Sie sind auf Seite 1von 13

1134

A simplified plane strain analysis of lateral wall


deflection for excavations with cross walls
Pio-Go Hsieh, Chang-Yu Ou, and Chiang Shih

Abstract: Previous studies have shown that installation of cross walls in deep excavations can reduce lateral wall deflection
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

to a very small amount. To predict the lateral wall deflection for excavations with cross walls, it is necessary to perform a
three-dimensional numerical analysis because the deflection behavior of the diaphragm wall with cross walls is by nature
three dimensional. However for the analysis and design of excavations, two-dimensional plane strain analysis is mostly used
in practice . For this reason, based on the deflection behavior of continuous beams and the superimposition principle, an
equivalent beam model suitable for two-dimensional plane strain analysis was derived to predict lateral wall deflection for
excavations with cross walls. Three excavation cases were employed to verify the proposed model. Case studies confirm the
proposed equivalent beam model for excavations with cross walls installed from near the ground surface down to at least
more than half the embedded depth of the diaphragm wall. For the case with a limited cross-wall depth, the proposed model
yields a conservative predicted lateral wall deflection.
Key words: deep excavation, lateral wall deflection, cross wall, equivalent beam model, plane strain analysis.
Rsum : Des tudes antrieures avaient dmontr que linstallation de murs de refends dans les excavations profondes peut
rduire la dflection latrale du mur jusqu des valeurs trs petites. Afin de prdire la dflection latrale dun mur pour des
excavations comportant des murs de refends, il est ncessaire de raliser trois analyses numriques dimensionnelles puisque
le comportement en dflection dun mur diaphragme avec des murs de refends est de nature tridimensionnelle. Cependant,
For personal use only.

lanalyse et la conception dexcavations est ralise en pratique plutt avec des analyses de dformation en plan en deux di-
mensions. Pour cette raison, un modle de poutre quivalente appropri pour lanalyse de dformation en plan en deux di-
mensions a t driv pour prdire la dflection latrale des murs dans les excavations ayant des murs de refends, ceci bas
sur le comportement en dflection de poutres continues et du principe de superposition. Trois cas dexcavations ont t utili-
ss pour vrifier le modle propos. Les tudes de cas confirment que le modle est appropri pour les excavations avec des
murs de refends installs prs de la surface du sol jusqu au moins plus de la moiti de la profondeur du mur diaphragme.
Dans le cas o la profondeur du mur de refend est limite, le modle propos prdit la dflection latrale du mur de faon
conservative.
Motscls : excavation profonde, dflection latrale des murs, mur de refend, modle de poutre quivalente, analyse de d-
formation en plan.
[Traduit par la Rdaction]

Introduction tion. The cross walls above the excavation surface are demol-
ished during the process of excavation. Cross walls function
Deep excavation may induce excess lateral wall deflections as lateral struts against lateral wall deflection but exist before
and ground settlements, which can damage adjacent build- excavation. If the cross wall is constructed using the dia-
ings. To avoid such damage, it is necessary to adopt remedial phragm wall construction technique, it would be of high
measures to limit the lateral wall deflection or ground settle- compressive strength and large axial stiffness. Under such
ment. circumstances, the diaphragm wall would be highly restrained
Ground improvement and strengthening of the retaining- from lateral movement and the ground movement would then
strut system are the common remedial measures to reduce ex- be reduced significantly. Ou et al. (2006, 2011) studied the
cavation-induced ground movements (e.g., Wong, et al. 1987; effectiveness of cross walls in restraining lateral wall deflec-
Gaba 1990; Liu et al. 2005; Parashar et al. 2007). Recently, tion and ground settlements and found that the cross wall has
cross walls have been widely used in some countries as an a corner effect similar to the diaphragm wall. In their studies,
alternative remedial measure. As shown in Fig. 1, cross walls they determined that the maximum wall deflection and
are constructed perpendicularly to the diaphragm wall, to ground settlement could be reduced by about 75% and 82%,
which both ends of cross walls are connected before excava- respectively.
Received 15 November 2011. Accepted 18 July 2012. Published at www.nrcresearchpress.com/cgj on 17 September 2012.
P.-G. Hsieh. Department of Assets and Property Management, Hwa Hsia Institute of Technology, New Taipei City, Taiwan 23568, R.O.C.
C.-Y. Ou. Department of Construction Engineering, National Taiwan University of Science and Technology, Taipei, Taiwan 10672, R.O.C.
C. Shih. Trinity Foundation Engineering Consultants Co. Ltd., Taipei, Taiwan 10685, R.O.C.
Corresponding author: Chang-Yu Ou (ou@mail.ntust.edu.tw).

Can. Geotech. J. 49: 11341146 (2012) doi:10.1139/T2012-071 Published by NRC Research Press
Hsieh et al. 1135

Fig. 1. Schematic diagram of the cross wall: (a) plan view (b) sec- Fig. 2. Plan view of lateral deflection of a diaphragm wall with and
tion AA. without cross walls.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

In fact, the deformation behavior of the diaphragm wall


and ground settlement are by nature three dimensional if the
cross wall is installed. Due to the elastic compression of the As shown in Fig. 3a, dh,0 is the lateral deflection of the di-
cross wall, the diaphragm wall at the location of cross wall aphragm wall at a depth at the cross wall, i.e., section AA;
still has a small amount of wall deflection, which will in- dh,mid is the lateral deflection of the diaphragm wall at this
crease with the distance to the cross wall, up to the midpoint depth at the midpoint of two cross walls, section BB; and
of two cross walls. In general the deflection at various sec- dh,d is the lateral deflection of the diaphragm wall at this
tions between two cross walls is much smaller than such depth at a distance d to the cross wall. Based on the above
cases without cross walls (Fig. 2). Moreover, Lin et al. assumptions, the deflection of the diaphragm wall at a spe-
(2012) found from analyses of four excavations that the cific depth, subject to uniform pressure, can be decomposed
amount of lateral deflection for excavations with cross walls into an elastic continuous beam supported by cross walls
where the deflections of the beam are all equal to dh,0 and a
highly depends on the cross wall interval, depth of cross wall
For personal use only.

