Vous êtes sur la page 1sur 11

Computers and Geotechnics 50 (2013) 55–65

Contents lists available at SciVerse ScienceDirect

Computers and Geotechnics


journal homepage: www.elsevier.com/locate/compgeo

A parametric study of wall deflections in deep excavations


with the installation of cross walls
Chang-Yu Ou a, Pio-Go Hsieh b,⇑, Yi-Lang Lin c
a
Department of Construction Engineering, National Taiwan University of Science and Technology, Taipei 10672, Taiwan
b
Department of Assets and Property Management, Hwa Hsia Institute of Technology, New Taipei City 23568, Taiwan
c
Hung Tung Construction Co., Ltd., Shin Long Group, Taipei 10668, Taiwan

a r t i c l e i n f o a b s t r a c t

Article history: Several case studies have revealed that the installation of cross walls in excavations can effectively reduce
Received 28 May 2012 the amount of wall deflection and ground settlement. However, the behaviour of the diaphragm wall due
Received in revised form 26 December 2012 to the installation of the cross walls is still unclear. This study performed a series of 3D numerical studies
Accepted 29 December 2012
of wall deflections for deep excavations with cross walls and studied the effects on the wall deflection of
Available online 26 January 2013
several parameters, including the number of cross walls, the distance to the cross wall, the cross wall
interval, the cross wall height and the cross wall embedment. The results presented in this study can
Keywords:
be used as a first approximation for cases in which cross walls are designed to reduce the wall deflection
Deep excavation
Cross walls
induced by deep excavation.
Parametric studies Ó 2013 Elsevier Ltd. All rights reserved.
Lateral wall deflection

1. Introduction mechanics is similar to that of a lateral strut, but the cross wall
is put into place prior to excavation. The cross walls installed in
Deep excavation in soft clay commonly causes large lateral wall excavations will be dismantled during the process of excavation
deflections and ground settlements. According to Ou et al. [1], the and will not create an obstacle to wale installation. The cross walls
maximum lateral wall deflection (dhm) for excavations in soft clay are expected to reduce the lateral wall deflection and ground set-
can reach as much as 0.005 times the excavation depth, and the tlement induced by excavation due to their high compressive
maximum ground settlement (dvm) can be as high as 1.0dhm. strength and axial stiffness. Because certain case histories and
Because the density of buildings in cities is relatively high, deep studies have shown that cross walls can effectively reduce the lat-
excavations often take place next to existing buildings or public eral wall deflection and ground settlement, use of cross walls in
facilities. Large ground settlements often damage the adjacent deep excavations is becoming more popular in selected Asian
buildings, which may result in a serious public safety issue. There- countries.
fore, necessary auxiliary measures such as ground improvement Ou et al. [6] found that cross walls created a corner effect sim-
are normally required to protect the adjacent buildings during ilar to that of diaphragm walls [8,9]. The diaphragm wall at the
deep excavation. Strengthening of the bracing system and/or cross wall still showed a small amount of lateral deflection, but this
increasing the passive resistance are two commonly adopted aux- amount was much smaller than in other locations. The lateral wall
iliary measures in engineering practice, and case histories of these deflection increased with increasing distance from the cross wall
measures with varying degrees of success have been reported [2– up to the midpoint between the two cross walls (Fig. 1). In general,
5]. the lateral wall deflections at various locations were smaller than
Cross walls are one of the auxiliary methods used to reduce the those in excavations without cross walls installed. Nevertheless,
ground settlements induced by deep excavation [6,7]. Fig. 1 shows only a few studies on the behaviour of cross walls have been per-
a schematic layout of the cross wall installation. Prior to excava- formed, and current designs and analyses rely mainly on past
tion, a concrete wall is transversely constructed in the excavation experience. No definitive conclusions exist regarding the influence
area using the same construction technique as that of a diaphragm of the cross wall layout on the reduction of lateral wall deflection
wall. Both ends of the cross walls are connected with the two or ground settlement, including the intervals of the cross walls
opposite diaphragm walls. The function of the cross wall in and the cross wall depth. In certain cases, the cross walls were de-
signed at relatively large intervals. In certain occasions, cross walls
⇑ Corresponding author. Tel.: +886 2 89415198; fax: +886 2 29415730. were installed from the ground surface to the bottom of diaphragm
E-mail address: spg@cc.hwh.edu.tw (P.-G. Hsieh). wall while others from the excavation surface down to the middle

0266-352X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compgeo.2012.12.009
56 C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65

Diaphragm wall (a)


A

Cross
wall

A
(a)

Cross wall
(dismantle along
with excavation)

Excavation
surface
Fig. 2a. (a) Profile of the excavation and subsurface soil conditions.
Diaphragm
wall
(b)
Cross wall
(b)
Fig. 1. Schematic diagram of the cross wall: (a) plan (b) section A–A.

of the embedded diaphragm wall, in still other instances, the cross


wall was designed to run only a few metres down from the exca-
vation surface. Because the layout of the cross walls in actual exca-
vations varies from case to case, it is necessary to further examine
the effectiveness of different cross wall layouts in reducing lateral
wall deflection.
In this study, a series of parametric studies using a 3D numeri-
cal method were performed with the plane strain ratio concept to
examine the cross wall influence factors that reduce the lateral
wall deflection, including adjacent cross wall bays, cross wall inter-
vals, cross wall height and cross wall embedment. It is expected
that with a better understanding of the behaviour of cross walls
in reducing the lateral wall deflection, cross walls in excavations
can be designed in a more compact manner and that the protection
of adjacent buildings can be ensured by designers.

