Modelling issues for numerical analysis of deep excavations

Helmut F. Schweiger
Computational Geotechnics Group, Institute for Soil Mechanics und Foundation Engineering Graz
University of Technology, Austria

ABSTRACT: The influence of different modelling assumptions on the results of numerical analyses of a
deep excavation problem is discussed. Based on a reference solution a comprehensive parametric study is per-
formed, identifying modelling assumptions which may have a significant influence on the calculated dis-
placement behaviour and the bending moments in the wall. The parameters investigated include wall friction,
domain chosen for the analysis, constitutive models and modelling of the grout body. In a second example the
influence of the design approaches defined in Eurocode7 for ULS-design are investigated in connection with
numerical methods. It can be concluded from this study that care must be taken when setting up a numerical
model because the sum of various assumptions, not considered being of large importance when looked at it
individually, may significantly influence the outcome of the numerical calculation.

water lowering is performed in steps in advance

1 PROBLEM DEFINITION – EXAMPLE 1 to the respective excavation step)
1.1 Geometry, basic assumptions and - anchors are modelled as rods, the grouted body as
computational steps membrane element (geotextile element in Plaxis
terminology) which guarantee a continuous load
The geometry of the problem follows from Figure 1. transfer to the soil
The domain analysed has been chosen as follows:
width = 150 m, depth = 100 m. The mesh consists of
approximately 1800 6-noded elements, which is re-
fined in areas where high stress gradients can be ex-
pected. The mesh was deliberately chosen to be rela-
tively fine in order to minimize the discretisation
error (Figure 2). The finite element code Plaxis is
used for all analyses presented in this paper.
The following assumptions have been postulated:
- plane strain
- influence of diaphragm wall construction is ne-
glected, i.e. initial stresses without wall, then wall
"wished-in-place" (weight of wall, γb = 24
kN/m3)
- diaphragm wall modelling: beam elements (E b =
30e6 kPa, ν b = 0.15, d = 0.8 m)
- interface elements between wall and soil
- horizontal hydraulic cut off at -30.00 m is not
considered as structural support, the same me-
chanical properties as for the surrounding soil are Fig 1. Geometry and excavation stages
assumed
- hydrostatic water pressures corresponding to wa-
ter levels inside and outside excavation (ground-

15
 are introduced in order to increase this effect and to
take into account the high stiffness at low strains,
which will be prevailing in most of the deeper layers
of the domain analysed, at least in a very approxi-
mate way. Rinter in Table 1 determines the reduction
of strength parameters ϕ and c in the interface ele-
ments as compared to the surrounding soil (tanϕinter
= Rinter tanϕ, cinter = Rinter c). The stiffness of the in-
terface is reduced as well. A value of 1 kPa is intro-
duced for the cohesion which improves numerical
stability, this is however not strictly required.

Table 1. Material parameters for HS-Model - reference solu-

tion
Depth of
0 - 20 20 - 40 > 40
layer (m)
ref
E50
45.000 75.000 105.000
Fig 2. Finite element mesh for reference solution (kPa)
Eurref
180.000 300.000 315.000
(kPa)
Eoedref
The following computational steps have been per- 45.000 75.000 105.000
(kPa)
formed: ϕ 35 38 38
stage 0: initial stress state (given by σv = γz, σh = ψ 5 6 6
Koγz, Ko = 0.43) c (kPa) 1,0 1,0 1,0
νur 0,2 0,2 0,2
stage 1: activation of diaphragm wall and ground- pref (kPa) 100 100 100
water lowering to -4.90 m m 0,55 0,55 0,55
stage 2: excavation step 1 (to level -4.80 m) Rf 0,9 0,9 0,9
stage 3: activation of anchor 1 at level -4.30 m and Rinter 0,8 0,8 -
prestressing
stage 4: groundwater lowering to -9.40 m
stage 5: excavation step 2 (to level -9.30 m) 1.3 Material Parameters for Structural Elements
stage 6: activation of anchor 2 at level -8.80 m and
prestressing
stage 7: groundwater lowering to -14.50 m Diaphragm wall
stage 8: excavation step 3 (to level -14.35 m) EA = 2.4e7 kN/m
stage 9: activation of anchor 3 at level -13.85 m EI = 1.28e6 kNm2/m
and prestressing ν = 0.15
stage 10: groundwater lowering to -17.90 m w = 7.5 kN/m/m
stage 11: excavation step 4 (to level –16.80 m)
Anchor row 1
Distance and prestressing loads for anchors follow EA = 2.87e5 kN
from Figure 1.
Anchor rows 2 and 3
EA = 3.22e5 kN
1.2 Material Parameters for Hardening Soil Model
Membrane elements for modelling grout body (an-
The so-called Plaxis Hardening Soil model (Brink-
chor row 1)
greve, 2002) has been used as reference model. As
the example is related to an actual project in Berlin, EA = 4.92e5 kN/m
simplified however for the exercise discussed here,
the basic set of material parameters used to obtain Membrane elements for modelling grout body (an-
the reference solution is based on data available in chor rows 2 and 3)
the literature and also on published experimental EA = 8.38e5 kN/m
data from triaxial and one-dimensional compression
test for Berlin sand. Although the Hardening Soil
model takes into account the stress dependency of
stiffness for primary loading as well as unload-
ing/reloading stress paths, three layers (see Table 1)

