You are on page 1of 8

Underground Singapore 2005

Practical Design Using FEM


T.G. Ng
GeoEng Consultants, Singapore

ABSTRACT: With the advancement in computing technology and availability of sophisticated but
user-friendly finite element programs, the emphasis of geotechnical design has shifted from simple
analysis or empirical approaches towards FEM modelling. Geotechnical design is a process which in-
cludes interpretation of site and ground conditions, understanding of soil behaviour, idealisation of the
physical system into numerical model, analysis and interpretation of numerical output, and engineer-
ing judgement. FEM analysis is only one step in the design process. It is important to recognise the
complexities and subtleties involved in the numerical method. Simple qualitative physical check can
be used to verify if the FEM is producing sensible results. To develop a robust design, sensitivity
analyses shall be carried out. Interpretation of the sensitivity analyses will improve the understanding
and appreciation of the geotechnical problem. Once the numerical model is verified, FEM can be used
to study feasibility of various engineering solutions to achieve more economical design.

1 INTRODUCTION

In land scarce Singapore, deep excavation is often carried to facilitate the construction of basement car
parks, underground expressway, MRT station, utilities and services tunnels etc. Except for single level
basement typically found in residential projects where cantilever or single propped retaining wall is of-
ten used, most of the deeper excavation projects require a multi-propped retaining wall system. The
problem of multi-propped wall is best solved using soil-structural interaction analysis (Gaba et al,
2003). Ravaska (2002) has also demonstrated that the use of numerical analysis can lead to more eco-
nomical design as opposed to the conventional methods. Amongst others, finite element method (FEM)
is the most widely used numerical analysis programme in Singapore for the design of temporary earth
retaining wall system.

This paper explains the concept of geotechnical design and discussed how to use FEM to study a geo-
technical problem and develop a robust solution. The latter is elaborated through the use a case study.

2 NUMERICAL MODELLING

Conventionally, cantilever or single propped earth retaining wall is design based on limit equilibrium
methods of calculation (BS 8002, 1994). For multi-propped retaining wall system, subgrade reaction
and pseudo-finite element methods such as those adopted in WALLAP, FREW and RIDO, are often
used. These analysis software are collectively known as 1-D programmes. Although the 1-D pro-
grammes do take the soil-structural interaction into account, they only predict the deflection and forces
in the wall element and loads in the props. The consolidation effects and ground settlement associated
with the excavation cannot be predicted using 1-D programmes.

Excessive ground movements behind the retaining wall can cause adverse impact to adjacent buildings
and utilities. Hence, one major concern in the design of earth retaining system for deep excavation
within the urban area is the estimation of the lateral as well as vertical ground movements. In order to
address the above concern, more advance numerical analysis such as finite element method (FEM) and
finite difference method (FDM) has to be used even for the simple cantilever or single propped retain-
ing wall system.

With the availability of more user-friendly FEM programmes, the emphasis of geotechnical design has
been shifted from simple analysis or empirical approaches towards numerical modelling. However,
due to the user-friendliness of these programmes, the users sometimes got carried away by the fancy
features and colourful output and forgot about the complexities and subtleties involved in the numeri-
cal methods. The potential of the numerical analysis in solving various geotechnical problems is enor-
mous (Potts, 2003). The potential for disaster is equally great if it is misused by operators who do not
understand the principal of soil mechanics and concept of geotechnical design.

3 GEOTECHNICAL DESIGN

It is important to recognise that geotechnical design involves much more than just a single run of FEM
analysis (Carter et al, 2000; Gaba et al, 2003; Barbour and Krahn, 2004). According to Gaba et al
(2003), there are five major elements necessary for geotechnical design. All the five elements are im-
portant and inter-related as shown in Figure 1.

Geological Processes


Site Investigation


Ground
stratigraphy &
groundwater
conditions


Empiricism,
Well-winnowed
experience
③ ④
Soil Applied
Behaviour Mechanics

Experiment Idealisation
Testing Modelling
Field Measurements Analysis

Figure 1. Elements of geotechnical design (after Gaba et al, 2003)

The five major elements are:


Element ① - understanding of geological setting and historical development of the site. Site investi-
gation, including relevant field and laboratory testing, is necessary to determine the geo-
logical profile of the site and provide soil parameters required for the design.
Element ② - establishment of ground stratigraphy and groundwater conditions. If the available in-
formation is not sufficient to establish the underlying ground profiles of the specific site,
more site investigation shall be carried out.
Element ③ - understanding soil behaviour. Knowledge of soil mechanics and soil behaviour under
loading condition anticipated over the lifetime of the retaining wall system is vital. More
laboratory testing and field measurement is required if necessary.
Element ④ - undertaking calculations and analyses. This includes idealisation of physical system into
mathematical system based on the understanding of soil behaviour. This can be carried
out using numerical method such as FEM. Sensitivity analyses should be carried out to
identify the critical element of the idealised model.
Element ⑤ - applying empiricism based on sound judgement and experience. Sound judgement and
experience are important element for any design. They are required at every stage of the
design process and grow with time and number of designs undertaken.