2. Numerical analysis

The UPIB building excavation with a seven-level basement was


adopted to validate the numerical model used in this study. The
details of the UPIB excavation, the subsurface soil conditions and
the monitoring results can be found in Ou et al. [10]; only a brief Fig. 2b. (b) Plan and instrumentation of the excavation.
summary is presented in this section.
Fig. 2a shows the profile of the excavation and the subsurface
soil conditions together with their physical properties and strength levels of concrete slabs, 1.5 m in thickness and extending from
parameters. The site was located in a 51-m thick soft soil overlying the ground surface down to a depth of 57.5 m (Ht), was used as
a stiff gravel formation. The ground water in the clay was located at an earth-retaining system. In brief, each floor slab was constructed
GL-3 m (GL refers to the ground surface level) but in the silty directly after an excavation stage, and no de-watering was in-
sands/well-graded gravels (SM/GW) layer at GL-10 m, the excava- volved prior to stage 6. At stage 7, the floor slab was constructed
tion depth was 32.5 m, which was completed in nine stages using at GL-20.2 m, the groundwater level in the SM/GW layer was low-
the top-down construction method. A diaphragm wall with seven ered from GL-10 m to GL-11.67 m and excavation proceeded down
C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65 57

Table 1
Data and parameters of structural members for the UPIB excavation: (a) diaphragm wall, (b) cross wall and buttress wall, (c) floor and steel strut.

Ht (m) t (m) q (t/m3) E (MPa) m


a
(a) Diaphragm wall
57.5 1.4 2.4 14,788 0.2

Depth (m) tc (m) q (t/m3) E (MPa) m


(b) Cross wall and buttress wallb
1.5–22 1.0 2.4 10,438 0.2
22–45 1.0 2.4 13,815 0.2

Stage Type tf (m) A (m2) s (m) E (MPa) m Iy (104 m4) Iz (104 m4) J (104 m4)
c
(c) Floor and steel strut
1 Floor 0.25 – 19,718 0.2 0.25 13.0 208.3
2 Floor 0.2 – 19,718 0.2 0.20 6.7 166.7
3 Floor 0.61 – 19,718 0.2 0.61 189.2 508.3
4 Floor 0.61 – 19,718 0.2 0.61 189.2 508.3
5 Floor 0.61 – 19,718 0.2 0.61 189.2 508.3
6 Floor 0.61 – 19,718 0.2 0.61 189.2 508.3
7 Floor 0.61 – 19,718 0.2 0.61 189.2 508.3
8 Steel strut 0.066 6.4 100,000 0.3 – 19.99 6.72
a
Note: Ht = Total depth of diaphragm wall; t = thickness of diaphragm wall; q = density of concrete; E = Young’s modulus; m = Poisson’s ratio.
b
Note: tc = thickness of cross wall.
c
Note: tf = thickness of floor; A = cross area of steel strut; s = spacing of temporary steel strut; Iy = moment of inertia in y-axial; Iz = moment of inertia in z-axial; J = polar
moment of inertia.

(a) Deflection (cm) Distance from the wall (m)


15 10 5 0 0 10 20 30 40 50 60 70 80

Settlement (cm)
0 0
-2
10
-4
Depth (m)

20 -6
30 -8
67m

-10
40
96 50
m

32
60
m
52 96m
m
32m
32m (b) Deflection (cm) Distance from the wall (m)
15 10 5 0 0 10 20 30 40 50 60 70 80

Settlement (cm)
Fig. 3. Mesh for the UPIB excavation. 0 0
-2
10
-4
Depth (m)

20 -6
to GL-26.05 m. At stage 8, the floor slab was constructed at GL- -8
30
24.8 m, the groundwater level in the SM/GW layer was lowered -10
from GL-11.67 m to GL-16.80 m and the excavation continued 40
down to GL-29.4 m. At stage 9, temporary steel struts with dimen- 50
sions 3-H400  400  13  21 mm, a spacing of 6.4 m and a slope
of 4.6:6 were installed at GL-29.0 m; the groundwater level in the 60
SM/GW layer was lowered from GL-16.80 m to GL-21.56 m and the
Deflection (cm) Distance from the wall (m)
excavation reached GL-32.5 m. (c)
Fig. 2b shows the excavation plan, the allocation of the cross 15 10 5 0 0 10 20 30 40 50 60 70 80
walls and buttress walls and the instrumentation. No buildings ex-
Settlement (cm)

0 0
isted near the excavation site. The cross walls, 1.0 m in thickness -2
10
and 45 m in depth, were constructed directly after the completion -4
Depth (m)

of the main diaphragm walls connected to the cross walls. Because 20 -6


the cross walls were dismantled during the process of excavation, 30 -8
no reinforcement steel bars exist in the cross walls above the final -10
excavation depth but temperature steel bars were arranged in the 40 Field observation
cross walls below the final excavation depth. The cross walls be- 50 Analysis with CW
tween GL+0 m and GL-1.5 m were backfilled with the in situ soil, Analysis w/o CW
60
those between GL-1.5 m and GL-22 m were cast with 13.7-MPa
concrete, and those below GL-22 m were cast with 24.0-MPa Fig. 4. Comparison of wall deflections and surface settlements at SI-8 from field
concrete. observation and from analysis: (a) stage 5, (b) stage 7, and (c) stage 9.
58 C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65

A three-dimensional finite difference computer program,


(a) Deflection (cm) Distance from the wall (m)
FLAC3D [11], was adopted for analysis. A pressure-dependent linear
15 10 5 0 0 10 20 30 40 50 60 70 80
elastic/perfectly plastic constitutive model with the plasticity pla- 0 0

Settlement (cm)
teau described by the Mohr–Coulomb strength criterion was used -2
to simulate the behaviour of soils in which the clayey soil was as- 10
-4

Depth (m)
sumed to be undrained. The undrained shear strength of the clay 20 -6
shown in Fig. 2a was adopted, and the Poisson’s ratio (m) and angle
30 -8
of dilation for the clay under the undrained condition were as-
-10
sumed equal to 0.495 and 0, respectively. The Young’s modulus 40
of Taipei silty clay (Es) was assumed equal to 400su where su is 50
the undrained shear strength [12].
The structural materials were simulated by the linear elastic 60
model and included the concrete diaphragm walls, cross walls,
buttress walls and temporary steel struts. Shell elements were (b) Deflection (cm) Distance from the wall (m)
used for the diaphragm walls, cross walls and buttress walls, and 15 10 5 0 0 10 20 30 40 50 60 70 80
beam elements were used for the temporary steel struts and con- 0 0

Settlement (cm)
crete floor slabs. The preloading of the steel struts was not simu- -2
10
lated in the analysis because this preloading was primarily -4