fixed-end beam, as shown in Figs. 3b and 3c, respectively.


below the excavation surface, and height of cross wall above
The deflection of the fixed-end beam at any distance d can
the excavation surface. To accurately predict the lateral wall
be expressed by (Fig. 3c)
deflection, it is necessary to perform three-dimensional (3D)
 
numerical analysis. wn L 02 d2 2d d2
However, 3D analyses are costly and in practice two- 1 Ddd 1  0 02
24EI L L
dimensional (2D) analyses are often used instead. With 2D
analyses, the cross wall interval, depth, size, and stiffness where wn is the earth pressure at a depth, L is the cross wall
should be appropriately simulated. Although the authors interval, d is the distance to the cross wall, E is the Youngs
have performed plane strain analysis for one case history, modulus of the beam, and I is the moment of inertia.
the analysis was only for the section at the cross wall, in Substituting d = L/2 into eq. [1], we obtain
which the diaphragm wall deflection was the smallest. As
shown in a previous study (Lin et al. 2012), the diaphragm wn L 04
2 Ddmid
wall deflection at the midpoint between two cross walls was 384EI
the largest, which should be analyzed when wall deflection where Ddmid is the deflection of the fixed-end beam at the
iss a concern. For this reason, the objective of this study is midpoint.
to derive an analytical model suitable for 2D plane strain Therefore, the deflection of a diaphragm wall at a distance
analysis for any section of diaphragm walls with cross walls d from the cross wall, dh,d, and at the midpoint between cross
using Winkler springs (Terzaghi 1955). The proposed walls, dh,mid , can be expressed, respectively, as (Fig. 3a)
method will be verified using three case histories.
3 dh;d dh;0 Ddd
Derivation of the analytical model
4 dh;mid dh;0 Ddmid
As shown in Fig. 2, the deflection of the diaphragm wall
varies from section to section due to the installation of cross Because the deflections of the continuous beam are all
walls. Deflection of the diaphragm wall is smallest at the lo- equal to dh,0 (Fig. 3b), for plane strain analysis, the continu-
cation of the cross wall and its value increases with increas- ous beam supported by cross walls is simulated as supported
ing distance to the cross wall, up to the midpoint of two by springs. The spring stiffness per unit depth per unit length
cross walls. For plane strain analysis of the deflection of the of the diaphragm wall, Kcw, equivalent to the effect of cross
diaphragm wall at any section, an equivalent beam method is walls across intervals of L, is thus equal to (Fig. 4b)
adopted and the following assumptions are made (refer to
Fig. 2): tcw Ecw
5 Kcw
1. The diaphragm wall is of infinite length. L0 Lcw
2. The cross walls are installed with equal intervals. where tcw is the thickness of the cross wall, Ecw is the
3. The cross walls are constructed from the near ground sur- Youngs modulus of the cross wall, and Lcw is the length of
face down to the bottom of the diaphragm wall. the cross wall.

Published by NRC Research Press


1136 Can. Geotech. J. Vol. 49, 2012

Fig. 3. Decomposition of the diaphragm cross wall system for a Fig. 4. Simulation of the deflection behavior for the diaphragm
section into a continuous beam supported by cross walls and a cross wall system for a section: (a) diaphragm wall with cross walls;
fixed-end beam: (a) diaphragm wall with cross walls; (b) continuous (b) continuous beam supported by cross walls; (c) fixed-end beam.
beam supported by cross walls; (c) fixed-end beam.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12
For personal use only.

Substituting Ddd in eq. [1] into eq. [6], we can obtain Kfeb,
d as
24EI
For plane strain analysis of the fixed-end beam, as shown 7 Kfeb;d
L 0 2 d 2  2L 0 d3 d 4
in Fig. 3c, the wall deflection at a distance d can be simu-
lated as a beam supported by springs whose deflection be- Substituting d = L/2 into eq. [7], we can obtain the equiv-
havior is exactly the same as that section of the fixed-end alent stiffness, Kfeb,mid, at the midpoint as
beam (Fig. 4c). Assuming that the deflection of the fixed- 384EI
end beam subjected to a uniform pressure, wn, at a distance 8 Kfeb;mid
d is equal to Ddd, the equivalent stiffness, Kfeb,d, of springs L0 4
supporting the continuous beam whose deflection behavior According to eq. [2], the deflection of a diaphragm wall at
is exactly the same as that section would be a distance of d to the cross wall (dh,d) is the sum of deflec-
wn tions obtained from Figs. 4b and 4c. By substituting Ddd =
6 Kfeb;d
Ddd wn/Kfeb,d and dh,0 = wn/Kcw into eq. [2], we can obtain the fol-
lowing expression
where Kfeb,d is the equivalent spring stiffness per unit depth
wn wn wn
per unit length of the diaphragm wall for a fixed-end beam 9
for the section at a distance d. Keq;d Kcw Kfeb;d

Published by NRC Research Press


Hsieh et al. 1137

where Keq,d is the equivalent stiffness of springs for the sec- Fig. 5. Excavation geometry and allocation of cross walls for case 1.
tion of the diaphragm wall with cross walls at a distance d
per unit depth per unit length of the diaphragm wall.
Equation [9] can be further simplified as
Kcw Kfeb;d
10 Keq;d
Kcw Kfeb;d
Therefore, the deflection behavior of the diaphragm with
cross walls at a distance d, as shown in Fig. 3a, can be con-
sidered as a beam supported by springs with the stiffness
equal to eq. [9] or [10]; that is, the springs with stiffness
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