The process of geotechnical design shall not stop at the five major elements described by Gaba et al
(2003). Carter et al (2000) suggested that geotechnical design shall also involve observation and moni-
toring during and following construction. In the next section, a case study is used to demonstrate how
to use FEM to study a geotechnical problem and develop a robust design based on the framework de-
scribed above.

4 CASE STUDY

4.1 Site and ground sconditions


A

Multi-storey Apartment
on raft foundation

Existing RC Wall

Figure 2. Site location and borehole location plan

Figure 2 shows a part plan of a housing development project. The site is bounded by a multi-storey
apartment and a road. The ground level of the proposed site is relatively flat at RL108.3. The site was
previously occupied by a 10-storey HDB flat which was founded on footings. The neighbouring multi-
storey apartment is founded on raft foundation. An existing reinforced concrete (RC) wall retaining
4.0m – 5.0m high of earth from RL108.3 is located between the site boundary and the multi-storey
apartment (Figure 3). From the information gathered above, it may be concluded that the soils at the
proposed site is relatively good in terms of strength and stiffness as it is competence enough to support
multi-storey buildings on shallow foundation.
Figure 3. Section A-A

Figure 3 shows the cross section of the site. Deep excavation is required to facilitate the construction
of thick pile cap for the tower block. The depth of excavation ranges from 4.7m to 6.5m from the ex-
isting ground level. Due to the short construction period and difficulty in strutting, cantilever retaining
wall system is proposed. Nonetheless, a robust design is required in view of the proximity of the deep
excavation to existing structure and building. This case study focuses on the design of the contiguous
bored pile wall abutting the exiting RC wall.

4.2 Ground stratigraphy and groundwater conditions

Based on the information revealed in the soil investigation boreholes, the proposed site is underlain by
2m thick of Fill which comprises of stiff sandy silt/dense silty sand followed by 9m thick of Residual
Soil (Grade VI) with SPT N values range between 15 and 40 blows/30cm. Underlain the residual soil
is 3m thick Completely Weathered Jurong Formation (Grade V) which composed of hard silt and very
dense silty sand with SPT N values range between 40 and 100 blows/30cm. This is followed by
Highly Weathered Jurong Formation (Grade IV) where the SPT N values are greater than 100
blows/30cm. Jurong Formation (Grade III & II) which composed of weak to moderately strong, mod-
erately to highly weathered siltstone and sandstone is found at a depth of 23.5m below the ground
level. The groundwater table obtained from water standpipes ranges between 2.0m and 3.0m below the
ground level. The idealised ground profile is shown in Figure 3. The soil parameters adopted for the
design are presented in Table 1.

Table 1. Summary of Soil Parameters


Soil Layer SPT Density Effective shear strength Deformation Permeability
N-value Modulus,
[kN/m3] c' [kN/m2] φ' [ ° ] E' [MPa] k [m/s]
Fill - 19 0 30 12 1E-7
Residual Soil
(Grade VI) 0 ≤ N < 20 19 0 30 1.3N 1E-7
Residual Soil
(Grade V) 20 ≤ N < 50 20 5 32 1.3N 1E-7
Completely Weathered
Jurong Formation 50 ≤ N ≤ 100 20 10 32 1.3N 1E-7
Highly Weathered
Jurong Formation N > 100 21 15 35 150 1E-8
Jurong Formation
Grade (III & II) N >> 100 21 25 35 250 1E-8
4.3 Finite element model

Finite element program SAGE-CRISP ver 4.3a was used in the design. Consolidation analyses using
effective stress parameters shown in Table 1 were carried out. In SAGE-CRISP programme, the time
dependent behaviour of the soil-structural interaction is solved using the fully coupled consolidation
equations based on Biot’s formulation. Seepage flow is considered together with the equilibrium equa-
tion, the strain-displacement equations and constitutive equations. In these analyses, all soils were
modelled using Mohr-Coulomb material. The initial groundwater level was assumed to be at the
ground surface. At every stage of excavation, the groundwater level within the excavation is re-defined
to the surface of the formation by setting the total pore pressure fixity equal to zero. The FE mesh of
the in-situ site condition and deformed mesh after the final excavation are shown in Figure 4 and 5, re-
spectively.
100kPa