16
2 RESULTS FOR REFERENCE SOLUTION the deformed mesh is shown and in Figure 6 the sur-
face settlements are plotted for the first and final ex-
cavation stage. Settlements increase from approxi-
mately 5 mm for the first stage to over 15 mm for
the final stage, which can be considered to be a very
plausible result. Figure 4 depicts the lateral dis-
placement of the wall together with the inclinometer
measurements, again for the first and final excava-
tion step. The measurements for the final stage have
been corrected for lateral movement of the base of
the wall which is not reflected in the inclinometer
measurement but most likely to occur (based on
measurements under similar conditions). Figure 5
shows calculated bending moments.

-1000 -800 -600 -400 -200 0 200 400 600

0

Deformed Mesh
Extreme total displacement 46.55*10-3 m
(displacements scaled up 100.00 times)
2

4
Fig 3. Deformed mesh (detail) – reference solution
6
-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
0 8

2 10

4 12

depth below surface [m]

6
14

8
16

10
18

12
depth below surface [m]

20

14
22

16 final stage
1. excavation stage 24

18
26

20
28

22
30

24
measurement 32
(final stage) -1000 -800 -600 -400 -200 0 200 400 600
measurement corrected 26
reference solution bending moments [kNm/m]
(final stage)
measurement
28 Fig 5. Bending moments – reference solution
(1. excavation stage)
reference solution 30
(1. excavation stage) distance from wall [m]

32 0 10 20 30 40 50 60 70 80 90 100
10
-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
vertical displacement of surface [mm]

5
horizontal displacement [mm]
0
Fig 4. Wall deflection – reference solution -5

-10

-15
In the following the most relevant results obtained
-20
for the reference solution are presented. Unlike oth- final stage
-25
erwise stated the last construction stage is consid- 1. excavation stage

-30
ered. In addition a few results for the first excavation
step (no anchors installed) are shown. In Figure 3 Fig 6. Surface settlements – reference solution

17
 3 INFLUENCE OF VARIOUS MODELLING
-70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
ASSUMPTIONS 0

2
In this section modelling assumptions such as the
dimensions of the domain analysed, modelling of 4

wall friction and grout body of anchors are investi- 6

gated. The influence of the constitutive model is ad-
dressed in section 4. 8

10

3.1 Influence of wall friction 12

depth below surface [m]

For the reference solution the interface elements im- 14
plemented in the code Plaxis have been used in a
standard way, i.e. the factor Rinter was used in order 16

to reduce the strength properties of the interface 18

elements with respect to the surrounding soil. The
elastic stiffness of the interface elements is governed 20

by a so-called "virtual thickness" which is based on 22

the average element size of the mesh adjacent to the reference solution
Rinter = 0.8 (final stage)
wall. For the reference solution Rinter = 0.8 has been Rinter = 0.5 (final stage) 24