Kcw are connected in series with the springs with stiffness


Kfeb,d, as shown in Fig. 4a.
With numerical plane strain analysis, cross walls should be
divided into many horizontal strips connected to finite ele-
ment nodes on the diaphragm wall. The equivalent stiffness
per unit depth per unit length, Kcw and Kfeb,d, for each hori- were cast with 24.0 MPa concrete. Both the cross walls and
zontal cross wall strip can be evaluated through eqs. [5] and buttress walls were demolished with excavation process.
[7], respectively. The equivalent stiffness Keq,d for the section Figure 6 also shows the subsurface soil profile, as well as
at a distance d can then be computed through eq. [10]. The the basic soil properties and strength characteristics. The
total stiffness for each strip of cross wall is obtained by the groundwater was 3 m below the ground surface.
multiplication of Keq,d and the depth of each strip, and then
lumped together as springs. Mesh and parameters
In actual 3D behavior, the different lateral wall deflection Figure 7 shows the 2D plane strain finite element mesh
along its length will cause shear stresses in the soil while the used for analysis of the center section of the excavation in
the northsouth direction. The vertical boundaries were re-
For personal use only.

plane strain simplification of this behavior does not. The in-


fluence of this simplification on the wall deflection will be strained from horizontal movement. The bottom was set at
discussed in the following case studies. the bedrock and considered to be restrained from both verti-
cal and horizontal movements. The left vertical boundary was
set at the center of the excavation zone, and the right boun-
Numerical verification dary was 130 m, about four times the final excavation depth,
Three case histories were used to verify the proposed away from the wall.
equivalent beam model. Each of the cases represents different In the analysis, the clay layers were simulated using the
settings of cross walls, such as cross wall interval and cross undrained soft clay (USC) model, as developed by Hsieh
wall depth. and Ou (2011, 2012) and employed by some investigators
(Kung et al. 2007; Hsiao et al. 2008; Ou and Hsieh 2011),
Case study No. 1 and the sand layers were simulated using the hyperbolic
model (Duncan and Chang 1970). Table 1 lists the clay pa-
Overview rameters for the USC model and sand parameters for the hy-
Figure 5 shows the plan of the excavation case, in which perbolic model. In the table, the strength parameters (su) of
three cross walls and 10 buttress walls were constructed. The clay were obtained directly from laboratory tests, and the de-
excavation depth was 32.5 m, which was completed in nine formation parameters (Ei, m, n) were obtained from the
stages using the top-down construction method (Ou et al. bender element test and the triaxial unloadingreloading test
2006). The thickness of the diaphragm wall was 1.5 m and on the soil at different depths near the construction site. The
the depth of the wall was 57.5 m. The diaphragm wall was strength parameter (c, f 0 ) of sandgravel was obtained di-
designed with 27.5 MPa of the compressive strength (fc0 ). As rectly from laboratory tests and the deformation modulus (K,
shown in Fig. 6, the thickness of concrete slabs 1FL and n) from an empirical correlation (Ou and Lai 1994).
B1FL were 250 and 200 mm, respectively, while the concrete The coefficients of earth pressure at rest (K0) for sand and
slabs B2FL to B6FL were 610 mm. After excavating down to normally consolidated clay are estimated using Jakys (1944)
GL29.4 m (GL refers to the ground surface level), three equation. When the clay is in the unloading or pre-consolida-
H400 400 mm 13 mm 21 mm steel struts, with a tion state, K0 is estimated using the equation proposed by Al-
spacing of 6.4 m and slope of 4.6:6, were installed as tempo- pan (1967).
rary supports. The construction sequence and location of the The diaphragm wall and lateral supports were assumed to
lateral supports are shown in detail in Fig. 6. behave as a linearelastic material. The struts and concrete
The cross walls in this case were 1.0 m thick and 45 m floor slabs were simulated as bar elements. The Youngs
deep with a 26 m interval (L) in the northsouth direction moduli of the concrete structures, including diaphragm walls,
and the 10 buttress walls were 1.0 m thick and 55 m deep floor slabs, and cross walls, were estimated from fc0 according
with their length varying from 6 to 15 m. The cross walls to ACI 318-95 (ACI Committee 318) and reduced by 20% as
and buttress walls between GL+0 to GL1.5 m were back- suggested by Hsieh and Ou (2012) and Lim et al. (2010). Ta-
filled with the in situ soil, between GL1.5 to GL22 m ble 2 lists the parameters of the diaphragm wall, concrete
were cast with 13.7 MPa concrete, and below GL22 m floor slabs, and struts.

Published by NRC Research Press


1138 Can. Geotech. J. Vol. 49, 2012

Fig. 6. Profile of the excavation sequence and subsurface soil layers for case 1. CL, silty clay; SM/GW, silty sand / well-graded gravel (clas-
sification according to ASTM 2006)..
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12
For personal use only.

Fig. 7. Finite element mesh used for analysis of case 1.

Table 1. Input parameters of soils for case 1.

Depth (m) Soil type K0 su/s v0 Rf Ei/su md nd n c (kPa) f () K = Kur n


0 to 3 Fill 0.93 0.32 0.9 2100 0.225 1.299 0.495
3 to 32.6 CL 0.50 0.34 0.9 2100 0.225 1.299 0.495
32.6 to 51 CL 0.55 0.38 0.9 2100 0.225 1.299 0.495
51 to 67 SM/GW 0.41 0.9 0.3 0.0 36 5600 0.5
Note: CL, silty clay; SM/GW, silty sand / well-graded gravel (ASTM 2006).

