Professional Documents
Culture Documents
and 2 = 0 30
= 9
Dilatancy angle 2
/
= 39
Cohesion c = 0 kN m
2
Young modulus E = 20X000 kN m
2
Poisson ratio # = 0X3
148 P. Oblozinsky et al. / Computers and Geotechnics 28 (2001) 145159
4. LEM in slurry trenching and advantages of FEM over LEM
Conventionally, the safety factor of slopes and vertical cuts is evaluated by LEM.
The LEM approach has also been applied to the slurry trench stability problem by
many researchers and numbers of methods have been suggested and developed.
Nash and Jones [7] proposed a 2D method for analyzing the slurry trench stability
problem, which was extended by Morgenstern and Amir-Tahmasseb [8]. Elson [9]
investigated the slurry trench stability in a model and summarized the main stabi-
lizing factors. Piaskowski and Kowalewski [10] introduced an idealized shape of the
3D sliding wedge in their pioneering work on investigation of slurry trench stability
under 3D conditions. In this analysis, the horizontal pressure developed in the slurry
lling was compared with the lateral earth pressure acting along the trench face
computed from the 3D sliding wedge, the shape of which was idealized with regard
to the arching eect. Washbourne [11] suggested the shape of the 3D sliding wedge
for the deep and shallow slurry trenches and analyzed the stability of the wedge in
the same manner as the Coulomb wedge was analyzed in the 2D retaining wall
problem. The safety factor was dened as the ratio of the horizontal force due to
slurry pressure to the active thrust of the 3D wedge. Tsai and Chang [12] presented
an improved 3D LEM method for slurry trench analysis in cohesionless soil in
which the shell-shaped sliding surface is found using the MohrCoulomb criterion.
The vertical stress acting on the soil element was calculated using Huder's formula
[13], which takes into account the horizontal arching eect introduced by Terzaghi
[14]. The safety factor was determined as the ratio of the stabilizing force of the
slurry to the horizontal driving force of the sliding wedge.
The essence of any LEM is in dividing the area above the critical sliding surface
into a number of columns and determining the vertical stress acting on the bottom
of each column. The vertical stress is de-composed into the normal and tangential
components, which are essential for obtaining the forces required for the evaluation
of either force or moment equilibrium. In the 3D analyses, the vertical stress acting
at the bottom of the column should be calculated with regard to the horizontal
arching in the soil. The method introduced by Tsai and Chang [12], which appears
to be the most sophisticated, oers a procedure for searching for the critical 3D
sliding surface. The horizontal stress component '
y
(parallel to the trench) was held
constant while '
x
(perpendicular to the trench) was changed from the `at rest' state
to the active state assuming that the movement was perpendicular to the trench wall.
However, the method does not consider the inuence of the slurry pressure when
searching for the critical sliding surface or when calculating the pressures acting on
the bottom of the soil column. The pressure in the slurry partly replaces the in-situ
stress in the soil, and this replacement will activate the stress redistribution behind
the trench face. It can be expected that the slurry pressure would aect the vertical
stress acting on the critical sliding surface, which also would have an inuence on
the horizontal arching in soil. Thus, the slurry density will have a great inuence on
the stress redistribution behind the trench face, on the failure pattern and on the
shape of critical sliding surface. This assumption is supported by Figs. 3 and 4,
which show the vertical stress in normalized form '
/
v
a'
/
vo
('
/
v
eective vertical stress,
P. Oblozinsky et al. / Computers and Geotechnics 28 (2001) 145159 149
'
/
vo
eective overburden stress) and the maximum shear strain increment on a section
through the centre of and normal to the trench wall as calculated by the SSR-FEM
at failure, i.e. when the soil strength is fully mobilized. The unit weight of the slurry
was 10.5 kN m
3
in Fig. 3, and 12.0 kN m
3
in Fig. 4. The soil parameters are given
in Table 1, the trench depth was 15 m, the length 6 m and the groundwater level was
Fig. 3. (a) Normalized vertical stress behind the trench wall at failure; (b) maximum shear strain incre-
ment.
Fig. 4. (a) Normalized vertical stress behind the trench wall at failure; (b) maximum shear strain incre-
ment.
150 P. Oblozinsky et al. / Computers and Geotechnics 28 (2001) 145159
at the ground surface. As can be seen from Figs. 3a and 4a, which show the nor-
malized vertical stress, '
/
v
a'
/
vo
, the gradient of reduction of vertical stress is higher
when the slurry density is lower; this reects the smaller supporting eect of the
slurry pressure in Fig. 3a. The critical sliding wedge can be estimated from the pic-
tures of maximum shear strain distribution obtained from the SSR-FEM solution,
as shown in Figs. 3b and 4b. Comparison of Figs. 3 and 4 highlights that dierent
densities of slurry (thus dierent pressures acting on the trench wall) will cause dif-
ferent failure patterns.