Rinter = 0.8 t_virt = 0.01

assumed, a value which is based on experience. In (final stage) 26
order to study the effect of this parameter an analy- reference solution
(1. excavation stage)
sis has been performed changing Rinter to 0.5. It fol- Rinter = 0.5
28

lows from Figures 7 to 9 that this parameter has a (1. excavation stage)
Rinter = 0.8 t_virt = 0.01 30
significant influence on the displacements. The hori- (1. excavation stage)

zontal displacement of the top of the wall increases -70 -65 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
32

by approx. 25 mm and the settlement behind the horizontal displacement [mm]

wall by approx. 15 mm. Bending moments to not
change significantly. In order to evaluate the influ- Fig 8. Wall deflection – influence of wall friction
ence of the elastic properties of the interface ele-
ments a calculation was performed with a reduced -1000 -800 -600 -400 -200 0 200 400 600
virtual thickness as compared to the default value set 0

in Plaxis, thus the stiffness of the interface is in- 2

creased. This results in a reduction of the maximum
horizontal displacement of the wall in the order of 5 4

mm (Figure 8). It is obvious from these results that 6

input parameters for modelling wall / soil interaction
have to be chosen very carefully, which is however a 8

difficult task because the elastic stiffness of an inter- 10

face is not a well defined mechanical property. Al-
though results presented here are related to the par- 12
depth below surface [m]

ticular interface element formulation implemented in 14

Plaxis it can be expected that other formulations will
show a similar sensitivity to input parameters. 16

18
distance from wall [m]
0 10 20 30 40 50 60 70 80 90 100 20
10
vertical displacement of surface [mm]

5 22

0
reference solution 24
-5
Rinter = 0.5
-10 26
-15
28
-20

-25 30
-30
reference solution
-35 Rinter = 0.5 32
-1000 -800 -600 -400 -200 0 200 400 600
-40
bending moments [kNm/m]
Fig 7. Surface settlements – influence of wall friction Fig 9. Bending moments – influence of wall friction

18
 -1000 -800 -600 -400 -200 0 200 400 600
3.2 Influence of domain analysed 0

2
For the reference solution the domain analysed was
chosen as 150 x 100 m for width (W) and depth (D) 4
of the mesh respectively. In order to study the influ-
ence of the discretized domain chosen the following 6

analyses have been performed: W x D = 150 x 70,

8
W x D = 100 x 100, W x D = 100 x 70 and W x D =
200 x 150 m. It follows from Figures 10 to 12 that 10
bending moments are hardly influenced but dis-
placements (horizontal displacements of wall and 12

depth below surface [m]

surface settlements) differ in the order of approx. 6
14
mm, the deeper meshes resulting in smaller surface
settlements. This is a result of the vertical upwards 16
displacements caused by the elastic unloading due to
excavation which increases with deeper meshes. 18
This effect would be even more pronounced with a
Mohr-Coulomb model but would practically vanish 20

when applying a constitutive model taking into ac- reference solution

D=100 W=150 22
count small strain stiffness effects. Similar observa- D=100 W=100
D=70 W=100
tions are usually made when analysing excavation of D=70 W=150 24
tunnels where surface settlements may depend sig- D=150 W=200

nificantly on the depth of the mesh below the tunnel 26

when linear elastic - perfectly plastic constitutive 28

models are used (Schweiger et al., 1999).
30

-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 32

0
-1000 -800 -600 -400 -200 0 200 400 600

2 bending moments [kNm/m]

Fig 11. Bending moments – influenced of domain analysed
4
final stage distance from wall [m]
6
0 10 20 30 40 50 60 70 80 90 100
5
vertical displacement of surface [mm]

8
0
1. excavation stage
10
-5

12
depth below surface [m]