For plane strain analysis of the diaphragm cross wall and 21.44 (MN/m)/m/m, respectively, in which tcw = 1.0 m,
system, Kcw values for the depths between GL1.5 to GL L = 26 m, Lcw = B/2 = 33.05 m, and Ecw = 13 917 MPa
22 m and GL22 to GL45 m were computed to be 16.20 (GL1.5 to GL22 m) and 18 420 MPa (GL22 to GL

Published by NRC Research Press


Hsieh et al. 1139

Table 2. Input parameters of structural members for case 1.

Diaphragm Kst of temporary steel struts or floor slabs ((MN/m)/m)


E (MPa) n Level 1 Level 2 Levels 37 Level 8
19 718 0.2 149.2 119.4 364.1 107.7

Table 3. Values of Kfeb,d and Keq,d for case 1.

Keq,d ((MN/m)/m/m)
d (m) Kfeb,d ((MN/m)/m/m) Kcw = 16.20 Kcw = 21.44
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

0 1 16.20 21.44
1 212.95 15.05 19.48
3 27.96 10.25 12.13
5 12.07 6.92 7.72
8 6.42 4.60 4.94
13 4.66 3.62 3.83

45 m). The equivalent stiffness of the fixed-end beam, Kfeb,d, Fig. 8. Comparison of wall deflections at SI-8 from field observa-
for the section at a distance d from the cross wall could be tion and those from analysis of case 1.
computed through eq. [7]. The equivalent stiffness of the dia-
phragm cross wall system, Keq,d, for the sections at various
distances d could then be computed using eq. [10]. Table 3
lists Keq,d and its corresponding Kfeb,d value for the sections
at various distances, where d = 0 denotes the section at the
For personal use only.

cross wall, i.e., at SI-8, and d = 13 m denotes the section at


the midpoint of two cross walls, i.e., at SI-7 or SO-1.

Analysis and results


The authors (Hsieh et al. 2010; Lin et al. 2012) performed
3D numerical analyses of this case and the same excavation
that was assumed with cross walls only, i.e., no buttress
walls, and found little difference in wall deflection between
the case with cross walls and buttress walls and the case
with cross walls only. This is because the lateral resistance
against the diaphragm wall lateral movement from the but-
tress wall was mainly from the side friction of the buttress
wall. Because the wall deflection at the midpoint between
the cross walls had been reduced to a very small amount, the
relative movement between the buttress wall and surrounding
soil was very small, which in turn caused the frictional resist-
ance from the buttress wall to be very small. Therefore, the
effect of buttress walls at SI-7 and SO-1 will be neglected in
the following study.
Figure 8 shows the comparison of observed wall deflec-
tions at SI-8 at stages 3, 5, 7, 9, and those predicted with
the proposed equivalent beam model, along with the 3D anal-
ysis results from Lin et al. (2012). In addition, analyses with
the assumption of no cross walls were performed and the re-
sults are also shown in the figure; these results are quite con-
sistent with the excavations in Taipei as studied in Ou et al.
(1993). As shown in Fig. 8, the wall deflections predicted Figure 9 shows the comparison of observed wall deflec-
from the proposed equivalent beam model at various stages tions at SO-1 at stages 3, 5, 7, 9, and those predicted with
were very close to those from the field observation and 3D the proposed equivalent beam model, along with the 3D anal-
analysis results. This may be attributed to the fact that little ysis results from Lin et al. (2012). The wall deflections pre-
difference in lateral wall deflection between the cross wall dicted from the proposed equivalent beam model at various
section and its nearby section exists. Compared with those stages were very close to those from the field observation
predicted from the case with the assumption of no cross and slightly larger than 3D analysis results. The slight differ-
walls, installation of cross walls can reduce lateral wall de- ence in lateral wall deflection near the center sections in the
flection at SI-8 by 75%. actual 3D condition may cause such a slight deviation. Com-

Published by NRC Research Press


1140 Can. Geotech. J. Vol. 49, 2012

Fig. 9. Comparison of wall deflections at SO-1 from field observa- Fig. 10. Comparison of the computed maximum lateral wall deflec-
tion and those from analysis of case 1. tions at various sections and those from field observation for case 1.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12
For personal use only.

Fig. 11. Excavation geometry and allocation of cross walls for case 2.

pared with those predicted from the case with the assumption
of no cross walls, installation of cross walls can reduce the
lateral wall deflection at SO-1 by 50%.
Figure 10 shows the variation of the predicted maximum
wall deflections (dhm) at various distances to the cross wall
(d) using the proposed equivalent beam model and 3D analy-
sis results from Lin et al. (2012). For comparison, the ob-
served maximum wall deflections at different locations, even used as an earth-retaining structure. The fc0 value of the dia-
in the adjacent bay, are also shown in the same figure. As phragm wall was 27.5 MPa.
shown in the figure, the predicted maximum wall deflections As shown in Fig. 11, four cross walls 0.8 m thick were
at different distances to the cross wall were close to the field constructed along the excavation length. Twelve buttress
observations though they were slightly larger than 3D analy- walls were constructed in front of the diaphragm wall be-
sis results. The restraining effect of the cross wall in reducing tween the cross walls and two buttress walls were constructed
the wall deflection was the largest at the location where the in back of the diaphragm wall along the short side. The but-
cross wall was installed. Such a restraining effect decreased tress walls were 0.8 m thick and 6.5 m long. The depth of
with increasing distance to the cross wall. This comparison the cross walls and buttress walls was between GL2.0 to
further confirms the proposed equivalent beam model. GL35.0 m as shown in Fig. 12. The cross walls and buttress
walls were backfilled with in situ soil above GL2.0 m, and
Case study No. 2 the cross walls and buttress walls between GL2.0 to GL
23.6 m and GL23.6 to GL35 m were cast with concrete
Overview with compressive strengths of 17.2 and 27.5 MPa, respec-
Figure 11 shows the plan of the excavation case, in which tively. Both cross walls and buttress walls inside the excava-
the excavation depth was 23.6 m, which was completed in tion above the excavation surface were demolished during
six stages using the top-down construction method as shown excavation process.
in Fig. 12. Floor slabs 1FL and B1FL used the beam-plate Figure 12 shows that the subsoil was mainly alternating
system and its slab thickness was 15 cm. Floor slabs B2FL silty sand (SM) and silty clay (CL) deposits. The gravel layer
to B4FL used the flat-slab system and its slab thickness was was located at about GL50.0 m and its SPT-N value was
35 cm. A 1.2 m thick and 43.0 m deep diaphragm wall was greater than 50. The groundwater table was located at GL