The main advantage of the SSR-FEM is its ability to calculate the stress state in
the soil, taking into account both, the arching eect and the slurry pressure in the
trench, while in the LEM it is necessary to adopt complicated assumptions to
account for the 3D eect. The advantages, which the 3D SSR-FEM oers over the
LEM, should be taken into account in the assessment of the slurry trench stability
and the prediction of safety factors in practical design.
5. Finite element analysis on centrifuge test
The slurry trench stability analyses were carried out using the mesh shown in Fig. 2
to investigate the capability of the shear strength reduction technique in slurry
trench problems and to examine the reliability of the calculated safety factor.
Results of analyses were veried by comparison with the available results of the
centrifuge experiments reported by Katagiri et al [15,16]. The trench length in the
FEM analysis varied from 1.5m to o(2D), the depth was 15 m and the groundwater
level was at the ground surface. The model soil was fully saturated and the analysis
was carried out in the terms of eective stress and drain condition was assumed. The
horizontal eective stress due to the slurry pressure acting on the trench face, as
indicated in Fig. 2 (unit weight of slurry
s1
= 10X5 kN m
3
), was replaced by the
equivalent nodal forces. The soil parameters are given in Table 1.
The aim of the centrifuge experiments on slurry trench stability [15,16] in fully
saturated sandy ground was to investigate and observe the failure mechanism during
the process of lowering the slurry level in the trench. The container set up is shown
schematically in Fig. 5. The experiments were conducted at a working acceleration
of 60G, at which the modelled trench lengths were 3, 6, 12 m and o (2D) and the
depth was 15 m in prototype scale. Since the problem was symmetrical along two
axes, thus only one quarter of the trench was modelled in the centrifuge tests. The
slurry, which was kept in a rubber bag with the shape of the trench, was modelled
using salt water of unit weight
s1
= 10X5 kN m
3
. The failure of the trench was
achieved by lowering the slurry level in the trench (the slurry was released from
the trench into the lower tank). The results are summarised in Table 2, in which
H is the height of the slurry level in the trench above the ground surface at
the instant when the ground settlement started to increase signicantly due to
lowering of the slurry level, i.e. the trench was on the edge of stability. The
calculated safety factors for the values of H obtained from the centrifuge tests,
given in Table 2, are around 1.00, in the range 0.961.19. This comparison
P. Oblozinsky et al. / Computers and Geotechnics 28 (2001) 145159 151
suggests that if F
= 1X2 had been chosen as the minimum value for the design,
the trench wall would be stable.
The comparison between the SSR-FEM and centrifuge experiments is shown in
Fig. 6. In this case the H, which was obtained from the centrifuge tests, is com-
pared with the prediction of the SSR-FEM analyses. It was assumed that the value
of the safety factor 1.00 indicates the limit of stability. The graphical comparison in
Fig. 6 shows that the relationship between the trench length and the minimum
height of the slurry lling from the FEM analyses is almost linear while the rela-
tionship obtained from the centrifuge test is not linear. This is an interesting nding
since from the nature of the 3D analyses it could be expected that the relationship
would be non-linear.
As was mentioned before, the angle of dilatancy 2 = 39
(Table 1) appears to be
too high, therefore an additional set of calculations was carried out in which the
dilatancy 2 = 0 30
= 9
(predicted by SSR-FEM)
3 0.18 1.03
6 0.54 1.19
12 0.92 1.18
2D 1.03 0.96
Fig. 5. Model of slurry trench in centrifuge container.
152 P. Oblozinsky et al. / Computers and Geotechnics 28 (2001) 145159
In the FEM analyses of the centrifuge tests described so far, the slurry trench was
modelled as shown in Fig. 2. This simplied idealisation has considered only the soil
behind the trench wall, but the soil beyond the ends of the trench was neglected. Aset of
analyses was carried out in order to investigate the error caused by such simplication.
The two meshes used for the comparison are schematically shown in Fig. 7. Case a is
the simplied idealisation of the trench, in which there is a rigid boundary at the
ends of the trench. Case b includes the soil beyond the ends of the trench. The trench
depth was 15 m in both cases, the length was 6 m and the width 1 m (in Case b). The
soil parameters are given in Table 1. Fig. 8 shows the safety factor obtained from the
SSR-FEM for the height of slurry lling, H, which varies from 0.0 to 0.75 m above
the ground level. At F
= 1X00. As seen from Fig. 10, the guide wall increases the stability of the trench
noticeably up to about H = 40 cm above the ground surface. If the slurry lling,
Table 3
Comparison of designed height, H, of slurry lling,
sl
= 10X5 kN m
3
Length L = 3 m Length L = 6 m Length L = 9 m
H (cm) F
H (cm) F
H (cm) F