3D finite element analyses of deep soil improvement

Florian Scharinger, Roland Lüftenegger, GDP ZT-OG civil engineers, Graz/Klagenfurt/Oberalm, Austria

Due to the fact that deep soil improvement by rigid inclusions is common practice when soft underground
conditions are explored, a study including analytical and numerical analyses was performed. For the full study,
a large field with different commercial constructions was investigated, whereas a local part with which was
classified as very sensitive relating to settlements and/or differential settlements was analyzed in detail.

» Furthermore, the constructions in this

area cause high loads on the foundation
slabs. Since the expected subsoil conditions in
0 m to 6 m, 8 m to 18 m below surface and below
21 m: sandy, silty, clay
consistency: soft to very soft
6 m to 8 m and 18 m to 21 m below surface:
gravel
consistency: loose
the concerned area have to be classified as soft
to very soft in most of the soil layers, a deep soil Eoedref= E50ref=4,000 kN/m², Eurref=12,000 kN/m², Eoedref= E50 ref=32,000 kN/m², Eurref=96,000 kN/m ²,
improvement by rigid inclusions was planned p =100 kN/m , m=0,85
ref ² pref=100 kN/m², m=0,5
and investigated by different analyses. In order
OCR=1.0, POP = 0 OCR=1.0, POP = 0
to estimate the prospective settlements of
the different components of the construction,
settlement analyses with analytical methods were
performed in advance and more complex finite-
element-analyses (3DFoundation) were carried out
in order to obtain more detailed information.

The performed analyses were initiated by similar

conditions for a real project, but it is clearly stated,
that the presented analyses do not represent
direct input from the real project and results
cannot be compared to monitoring data. The
aim of the numerical analyses is to show how the
estimation of the final settlements and of the
required consolidation time can be performed.

Underground conditions and modelling

Based on the results from underground
investigations in the vicinity of the investigated
area by cone penetration tests and core drillings
an underground model consisting of five
layers was developed for the 3D finite element
analysis. Generally, the subsoil conditions in the
first 35 metres below the top ground surface
(approximately flat) are described as an alternating
layering of sandy, silty clay and gravel with Figure 1: Bottom view of deep gravel layer below 18 m, visualisation
different thickness: of the varying layer thickness (calculation model without lowest clay layer)

10 Plaxis bulletin l Autumn issue 2010 l [Link]

Figure 2: Partial view on the structural elements

For the second gravel layer in between of 18 and maximum depth of 35 metres. The groundwater Loading situation
21 metres below surface a very local changing table was explored in a depth of approximately Based on a possible configuration of the
of the thickness was adopted in order to model 1 metre below the to ground surface. For the 3D construction the following loading situation was
a local weakness directly below the foundation finite-element-analyses the Hardening Soil Model developed for the numerical analysis. Due to
slab. The shape of the area with reduced thickness was used for modelling the deformation behavior the fact that a large number of single loads with
of the gravel layer was estimated by almost of the soil layers. minor influence on the overall settlement behavior
concentrical rectangles, whereas the thickness of the slab would be present in the adopted
of the layer reduces from 3 metres to 2, 1 and 0.5 The foundation slab, which has an irregular construction, a simplified pattern with respect to
meter(s) (Figure 1). In reality, the shape would shape in plan view and the nearly 100 rigid the major loads was incorporated. Therefore, the
show a more continuous decrease of thickness. inclusions directly below the slab were modelled top surface of the foundation slab was divided into
For the sake of a simple finite-element-mesh this as volumetric elements assuming linear elastic 5 zones for modelling uniform distributed loads in
type of modelling was chosen to allow for more behavior. The rigid inclusions were modelled each of them.
complexity in the area of the rigid inclusions. according to an almost regular layout from the
A smooth shape was modelled in the beginning bottom of the slab to the top surface of the deep The value for each zone represents the mean
(first trials) by using 18 “instant” boreholes, but gravel layer. The used finite-element-model value of the corresponding loads. The centre
the entire mesh with the rigid inclusions (volume consists of 63,750 15-noded elements (prisms and zone contains the main part with a high-rise
elements) resulted in a six-digit number of finite- tetrahedrons) and involves approximately 170,000 construction. In zone II additional point loads were
elements and was therefore not followed. nodes. Figure 2 gives an impression of the finite- considered, whereas lower loads are present in
The chosen model for calculation has dimensions element model incorporating the high number of zones III, IV and V. Figure 3 shows a sketch of the
of 60 by 60 metres in plan view and reaches to a rigid inclusions. modelled load pattern in plan view.