Published by NRC Research Press


Hsieh et al. 1141

Fig. 12. Profile of the excavation sequence and subsurface soil layers for case 2.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12
For personal use only.

Table 4. Input parameters of soils for case 2.

Depth (m) Soil type K0 su/s v0 Rf Ei/su md nd n c (kPa) f () K = Kur n


0 to 5.0 CL 0.93 0.32 0.9 2100 0.225 1.299 0.495
5.0 to 10.3 SM 0.50 0.9 0.3 0 30 1101 0.5
10.3 to 31.4 CL 0.52 0.34 0.9 2100 0.225 1.299 0.495
31.4 to 37.2 SM/ML 0.50 0.9 0.3 0 30 1580 0.5
37.2 to 45.0 CL 0.56 0.38 0.9 2100 0.225 1.299 0.495
45.0 to 50 SM 0.44 0.9 0.3 0 34 2681 0.5

2.5 m and was generally considered to be hydrostatic. How- Table 5. Input parameters of structural members for case 2.
ever, the piezometric water pressure in the sixth layer of soil,
Diaphragm Kst of floor slabs ((MN/m)/m)
silty sand or Chingmei gravel layer, was located at about
GL10 m. The test data for each soil layer are as shown in E (MPa) n Level 12 Levels 35
Fig. 12. 19 718 0.2 124.2 268.7

Mesh and parameters


The plane strain finite element mesh for the central section have as linearelastic materials. The parameters for the soils,
of the excavation was similar to Fig. 7. The right boundary retaining walls, and concrete floor slabs were determined in
was 105.0 m from the wall and the bottom was set at the ways similar to those employed in case 1. Tables 4 and 5 list
gravel layer, which was 50.0 m below the surface. The left the parameters of the soils and structural members, respec-
boundary was set at the center of the excavation zone and re- tively.
strained from horizontal movement. The equivalent stiffness, Keq,d, of the diaphragm cross
Similar to case 1, the clay layers were simulated using the wall system for the sections at various distances was calcu-
USC model and the sand layers using the hyperbolic model. lated following a method similar to case 1, in which the val-
The diaphragm wall and lateral supports were assumed to be- ues of Kcw between GL2.0 to GL23.6 m and GL23.6 to

Published by NRC Research Press


1142 Can. Geotech. J. Vol. 49, 2012

Table 6. Values of Kfeb,d and Keq,d for case 2.

Keq,d ((MN/m)/m/m)
d (m) Kfeb,d ((MN/m)/m/m) Kcw = 11.91 Kcw = 15.06
0 1 11.91 15.06
1 40.73 9.21 10.99
3 5.00 3.52 3.76
6.5 1.29 1.16 1.19
12 0.53 0.51 0.51
18.5 0.36 0.35 0.36
20.95 0.35 0.34 0.35
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

GL35 m were computed to be 11.91 and 15.06 (MN/m)/m/ Fig. 13. Comparison of wall deflections from field observation and
m, respectively, in which tcw = 0.8 m, L = 41.9 m, Lcw = B/ those from analysis for case 2: (a) SID4; (b) SID5.
2 = 25 m, and Ecw = 15 594 MPa (GL2.0 to GL23.6 m)
and 19 718 MPa (GL23.6 to GL35 m). Table 6 lists the
equivalent stiffnesses for the sections at various distances
and their corresponding Kfeb,d values. In the table, d =
18.5 m denotes the section at SID4 and d = 20.95 m denotes
the central section.

Analysis results
Similar to case 1, it was found from the parametric study
that the buttress wall contributed little in reducing the lateral
wall deflection, thus only the effectiveness of cross walls in
reducing lateral wall deflections was studied.
For personal use only.