-10

14 -15 reference solution

D=100 W=150
-20 D=100 W=100
16
D=70 W=100
D=70 W=150
-25 D=150 W=200
18

-30
20
Fig 12. Surface settlements – influenced of domain analysed
reference solution 22
D=100 W=150
D=100 W=100
D=70 W=100 24 3.3 Influence of modelling ground anchors
D=70 W=150
D=150 W=200
reference solution
26
When using the code Plaxis the load transfer from
D=100 W=150
D=100 W=100 28 the free length of the ground anchors into the ground
D=70 W=100 can be conveniently modelled with membrane ele-
D=70 W=150
D=150 W=200
30 ments. These elements, which have no bending stiff-
32
ness but axial stiffness only, allow a continuous load
-50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0 transfer from the membrane element to the ground
horizontal displacement [mm] along its entire length and avoid a concentrated point
load at the end of the free anchor length. Of course
Fig 10. Wall deflection – influenced of domain analysed this modelling technique is only applicable for
working load conditions because the limiting pull

19
out force cannot be taken into account correctly with To emphasize the importance of a continuous load
this simple model. transfer along the grout body two analyses without
-140 -120 -100 -80 -60 -40 -20 0
membrane elements have been performed. In the
0 first one the free anchor length has been kept the
2
same as in the reference solution and in the second
one the free anchor length has been increased by
4 half of the length of the grout body in order to com-
6
pensate for not modelling the load transfer in more
detail. Figures 13 to 15 clearly show that care must
8
be taken when choosing the model representing the
10 ground anchor and grout body.
12

depth below surface [m]

distance from wall [m]
14
0 10 20 30 40 50 60 70 80 90 100
10

vertical displacement of surface [mm]

16
0

18 -10

-20
20
-30
22
-40

24 -50 reference solution

no membrane element
-60 for load transfer
26 no membrane element
-70 free anchor length increased
reference solution 28 -80
no membrane element
for load transfer
no membrane element 30 Fig 15. Surface settlements – influence of modelling grout
free anchor length increased
32
body
-140 -120 -100 -80 -60 -40 -20 0

horizontal displacement [mm]

Fig 13. Wall deflection – influence of modelling grout body
-1250 -1000 -750 -500 -250
4 INFLUENCE OF CONSTITUTIVE MODEL
Because linear elastic - perfectly plastic constitutive
0 250 500 750 1000
0
models are still widely used in practice some analy-
ses are performed with a Mohr-Coulomb model and
2
compared to the reference solution. The Young's
4 moduli are back-calculated for each layer corre-
sponding to the initial stiffness introduced in the
6
Hardening Soil model of the reference solution in
8 the middle of each layer (Table 2). MC_3 uses the
loading stiffness (E50ref) and MC_4 the unloading
10
stiffness (Eurref). It follows from the results shown in
12 Figures 16 to 18 that high differences are obtained
depth below surface [m]

for displacements, with the reference solution of the

14
HS-Model being in between the two MC-solutions.
16 A similar picture is obtained for bending moments.
18
The deficiencies of elastic-perfectly plastic models
become apparent when looking at the displacements
20 behind the wall in Figure 18 where all MC-models
reference solution
22
show unrealistic heave of the surface.
no membrane element
for load transfer
no membrane element
free anchor length increased
24
Table 2. Material parameters MC-model 3 and 4
depth of E E
26 ϕ ψ c ν Rinter
layer (MC_3) (MC_4)
28 kP
m kPa kPa ° ° - -
a
30
0 - 20 32 000 128 000 35 5 1.0 0.3 0.8
32 20 - 40 90 000 360 000 38 6 1.0 0.3 0.8
-1250 -1000 -750 -500 -250 0 250 500 750 1000 > 40 196 000 588 000 38 6 1.0 0.3 -
bending moments [kNm/m]

Fig 14. Bending moments – influence of modelling grout body

20
 -60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
distance from wall [m]
0
0 10 20 30 40 50 60 70 80 90 100
30
2

vertical displacement of surface [mm]

25

4 20
15
6 10
5
8 0
-5
10 -10
-15 reference solution
12 MC_3 (E-loading)

depth below surface [m]