Depth (m)
assumed to push the steel struts into tight connections with the 20 -6
wales and diaphragm walls. The stiffness of the wales was much 30 -8
lower than that of the diaphragm wall, and therefore the wales -10
were not considered in the analysis. The joints between the main 40
diaphragm walls and the cross walls were simulated as rigid con- 50
nections. Table 1 lists the material properties of the structural
60
members used in the analysis.
For simplification, only the hatched area shown in Fig. 2b was
adopted for analysis. When planning the mesh, the boundaries in (c) Deflection (cm) Distance from the wall (m)
the horizontal directions (x- and y-directions) were placed at a dis- 15 10 5 0 0 10 20 30 40 50 60 70 80
tance of four times the excavation depth behind the wall, and the 0 0

Settlement (cm)
boundary in the depth direction (z-direction) was placed at the 10 -2
bedrock. The boundary of the bottom surface was set as hinge sup- -4
Depth (m)

ports, and the vertical boundaries were set as roller supports. Fig. 3 20 -6
shows the mesh of the UPIB excavation used in the analysis. 30 -8
As shown in Fig. 2b, the SI-8 and SO-1 inclinometers can repre- Field observation -10
40
sent the deformation behaviour of the diaphragm wall at the cross Analysis with CW and BW
wall and at the midpoint of the two cross walls, respectively. Be- 50 Analysis w/o CW and BW
cause the other inclinometers show deformation behaviours simi- Analysis with CW, w/o BW
60
lar to those of SI-8 or SO-1, they will not be discussed in this study
for the sake of brevity. Fig. 5. Comparison of wall deflections and surface settlements at SO-1 from field
Figs. 4 and 5 show the comparison of the observed wall deflec- observation and from analysis: (a) stage 5, (b) stage 7, and (c) stage 9.
tions and the surface settlements at SI-8 and SO-1, respectively, at
the 5th, 7th, and the 9th excavation stages, as well as those com-
without cross walls. The maximum wall deflection and ground set-
puted from the numerical analyses. The analyses follow the exact
tlement at the final excavation stage were reduced by 67% and 81%,
construction sequence and the allocation of cross walls and but-
respectively, via the installation of cross walls. Therefore, installa-
tress walls. In addition, analyses were performed with the assump-
tion of cross walls can substantially reduce the lateral wall deflec-
tion of no cross walls and buttress walls, and these results are also
tions and surface settlements.
shown in the same figures.
The above-mentioned numerical model is able to accurately
Figs. 4 and 5 show that the computed lateral wall deflections for
compute both the wall deflection and the ground settlement for
the excavation without cross walls are quite consistent with those
excavations with cross walls. Additional case verifications can be
of the excavations observed in Taipei, as studied by Ou et al. [1].
found in Hsieh et al. [13]. Although the constitutive model applied
The analysis resulted in a reasonable computed lateral wall deflec-
was not the most updated version, this paper aims to study the
tion for the excavation with cross walls for different excavation
effectiveness of different layouts of cross walls in deep excavations
stages and a less accurate computation for the surface settlement
because the small strain behaviour of the soil far from the dia-
phragm wall was not taken into account. The adopted simulation
of the construction sequence, the constitutive models and the
selection of parameters in this study can be justified with respect
to the lateral wall deflection.
As shown in Fig. 4, the computed wall deflections and surface
settlements at the cross wall, i.e., SI-8, for the different stages were
much smaller than those without cross walls. The maximum wall
deflection and ground settlement at the final excavation stage
were reduced by 77% and 78%, respectively, by the installation of
cross walls. Similarly, Fig. 5 shows that the computed wall deflec-
tions and surface settlements at the midpoint of two cross walls
(i.e., SO-1) for the different stages were much smaller than those Fig. 6. Hypothetical excavation.
C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65 59

(a) B (b) B (c) B

GL-3 GL-3 GL-3


He GL-6
GL-6 GL-6
=10 GL-9
He GL-9 GL-9
=20 GL-12 He=10 GL-12
GL-15 Hp He GL-15
GL-18 =10 =30 GL-18
Unit : meter
GL-21
He=20
GL-24
Unit : meter GL-27
Hp
=20 He=30
Hp Unit : meter
=30

Fig. 7. Schematic diagram for the construction sequence of the hypothetical excavations: (a) He = 20 m, (b) He = 10 m, and (c) He = 30 m.

Table 2
Data and parameters of structural members for the hypothetical excavation: (a) diaphragm wall, (b) cross wall, (c) steel strut.

Hp (m) t (m) q (t/m3) E (MPa) m


a
(a) Diaphragm wall
He 1.0 2.4 14,788 0.2

tc (m) q (t/m3) E (MPa) m


(b) Cross wall
1.0 2.4 14,788 0.2

Type A (m2) s (m) E (MPa) m Iy (104 m4) Iz (104 m4) J (104 m4)
(c) Steel strut
H400  400  13  21 0.022 4 100,000 0.2 6.66 2.24 8.90
a
Note: Hp = embedded diaphragm wall.

rather than to perform an evaluation of the constitutive model. If Diaphragm wall


the computed lateral wall deflections are reasonably accurate,
the evaluation of the effectiveness of the cross walls using this
Cross wall
method can be justified. Thus, the model will be used to assess He
(dismantle along
the wall deflection behaviour for excavations with various cross H ch
with excavation)
wall allocations.
As shown in Fig. 6, a basic hypothetical excavation scenario was Final excavation Ht
constructed for the parametric study. This case site was character- surface
ised by a rectangular shape with an excavation width (B) of 40 m
and a rather large to nearly infinite excavation length (L). The final Cross wall H ce
excavation depth (He) was 20 m, and this depth was completed in Hp
seven stages. The embedded diaphragm wall (Hp) was equal to He.
The H400  400  13  21 temporary steel supports were installed
1.0 m above the excavation surface, and excavation proceeded to
the next stage through stages 1–6. Fig. 7a presents the temporary Fig. 8. Definition of parameters that describe the diaphragm wall and cross wall.
struts, excavation levels and excavation sequence of the basic
hypothetical excavation. The thickness of both the diaphragm wall
and the cross walls was 1.0 m. The interval between the cross walls The subsoil was assumed to be clay. The undrained shear
(L0 ) was 20 m, and the cross walls were constructed using the same strength (su) for the basic hypothetical excavation was assumed
method used for the diaphragm wall. Table 2 lists the structural equal to 0:3r0v where r0v is the effective vertical pressure. The
dimensions and the material properties used in the analysis. groundwater table was located at the ground surface and assumed
As shown in Fig. 8, the cross wall height (Hch) refers to the to be in the hydrostatic condition. The numerical model was sim-
height of the cross wall above the final excavation surface, and ilar to the validation of the previous case, including the constitu-
the cross wall embedment (Hce) refers to the depth of the cross tive model of clay and the structures and simulation of the
wall below the final excavation surface. The cross walls were re- excavation sequence. Table 3 also lists the clay parameters for
moved concurrently with the excavation stages. the numerical analysis.
60 C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65

Table 3
Soil parameters for the hypothetical excavation.

