A numerical study on ground displacement and stress

during and after the installation of deep circular

diaphragm walls and soil excavation
Yasushi Arai
a,
*
, Osamu Kusakabe
b
, Osamu Murata
c
, Shinji Konishi
d
a
Railway Technical Research Institute, Structure Technology Division, 2-8-38, Hikari-cho, Kokubunji-shi, 185-8540, Tokyo, Japan
b
Department of Civil & Environmental Engineering, Tokyo Institute of Technology, Japan
c
Marketing and Business Development Division, Railway Technical Research Institute, Japan
d
Railway Technology Promotion Division, Railway Technical Research Institute, Japan
Received 31 May 2006; received in revised form 5 November 2007; accepted 5 November 2007
Available online 21 February 2008
Abstract
Three-dimensional total stress elasto-plastic FEM analysis was conducted to examine ground movement and stress after the instal-
lation of circular diaphragm walls and soil excavation within the walls.
Combinations involving three dierent wall thicknesses and two dierent excavation sequences within the wall were adopted to inves-
tigate dierences in the nal ground movement and the lateral stress in the ground after wall installation and excavation within the wall
in multi-layered ground.
The analysis results showed that the construction sequence signicantly aects ground displacement and lateral stress behind the wall,
implying that the situation even before excavation within the wall is no longer axisymmetric.
Reducing the wall thickness also reduced the vertical and circumferential section forces of the wall after excavation within it.
2007 Elsevier Ltd. All rights reserved.
Keywords: Circular shaft; Excavation; Diaphragm wall; Installation eects; Three-dimensional FEM analysis
1. Introduction
Any large-scale excavation requires retaining structures
(such as diaphragm walls) to be installed before soil exca-
vation is commenced. Although the inuence of excavation
work on the surrounding ground and on existing structures
has been commonly evaluated through numerical simula-
tion, the main interest focuses on the inuence of the exca-
vation process on these areas after retaining structures are
installed in the soil. In other words, numerical simulation
of the excavation process has generally been carried out
under the assumption that the construction of retaining
structures prior to excavation will not aect in situ stress
conditions. The corresponding numerical technique is
sometimes referred to as wished-in-place, where the wall is
placed without any change to the in situ stress (e.g. De
Moor and Stevenson [1]). This common assumption, how-
ever, has been questioned in practice as the inuences of
wall installation (referred to as installation eects) are
thought to aect the subsequent behavior of the wall and
the nal overall condition.
The process of constructing diaphragm walls is rather
complicated. The wall normally consists of a series of pan-
els whose full construction sequence includes excavation
under bentonite slurry followed by concreting and
hardening.
Gunn and Clayton [2] and Kutmen [3] stressed the
importance of the eects of installing the retaining wall.
Gunn et al. [4] performed two-dimensional FEM analysis,
0266-352X/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2007.11.001
*
Corresponding author. Tel.: +81 42 573 7266; fax: +81 42 573 7248.
E-mail address: araton@rtri.or.jp (Y. Arai).
www.elsevier.com/locate/compgeo
Available online at www.sciencedirect.com
Computers and Geotechnics 35 (2008) 791807
modeling the full construction sequence of the diaphragm
wall followed by concreting and subsequent hardening.
De Moor [5] carried out two-dimensional FEM analysis
series on a plan (horizontal) section through a series of wall
panels for a given depth. Ng et al. [6] conducted pseudo
three-dimensional FEM analysis on the eects of dia-
phragm wall installation to examine load transfer mecha-
nisms, horizontal arching and vertical load transfer. Ng
and Yan [7] performed three-dimensional nite dierence
analysis of a single-diaphragm wall panel construction of
15 m deep, 8 m long and 0.6 m wide, and later extended
this to three-panel construction (Ng and Yan [8]). Gourve-
nec and Powrie [9] published the results of three-dimen-
sional FEM analysis of a straight diaphragm wall
consisting of several wall panels and measuring 15 m deep,
2.57.5 m long and 1 m wide contained in sti clay.
All the methods of analysis outlined above, however,
model the wall installation stage but not the subsequent
soil excavation stage. Furthermore, the notional prototype
of a diaphragm wall in the previous analysis involved a
long, straight wall of moderate depth installed in sti clay.
In recent years, the construction of axisymmetric vertical
shafts for shield tunnels and large-scale circular under-
ground facilities as deep as 70-100 m has increased (Goto
et al. [10], Ariizumi et al. [11]). Muramatsu and Abe [12]
carried out eld measurements focusing on the peripheral
ground deformation when a diaphragm wall was installed
and when soil within the wall was excavated.
Little attention has been paid to examining the eects
that deep circular diaphragm wall construction has on
ground conditions prior to the excavation process in prac-
tical design. Analysis covering the complete process is
required in order to examine the eects of the construction
sequence on ground conditions and stress distribution in
the wall, including the eects of installation.