-20 MC_4 (E-unloading)
-25
14
-30

16
Fig 18. Surface settlements - MC-models 3 and 4

18
4.1 Parameter variation with MC-model 1
20
In this section a limited parametric study, similar to
22 the one presented in section 3, is performed with the
parameter set of MC_1 (Table 3) as basic analysis. It
24
is interesting to see that with certain, however not
26 very realistic, assumptions the Mohr-Coulomb
model calculates a similar lateral deflection of the
28 wall as the Hardening Soil model (Figure 19). A
reference solution
MC_3 (E-loading) 30
match of surface settlements however cannot be
MC_4 (E-unloading) achieved (Figure 21). Bending moments are not in-
32 fluenced so much (Figure 20). Again the strong in-
-60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
fluence on the results of the assumptions made for
horizontal displacement [mm]
Fig 16. Wall deflection – MC-models 3 and 4 wall friction is obvious.
-60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
0
-1000 -800 -600 -400 -200 0 200 400 600
0
2

2
4

4
6

reference solution 6
MC_3 (E-loading) 8
MC_4 (E-unloading)
8
10

10
12
depth below surface [m]

12
depth below surface [m]

14

14
16

16
18

18
20

20
22

22
24

24
MC_1 26
Poisson = 0.2
26 Poisson = 0.4
Rinter = 0.5 28

28 reference solution
(Hardening Soil) 30

30
32
-60 -55 -50 -45 -40 -35 -30 -25 -20 -15 -10 -5 0
32
-1000 -800 -600 -400 -200 0 200 400 600 horizontal displacement [mm]
Fig 19. Wall deflection – MC-variations
bending moments [kNm/m]
Fig 17. Bending moments - MC-models 3 and 4

21
 5 PROBLEM DEFINITION – EXAMPLE 2
-1000 -800 -600 -400 -200 0 200 400 600
0

2
The second, simpler example briefly addresses the
influence of the design approach according to Euro-
MC_1
Poisson = 0.2
4 code7 when performing ULS-design with finite ele-
Poisson = 0.4 ments. Eurocode7 allows for three different design
6
Rinter = 0.5
reference solution approaches DA1 to DA3 which differ in the applica-
(Hardening Soil)
8 tion of the partial factors of safety on actions, soil
properties and resistances.
10
EC7 states: "It is to be verified that a limit state of
12 rupture or excessive deformation will not occur with

depth below surface [m]

the sets of partial factors" as given in Tables 4 and 5
14 for all three approaches. It is noted that 2 separate
16
analyses are required for design approach 1. The
problem which arises for numerical analyses is also
18 immediately apparent because DA1/1 and DA2 re-
quire permanent unfavourable actions to be factored
20
by a partial factor of safety, e.g. the earth pressure
22 acting on retaining structures. This is however not
readily taken into account in numerical analyses be-
24
cause the earth pressure is not an input but a result
26 of the analysis.
28
Table 4. Partial factors for actions according to EC7
30 Actions γF
Design
approach Permanent
32 unfavourable 1) Variable 2)
-1000 -800 -600 -400 -200 0 200 400 600
γG γQ
bending moments [kNm/m]
DA1/1 1.35 1.50
Fig 20. Bending moments – MC-variations DA1/2 1.00 1.30
DA2 1.35 1.50
Geot.3): 1.00 1.30
DA3
Table 3. Material parameters MC-model 1 Struct.4):1.35 1.50
depth of layer E ϕ ψ c ν Rinter
kP
m kPa ° ° - - Table 5. Partial factors for soil properties and resistances ac-
a
cording to EC7
0 - 20 47 000 35 5 1.0 0.3 0.8
Soil properties γM Resistances
20 - 40 244 000 38 6 1.0 0.3 0.8 Design
> 40 373 000 38 6 1.0 0.3 - approach tanϕ’ c’ cu Unit weight Passive Anchor
γϕ γc γcu γF γR;e γa
DA1/1 1.00 1.00 1.00 1.00 1.00 1.10
DA1/2 1.25 1.25 1.40 1.00 1.00 1.10
distance from wall [m]
DA2 1.00 1.00 1.00 1.00 1.40 1.10
0 10 20 30 40 50 60 70 80 90 100 DA3 1.25 1.25 1.40 1.00 1.00 1.00
30
vertical displacements of surface [mm]