Fill
Residual Soil (N=35)

Residual Soil (N=17)


Residual Soil (N=23)
Residual Soil (N=44)

H.W. Siltstone (N>100)

Siltstone

Figure 4. FE mesh modeling the in-situ site condition

Figure 5. Deformed mesh after final excavation

4.4 Idealisation of physical system into numerical system

As shown in Figure 4, the existing RC wall was idealised as an L-shape wall using quadrilateral con-
crete elements. The heel of the wall was ignored in the numerical model to achieve conservative pre-
diction for sliding. The multi-storey building was idealised as a concrete box with a uniform 1.2m
thick raft foundation. The internal walls and rainwater tanks within the basement structure were not
modelled. The structural elements of the building above the 1st storey slab, which has no significance
to the deep excavation, were also omitted from the numerical model. The vertical load of the building
was simulated by applying a 100kPa surcharge on the raft foundation. The model was then subjected
to a consolidation analysis for a period of 10 years to establish the in-situ stress condition before the
commencement of the proposed excavation works.

The proposed temporary retaining wall is 900mm diameter contiguous bored piles spaced at 1.0m c/c.
In a 2-D plane strain analysis, circular bored piles are idealised as rectangular diaphragm wall with an
equivalent thickness t. It can be proven mathematically that the flexural stiffness (EI) and axial stiff-
ness (EA) between the physical and numerical system can be maintained if
t = 0.866D (1)
where D=diameter of the bored pile, and equivalent Young’s modulus, Em calculated using the follow-
ing equation
πD 2
Em = E p (2)
4 st
where Ep = Young’s modulus of bored pile; s = spacing of between bored piles. Equations 1 and 2
could also be used to simulate the foundation bored piles which are 1.5m in diameter spaced at 2.8m
c/c. It should be noted that the installation process of the bored piles and contiguous bored piles could
not be modelled using the FEM programme. The piles are wish-in-place by replacing the soil elements
with the respective pile elements at the predetermined calculation stage. Once the basic components of
the FEM numerical model are setup, it is prudent to test run the model to identify any mistake in the
input or meshing. The complexity of the FEM model should be built progressively as the understand-
ing of the numerical system and its representation of the physical system improves.

4.5 Sensitivity analyses

Sensitivity analyses are a series of simulations in which one aspect of the numerical model is varied at
a time and the effect of this variation to the system are compared and reviewed. The sensitivity analy-
ses carried out during the design stage for the case study include varying the side boundary of the FE
mesh, soil parameters, CBP penetration depth, groundwater level and types of analysis. The results of
the latter are presented here. The cases analysed are summarised in Table 2.

Table 2. Cases of sensitivity analyses


Analysis Descriptions
Case 1 Consolidation analysis using soil parameters shown in Table 1 with a total excavation
and construction period of 3 months
Case 2 Consolidation analysis but the duration for every excavation and construction stage is
set to one second (Pseudo-undrained)
Case 3 Consolidation analysis same as Case 1 but with soil permeability 10 times higher than
that shown in Table 1
Case 4 Consolidation analysis same as Case 1 but with soil permeability 100 times higher than
that shown in Table 1 (Pseudo-drained)

Figure 6 shows the predicted wall deflections and forces in the CBP wall at various stages of excava-
tion for Case 1 analysis. The corresponding settlement predicted at the raft foundation is shown in
Figure 7. The deflection of the wall is largest at the top and reduces with depth as the fixity in the
ground increases. The maximum bending moment is occurring at the location where the curvature of
the wall deflection is the largest. The response resembles typical behaviour of a cantilever wall. The
qualitative physical check above is an important way to verify if the FEM is producing sensible results.
Based on the results shown in Figure 7, the maximum total and differential settlement predicted at the
raft foundation are 15mm and 1:1100 respectively. Comparison of the maximum wall deflection and
bending moment from the sensitivity analyses are presented in Figure 8.
110 110 110

Excavate to RL106.8

Excavate to RL104.8
108 108 108
Excavate to RL102.4

Over excavation
106 106 106
Excv ext pilecap

104 104 104


Elevation (mRL)

Elevation (mRL)

Elevation (mRL)
102 102 102

100 100 100

Excavate to RL106.8 Excavate to RL106.8


98 98 98
Excavate to RL104.8 Excavate to RL104.8

Excavate to RL102.4 Excavate to RL102.4

96 Over excavation
Over excavation 96 96
Excv ext pilecap
Excv ext pilecap

94 94 94
-40 -30 -20 -10 0 -400 -300 -200 -100 0 100 -100 0 100 200 300
CBP Wall Deflection (mm) CBP Wall Bending Moment (kNm/m) CBP Wall Shear Force (kN/m)