Figure 13 shows the comparison of the observed wall de-


flections at SID4 and SID5 at the fourth and sixth excavation
stages and those predicted using the proposed equivalent
beam model, along with the 3D analysis results from Lin et
al. (2012). In addition, analysess with the assumption of no
cross walls were performed and the results are also shown in
the same figure, which are also consistent with the excava-
tions in Taipei as studied in Ou et al. (1993). As shown in
the figure, the predicted wall deflections using the proposed
model were slightly larger than field observations and 3D
analysis results. The slight difference in lateral wall deflec-
tion near the center sections in the actual 3D condition may
cause such a slight deviation. The figure also shows that the
predicted wall deflection at SID4 was moderately smaller
than that for the no cross wall case. It is justified that the
cross wall interval was too large, namely 41.9 m, causing
the corner or cross wall effect to have insignificant influence
on the deformation for the section near the center of the
span. Although SID5 was located in the span of the 45.9 m
cross wall interval, it was only 6.5 m away from the cross
wall. The corner or cross wall effect had a significant influ-
ence on the deformation behavior at that section. The analy-
sis result reveals that the predicted wall deflections for the
cross wall case were much smaller than those without the
cross wall. near the wall for a considerable time at this stage according
Figure 14 shows the variation of the predicted maximum to the construction record, which might have resulted in an
wall deflection (dhm) at various distances to the cross wall abnormal increment of dhm. The restraining effect of the cross
(d) using the proposed equivalent beam model and 3D analy- wall in reducing the wall deflection was the largest at the lo-
sis results for L = 41.9 m. For comparison, the observed cation where the cross wall was installed. Such a restraining
maximum wall deflections at different locations, even in the effect decreased with increasing distance to the cross wall, up
other bays, are also shown in the same figure. As shown in to the midpoint of two cross walls. The difference of dhm at
the figure, the predicted maximum wall deflections at differ- the central section between no cross walls and with cross
ent distances to the cross wall were close to the field obser- walls was minimal because of the large cross wall interval.
vations except for SID3 and SID9, in which some heavy This comparison further confirms the proposed equivalent
construction vehicles entered the construction site and stayed beam model.

Published by NRC Research Press


Hsieh et al. 1143

Fig. 14. Comparison of the computed maximum lateral wall deflec- Fig. 15. Excavation geometry and allocation of cross walls for case 3.
tions at various sections and those from field observation for case 2.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

the surface. The restraint of the boundaries and mesh genera-


tion were similar to Fig. 7.
Similar to case 1, the clay layers were simulated using the
USC model and the sand layers using the hyperbolic model.
The diaphragm wall and lateral struts were assumed to be-
have as linearelastic materials. The parameters for the soils,
Case study No. 3
For personal use only.

retaining wall and struts were determined in a way similar to


those employed in case 1. Tables 7 and 8 list the parameters
Overview
of the soils and structural members, respectively.
Figure 15 shows the plan of the excavation case, in which The equivalent stiffness, Keq,d, of the diaphragm cross
the excavation depth was 13.5m, which was completed in
wall system for the sections at various distances was com-
five stages using the bottom-up construction method. Figure
puted following a method similar to case 1, in which the Kcw
16 shows the construction sequence and depth of the cross
value between GL13.5 to GL20.0 m was found to be
walls. The spacing between temporary steel strut was 4.6 m
38.78 (MN/m)/m/m in which tcw = 0.7m, L = 28.7m, Lcw =
in the northsouth direction and 8.0 m in the eastwest direc-
tion. A 0.7 m thick and 27 to 30 m deep diaphragm wall was B/2 = 12.4 m, and Ecw = 19 718 MPa. Table 9 also lists the
adopted as the earth retaining structure. The fc0 value of the equivalent stiffness for the sections at various distance and
diaphragm wall was 27.5 MPa. their corresponding Kfeb,d values. In the table, d = 10 m de-
notes the section at SID1 and d = 14.35 m denotes the cen-
As shown in Fig. 15, two cross walls 0.7 m thick with
28.7 m spacing were constructed in the northsouth direc- tral section.
tion. Each cross wall was constructed from GL13.5 to GL Analysis results
20.0 m, i.e., from the excavation surface down to 6.5 m be-
low it (Fig. 16). The fc0 value of the cross wall was 27.5 MPa. Figure 17 shows the comparison of the observed wall de-
The cross walls were backfilled with the in situ soil above flections at SID1 at the second to the final excavation stages
GL13.5 m. The construction method for the cross walls and those predicted using the proposed equivalent beam
was the same as that for the diaphragm wall. model, along with the 3D analysis results from Lin et al.
Figure 16 shows the subsurface soil profile, which was (2012). In addition, analyses with the assumption of no cross
mainly alternating silty sand (SM) and silty clay (CL) depos- walls were performed and the results are also shown in the
its. The groundwater level was located at GL1.7 m and was figure, which were also consistent with the excavations in
generally in the hydrostatic condition. The piezometric water Taipei as studied in Ou et al. (1993). As shown in Fig. 17,
pressure in the fourth and sixth silty sand layers was at about the predicted wall deflections at SID1 at the second and third
GL7 and GL10 m, respectively. The piezometric water stages using the proposed equivalent beam model were very
pressure in the eighth layer, silty sand or the Chingmei close to those of 3D analysis results and the assumption of
gravel, was located at GL10 m. The test data of soil for no cross walls. Predicted wall deflections for the cases with
each layer are as shown in Fig. 16. and without cross walls were also close to the field observa-
tions, implying that cross walls had little effect in restraining
Mesh and parameters the lateral wall deflection. This is because cross walls in this
The plane strain finite element mesh for the central section case were set from the excavation surface down to 6.5 m
of the excavation was similar to Fig. 7. The left vertical below the excavation surface. There were no cross walls
boundary therefore was set at the center of the excavation above the excavation surface, so the cross walls could not
zone; the right boundary was 70 m from the wall; and the provide a strong lateral resistance against the wall movement
bottom was set at the gravel layer, which was 52.5 m below at the earlier stages of excavation.

Published by NRC Research Press


1144 Can. Geotech. J. Vol. 49, 2012

Fig. 16. Profile of the excavation sequence and subsurface soil layers for case 3.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12
For personal use only.

Table 7. Input parameters of soils for case 3.