25
20
15 5.1 Geometry, parameters and computational steps
10
5 The geometry of the problem follows from Figure
0
22. The following construction steps have been con-
-5
-10 MC_1
sidered in the analysis:
-15 Poisson = 0.2 - initial phase (K0 = 0.5)
Poisson = 0.4
-20
Rinter = 0.5 - activation of diaphragm wall (wished-in-place)
-25
-30
reference solution
(Hardening Soil)
- activation of surcharge loads
- excavation step 1 to level -2.0 m
Fig 21. Surface settlements – MC-variations
- activation of strut at level -1.50 m, excavation step
2 to level -4.0 m,
- groundwater lowering inside excavation to level -
6.0 m
- excavation step 3 to level -6.0 m

22

Fig 22. Geometry and excavation stages for example 2
The surcharge of 10 kPa is a permanent load, the
surcharge of 50 kPa is a variable load. Bedrock was
assumed at a depth of 20 m below ground surface.
The following parameters have been used.
Soil properties:
E = 30 MPa
ν = 0.3
ϕ = 27.5°
c = 10 kPa
γsaturated = 20 kN/m3
γabove water table = 19 kN/m3
Diaphragm wall:
E = 3.0E7 kN/m2
ν = 0.18
γ = 24 kN/m3
d = 0.8m
Strut:
EA = 1.5E6 kN/m
6 RESULTS – EXAMPLE 2
The following results were evaluated applying DA2
and DA3 as described above: required embedment
depth, design strut force, design bending moment. It
is acknowledged that in general the required em-
bedment would not be determined be means of finite
element analyses but by employing a more conven-
tional approach. However, it is done here for high-
lighting differences in design approaches.
In order to determine the embedment depth
analyses were performed with different wall lengths.
Starting with an embedment depth of 5 m the wall
23
was shortened in 0.5 m intervals and for each wall
length a new analysis was performed. Horizontal
displacements of the base of the wall, bending mo-
ments, strut forces and the factor of safety, obtained
by means of a strength reduction technique, were
evaluated. Figure 23 shows the increase of horizon-
tal deformation of the base of the wall when decreas-
ing the length of the wall for DA2, DA3 and for an
analysis with characteristic parameters. The charac-
teristic analysis and DA2 is almost the same because
the only difference is the factor of 1.11 for the vari-
able load in DA2. Numerical convergence could not
be achieved for an embedment depth of 1.5 m for
DA2 and characteristic parameters and a depth of
3.0 m for DA3 respectively.
40.0
no convergence
35.0
Horizontal displacement [mm]
30.0
25.0
20.0
15.0
10.0
characteristic
5.0 DA2
DA3
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Embedment depth [m]
Fig 23. Horizontal wall displacement vs. embedment depth
Figure 24 plots the safety factor obtained from a
strength reduction technique for different embed-
ment depths. Again values for DA2 and characteris-
tic parameters do not vary much and values for DA3
are much smaller because the soil strength is already
factored at the beginning of the analysis. Figure 24
features an additional line, namely the value ob-
tained for DA2 divided by the partial factor which
has been put on the strength parameters in DA3.
They compare well for embedment depths between
3.5 and 4.5 m but for factors around or below 1 they
differ. To some extent this can be attributed to de-
tails of the iteration procedure and convergence set-
tings which become more sensitive for states at or
very near to failure. No attempt has been made to
achieve a closer matching by tightening tolerance
factors because this was not the main goal of this in-
vestigation.
 1.6
minimum embedment depth of approximately 3.5 m
characteristic
is required for a factor of safety > 1.0 (Figure 24).
1.5
DA2 However this result is considered to be quite accept-
DA3
1.4 DA2 / 1.25
able given the various modelling assumptions in-
volved in the two approaches.
Safety factor [-]

1.3
Discussion on the merits of numerical analyses in
1.2
combination with Eurocode7 design approaches can
also be found in Schweiger (2005), Bauduin, De Vos
1.1
& Frank (2003) and Simpson (2000).
1.0

400.0
0.9
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
350.0
characteristic
Embedment depth [m] DA2
300.0
Fig 24. Factor of safety vs. embedment depth DA3

strut forces [kN/m]

DA2*1.35
250.0

200.0

400.0
150.0

350.0
characteristic 100.0
Bending moments [kNm/m]

DA2
300.0
DA3 50.0
DA2*1.35
250.0
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
200.0
Embedment depth [m]
150.0
Fig 26. Strut forces vs. embedment depth
100.0

