Sensitivity study of the hardening soil model parameters

based on idealized excavation

B. Gebreselassie, H.-G. Kempfert lnstitute of Geotechnics and Geohydraulics, UniversihJ of Kassel, Germany

Keywords: excavation, constitutive soil model, drained, undrained, finite element method

ABSTRACT: The paper presents a study of the sensitivity of the hardening soil model (HSM) parameters to a change of values in an idealised excavation in normally consolidated soft clay soil. The HSM is a constitutive elasto-plastic-cap model which is presently impiemented in the PLAXIS finite element program. By varying one parameier and keeping the other parameters constant, the influence of each parameter on the performance of the excavation can be studied. ln this way, ii can be proven whether the model parameters did pedorm exactly as they may theoretically be expected to pedorrn

1 lntroduciion
The influence of the hardening soil model parameters under drained conditions on the stress-strain and volume change behaviour has been discussed in the forgoing arlicle in this proceeding (Gebreselassie & Kempferl, 2005) for a triaxial and one dimensionai compression loading condition. The influence of these parameters on the per-formance of an idealised excavation are investigated in this paper. By varying one parameter and keeping the other parameters constant, the influence of each parameter on the pedormance of the excavation can be studied. This study restrict itself to normaliy consolidated soft clays only, however, the result of the study may also apply to other type of soils. The outcome of the sensitivity study may help the user to have a clear picture of the influence of each parameter of a hardening soil model on the pedormance of an excavation. lt helps to judge the confidence interval of the variation of the soil parameters

The idealised excavation problem

ln order to perform the sensitivity study of the model parameiers, an ideaiised excavation shown in Figure 1a has been chosen. The ground is assumed to be a deposit of a homogeneous lacustrine soft soil with the ground water table located at 1.5 m below the ground surface. The excavation 6.0 m deep is supporled by a sheet pile wall of the type Hoech 134 which has a total length of 12.0 m and an embedment depth of 6 m. The wall is suppofied by two level of struts of the type IPB 360 St 37. A buiiding load o{ 50 kN/m2 ai a distance of 3 m behind the wall and a traific load of 10 kN/m2 are assumed at the ground sudace.

327

a)L, t l<__',, A__+?

ro =0-q-l *;sr ;:- -- ---l1l
X
E

.l

o*-g.o
GWI _______.1
I I I

-*----=----r tiYli7.li

: "' Excav. I
'1

-4Q

=-

^flO 3 txcav.

I I

Lacustrine soft soil

txcav.
- all dimensions are in m - dimensions are not to scale

-12.0
-=]

F--36

No. of elements = 2009, No. of Nodes = 16499 No. of stress points - 241A8

Figure 1. a) tdealized excavation, b) the finite element model and its mesh

The constitutive soil model

The constitutive model to which the soil parameters are being calibrated in this paper is an elastoplastic-cap soil model known as the hardening soil model (HSM). The HSM is implemented in the finite element code for soils and rocks "PLAXIS" (Brinkgreve,2002).lt is originally developed based on the so called the Duncan-Chang hyperbo{ic model. lt, however, supersedes the hyperboiic model, because it uses the plasticity theory instead of the elasticity theory, it includes ihe dialatancy soil behaviour and it introduces the yield cap. The HSM also considers the stress dependant stiffness of the soil according to the power law. For detail lnformation on the constitutive model and the program PLAXIS refer to Brinkgreve (2002) (see also Gebreselassie & Kempferl, 2005). As mentioned above, the HSM is used to simulate the soil behaviour, whereas a Mohr-Coulomb Model (t\4CM) is used for the intefface elements. A drained type of analysis is chosen, because it is believed that this condition is most unfavourable condition for excavations in soft deposits. The reference soil parameters required for the HSM are given in Table 1 and the soil parameteas for the interJace elements according to the MCM are given in Table 2. The siiffness of the soil is adopted as it is for the intedace element, whereas the shear strength parameter are reduced by a factor of 1/3. The wall and the struts are assumed to behave linear elastic with the following material

propediesat: EA= 3.5g1 x 106 kN/m, El = 5.355 x 104 kN-m2/m, w.= 1.34 kN-m/ffi. v = 0.30 Strut: EA = 3.801 x 106 kN, Lsp"cins = 2.0 m
Table 1. Reference soil parameters for the HSM
7"",

IkNim3]

[']
25.3

lkNim,l
13.2

-)u IkNim2] 3253

-rci tr ^-.
IkNim2]