In this research, we performed three-dimensional total
stress elasto-plastic FEM analysis to examine ground
movement and stress in the ground after installation of
circular diaphragm walls and soil excavation within the
walls. We selected a notional prototype for this analysis
to model a previous deep circular diaphragm wall con-
struction for an underground river project in Tokyo,
with 32 m diameter, 85.5 m wall length and 1.2 m wall
thickness. The ground conditions selected are those typi-
cally encountered in the Tokyo Bay area. Two factors
considered are the thickness of the diaphragm wall and
the type of soil excavation within the wall, and we dis-
cussed the results to examine the eect due to construc-
tion sequence.
In the foregoing project, earth and water pressures, as
well as stresses in the diaphragm wall are being monitored
to verify and improve the current design calculations.
The monitoring data has attracted the attention of those
designing the future Central Linear Shinkansen and Out-
Ring expressway projects, and this study will provide fur-
ther insight into the properties of circular diaphragm wall
construction.
2. Numerical modeling of diaphragm wall construction and
soil excavation within the wall
2.1. Aims of this research
Numerical modeling of the full construction sequence
for diaphragm walls and subsequent soil excavation
involves at least three numerical modeling phases. The rst
phase is a modeling process to construct a single wall panel,
including excavation under bentonite slurry, followed by
concreting and hardening. The second phase involves
installing the other wall panels and joining several single
panels to form a complete diaphragm wall structure. The
third phase is the main soil excavation in front of or within
the wall.
2.2. Notional prototype
The notional prototype for the wall adopted in this
study involves a circular diaphragm wall constructed in
typical ground conditions in the Tokyo Bay area, where
a layer of soft, normally consolidated clay is underlaid by
a sti sand layer on base mud rock.
The dimensions and soil parameters used in the analysis
were selected on the basis of previous experience and the
current design practice for axisymmetric shafts in Japan.
The notional reference prototype has a circular shaft with
an internal diameter of 32 m, a wall thickness of 1.2 m
and an embedded depth of 85.5 m. The depths of the clay
and sand layers are 25.6 m and 40.8 m respectively, and
the layer below the sand layer is base mud rock. The under-
ground water level is assumed to be the surface ground
level referred to in previous cases of construction.
In this analysis, the circular wall was composed of
twelve initial panels and twelve subsequent panels, and
the circular wall was modeled as a polygonal shape with
48 sides. The initial panels consist of three gutters while
the subsequent ones are comprised of a single gutter, mean-
ing that the circular diaphragm wall consists of 48 gutters
in total. Each gutter has a length of 2.09 m, a depth of
85.5 m and a typical thickness of 1.2 m. Japanese practice
is to construct the subsequent panels (which join the initial
panels to form a diaphragm wall) using an eective con-
crete cutting method. This results in a joint so ecient at
transferring circumferential forces that the circumferential
stiness of the wall (both axial and bending) can be
assumed to match the vertical stiness, referring to Japa-
nese Standard Design Code [13]. Consequently, such an
assumption has been made in the examples of analysis pre-
sented in this paper, and the wall was modeled with isotro-
pic stiness properties.
2.3. FEM analysis
2.3.1. Outline of FEM model
Three-dimensional numerical simulations were carried
out using a commercially available FEM analysis code
792 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
named MARC. Fig. 1a shows the FEM mesh with a vol-
ume of 200 m in depth and 400 m 400 m in area, and
the circular shaft with an internal diameter of 32 m is
located in the center of the analysis zone (see Fig. 1b and
c). Such a large zone was selected to avoid any measurable
eects from the boundary in the nal results. In this
research, we use hexahedral elements for three-dimensional
FEM. These elements have eight nodes and eight integra-
tion points, and the Gaussian integration method was
applied to them. The analyzed zone has 23,775 nodes and
21,075 elements in total. The displacement of nodes along
the side boundaries is xed in the X-direction, while the dis-
placement of those along the bottom boundary is xed in
the Z-direction. The clay and sand were assumed to behave
as an elasto-perfectly plastic body with the MohrCou-
lomb failure criterion, obeying the associated ow rule.
The base mud rock was assumed to be a linear elastic body.
The ground is divided into 14 horizontal layers whose soil
properties are listed in Table 1. These properties were
selected to represent realistic values on the basis of previ-
ous literature (Waseda University [14]).
The way in which multiple wall panels are connected is
also an important factor to consider. Large-scale circular
diaphragm walls are typically formed by rst placing a
number of initial panels with equal spacing, and then con-
necting them with subsequent panels to ll the spaces (sche-
matically illustrated in Fig. 2).
Each initial panel is commonly composed of three gut-
ters, and is constructed in four stages as shown in Fig. 3.
The rst gutter (Gutter 1) is excavated and lled with
slurry, the third gutter (Gutter 3) is formed in the same
way, and the second gutter (Gutter 2) is then used to con-
nect the three gutters. A reinforced steel cage is then
inserted, and concrete slurry is poured into it. The subse-
quent panel (consisting of a single gutter) is made in such
a way that the gutter is excavated and lled with slurry,
(a) An entire finite-element mesh
(b) A half finite-element mesh (c) Detail finite-element mesh
of the circular shaft
200m
Z Y
X
400m
400m
0.