Soil ct (kN/m3) K0 su =r0v Es/su m w (°)


Clay 18.25 0.5 0.3 400 0.49 0

Note: ct = total unit weight of soil; K0 = coefficient of earth pressure at rest;


su =r0v = normalised undrained shear strength; su = undrained shear strength;
r0v = effective vertical pressure; Es = Young’s modulus of soil; m = Poisson’s ratio;
w = dilation angle.

To simplify the analyses, the excavation sites were assumed


symmetric with respect to the centrelines of the width and length
of the sites, and therefore, only 1/4 of the excavation area, marked
with hatching and labelled ‘‘analysis block’’ in Fig. 6, was adopted Fig. 9. Lateral defection of the diaphragm wall (plane view).

for analysis. When planning the meshes, the meshes in the depth
direction (z-direction) were continued down to a depth of 1.3
times the depth of the diaphragm wall, and in the width direction midpoint between two cross walls (i.e., section B–B). For a conser-
(x-direction), the meshes were extended up to four times the exca- vative approach, the lateral wall deflection at the midpoint of two
vation depth behind the wall. When the excavation length was cross walls will be used to evaluate the effectiveness of the cross
equal to 1.0L0 or 5L0 , the principle of mesh generation in the width wall installation. The plane strain ratio at the midpoint of the
direction (x-direction) was basically similar to that in the length two cross walls (PSRC) is defined as
direction. When the excavation length approached infinity, i.e.,
dhm;mid
L/L0 = 1 (Fig. 6), the meshes were extended up to four times the PSRC ¼ ð1Þ
cross wall interval. The boundaries of the meshes were set in a dhm;ps
manner similar to that of the UPIB case study. where dhm,mid is the maximum lateral wall deflection at the mid-
For the parametric studies, the values of He, B, L0 , L, Hch and Hce point of the two cross walls, and dhm,ps is the maximum lateral wall
were varied as shown in Table 4. Of these, L0 = 1 implies that the deflection under the plane strain condition.
excavation was not installed with cross walls and that wall deflec- A smaller PSRC value implies that the cross walls have an ade-
tion would occur under the plane strain condition, in which the quate restraining effect in reducing the lateral wall deflection
maximum lateral wall deflection was represented by dhm,ps. When and vice versa. When the PSRC is equal to 1.0, the cross walls have
L/L0 = 1.0, the analysed wall deflection would occur under the con- no effect in restraining the lateral deflection of the wall at the mid-
dition of no cross walls. Fig. 7b and c show the temporary struts, point between the two cross walls.
excavation levels and excavation sequence for He = 10 m and For the cases of Hch/He = 1.0 and Hce/HP = 1.0, the computed
He = 30 m, respectively, in which H400  400  13  21 temporary plane strain ratio at the midpoint of two cross walls (PSRC) is listed
steel supports were installed 1.0 m above the excavation surface. A in Table 5 and categorised by a volume of excavated soil per unit
total of 330 numerical analyses were performed in which the basic length (B  He) for each analysis ID, and this table will be used
values of He, L0 , B, L, Hch/He, Hce/Hp and their variations were used as for further analysis.
listed in Table 4. Each value of He, L0 , B, and L/L0 in Table 4 is marked
by a number in parentheses near its entry. A combination of these
numbers was used for an analysis ID. For example, ID = 2413 3. Effect of the location of the cross wall bay
represents He = 20 m, L0 = 36 m, B = 20 m, L = 1.
Ou et al. [8] used the concept of the plane strain ratio (PSR) to As studied by many researchers [8,9,14,15], a section close to
evaluate the impact of a diaphragm corner on the lateral wall the diaphragm wall corner would have a smaller lateral wall
deflection. The PSR is defined as the maximum lateral wall deflec- deflection than that under the plane strain condition. Because the
tion at a distance (d) to the diaphragm wall corner divided by the cross wall was constructed using the same method as the dia-
maximum lateral wall deflection under the plane strain condition. phragm wall, a section near the cross wall should have a similar
Because the cross wall had similar effects on the lateral wall deflec-
tion as those of the diaphragm wall corner [6], the concept of the
PSR will be adopted to evaluate the effect of the cross walls on
Table 5
the lateral wall deflection. As shown in Fig. 9, with the installation Computed plane strain ratio at the midpoint of two cross walls for each analysis ID for
of cross walls, the lateral wall deflection was the smallest at the Hch/He = 1.0 and Hce/HP = 1.0.
location where a cross wall was installed (i.e., section A–A, referred
BHe = 200 m2 BHe = 400 m2 BHe = 600–800 m2 BHe = 1200–2400 m2
to as the ‘‘cross wall section’’ in this study), and it was largest at the
ID PSRC ID PSRC ID PSRC ID PSRC
1113 0.085 1123 0.096 1133 0.091 2133 0.107
1213 0.263 1223* 0.206 1233 0.174 2233* 0.147
Table 4
1313 0.434 1323 0.343 1333 0.293 2333 0.228
Basic values of parameters and variations applied in the parametric studies.
1413* 0.634 1423 0.518 1433 0.438 2433 0.349
Parameters Basic value value (No.) Variation value (No.) 2113 0.109 2123* 0.107 3123 0.119
2213* 0.181 2223* 0.159 3223* 0.160
He (m) 20 (2) 10 (1), 30 (3) 2313 0.336 2323* 0.276 3323 0.242
L0 (m) 20 (2) 12 (1), 28 (3), 36 (4), 1 (5) 2413* 0.514 2423* 0.431 3423 0.333
B (m) 40 (2) 20 (1), 80 (3) 3113 0.110 3133 0.127
L/L0 1 (3) 1 (1), 5 (2) 3213 0.164 3233 0.158
Hch/He 1.0 0, 0.2, 0.4, 0.6, 0.8 3313 0.276 3333 0.214
Hce/Hp 1.0 0, 0.2, 0.4, 0.6, 0.8 3413 0.425 3433 0.306
Note: the number in the parenthesis used to identify the value for each variable. A *
Indicate the scenarios were used to compute critical height and embedment of
combination of the numbers of He L0 BL was given to an analysis ID. the cross wall.
C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65 61

0.5 0.5

0.4 0.4

0.3 0.3

PSR
PSR

0.2 0.2

0.1 0.1

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 10. Relationship between the plane strain ratio and the distance in the central Fig. 11. Relationship between the plane strain ratio and the distance in each bay.
cross wall bay.