-ur

-.2t

mRl

K;"
t-l

V,,

lkNlm2l
1

[kN/rn2]
1

It

z3+

9170

00

0.63

0.83

it 0.573

t-l
A.20

)'a

Table 2. Reference soil parameters for ihe iniedace elements according to the MCM
Tsat

a=1
LJ

a [kN/m,]
AA

tr
lkNim2l

tr

tncreamenl

J tei

U.

ret

IkN/m3]

[kNim2]

[kNim2]

ikNim2l

I-l

lq 5

a^a

etAe

The finite element model and the ealculation stages

An imporlant parl of the finite element model is shown in Figure 1b, The model is extended to a depth oI 42m where afixed boundary is imposed. At a distance of 36 m behindthe wall and atthe symmetry axis, a zero horizontal displacement is imposed. The size of the model as a whole is 48 m wide and 42 m high. Triangular elements with '15 nodes are used in generating the mesh. This element provides a fourlh order interpolation for displacements and it involves twelve numerical integration stress points (Gauss points).The model consists of 2009 elements, 16499 nodes and
24108 stress points. For the drained analysis of the idealised excavation problem, the HSM parameters in Table 1 are adopted as a reference parameiers for the soil body. ln order to study the sensitivity of the soil parameters, their values are varied above and below the reference values. The contact between the wall and the soil body is simulated by mean of interJace elements whose material properties are given in Table 2. The following construction stages has been followed in the computation:
Stage 0: Stage 1:
Stage 2: Stage 3: Stage 4: generation of the initial stresses (K6 - method) application of the surcharge and traffic loads installation of the wall first excavation installation of the 1st strut and 2nd excavation installation of the 2nd strut and 3rd excavation

Stage 5;

Analysis of ihe computation results

of the Poisson's ralio vur

5.1 The effect of the variation

As it can be seen from Figure 2, the parameter yur seems to be a pure deformation parameter. ln other words, vur may affect the deformation of the wall and soil movements but not the eafth pressure and bending moment of the wall. A change in v* trom its reference value of 0.2 to a smaller
0r
-1
I
L-

10
0

_f
'E -c

()
--Reference

130i
I

-1C

cL

"f
_l

-6 t-

* o
0.)

x
E

tr F

120 110 100

'cn

:/f
-8r
I

---*-

v,. = o.so

6 -so
E
c)

= c)
CD

-AA

-o! "f
- lLi -

-50
-b( l

-11 -- tl -

,^| 7C -50 -30 -10 10-150

wall deformation [mm]

horizontal

earlh

-50 50 -225 -150 -75 0 75 pressure Bending moment [kN/m1 [kN-m/m]

0 3 6 I

12

i2

distance from ihe symmetry axis towards the wall [m]

18 24 3A 36 42 4l disiance behind the wall [m]

Figure 2. The effect of the variation of y,. on a) deformation of the wall, b) earlh pressure, c) bending moment of the wail, d) heave of the bottom of excavation, and e) settlement at the sudace

323

value of 0.05 and a larger t'alue of 0.3 has resulted in a uniform change of the wall deflection by about -7 and 4% respectively. Simiiarly, a change of the heave by about i5 and -i}"k, and a change of the suijace settlement by about -21 and 15% respectively are calculared.

5.2

The effect of the variation of the coefficient of the eafth pressure at resi ,ri"

The HSM treats Ki' and K6separately. Whereas Ki" is a model parameter which is ciosely reiated to the stiffness parameters Esl, Er,, Eoea a,od v*, Ko is purely used to define the initial state of the stresses. For normally consolidated soft soils, however, these values are more or iess the same. The value of Ki" as a model parameter can not be varied indefinitely. For example, for the given reference parameters, the minimum and maximum possible values

of Ki"

are 0.437 and 0.71

respectively. Here Ko is assumed to vary with Ki". As it can be seen from Figure 3, the parameter

