2064
Proceedings of the 18
th
International Conference on Soil Mechanics and Geotechnical Engineering, Paris 2013
elastic, yield and failure surfaces. Thus, A total of seven
parameters are required for the USC model, i.e.,
s
uc
,(undrained
shear strength from
K
0
-concolidated undrained compression
test),
E
i
,(Young’s modulus at small strain)
i
,(threshold of
small strain)
R
f
,(failure ratio)
K
s
,(ratio of the undrained shear
strength from undrained shear strength from
K
0
-concolidated
undrained compression test to that from undrained shear
strength from
K
0
-concolidated undrained extension test) as well
as
m
and
n
(degradation parameters).
The TNEC case history was used for calibration (Ou et al.
1998). Figure 2 shows t
he comparison of wall deflections
and ground movements obtained from field observation
and those from finite element analysis using different
models. Except for the
=0 MC model where
E
u
/
s
u
was
assumed to be 400 according to the local experiences, other soil
parameters such as undrained shear strength,
E
50
ref
,
E
ur
ref
,
E
oed
ref
,
G
0
ref
were determined from laboratory tests.
0.7
were calibrated
to be 5
×
10
-5
. Details of the soil parameter evaluation can be
found in Lim et al. (2010). Though wall deflections can be
predicted well for all models, only the USC model can yield
ground settlements close to field observations (Figure 2).
Moreover, a hypothetical excavation with an 80 m thick soft
clay where its properties were assumed to be the same as the
third soil layer of the TNEC case was used for further
calibration. The USC model gives a more reasonable prediction
in wall deflection and ground settlements than other three soil
models (Figure 3). The USC model is thus adopted for
parametric studies.
Figure 1. Concept of the USC model (a) Stress-strain behavior (b)
Relationship of failure, yield and elastic surfaces.
Figure 2. Comparison of settlements from field observation with those
from analyses for TNEC
Figure 3. Comparison of settlements from various soil models for a
hypothetical excavation with 80 m thick soft clay
3 PARAMETRIC STUDIES AND MECHANISM OF
SETTLEMENT
A wide range of assumed excavation cases including excavation
depth of 9 to 20 m, excavation width of 20 to 60 m, normalized
undrained shear strength (CK
0
UC) of 0.28 to 0.34, depth to hard
rock of 25 to 50 m was analyzed using the USC model. A
typical parametric result, variation of movements with the
excavation width, is shown in Figure 4, indicating that the
PIZ
changes with the excavation width. The excavation depth,
excavation depth, thickness of soil clay and depth to hard rock
are all affecting the
PIZ
. Based on the parametric results, we
have found the following relationship
Figure 4. Variation of ground settlement with excavation width (B) for
s
uc
/σ'
v
=0.3.
When the rock-like soil is very deep, i.e., H
g
is very large
If
2
2
e
H B
2 ,
e
H
e
H PIZ
2
If
2
2
e
H B
2 ,
e
H
2
2
e
H B PIZ
When the rock-like soil is of the limited depth, i.e.,
H
g
is
relatively small
If
2
2
e
H B
2 ,
e
H
2
e
H
g
H
,
; 2
,
e
H PIZ
2
e
H
g
H
g
H PIZ
If
2
2
e
HB
2 ,
e
H
2
2
e
H B
g
H
,
2
2
e
H B PIZ
;
2
2
e
H B
g
H
,
.
g
H PIZ
The above results are summarized below:
When
2
2
e
HB
> 2 (wide),
e
H
) ,
min(
2
2
g
e
HH B
PIZ
(3)
When
2
2
e
H B
2 (narrow),
(4)
e
H
) , 2min(
g
e
HH
PIZ