50.0
7 SUMMARY AND CONCLUSION
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Embedment depth [m] A reference solution for a deep excavation problem

Fig 25. Bending moments vs. embedment depth utilizing the finite element method and an elastic-
plastic constitutive model as been presented. The
problem has been specified by the German Working
Figure 25 plots bending moments versus embedment Group 1.6 of the DGGT and has been used, in addi-
depths. It is interesting to see that the design bending tion to the work presented here, as benchmark prob-
moment of DA2 (which is the result of the DA2 lem (Schweiger 2002). Based on the reference solu-
analysis multiplied by the partial factor of 1.35) co- tion a comprehensive study was performed in order
incides with DA3 for an embedment depth of 4 m to evaluate quantitatively the influence of various
but is different for 3.5 and 4.5 m. The same holds for modelling assumptions on calculated displacements
the strut forces (Figure 26), with DA3 resulting in and bending moments. Some analyses assuming
higher strut forces than DA2 where the resulting elastic-perfectly plastic material behaviour con-
force is factored. When applying DA2 an additional firmed the well known fact that these very simple
check has to be made with respect to the passive re- constitutive models are not well suited for predicting
sistance for which a partial factor of 1.4 (see Table realistic deformations for these types of problems.
5) has to be applied. One way of checking this is to Secondly, the influence on the design approaches
compare the passive earth pressure obtained from as defined in Eurocode7 in ULS-design employing
the finite element analysis (multiplied by the partial the finite element method has been discussed. It is
factor of 1.35) to the maximum theoretical passive shown that different design approaches can be read-
pressure divided by 1.4. This corresponds roughly to ily accommodated for in finite element analyses but
consider a mobilisation of 50% of the passive resis- design strut forces and design bending moments de-
tance which is sometimes assumed in practical de- pend on the design approach used. Based on the re-
sign. Of course different theoretical solutions can be sults shown here and from other investigations it can
used to obtain the maximum passive resistance but be concluded that no general trend can be observed
here Coulomb's solution is employed for simplicity. which design approach (DA1, DA2 or DA3) leads to
By doing so the required embedment depth is around higher design forces. This depends to a large extent
3.0 m for DA2, which is less than for DA3 where a on the problem under consideration and in particular

24
on the relative stiffness between support system
(wall and struts/anchors) and soil.
Finally it is emphasized that numerical methods
provide a valuable tool in assessing serviceability
limit states (SLS) because it is possible to model soil
structure interaction adequately. However, care must
be taken in setting up the model because a number
of assumptions which have to be made by the mod-
eller will not be found in any code of practice or
geotechnical report. As far as Ultimate Limit State
(ULS) design is concerned, numerical methods can
be used but more experience is still required with re-
spect to model soil structure interaction in the ulti-
mate limit state in order to guarantee a consistent
level of safety factor in structural elements and soil.

REFERENCES

Bauduin, C., De Vos, M. & Frank, R. (2003). ULS and SLS

de-sign of embedded walls according to Eurocode 7.
Proc.XIII ECSMGE, Prague (Czech Republic), Vol. 2, 41-
46
Brinkgreve, R.B.J. (2002). Plaxis, Finite element code for soil
and rock analyses, users manual. Rotterdam: Balkema.
Schweiger, H.F., Kofler, M. & Schuller, H. (1999). Some re-
cent developments in the finite element analysis of shallow
tunnels. Felsbau 17, 426-431.
Schweiger, H.F. (2002). Results from numerical benchmark
exercises in geotechnics. Proc. 5th European Conf. Nu-
merical Methods in Geotechnical Engineering (P. Mestat,
ed.), Presses Ponts et chaussees, Paris, 305-314.
Schweiger, H.F. (2005). Application of FEM to ULS design
(Eurocodes) in surface and near surface geotechnical struc-
tures. Proc. 11th Int. Conference of IACMAG, Turin, Italy,
Bologna: Patron Editore. 419-430.
Simpson, B. (2000). Partial factors: where to apply them? Proc.
Int. Workshop on Limit State Design in Geotechnical En-
gineering, Melbourne, 145-154.

25
26