[Link] l Autumn issue 2010 l Plaxis bulletin 11

Plaxis Practice: 3D finite element analyses of deep soil improvement

Figure 3: Sketch of the different loading zones Figure 4: contour plot of settlements, partial geometry

Figure 5: Deformation of deep sand layer, bottom view, partial geometry

12 Plaxis bulletin l Autumn issue 2010 l [Link]

 Plaxis Practice: 3D finite element analyses of deep soil improvement

Figure 6: Estimated band width of calculated settlements with consolidation time

For the finite-element-analysis, the following Modelling of the discontinuity of deep gravel However, failure by punching of the columns in
uniform pressures have been considered: layer: the partly very thin gravel layer below the toes of
Zone I: ca. 138 kN/m² Reduction of thickness by changing the the rigid inclusions is not indicated by the analysis
Zone II: ca. 37 kN/m² corresponding material data sets. (plastic points plot, stress distributions – not
Zone III: ca. 24 kN/m² shown here).
Zone IV: ca. 21 kN/m² Installation of rigid inclusions:
Zone V: ca. 22.5 kN/m² Effects due to the installation process have not Regarding the consolidation time, the simplified
Due to the fact that the concrete slab was been taken into account; the activation of the rigid model (no detailed discretisation of columns)
modelled by using volumetric elements, the inclusions happened by changing the material resulted in approximately 90 days (3 months) to
weight of the concrete slab with a thickness of 0.9 parameters (wished-in-place). reach 95% of the final settlements. Approximately
meters is not included in the given values. The 50% of the estimated settlements were predicted
additional weight due to the installation of the Installation of foundation slab: after approximately 30 days (1 month).
rigid inclusions was modelled by increasing the Activation of the concrete slab with a thickness of
unit weight of the corresponding material data set. 0.9 metres. In order to define the estimated development
of settlements with time and to form a basis for
Calculation procedure and results Loading situation: comparison with monitoring data, the following
As already mentioned, the aim of the The load distribution presented before was diagram was derived from the calculation results.
calculations was an estimation of settlements activated in two calculation phases (50% and Figure 6 shows the results of the drained and
and consolidation time. In order to fulfil both 100%). undrained analysis and gives a band width for the
requirements in one analysis, undrained estimated settlements in the centre of the slab
conditions and corresponding consolidation steps As already mentioned, the main focus in this and points along the boundary, respectively.
were defined. drained analysis was the determination of
the prospective settlements in the final state Conclusions
Due to the size of the model, calculation time (after decisive consolidation phase). Due to the The presented boundary value problem could
and stability were problematic and therefore, a calculation results, the maximum settlement in the be solved by using the finite-element-method
different way of solving the problem was chosen. centre of the foundation slab is approximately 33 incorporating 3DFoundation. Despite of the
This means, that finally an analysis with drained mm, whereas the settlements along the boundary problems with too large systems (too many
conditions was performed with the presented of the slab are in the range of 24 to 30 mm. elements) for undrained analysis, the presented
model for an estimation of the settlements and Differential settlements in the centre part are in boundary value problem could be solved and an
another model using a less accurate discretisation the range of 2 to 3 mm. estimation of final settlements and consolidation
of the rigid inclusions – blocks with mean material time could be derived. With the presented
parameter sets – and undrained subsoil conditions Figure 4 shows a contour plot of the settlements diagram in Figure 6, the basics for comparison
was used for estimating the consolidation time. in the area of the foundation slab and the rigid of calculation results and monitoring data were
inclusions. Looking at the columns closely, it prepared.
For the analysis with the presented model, the can be seen that the compression of the rigid
following calculation steps have been defined: inclusions is very small and that local dents in the
Initial state (K0-procedure): deep sandy layer are visible (Fig. 5).
K0 = 1-sin { , deep sand layer with constant
thickness of 3 metres

[Link] l Autumn issue 2010 l Plaxis bulletin 13