Depth (m) Soil type K0 su/s v0 Rf Ei/su md nd n c (kPa) f () K = Kur n


0 to 4.4 CL 0.93 0.32 0.9 2100 0.225 1.299 0.495
4.4 to 14.4 SM 0.48 0.9 0.3 0 31 1001 0.5
14.4 to 24.1 CL 0.48 0.34 0.9 2100 0.225 1.299 0.495
24.1 to 30.8 SM 0.47 0.9 0.3 0 32 2120 0.5
30.8 to 35.2 CL 0.48 0.34 0.9 2100 0.225 1.299 0.495
35.2 to 42.3 SM 0.47 0.9 0.3 0 32 2568 0.5
42.3 to 44.5 CL 0.57 0.38 0.9 2100 0.225 1.299 0.495
44.5 to 52.5 SM 0.46 0.9 0.3 0 33 3588 0.5

Table 8. Input parameters of structural members for case 3. Figure 17 also shows that the predicted wall deflections at
the fourth and fifth stages were larger than field observation
Diaphragm Kst of temporary steel struts ((MN/m)/m) and the 3D analysis results. The proposed model seemed to
E (MPa) n Stage 1 Stage 2 Stage 3 Stage 4 yield a conservative result. Similar to cases 1 and 2, the dif-
19718 0.2 45.3 90.6 90.6 114.5 ference in lateral wall deflection near the sections at SID1, in
the actual 3D condition, may cause such a deviation.

Published by NRC Research Press


Hsieh et al. 1145

Table 9. Values of Kfeb,d and Keq,d for case 3. Fig. 18. Comparison of the computed maximum lateral wall deflec-
tions at various sections and those from field observation for case 3.
Keq,d for Kcw =
38.78 ((MN/m)/
d (m) Kfeb,d ((MN/m)/m/m) m/m)
0 1 38.78
1 17.63 12.12
3 2.28 2.15
5 0.96 0.94
10 0.39 0.38
14.35 0.32 0.32
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

Fig. 17. Comparison of wall deflections at SID1 from field observa-


tion and those from analysis of case 3.

Conclusions
For personal use only.

Cross walls are installed perpendicular to a diaphragm


wall, to which both ends of the cross walls are connected be-
fore excavation. Cross walls function as lateral struts against
lateral wall deflection, but exist before excavation. Previous
studies have shown that installation of cross walls in deep ex-
cavations can reduce lateral wall deflection to a very small
amount. In this study, the effectiveness of cross walls in re-
ducing lateral wall deflection was found to be very signifi-
cant. Although the deformation behavior of a diaphragm wall
with cross walls is by nature three dimensional, 3D finite el-
ement analyses are costly and therefore 2D analyses are usu-
ally preferred by practicing engineers.
By decomposing the diaphragm cross wall system into an
elastic continuous beam supported by cross walls where
the deflections of the diaphragm wall are all equal and a
fixed-end beam, an equivalent beam model suitable for 2D
plane strain analysis was derived. The 3D behavior of a dia-
phragm wall with cross walls could then be analyzed with 2D
plane strain methods.
Three excavations with cross walls, with different cross
depths, spans, and intervals, were used to confirm the pro-
Figure 18 shows the comparison of the predicted maxi- posed equivalent beam model. Results show that the pro-
mum wall deflections (dhm) at various distances to the cross posed method can result in good prediction of lateral wall
wall (d) using the proposed equivalent beam model, 3D anal- deflection at the cross wall section where minimal difference
ysis results from Lin et al. (2012), and the observed maxi- in wall deflection between the cross wall section and its
mum wall deflections at different locations. As shown in the nearby section exists. For the section where the lateral wall
figure, although the predicted wall deflections were still deflection has a slight deviation in wall deflection with its
smaller than those without cross walls, the restraining effect nearby section, the proposed equivalent beam model pre-
of cross walls in reducing the lateral wall deflection was just dicted a slightly conservative lateral wall deflection.
to a certain extent. Similar to Fig. 17, 3D analysis results and
the observed maximum wall deflections at different locations Acknowledgement
were mostly smaller than those predicted and the proposed The authors wish to acknowledge the support provided by
equivalent beam model was conservative for the case with the National Science Council in Taiwan with grant No.
cross walls with limited depths. NSC98-2221-E-146-007.

Published by NRC Research Press


1146 Can. Geotech. J. Vol. 49, 2012

References excavation with cross alls and buttress walls. Journal of


GeoEngineering, 1: 7986.
ASTM. 2006. Standard practice for classification of soils for
Ou, C.Y., Hsieh, P.G., and Lin, Y.L. 2011. Performance of
engineering purposes (Unified Soil Classification System).
excavations with cross walls. Journal of Geotechnical and
ASTM standard D2487. American Society for Testing and Geoenvironmental Engineering, 137(1): 94104. doi:10.1061/
Materials, West Conshohocken, Pa. (ASCE)GT.1943-5606.0000402.
ACI Committee 318. 1995. Building code requirements for structural Parashar, S., Mitchell, R., Hee, M.W., Sanmugnathan, D., Sloan, E.,
concrete (ACI 318-95) and Commentary (ACI 318R-95). Amer- and Nicholson, G. 2007. Performance monitoring of deep
ican Concrete Institute (ACI), Farmington Hills, Mich. excavation at Changi WRP project, Singapore. In Proceedings of
Alpan, I. 1967. The empirical evaluation of the coefficient K0 and K0R. the 7th International Symposium on Field Measurements in
Soils and Foundations, 7(1): 3140. doi:10.3208/sandf1960.7.31. Geomechanics. Edited by J. DiMaggio and P. Osborn. ASCE,
Duncan, J.M., and Chang, C.Y. 1970. Nonlinear analysis of stress and Reston, Va. pp. 112.
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by DALHOUSIE UNIVER on 11/09/12