The lateral wall deflection was minimally affected by the con-


effect as the diaphragm wall corner. The differences in corner ef- straint of the adjacent cross wall bays, regardless of whether it
fects between the cross wall and diaphragm wall are examined was created by the corner diaphragm wall or the cross wall. How-
in this work. ever, the lateral wall deflection in a cross wall bay formed by the
Fig. 10 shows the relationships between the computed plane corner diaphragm wall and cross wall may be heavily affected by
strain ratio (PSR) in different cross wall bays and the distance (d) the corner diaphragm wall. The lateral wall deflection near the cor-
from an evaluated section to the corners, where Hch/He = 1.0, Hce/ ner diaphragm wall should be considerably smaller than that near
HP = 1.0, and L/L0 = 1, 5 (ID = 2221 and 2222). As shown in the fig- the cross wall.
ure, for L/L0 = 5, the tendency for variation of the PSR in bay 2
(the two ends are cross walls) was similar to that in bay 3 (central 4. Effect of the cross wall interval
cross wall bay, the two ends are cross walls). The PSR values at d/
L0 = 0 in bays 2 and 3 differ slightly because the lateral wall deflec- In a fixed-end beam, the deflection of the beam is proportional
tion in bay 2 was slightly affected by the diaphragm wall corner to the span to the power of four. The deflection of the diaphragm
whereas that in bay 3 was minimally affected by the diaphragm wall per unit depth between two cross walls behaves similarly to
wall corner. Obviously, the lateral wall deflections in bays 2 and that of a fixed-end beam but is not exactly the same because the
3 were constrained primarily by the cross wall and affected mini- cross wall will be demolished stage by stage with the progress of
mally by the diaphragm wall corners. However, the PSR in the ut- excavation. Although the deflection behaviour of the diaphragm
most left or right bay, i.e., bay 1, was seriously affected by the wall per unit depth between two cross walls may not be exactly
diaphragm wall corner, and the tendency for variation of the PSR the same as that of a fixed-end beam, i.e., proportional to the inter-
was different from that observed in bays 2 and 3. The PSR at d/ val to the power of four, it should still show certain deformation
L0 = 0 in bay 1 for L/L0 = 5 was affected primarily by the diaphragm characteristics of a fixed-end beam.
wall corner, which was close to the PSR at d/L0 = 0 for L/L0 = 1. More- Fig. 12 plots the relationship between the PSRC and the square of
over, for L/L0 = 5, the PSR at d/L0 = 1 in bay 1 was close to those of the cross wall interval (L0 2). The figure shows that the PSRC value
bays 2 and 3, which were all mainly constrained by cross walls. increased with increasing L0 2 for a specific B  He.
The above observations can be explained by the fact that, in the
case of L/L0 = 1, the corner formed by the two diaphragm walls
functioned as an arch when it was subjected to external earth pres- 1
2
sure during the excavation. Such an arch effect would cause the BHe = 200m
diaphragm wall to move minimally and would result in a notably 0.9 2
BHe = 400m
small PSR value near zero. However, for L/L0 = 3 to 1, the cross 0.8 2
walls were removed with the excavation stages, and the remaining BHe = (600~800)m
2
parts of the cross walls were subject mainly to high compressive 0.7 BHe = (1200~2400)m
forces that induced a certain amount of elastic compression. A sec-
tion near the cross walls displayed a relatively larger PSR than that 0.6
PSRC

near the diaphragm walls. 0.5


Fig. 11 shows the relationship between the PSR and d in the cen-
tral cross wall bay for the cases of L/L0 = 1, L/L0 = 5 and L/L0 = 1 0.4
(ID = 2221, 2222 and 2223, respectively). The results for the PSR
0.3
in bay 2 for L/L0 = 5 in Fig. 10 suggest that the PSR in the central
cross wall bay for L/L0 = 3 displays a behaviour similar to those of 0.2
L/L0 = 5 and L/L0 = 1, although there are no data for L/L0 = 3 dis-
played in Fig. 11. The figure further demonstrates that the PSR in 0.1
the central cross wall bay was nearly the same for L/L0 = 3 to 1,
0
which were parallel to that for L/L0 = 1 but differed by a constant 0 400 800 1200 1600
that approximately equals the PSR value at the cross wall section
in the central cross wall bay.
In conclusion, the lateral wall deflection in a cross wall bay Fig. 12. Relationship between the plane strain ratio at the midpoint of two cross
formed by two cross walls was constrained only by the cross walls. walls and the cross wall interval.
62 C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65

