such foundations relatively numerical analyses with the objective assessing
complex interactions between the different the settlement behaviour of the tower are
foundation elements occur and as a consequence presented. A number of 3D analyses have been

1
Computational Geotechnics Group, Institute for Soil Mechanics and Foundation Engineering, Graz University of Technology,
Rechbauerstrasse 12, 8010 Graz, Austria. Franz.tschuchnigg@tugraz.at
performed where different arrangements of The two egg-shaped areas represent the
diaphragm wall panels have been investigated to regions where high point loads, up to 14900 kN
find an economical and technical feasible are acting. The thickness of the foundation slap
solution for the layout of the foundation is 2.5 m in the inner region of the excavation and
elements. 1.6 m in the outer areas. The diaphragm wall
panels have a thickness of 0.8 m beneath the high
loaded areas and the sensitive zones and 0.6 m in
the other regions. The diaphragm wall, which
acts as a retaining wall for the excavation and
also as a foundation element, has a thickness of
1.0 m.
It is planed to install the deep foundation
elements from the ground surface and to realise
the excavation afterwards with the top-down
method. As a consequence the panels are acting
as tension elements and minimise the heave
during excavation. Detailed information about
Figure 1. Floreasca City Center (c) beyer.co.at
the construction sequence is given in [3].
Due to the high loads in the core of the
For the executed foundation concept a
construction large differential settlements of the
parametric study based on recently published soil
foundation slab are expected. The aim of the 3D
data has been carried out.
finite element analysis is to minimise both the
Additionally results from finite element
total deformations of the construction and, even
simulations of an in situ load test are presented.
more important, the differential displacements of
This test was performed to obtain additional
the slab.
information on the settlement behaviour of the
deep foundation elements and to confirm the
assumptions made related to soil properties used
for the 3D analysis of the entire foundation. The
test was conducted on a diaphragm wall element
(barrette) using the "Osterberg Method" (O-
Cell) [1].
All calculations in the paper are performed
with the Finite element code Plaxis 3D
Foundation [2]. The mechanical behaviour of the
soil is described with both the Hardening Soil
and the Hardening Soil Small model, which are
double hardening models available in the Plaxis
model library. Figure 2. Top view of construction

2 PROJECT OVERVIEW 3 SOIL CONDITION AND MODELLING

The top view of the project layout is shown in The soil profile for the finite element simulation
Figure 2. The excavation has a maximum length is based on core drillings with depths down to
of 93.4 m and a maximum width of 61.7 m. The -60.0 m from the surface. All borehole logs show
bottom of the foundation slab is -20.4 m below alternate layers of either sands or silty clays. For
the ground surface. the calculations presented in this paper either the
so-called Hardening Soil model [4] or the
Hardening Soil Small model [5] was used to is modelled (Model B). For the calculations with
model the soil behaviour. Both models are the final layout of the deep foundation elements
elasto-plastic constitutive models, which enable another symmetry axis is defined and only one
to model both deviatoric and volumetric quarter of the construction is modelled
hardening and take the stress dependency of (Model A). This enables a much finer
stiffness into account. discretization of the mesh. Figure 2 also shows
The Hardening Soil Small (HSS) model the symmetry used for Model A. As an example
additionally allows for modelling the high Figure 3 shows one finite element model of
stiffness at very low strains. As a consequence Model B. All models analysed consist of around
the obtained soil displacements at deeper depths 52000 elements. The model depth of all 3D
are automatically reduced and a more realistic models is 80.0 m. The deepest borehole reached
settlement profile with depth can be computed only to a depth of -60.0 m and it is therefore
[6]. assumed that the alternating layers of sands and
Compared with the Hardening soil model silty clays continue. Nevertheless sensibility
(HS) the HSS model needs two additional analyses were performed to assess the influence
parameters to describe the stiffness behaviour at of the uncertainties in the soil profile.
small strains. Namely the initial shear modulus All calculations in this paper are drained
G0 and the shear strain level 0.7, which analysis, which means final settlements are
represents the amount of shear strains where the presented. This seems to be justified because of
secant shear modulus is reduced to 70% of its the alternate layers of sands and silty clays,
initial value. which speeds up the consolidation procedure.
The stress dependency of stiffness is taken To obtain realistic deformations of the
into account in the constitutive model as excavation pit and a reliable stress distribution in
proposed by Ohde [7], but slightly modified the soil after the excavation it is necessary to
according to the following equation: model the real construction sequence. To include
the high stiffness of the superstructure, which
ref c cos + 3 sin m influences both the stress distribution in the
E50 = E50 ( ) (1)
c cos + p ref sin foundation slab and the calculated settlements,
the core walls of the basement floors are also
where E50ref is the reference secant stiffness modelled.
modulus in a drained triaxial test at a reference
pressure pref. and c are the effective strength
parameters of the soil, '3 is the effective minor
principle stress and m is a parameter, which
controls the rate of stress dependency.