Kj" would

affect the deformation of the wall, the soil movements, the eafth pressure and bending

moment of the wall, although the magnitude of its influence is moderate as compare to the triaxial case of loading (Gebreselassie & Kempferl, 2005). Varying the value of Ki" from the reference value of 0.573 to those extreme values has resulted in a change of the maximum wall deformation of about -21 and 3% respectively. Similarly, a change of the heave by about 2 and -9"/", a change of the sudace settlement by about -5 and 7", a change of the earth pressure by about -21 and B"k, and a change of the bending moment by about -18 and 2% respectively are calculated. From the above percentage difference presentation and the Figure 3, it appears that varying lhe Ki" value towards the lowest limit is more sensitive than varying its value towards the upper iimit, although the difference between the reference vaiue and the extreme values is almost the same.
tt/t I -t- ---:-;--ffffi I , tlt .i i2\i &1 tlD): r'i\ -T \-/. : fr+l &1 t II ^l : , //M', I l' I 1ll' ///',1 'l 64tr l, cL . 4 I i I ; i // '/i /i i /rtt! .f : 'dd b: i T , I I i:, , tI :L I 6 /'(---+:Reference i l :I -5i l+ // v iiI I A l-__ l--e-K=0.4371 irl rt -i -b - b, O
I

-rT

r E

l5U

10 0

ir

7 O
cg

tqo

-al

//

'.4

> (n rcn O 'uu x

-tu

t o c) D

r+'J9 zlfl f LI f t' { b :I r Tt 1| i .& -r 8it tr : : , la i+l' .# ^t 4:Qi "i i & i trld) fl t , lrl l rrltT iri toi ,:.i ,t il9: : i I S{ &O: .,-i it -f1I i&l 1]i ij I I f\ t:l \\ 12 [t] . t j'- ji+
i

I ll i

4>'

;l-----a-- Ko l---a-- =

I-_T 0.71 I E t

*t o i

c)

E Fon

20
1

o
E
(f)

-30

-i

iI

-o c)1

-9 o

10

E o

AJ

-40

i
I

o9C
(d

c0

-60

c)

-70 -50 -30 -10 i0 -10

10-150

-50

50

0 3 6
Bending moment [kN-m/m]

horizontal :al wall wall

deforrnation fmm

1 press U re eafth lkN/ml

9 12 disiance from ihe symmetry axis

-74

121821 30364248
disiance behind the wall [m]

iowards the wall [m]

Figure 3. The effect of the variation

of Ki" on a) deformaiion of the wail, b) eafth pressure,

c)

bending moment of the wall, d) heave of the bottom of excavation, and e) settlement at the surface

5.3 The effect of ihe variation

of the failure factor R;

ln triaxial and oedometer loading condiiions, ii has been proved thai the failure factor Rl plays an imporlant role in enhancing or retarding the failure of the soil body (Gebreselassie & Kempferl, 2005). lts influence on ihe iciealised excavation, however. seems to be minimum, wiih excepiion of the settlement behind ihe wall (Figure a)). Varying the value of ,Qrfi'om ihe reference vaiue of 0.83

tc 0.67 and 0.97 has resulted in a change of the surface settlement by about -8 and
tively. For all the other cases, the difference remains below +5%.
-r
a'

69," respec-

::: -ql)
^i:l i:j

(a)l i',i
t|

E E

l5u
14A
'130 - IU

.o
(g

tr
.C
c)

Reference ii--e:D-_n07
v ilIv,J,

X q) 120 E
_o

tr
E

ar.

c^o

---*-si

Rr

= 0.67

O -JU 110 100

:t
-50 -30 -10 10-150 -50
horizontal wall deformation [mm] IkNim]

o o

5
(6

G)

= 0)

_tA

-an
-60

c)

50

-225 -150

-75 0

-a
75

036912
distance from the

-74

12 18 24 30 36 42 48

eadh pressure

Bending moment
IkN-m/m1

distance behind

symmetry axis
towards the wall [m]

the wall [m]

Figure 4. The effect of the variation of Eron a) deformation of the wall, b) earth pressure, c) bending moment of the wall, d) heave of the bottom of excavation, and e) settlement at the surface