1 0.5
2
0.9 BHe = 200m
2 0.4
BHe = 400m
0.8
2
BHe = (600~800)m
0.7 0.3
2

PSR
BHe = (1200~2400)m
0.6 Upper bound 0.2
PSRC

0.5 Lower bound


0.1
0.4

0.3 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.2

0.1 Fig. 14. Relationship between the plane strain ratio in the central bay and the
distances for different intervals of adjacent bays.
0
0 2 4 6 8 10 12 14 16 18 20

cross wall bay. As shown in the figure, the PSRC in the central cross
Fig. 13. Relationship between the plane strain ratio at the midpoint of two cross wall bay decreases with increasing intervals of the adjacent cross
walls and the bay factor. wall bays. These results were reasonable because the diaphragm
wall per unit depth can be simulated as a continuous beam, and
the cross walls can be treated as supports. During excavation, a
The PSRC value was approximately linearly proportional to L0 2 cross wall bay adjacent to the central cross wall bay but at a larger
for each excavation. When L0 approached 0, the diaphragm wall interval than the central cross wall bay would result in a larger
still showed a certain amount of movement, and the PSRC was wall deflection in this cross wall bay, which in turn causes the
not equal to zero. Moreover, the PSRC generally decreased with smaller wall deflection in the central cross wall bay compared with
increasing excavation volume, BHe, for a specific cross wall interval. those from equal cross wall bay intervals. Such a decrease in the
This effect occurred because both dhm,ps and dhm,mid increased with PSRC in the central cross wall bay due to the unequal adjacent cross
BHe, but the former increased at a faster rate than the latter. wall bay intervals tended to converge when L0a =L0 > 2.
Fig. 12 also shows that when the cross wall interval was rather The adjacent cross wall bay intervals clearly had a small influ-
small, e.g., L0 = 12 m (L0 2 = 144 m2), the PSRC remained nearly the ence on the PSRC in the central cross wall bay. For example, the
same, approximately equal to 0.1, for different values of BHe. This PSRC in the central cross wall bay for L0a =L0 ¼ 1 differed by only
result implies that if L0 is small, the cross wall corner (with suffi- 0.05 compared with that of the equal cross wall interval. In prac-
cient rigidity) has more influence than the excavation volume on tice, such a difference is most likely negligible.
the wall deflection at the midpoint of two cross walls. The results
were consistent with the diaphragm wall corner effect, as studied
6. Effect of the cross wall depth
in Ou et al. [8], implying that when the distance to the diaphragm
wall corner was small, the diaphragm wall corner had more influ-
As shown in Fig. 8, the influence of the total depth of the cross
ence on the PSR than the excavation geometry.
walls on the lateral wall deflection will be studied in terms of the
Because the PSRC was approximately linearly proportional to L0 2
cross wall height, which refers to a measurement taken above the
and decreased with increasing BHe, a bay geometry factor (Fg) is
excavation surface (Hch), and the embedded cross wall, which re-
subsequently defined as BHe divided by L0 2, i.e., F g ¼ BHe =L02 , repre-
fers to a measurement taken below the excavation surface (Hce).
senting the amount of soil excavated in a cross wall bay. Fig. 13
In practice, cost is an important concern when designing cross
shows the relationship between PSRC and Fg. As shown in the fig-
walls. Therefore, Hch/He and Hce/Hp should be designed to be as
ure, when Fg approached zero, indicating that L0 was quite large rel-
small as possible while maintaining a PSRC close to that of walls
ative to the volume of excavated soil (BHe), the PSRC was minimally
with full height and full embedment, i.e., Hch/He = 1 and Hce/
affected by the cross wall corner. The section was near the plane
Hp = 1. Therefore, the critical height ratio, (Hch/He)cr, is defined as
strain condition, and the PSRC was close to 1. When Fg was varied
the height ratio (Hch/He) at which the resulting PSRC is the same
from 0 to 2, the PSRC decreased rapidly with increasing Fg. Further-
as that when Hch/He = 1 for a specific He, L0 , and B. Any further in-
more, when Fg was large, e.g., larger than 2 or indicating that L0 was
crease in Hch/He over the critical value does not cause a decrease
small relative to the volume of excavated soil, the PSRC was greatly
in the PSRC. Similarly, the critical embedment ratio, (Hce/Hp)cr, is de-
affected by the cross wall interval, showing that the cross walls can
fined as the embedment ratio (Hce/Hp) at which the obtained PSRC
effectively reduce the lateral wall deflection. Under such circum-
is the same as that when Hce/Hp = 1. Again, any further increase in
stances, the PSRC decreased slowly with increasing Fg for all bay
Hce/Hp over the critical value does not cause a decrease in the PSRC.
factors.
Therefore, (Hch/He)cr and (Hce/Hp)cr will be used as the main param-
eters in the study of the effect of the cross wall depth on the lateral
5. Effect of the interval of adjacent cross wall bays wall deflection.
By plotting the relationship between the PSRC and Hch/He for a
The interval of the cross walls might vary from bay to bay. To certain set of He, L0 , and B when Hce/Hp = 1.0 and L/L0 = 1, we can
investigate the effect of the interval of adjacent cross wall bays identify a possible range of critical heights, (Hch/He)cr, as shown
on the deflection behaviour, variations of the computed PSR in in Fig. 15. Similarly, for a certain set of He, L0 , and B when Hch/
the central cross wall bay for L0a ¼ L0 ; 2L0 ; 3L0 and 1 (ID = 2223) He = 1.0 and L/L0 = 1, the (Hce/Hp)cr can be obtained by plotting
are shown in Fig. 14, where L0a denotes the interval of the adjacent the relationship between the PSRC and Hce/Hp.
C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65 63

1.0 Table 6
Plane strain ratios for the case of He = 20 m, L0 = 20 m and B = 40 m and L/L0 = 1
(ID = 2223) with various values of Hch/He and Hce/Hp.

Hch/He Hce/Hp PSRC PSR at the cross wall


0.8
1.0 1.0 0.159 0.138
Range of 0 1.0 0.397 0.395
(Hch /He)cr 0 0.3 0.418 0.416
0.6 1.0 0 0.566 0.565
PSRC