4 COMPUTATIONAL MODEL AND


CALCULATION PROCEDURE
Figure 3. 3D Finite element model of Model B

To reduce the complexity and size of the 3D


models symmetry axes are defined. This is
possible because the high loaded area is almost 5 RESULTS OF NUMERICAL ANALYSIS
symmetric and the influence of the non-
symmetric outer part of the construction is 5.1 Optimisation of the foundation concept
expected to be small. In all 3D calculations 15 In the first three layouts studied the diaphragm
noded wedge elements with quadratic shape wall panels are arranged radially from the high
functions are used. Figure 2 shows the symmetry loaded egg shaped area. All panels beneath the
axis for the 3D models where half the excavation
2.6 m thick slab have a width of 0.8 m and a modelled, which allows a finer discretization but
length of 25.0 m the 0.6 m thick panels in the keeping the model size within acceptable limits.
outer zone have a reduced length of 15.0 m.
Figure 4 shows the deep foundation elements of
layout 2.

Figure 4. Diaphragm wall panels of layout 2


Figure 5. Top view of the final foundation concept

All results discussed in this section are One can see in Figure 5 that also the inner parts
obtained with the Hardening Soil model. The of the diaphragm wall panels are disconnected.
computed maximum settlements of the first three This has the effect that the global stiffness of the
layouts are between 85 and 100 mm and the foundation system is decreased, which
differential settlements between point A, in the automatically leads to higher differential
middle of the construction, and point B which is settlements but this should be compensated with
located in the upper right corner (Figure 2) of the a second circle of barrettes installed beneath the
1.5 m thick foundation slab are between 42 and high loaded area. The vertical settlements
60 mm. The problem with these foundation calculated are about 105 mm. Figure 6 shows the
concepts is that a stiff connection between the contour lines of vertical displacements of the
panels, as it is assumed in the finite element entire 3D model. Between point A and B about
calculation is technically very difficult to 65 mm of differential settlements are expected
accomplish. Another problem is that the radially and approximately 47 mm within the 2.6 m thick
oriented diaphragm wall panels lead to stress slab (point A to point C).
concentrations in the panels. Therefore
alternative geometries with radially and
tangentially arranged panels were investigated.
These arrangements have the advantage that the
high point loads coming from the superstructure
are directly transferred to deep foundation
panels. Detailed information of the different
layouts studied is given in [8].

5.2 Final foundation concept


Due to the above mentioned shortcomings
another foundation system is studied where two
discontinuous circles of panels are arranged in
the area of concentrated loads (Figure 5). For this
variation only a quarter of the structure was Figure 6. Contour of vertical displacements Final concept
6 PARAMETRIC STUDY BASED ON Figure 8 shows the computed settlement troughs
PUBLISHED SOIL DATA along cross section A-A (Figure 5). The results
of the worst and best case scenario are compared
The soil parameters used for the settlement with the ones obtained with the parameters used
predictions discussed above are based on the soil for the settlement predictions. The results for
description in the geotechnical reports and both the HS and the HSS model are presented.
laboratory tests. Due to the fact, that high order
constitutive models, as used for the numerical
analysis, need in general more input parameters
than given in standard geotechnical reports also
profound knowledge of the used soil models and
experience with correlations between certain soil
parameters play a significant role. In this section
a parametric study based on recently published
soil data, not available at the time of analysis
presented earlier, is presented. The parameters
presented by Saidel et al. [9] are typical for the
soil conditions in Bucharest. The published data
is based on drained and undrained triaxial tests,
oedometer tests, direct shear tests, insitu tests,
cross-hole seismic survey, cyclic triaxial tests,
observational method and finally experience. In
[9] a range of most of the input parameters
Figure 7. Soil parameters for sand layers
necessary for the HS and HSS Model are given.
The aim of this study was firstly to see how
the parameters derived from the geotechnical
reports and used for the settlement predictions fit
to the parameters presented by Saidel et al. and
secondly, and mainly, to work out the influence
on the computed settlement troughs. For the
comparisons presented in this section the
executed foundation concept was used where the
barrettes beneath the core have a length of 30 m
instead of 25 m and the diaphragm wall which
acts as retaining wall and as foundation element
has altering lengths of either 15 or 25 m. This Figure 8. Settlement troughs along cross section A-A
modification of the foundation system yields to a
reduction of both maximum and differential The difference between the HS and the HSS
settlements of about 6 mm, compared to the model shows clearly the influence of the small
layout presented above. strain stiffness, where the settlements from
Figure 7 shows the soil parameters for the deeper depths are automatically reduced. One
sand layers. The input parameters for the HS and can also see that the predicted settlement trough
HSS model are presented along the horizontal computed with the HS model, where the input
axes. The bars represent the range of the soil parameters were derived from the geotechnical
parameters, where the values at the horizontal report is between the worst and the best-case
axes are the used values for the settlement scenario. The "best case" calculation with the
predictions and the light and dark grey bars HSS yields only 25 mm maximum
indicate the deviation of the best and the worst- displacements. This comes from the fact, that the
case scenario according to [9] respectively.
additional parameters for the HSS model, namely sand fill and its very limited friction against the
the G0 and the 0.7, are very optimistic. natural soil for the panel above the barrette. For
the lower part of the barrette the agreement can
also be considered as reasonable. But the
7 VALIDATION OF THE ANALYSIS WITH measured kink of the load displacement curve at
AN IN SITU LOAD TEST a load level of about 5 MN was not captured with
the FE analysis. Also numerical studies with
To obtain additional information of the modified soil conditions did not show this
settlement behaviour of the deep foundation an profile. It is also found that differences between
in situ test was conducted on a diaphragm wall drained and undrained conditions are not
element (barrette) using the "Osterberg Method". significant although the undrained analysis
The barrette is installed from the surface. The top results in somewhat smaller settlements.
is located 15.5 m beneath the surface and the
panel has a total length of 25.0 m. The
"Osterberg Cell" is installed 9.0 m above the 8 CONCLUSION
bottom of the barrette. A numerical model was
set up in close agreement to the real test. The soil Results from numerical analyses with the
profile used and the parameters are the same as objective to assess the settlement behaviour of
mentioned above. The barrette is discretized by the Sky Tower have been presented. A number
means of volume elements and the soil-structure of 3D analyses have been performed
interaction is defined with a strength reduction investigating different arrangements of barrettes
factor Rinter of 0.8, which determines the interface forming the deep foundation. The final concept
strength with respect to the strength of the of the deep foundation consists of two
surrounding soil. discontinuous circles of diaphragm wall panels
and saves 10% volume of deep foundation
elements compared to the preliminary designs.
Additionally a numerical back calculation of an
O-Cell test has been presented. The agreement
between back analysis and the in situ test can be
considered as reasonable. Finally a parametric
study based on soil data published in the
literature shows that the used soil parameters
derived from the geotechnical report yield to
good, albeit somewhat conservative settlement
prediction of the Sky Tower.