strain in soils. Journal of the Soil Mechanics and Foundations Terzaghi, K. 1955. Evaluation of coefficients of subgrade reaction.
Division, ASCE, 96: 637659. Gotechnique, 5(4): 297326. doi:10.1680/geot.1955.5.4.297.
Gaba, A.R. 1990. Jet grouting at Newton station. In Proceedings of Wong, K.S., Wong, I.H., and Broms, B.B. 1987. Methods of improving
the 10th Southeast Asia Geotechnical Conference, Taipei, Taiwan. the stability of deep excavations in soft clay. In Proceedings of the
pp. 7779. 8th Asian Regional Conference on Soil Mechanics and Foundation
Hsiao, E.C.L., Schuster, M.J., Juang, C.H., and Kung, G.T.C. 2008. Engineering, Kyoto, Japan. pp. 321324.
Reliability analysis and updating of excavation-induced ground
settlement for building serviceability assessment. Journal of List of symbols
Geotechnical and Geoenvironmental Engineering, 134(10): c effective cohesion
14481458. doi:10.1061/(ASCE)1090-0241(2008)134:10(1448). d distance to the cross wall
Hsieh, P.G., and Ou, C.Y. 2011. Analysis of nonlinear stress and E Youngs modulus of the diaphragm wall
strain in clay under the undrained condition. Journal of Ecw Youngs modulus of the cross wall
Mechanisms, 27(2): 201213. doi:10.1017/jmech.2011.24. Ei initial Youngs modulus or Youngs modulus at small
strain
Hsieh, P.G., and Ou, C.Y. 2012. Analysis of deep excavations in clay
fc0 compressive strength of concrete
under the undrained and plane strain condition with small strain I moment of inertia of the diaphragm wall per unit depth
For personal use only.

characteristics. Journal of the Chinese Institute of Engineers, 35 K Stiffness modulus number in the Duncan-Chang model
(5): 601616. doi:10.1080/02533839.2012.679115. K0 coefficient of earth pressure at rest
Hsieh, P.G., Lin, Y.L., and Ou, C.Y. 2010. Three-dimensional Kcw equivalent spring stiffness per unit depth per unit
numerical analysis and performance of deep excavation with cross length of the diaphragm wall
walls and buttress walls. Journal of the Chinese Institute of Civil Keq,d equivalent stiffness of springs for the section of the
and Hydraulic Engineering, 22: 1122. [In Chinese.] diaphragm wall with cross walls at a distance of d per
Jaky, J. 1944. The coefficient of earth pressure at rest. Journal of the unit depth per unit length of the diaphragm wall
Society of Hungarian Architects and Engineers, 78: 355358. Kfeb,d equivalent spring stiffness per unit depth per unit
Kung, T.C., Juang, C.H., Hsiao, C.L., and Hashash, Y.M.A. 2007. length of the diaphragm wall for a fixed end beam for
Simplified model for wall deflection and ground surface settlement the section at a distance of d
caused by braced excavation in clays. Journal of Geotechnical and Kfeb,mid equivalent stiffness of springs for the section of the
Geoenvironmental Engineering, 133(6): 731747. doi:10.1061/ fixed-end beam at the midpoint of two cross walls per
unit depth per unit length of the diaphragm wall
(ASCE)1090-0241(2007)133:6(731).
Kst stiffness per unit width of temporary steel struts or
Lim, A., Ou, C.Y., and Hsieh, P.G. 2010. Evaluation of clay
floor slabs
constitutive models for analysis of deep excavation under Kur unloadingreloading stiffness modulus number in the
undrained conditions. Journal of GeoEngineering, 5(1): 920. DuncanChang model
Lin, Y.L., Ou, C.Y., and Hsieh, P.G. 2012. Three dimensional L cross wall interval,
numerical analysis of deep excavations with cross walls. Depart- Lcw length of the cross wall
ment of Construction Engineering, National Taiwan University of md, nd degradation parameters
Science and Technology, Taipei, Taiwan. Research Report n stiffness modulus exponent in the Duncan-Chang
GT2012001. model
Liu, G.B., Ng, C.W.W.M., and Wang, Z.W. 2005. Observed N standard penetration number
performance of a deep multistrutted excavation in Shanghai soft Rf failure ratio
clays. Journal of Geotechnical and Geoenvironmental Engineer- su undrained shear strength
ing, 131(8): 10041013. doi:10.1061/(ASCE)1090-0241(2005) tcw thickness of the cross wall
131:8(1004). wn earth pressure per unit depth
Ou, C.Y., and Hsieh, P.G. 2011. A simplified method for predicting gt total unit weight of soil
dh,0 deflection of the diaphragm wall at the cross wall
ground settlement profiles induced by excavation in soft clay.
dh,d deflection of the diaphragm wall at a distance, d, to the
Computers and Geotechnics, 38(8): 987997. doi:10.1016/j. cross wall
compgeo.2011.06.008. dhm maximum wall deflection
Ou, C.Y., and Lai, C.H. 1994. Finite element analysis of deep dh,mid deflection of the diaphragm wall at the midpoint of two
excavation in layered sandy and clayey soil deposits. Canadian cross walls
Geotechnical Journal, 31(2): 204214. doi:10.1139/t94-026. Ddd deflection of the fixed-end beam at any distance of d
Ou, C.Y., Hsieh, P.G., and Chiou, D.C. 1993. Characteristics of Ddmid deflection of the fixed-end beam at the midpoint
ground surface settlement during excavation. Canadian Geotech- n Poissons ratio
nical Journal, 30(5): 758767. doi:10.1139/t93-068. s v0 effective overburden stress
Ou, C.Y., Lin, Y.L., and Hsieh, P.G. 2006. Case record of an f0 effective friction angle

Published by NRC Research Press

Das könnte Ihnen auch gefallen