5.4

The effect of the variation of the consirained modulus

Eo"6

As shown in Figure 5, a variation of the constrained modulus Eo,a by about +50% its reference value, has resulted in a change of the maximum wall deflection by about -30 and 5% respectively. Similarly, a change of the heave by about -'17 and -4"/", a change of the sudace settlement by about -24 and 2"/", a change of the maximum eafth pressure above the bottom of excavation by about -17 and 2o/", and a change of the bending moment by about -23 and 3% respectively has been observed (Figures 5). Hence, the following conclusion may be drawn with regard to the
response of the excavation to the change of Eoea. a) ln all cases Eoea is more sensible to a change of value below the reference value than to value greater than the reference. lt can be seen from Figure 6 that a reduction of the reference .;alue of Eoea by 50% has caused a reduction of the wall and soil movements, the active earth pressure and the bending moment by about 17 1o 307o, whereas increasing the reference value by same amount (50%) show no significant influence (2 to 5%), lt seems that the ratio of EsdEo"a is more impoftant than the absolute value of the Eo"6.For the reference case, this ratio becomes 1.1. lf lhe Eo"a is increased or decreased by about 50%, the ratio becomes 0.73 and 2.20 respectively. The ratio in the case of increasing Eo"a is more closer to the reference ratio than the other way round. This might be the reason why the change oI Eoea is more sensible to a value below ihe reference than above the reference value. b) Contrary to expectation, a reduced value of Eo"6 has resulted in a reduction of wall and soil movements. c) Figure 6b shows a reduced active pressure and an increased passive pressure for the case of Eoea smaller than the reference value. This again contradicts with the reduced wall movement that is discussed in (b). A reduced wall movement would have resulted a higher active pressure and lower passive pressure. d) A reduced active pressure on one side and an increased passive pressure on the other side has resulted in a reduced bending moment, which seems logical in respect to the given loading condition but not in a general sense.

325

(d

IJU

10m

-a'-L
^l:

|
,

I
I

--^

o -=
O

c
X

140

Reference
n<vFref Loed v.! ^

-4

1
I

> r2n

-o
_1

-+I

1.s x EJd

T
"L

* o I
E
O

120 110

'E -zv tr :_30

T
I

-8 ,9
10
11

+ I
I

-o a) 100 !
cd

(D -60 -7n -80

to

6 -qo F o -cn

o9o
o

_T

12

-100 -75 -50

0 -150 -50 50 wall eafth pressure deformation [mm] [kN/m]

-25
horizontal

-250 -150
Bending

-50 50 moment [kN-m/m]

0 3 6 9 12 cjistance from the symmetry axis

towards the wall [m]

12 .,8 24 30 36 42 48 distance behind

the wall [m]

Figure 5. The effect of the variation at Ess6 on a) deformation of the wall, b) eanh pressure, c) bending moment of the wall, d) heave of the bottom of excavation, and e) settlement at the surface

5.5 The effect of the variation

of the un/reloading modulus of elasticity Eu,

The reference value of Eur was directly taken from triaxial test result and it is equal to s.s.rjir (Gebreselasie,2003). Lowering the reference value Io z.tllr, which is usually recommended in
praciice with the absence of a test result, and further lowering the reference value to z.flij have resulted in an increase of the displacement of the toe of the wall by about 27 and 637" respectively. Similarly, a change of the heave of the bottom of excavation by about 75 and 15A7", a change of the suface settlement by aboui -15 and 3Ok, a change of the active pressure above the bottom of excavation by about -16 and 22"/" respectively, and an insignificant change of the maximum bending moment (below 2.5 %) are calculated (Figure 6). The earth pressure below the excavation level on both active and passive side also shows no significant change relative to the reference value. Contrary to the expectation, the settlement at ihe surface for the reduced values of Eur is less
10

0 -l

S zeo
-3 -4 E
-
_A

-i0

F-2n r*'
tr 5-30

t
o_

E
E

-oo

c)

-7 -8 -9 -10
-t
I

0)Fn

u) -60 i

-100-80 -60 -40

0 -150 -50 5C -250 -'150 ,50 50 horizontal wall earih pressure Bending moment deformation [mm] [kN-m/m] ikNiml
-20

-c

80

0 3 6 I 12 distance from the symmetry axis

i2

towards the wall lml

18 24 30 36 42 48 distance behind the vuall [m]

Figure 6. The effect of ihe variation of Euron a) deformaiion of the wall, b) earlh pressure, c) bending moment of the wall, d) heave of ihe boitom of excavation, and e) seitlement at ihe sudace

326

than iha'i fi'om the reference rvalue. This is mainly due io the upward displacement of ihe wall. The whole soil body seems to heave upwards due to lower values of Eu,

5.6 The effect of the variation

of the secant modulus of elasticity

Eso

by t50% from its reference value. These variations of Eso have resulted in a change of the maximum wall deflection by about 45 and -24"/"
Figure 7 shows the effect of the variation
Esc

of

respectiveiy. Similar change of ihe heave by about 21 and 11"h, the sudace seirlement by about 71 and -37"/o, the maximum earth pressure above the botiom of excavation by about'.l9 and -15%, and the maximum bending moment by about 27 and 1B% respectively has been observed. At first glance, it seems that an increased wall movement shouid result in higher passive resistance, because the soil is more close to the passive limit state. However, as the numerical study of the mobilisation of the passive resistance (Gebreselassie, 2003; Gebreselassie & Kempfed, 2005) also shows, the passive resistance is lower for lower values of the modulus for a given displacement of the wall and keeping the shear parameter constant. This is exactly what one can observe in Figure 7. Lower value of Esa leads to higher wall movement but a lower passive resistance and vice versa.
'l