Figure 9. Displacement of upper and lower part of the


barrette during the loading test REFERENCES

Figure 9 shows the displacements of the load [1] Fellenius, B.H., The O-Cell-A brief introduction to an
innovative engineering tool, Vg- och Vattenbyggaren,
test for the upper and lower part of the barrette. It Stockholm, Vol. 47, No. 4, 11-14, 2000.
follows that displacements from the finite [2] Brinkgreve, R.B.J. & Swolfs, W.M., Plaxis 3D
element analysis are somewhat over predicting Foundation. Finite element code for soil and rock
the test results. Also when using the HSS model analyses, Users manual, The Netherlands, 2007.
[3] Zehentner, H., Tschuchnigg, F. & Schweiger,H.F., Die
the global stiffness response of the O-Cell test is Baugrube des Sky Tower in Bukarest, Proc. of 26th
underestimated. The sudden increase of the Christian Veder Kolloquium, Graz, 147-162, 2011.
upper part of the barrette can be captured [4] Schanz, T., Vermeer, P.A. & Bonnier, P.G., The
approximately with the finite element analysis Hardening-Soil Model: Formulation and Verification,
In R.B.J. Brinkgreve, Beyond 2000 in Computational
but little effort has been put in modelling the Geotechnics, pp. 281-290. Rotterdam: Balkema, 1999.
[5] Benz, T., Small-strain stiffness of soils and its
numerical consequences, Dissertation, Mitteilung 55
des Instituts fr Geotechnik. Universitt Stuttgart, 2007.
[6] Tschuchnigg, F. & Schweiger, H.F. Study of a complex
deep foundation system using 3D finite element
analysis. In T. Benz & S. Nordal (eds.), Conf. of num.
methods in geot. eng.: 679-684, Rotterdam: Balkema,
2010.
[7] Ohde, J., Zur Theorie der Druckverteilung im
Baugrund, Der Bauingenieur, Vol. 20, pp. 451-459,
1939.
[8] Tschuchnigg, F.; Schweiger, H. F; Frhlich, K., 3D
Finite Element analysis of a deep foundation with
diaphragm wall panels, Geotechnical Challenges in
Megacities, Moscow, 471-478, 2010.
[9] Saidel, T, Cpraru, C., Marcu, A., Influence if
constitutive laws and geotechnical parameters on deep
excavations design, and evaluation of their influence on
neighbouring buildings: examples from recent projects
in Bucharest, From Research to Design in European
Practice, Bratislava, 2010.

Potrebbero piacerti anche