0.4 with Hch/He = 1 and Hce/Hp = 1 yields the best reduction in the lat-
eral wall deflections and produces a minimum PSR.
As shown in Fig. 16, the lateral deflections of the wall below the
0.2 excavation surface for the case of Hch/He = 0 and Hce/Hp = 1.0 were
similar to those of the case of Hch/He = 1 and Hce/Hp = 1, but those
above the excavation surface for the former case were larger than
for the latter case. Near the ground surface, the deflections for the
0.0
0.0 0.2 0.4 0.6 0.8 1.0 case of Hch/He = 0 and Hce/Hp = 1.0 were close to those of the case
with no cross walls and much larger than those of the case of
H ch / H e
Hch/He = 1 and Hce/Hp = 1. Moreover, the PSR in the case of Hch/
Fig. 15. Schematic diagram of the relationship between the plane strain ratio at the He = 0 and Hce/Hp = 1.0 was certainly larger than that in the case
midpoint of two cross walls and the ratio of the cross wall height for differential of Hch/He = 1 and Hce/Hp = 1 (Table 6). Although the case of Hch/
ratios of cross wall embedment. He = 0 and Hce/Hp = 1.0 is able to reduce a certain amount of the lat-
eral wall deflection or the PSR, it is necessary to install the cross
walls above the excavation surface to further reduce the deflection.
Fig. 16a and b shows the lateral wall deflections at the midpoint Fig. 16 also shows that the lateral wall deflections above He/2
section of two cross walls and at the cross wall section for a basic for Hch/He = 1.0 and Hce/Hp = 0 were similar to those for Hch/
excavation, i.e., He = 20 m, L0 = 20 m, B = 40 m and L/L0 = 1 He = 1.0 and Hce/Hp = 1.0. However, the lateral wall deflections be-
(ID = 2223), but with four different cross wall depths: Hch/He = 1.0 low He/2 for Hch/He = 1 and Hce/Hp = 0 were much larger than those
and Hce/Hp = 1.0, Hch/He = 0 and Hce/Hp = 1.0, Hch/He = 0 and Hce/ for Hch/He = 1 and Hce/Hp = 1 but still considerably less than those
Hp = 0.3, and Hch/He = 1.0 and Hce/Hp = 0. For comparison, the re- with no cross walls. Table 6 also shows that the PSR for Hch/
sults from the analysis in the absence of cross walls (ID = 2523) He = 1 and Hce/Hp = 0 was larger than that for Hch/He = 1 and Hce/
are also shown on the same figure. Table 6 lists the corresponding Hp = 1. The overall reduction in lateral wall deflection or the PSR
plane strain ratios at the midpoint section of two cross walls and was even less than those for Hch/He = 0 and Hce/Hp = 1.0. Although
those at the cross wall section. The lateral wall deflections or PSRs the case involving Hch/He = 1.0 and Hce/Hp = 0 is able to reduce a
for those scenarios with different depths of cross walls were all certain amount of the wall deflection or the PSR, it is necessary
much smaller than those for the scenario with no cross walls. to install the cross walls below the excavation surface to further re-
The installation of cross walls obviously reduces the lateral wall duce the lateral wall deflections.
deflections or the PSRs, both at the midpoint of the two cross walls Selected construction projects were installed with cross walls
and at the cross wall section. Of these scenarios, an excavation from the excavation surface down to a depth of only 5–6 m into
the soil, a scenario similar to the case of Hch/He = 0 and Hce/
Hp = 0.3, as shown in Fig. 16. Because no cross walls were installed
(a) Wall deflection (cm) (b) Wall deflection (cm) above the excavation surface, the lateral wall deflections above the
excavation surface or the PSRs were similar to those in the case of
25 20 15 10 5 0 25 20 15 10 5 0
0 0 Hch/He = 0 and Hce/Hp = 1.0. Although the overall lateral wall deflec-
tions or the PSRs were moderately larger than those for Hch/He = 1
5 5
and Hce/Hp = 1, the lateral wall deflections below the surface were
10 10 considerably reduced compared with those in the scenario with no
cross walls. Moreover, the lateral deflections at the bottom of the
Depth (m)
Depth (m)

15 15
wall were close to those in the scenario with no cross walls, which
20 20 caused relatively larger bending moments and shear forces than
25 25 those in the other cases. Special consideration should be given to
the design of the thickness of the wall and its reinforcements when
30 30 Hch/He = 0 and Hce/Hp = 0.3 are adopted.
35 35 To obtain the inclination of the critical height and the embed-
ment ratios of the cross walls, as marked by  and listed in Table 5,
40 40 ten scenarios were established based on a combination of three ba-
sic and one varied value from He L0 BL in Table 4, and eight addi-
tional scenarios were constructed with consideration of different
soil strengths and structural stiffnesses by combining basic values
of HeL0 BL (ID = 2223) with the variation of su ¼ 0:25r0v and 0.35 r0v ,
0.2 times and 3 times the wall system stiffness, 5 times and 20
times the axial rigidity of the struts, and 0.6 and 0.8 times the
thickness of the cross wall. A total of 18 scenarios, each combined
Fig. 16. Comparison of the lateral wall deflections for different ratios of cross wall
height and embedment for ID = 2223 and 2523: (a) at the midpoint of two cross
with various values of Hch/He and Hce/Hp, were performed to obtain
walls, (b) at the cross wall section. the values of (Hch/He)cr and (Hce/Hp)cr. The variation of the (Hch/He)cr
64 C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65

(a) 1 He = 0.6 and Hce/Hp = 0.6 than for Hch/He = 1.0 and Hce/Hp = 1.0. This
0.9 situation occurred because the entire diaphragm wall, from top to
bottom, was directly restrained from movement by the cross wall
0.8
in the case of Hch/He = 1.0 and Hce/Hp = 1.0 whereas only part of the
0.7
9 points diaphragm wall was restrained from movement in the case of
0.6
Hch/He = 0.6 and Hce/Hp = 0.6. Similarly, the lateral wall deflection
0.5 at the midpoint of the two cross walls for Hch/He = 0.6 and
2 points
0.4 2 points Hce/Hp = 0.6 was almost equal to that for Hch/He = 1.0 and
0.3 Hce/Hp = 1.0. For the maximum lateral wall deflection, which usu-
0.2 ally occurred at the midpoint of two cross walls, both the full cross
0.1 wall depth and the critical wall depth resulted in approximately
0 the same reduction in the lateral wall deflection.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
7. Conclusions