E E
( =
'1

50

'10

c 1 40
----r---

0 10

---AE

cd

30 20 10

x q)
o
E
-o
1

tr i-30

c--

-uIi 5 oo "7 o 1:
s[-.

t
I
I

-l-

F
1

-+o

l_

o o
c)

E O.^ F

T
I

-c.
cd

u) -60
-70
-80

c)

T
i

0)
cd

o)

horizontal wall deformation [mm]

-iso -50 50 -2sa -150 -50 50 earth pressure Bending moment [kN/m] [kN-m/m]

-C

036912
distance {rom the symmetry axis towards the wall [m]

-90
1

2182430364248
distance behind the wall [m]

Figure 7, The effect of the variation of Eso on a) deformation of the wall, b) eadh pressure, c) bending moment of the wall, d) heave of the bottom of excavation, and e) settlement at the sudace

Sunrmary

A sensitivity study of the soil parameters for HSM has been conducted based on an idealised excavation in normally consolidated cohesive soils in order to study ihe influence of these parameters on the pedormance of an excavation. The result of the study may be summarised as follows: E56 Seerns to lead the role of influencing the wall displacement, the earth pressure, the bending moment and the settlement behind the wall. Due to the non-linearity, however, increasing its value does not necessarily produce the same effect as the other way round . Its influence on bottom heave is limited only to the heave near the wall and its influence ceases towards the middle of the excavation. E ,' plays a dominant role on the heave of the excavation and the displacement of the wall toe, but it has insignificant inf{uence on the bending moment. vu, has oniy an etfect on the botiom heave and settlement at the sudace. Fr shows a negligible etfect on all the cases. Ki" value may affect the deformations, bending moment and the eafth pressure. although

)/

aa1

the magnitude of its influence is moderate as compare to the triaxial state of stress. Kf" value towards the lowest limit is more sensitive than varying its value towards the upper Iimit, although the diiference between the reference value and the extreme values is almost the same. The sensitivity study of the Eoea shows that the ralio EsdEo"a is more impoftant than the absolute value of Eo"a.

List of Symbols and Abbreviations

Elit = .""unt modulus ai 50% of the

Ef,'o=constrainedmodulus at

failure

stress and at effeciive reference pressure of

p'ef

&
K;"

= ratio of the stress at failure and the

ultimate stress

p'4 Eff' = unlreloading modulus p'{ ^t E = modulus of elasticity 7",, = saturated unit weight of soil E = effective angle of intemal friction = wallfriction c' = effective cohesion

u, v m
v

=coefficientof theearthpressureat rest for normally consolidated soils = Poisson's ratio for un/reloading = Poisson's ratio

= exponent in the power law HSM = Hardening Soil Model MCM = Mohr-Coulomb Model

References
Rotterdam.

Brinkgreve R.B.J. 2002. Hand book of the finite element code for soil and rock analysis "PLAXIS". Balkema Publisher,
Gebreselassie B. 2003. Experimental, analytical and numerical investigations of excavations in normally consolidated soft soils. Dissertation, University of Kassel,.Schriftenreihe Geotechnik, Heft 1 4. Gebreselassie B., Kempfert H.-G. 2005. Mobilisation of the erth resistance of a normally consolidated cohesive soils. The Proceedings of the 16tn lnternational conference on Soil Mechanics and Geotechnical Engineering, Osaka, Japan (in press). Gebreselassie 8., Kempfert H.-G. 2005. Calibration of soil parameters for an elasto-plastic cap model under drained condiiion. The Proceedings of the 1 1s lnternational conference of IACMAG, Turin, ltaty (in this volume).

328

Proceedings of the Eleventh lnternational Conference on Computer Methods and Advances in Geomechanics

TORINO

/ ITALY I 19.24 JUNE 2OO5

Prediction, analysis and design ln geomechanical applications

Edited by:

Giovanni Barla and Marco Barla

Department of Structural and Geotechnical Engineering, Politecnico di Torino, ltaly

VOLUME

PATRON EDITORE BOLOGNA 2OO5