Figure 6. Predicted response of CBP wall … Case 1

Distance from excavation (m)


0 2 4 6 8 10 12 14 16 18 20
0

-4
Settlement (mm)

-8
Excavate to RL106.8

-12 Excavate to RL104.8


Excavate to RL102.4
Over excavation
-16
Excv ext pilecap

-20

Figure 7. Predicted settlement of raft foundation … .Case 1


110
110

108
108

106 106

104 104
Elevation (mRL)

Elevation (mRL)

102 102

100 100

Case 1
98 98 Case 1
Case 2
Case 2
Case 3
96 96 Case 3
Case 4
Case 4
94 94
-60 -50 -40 -30 -20 -10 0 10 -800 -600 -400 -200 0 200 400
CBP Wall Def lection (mm) CBP Wall Bending Moment (kNm/m)

Figure 8. Comparison of sensitivity analyses


As shown in Figure 8, the response of the consolidation analysis falls between the undrained and
drained analyses as expected. The negative excess pore pressure generated behind the CBP wall under
the undrained loading condition is the possible cause for the much smaller wall deflection. It is also
noteworthy that the CBP wall responses in Case 1 and Case 3 are about the same. This seems to indi-
cate that the soil permeability is not very sensitive to the overall performance of the retaining system in
this case if the uncertainty in the soil permeability is within one order of magnitude. Figure 8 shows
that the maximum wall deflection and bending moment are the largest in Case 4 (drained condition).
Despite the fact, the sensitivity analysis also shown that the bending moment at the top 3-4m is actu-
ally largest in Case 2 (undrained condition). Hence, unless the maximum bending moment from the
drained analyses is used to design full length reinforcement of the CBP wall, care shall be taken not to
assume that the case produces most conservative wall deflection will results in most conservative
bending moment envelop. It is also important to remember that the sensitivity analysis carried out
above is only one step in the modelling process. The final decision on the design of the wall lies on the
judgement and experience of the designer.

5 CONCLUSIONS

A robust geotechnical design starts from geotechnical interpretation of the site and ground conditions.
Coupled with the knowledge of soil behaviour, the physical system is idealised into numerical system.
However, not all physical complexity of the geotechnical problem can be idealised into FE model.
Simplifications and assumptions have to be made based on sound engineering judgement. In view of
the above, sensitivity analyses shall be carried out to verify the numerical model. Interpretation of the
sensitivity analyses will improve the understanding and appreciation of the problem. Once the numeri-
cal model is verified, FEM can be used to study feasibility of various engineering solutions to achieve
more economical and robust design.

The process of geotechnical design should continue throughout the construction period. Observation
and monitoring make during actual construction provide the best verification of the responses pre-
dicted by the FEM analyses. With the aid of proper planned and instrumented monitoring data, back-
analysis using FEM can be used to reproduce or interpret the actual field performance. This help to in-
crease the understanding of soil behaviour and improve the skill of using FEM. The experience gained
will also improve the confidence level of the geotechnical engineer in making decisions and judge-
ments over similar problems in the future.

With the realisation of the enormous potential of numerical modelling and FEM in geotechnical design,
the civil engineering fraternity in general shall also be wary about the potential pitfalls if the tool was
misused by untrained operators. To quote Potts (2003), useful numerical modelling requires skilled
operators who
i. have a detailed understanding of soil mechanics and the underpinning theory for the numerical
algorithms;
ii. understand the limitations of constitutive models; and
iii. are familiar with the software that is being used for the numerical modelling.

REFERENCES

Barbour, S.L. and Krahn, J. 2004. Numerical Modelling – Prediction or Process?. Geotechnical News. December
2004. pp 44-52.
BS 8002:1994. Code of practice for earth retaining structures. London. British Standards Institution.
Carter, J.P.; Desai, C.S.; Potts, D.M., Schweiger, H.F.; and Sloan S.W. 2000. Computing and computer model-
ling in geotechnical engineering. Conference proceedings GeoEng2000,Melbourne, Australia.
Gaba, A.R; Simpson, B; Powrie, W.; Bead,am, D.R. 2003. Embedded retaining walls – guidance for economic
design. Publication C580. London. CIRIA.
Potts, D.M. 2003. Numerical analysis: a virtual dream or practical reality? Geotechnique. Vol 53, No. 6, pp 535-
573.
Ravaska, O. 2002. A sheet pile wall design according to Eurocode 7 and Plaxis. Numerical methods in geotech-
nical engineering. (ed. P. Mestat), pp 649-654. Paris: Presses de I’ENPC/LCPC.

You might also like