(b) 1
3 points
This study performed a series of 3D numerical analyses to
0.9 investigate the effect of different cross wall installation layouts
0.8 on the lateral wall deflection in excavations and included the cross
6 points
0.7 wall interval, height and embedded depth as variables. The influ-
0.6 ence of the adjacent cross wall intervals on the lateral wall deflec-
0.5 tion in the central cross wall bay was also studied. The results can
2 points be used as a first approximation for the design of a reasonable cross
0.4
wall interval and an economically feasible cross wall depth to facil-
0.3
itate a reduction in the lateral wall deflection equal to that pro-
0.2 vided by a full cross wall depth. The following conclusions can
0.1 also be drawn:
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
1. The cross wall produced a corner effect similar to that of a dia-
phragm wall corner, near which the lateral wall deflections
Fig. 17. (a) Critical ratio of the cross wall height for various sets of excavations, (b) were much smaller than those occurring at other sections of
critical ratio of cross wall embedment for various sets of excavations. the wall. However, a small amount of lateral wall deflection
was observed at the cross wall section due to the elastic com-
pression of the cross walls, and the removal of cross walls dur-
and (Hce/Hp)cr values with L0 /B are shown in Fig. 17, implying that ing excavation in close proximity to the diaphragm wall corner
(Hch/He)cr and (Hce/Hp)cr generally decrease with increasing L0 /B. caused minimal lateral wall deflection. The difference in the
Fig. 18a and b displays the lateral wall deflections both at the plane strain ratio between the cross wall and the diaphragm
midpoint of the two cross walls and at the cross wall section for wall corners is approximately equal to the PSR at the cross wall
the cases involving full cross wall depth (i.e., Hch/He = 1.0, section.
Hce/Hp = 1.0) and the case (Hch/He = 0.6, Hce/Hp = 0.6) whose cross 2. The lateral wall deflection in a cross wall bay formed by two
wall height and embedment close to their critical values, respec- cross walls was constrained only by the cross walls and the
tively, when He = 20 m, L0 = 28 m and B = 40 m (ID = 2323). The cross wall interval in this bay and was affected minimally by
lateral wall deflections at the cross wall section were quite close the adjacent cross wall bays; thus, the effect of different dia-
in both cases, although the lateral deflections at the top and phragm walls or cross wall intervals would be negligible.
bottom of the diaphragm wall were slightly higher for Hch/ 3. The lateral wall deflection or plane strain ratio at the midpoint
of two cross walls was approximately linearly proportional to
(a) Wall deflection (cm) (b) Wall deflection (cm) the square of the cross wall interval. When the cross wall bay
25 20 15 10 5 0 25 20 15 10 5 0
0 0 geometry factor was varied from 0 to 2, the plane strain ratio
at the midpoint of two cross walls decreased rapidly with an
5 5 increasing bay geometry factor. When the bay geometry factor
10 10 was larger than 2, the plane strain ratio was greatly affected by
the cross wall interval, implying that the cross walls can effec-
Depth (m)
Depth (m)

15 15
tively reduce the lateral wall deflection. Under such circum-
20 20 stances, the plane strain ratio increased slowly with
25 increasing Fg for all bay factors.
25
4. The installation of a full cross wall depth, i.e., from the ground
30 30 surface down to the bottom of the wall, can produce the best
35 35 reduction in the lateral wall deflection. However, use of the crit-
ical cross wall height and embedment can yield a lateral wall
40 40 deflection quite close to that provided by the full cross wall
depth in a more economical manner.

Acknowledgement
Fig. 18. Comparison of the lateral wall deflections between the cases of critical wall
height and embedment and of full depth for ID = 2323: (a) at the midpoint of two The authors acknowledge the support provided by the National
cross walls, (b) at the cross wall section. Science Council in Taiwan via Grant No. NSC100-2221-E-146-007.
C.-Y. Ou et al. / Computers and Geotechnics 50 (2013) 55–65 65

References [8] Ou CY, Chiou DC, Wu TS. Three-dimensional finite element analysis of deep
excavations. J Geotech Eng 1996;122(5):337–45.
[9] Finno RJ, Blackburn JT, Roboski JF. Three-dimensional effects for supported
[1] Ou CY, Heish PG, Chiou DC. Characteristics of ground surface settlement during
excavations in clay. J Geotech Geoenviron Eng 2007;133(1):30–6.
excavation. Can Geotech J 1993;30(5):758–67.
[10] Ou CY, Lin YL, Hsieh PG. A case record of an excavation with cross walls and
[2] Wong KS, Wong IH, Broms BB. Methods of improving the stability of deep
buttress walls. J GeoEng 2006;1(2):579–86.
excavations in soft clay. In: Proceedings of the 8th Asian regional conference
[11] Itasca Consulting Group, Inc. FLAC3D: fast Lagrangian analysis of continua,
on soil mechanics and foundation engineering, Kyoto; 1987. p. 321–4.
version 2.1. Minnesota, USA; 1997.
[3] Gaba AR. Jet grouting at Newton station. In: Proceedings of the 10th Southeast
[12] Lin YL. The effects of cross walls on the movements of excavations in clay. PhD
Asia geotechnical conference, Taipei; 1990. p. 77–9.
dissertation. Taipei, Department of Construction Engineering, National Taiwan
[4] Liu GB, Ng CWW, Wang ZW. Observed performance of a deep multistrutted
University of Science and Technology; 2010.
excavation in Shanghai soft clays. J Geotech Geoenviron Eng 2005;131(8):
[13] Hsieh PG, Ou CY, Lin YL. Three dimensional numerical analysis of deep
1004–13.
excavations with cross walls. Acta Geotech 2012. http://dx.doi.org/10.1007/
[5] Parashar S, Mitchell R, Hee MW, Sanmugnathan D, Sloan E, Nicholson G.
s11440-012-0181-8:1-16.
Performance monitoring of deep excavation at Changi WRP project. In:
[14] Chew SH, Yong KY, Lim AYK. Three-dimensional finite-element analysis of a
Proceedings of the 7th international symposium on field measurements in
strutted excavation. In: Yuan, editor. Computer methods and advances in
geomechanics, Singapore; 2007. p. 1–12.
geomechanics. Rotterdam, The Netherlands: Balkema; 1997.
[6] Ou CY, Hsieh PG, Lin YL. Performance of excavations with cross walls. J Geotech
[15] Lee FH, Yong KY, Quan KCN, Chee KT. Effect of corners in strutted excavations:
Geoenviron Eng 2011;137(1):94–104.
field monitoring and case histories. J Geotech Geoenviron Eng 1998;124(4):
[7] Hsieh HS, Lu YC, Lin TM. Effects of joint details on the behavior of cross walls. J
339–49.
GeoEng 2008;3(2):55–60.

Vous aimerez peut-être aussi