Layer 1 G.L.0.0
Layer 2
Layer 3
Layer 4
Layer 5
Layer 6
Layer 7
Layer 8
Layer 9 G.L. 52.7
Soil excavation layer
400m
200m
Z Y
X
85.5m
32m
200m
Detailed in Fig.1(c)
Fig. 1. Three-dimensional nite-element mesh.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 793
and then a reinforced steel cage is inserted and concrete
slurry is poured into it.
2.3.2. Modeling the wall panel construction
In the numerical simulation, the construction sequence
of the initial panel is modeled in the following stages:
Stage 1 (Ben1): Soil elements in Gutter 1 are removed,
and bentonite slurry with a pressure
increasing linearly with depth is
applied along the excavated area. The
slurry has a unit weight of 11.0 kN/m
3
.
Stage 2 (Ben13): Soil elements in Gutter 3 are removed,
and the slurry pressure is applied
along the excavated area.
Stage 3 (Ben123): Soil elements in Gutter 2 are removed,
and the slurry pressure is applied
along the excavated area.
Stage 4 (Con123): The slurry pressure is removed, and
elastic elements are placed in the exca-
vated area to model the solid concrete
wall. The self-weight of a unit weighing
24.0 kN/m
3
with a Youngs modulus of
27 MN/m
2
and a Poisson ratio of 0.2 is
then dened in the elastic elements.
From Stage 3 (Ben123) to Stage 4 (Con123), the pres-
sure along the excavated wall may vary with time. Lings,
Table 1
Soil properties
Soil type (index) Layer index Depth (G.L.- m) E (MN/m
2
) Poisson ratio C (kN/m
2
) / (deg)
Clay (C) C1 0.05.1 4.9 0.48 19.6 0
C2 5.110.2 6.1 0.48 24.2 0
C3 10.215.3 7.2 0.48 28.8 0
C4 15.320.4 8.4 0.48 33.4 0
C5 20.425.6 9.5 0.48 38.0 0
Sand (s) S1 25.630.7 174.1 0.30 30.0 35
S2 30.735.8 190.4 0.30 30.0 35
S3 35.842.9 212.1 0.30 30.0 35
S4 42.952.7 240.8 0.30 30.0 35
S5 52.766.4 278.2 0.30 30.0 35
Base mud rock (B) B1 66.485.5 400.0 0.48
B2 85.5112.0 400.0 0.48
B3 112.0148.8 400.0 0.48
B4 148.8200.0 400.0 0.48
Initial panel Subsequent panel
Thinner circle: Three construction areas
85.5m
32m
I
I I
I I I
Fig. 2. Dimensions of the deep circular diaphragm wall (not to scale).
Gutter 1 Gutter 2 Gutter 3
Stage1 Stage2 Stage3 Stage4
Gutter 1
Bentonite (Ben1)
Gutters 1, 3
Bentonite (Ben13)
Gutters 1, 2, 3
Bentonite (Ben123)
Gutters 1, 2, 3
Concrete (Con123)
Fig. 3. Modeling construction sequence of the single initial panel (not to scale).
794 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
Ng and Nash [15] adopted bi-linear distribution with
depth, concrete slurry pressure at the upper part and
bentonite slurry pressure at the lower part using the con-
cept of critical height suggested in CIRIA report 108 (Clear
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
1
4
8
35
34
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1 5
1 4
1 31 2
1 1
1 0
9
8
7
6
5
4
3
2
1
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
39
38
37 36
S-9
I
-
1
1
S
-
8
S
-
7
S
-
6
S
-
5
S
- 4
S - 3
S
-
2
S
-
1
S
-
1
2
S
-
1
1
S
-1
0
I-10
I-9
I
-
8
I
-
7
I
-
6
I
-
5
I - 4
I - 3
I
-
2
I
-
1
I
-
1
2
(Legends) I: Initial Panel, S: Subsequent Panel
(Step F: Initial panels complete Step L: Diaphragm wall complete)
Step A (Serial Step 4) Step B (Serial Step 5) Step C (Serial Step 6)

I
II
III
Step D (Serial Step 10) Step E (Serial Step 11) Step F (Serial Step 12)
Step G (Serial Step 16) Step H (Serial Step 17) Step I (Serial Step 18)
Step J (Serial Step 22) Step K (Serial Step 23) Step L (Serial Step 24)
Fig. 4. Installation of panels at selected steps of the diaphragm wall.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 795
and Harrison [16]). In the present simulation, this method
was not adopted as the concept of critical height is consid-
ered applicable only up to a depth of 30 m. From experi-
ence gained in Japan, the critical height may also change
with the pouring rate of the concrete slurry, the slump
value and the temperature when concreting is carried out
(Arii [17], JSCE [18]).
The construction sequence of the subsequent panels is
modeled only in Stages 1 (Ben1) and 4 (Con123). In Stage
4, the boundaries between the concrete elements and the
surrounding soil elements are assumed to be continuous.
Fig. 4 shows the status of panel installation at selected
steps for discussion of the FEM analysis. The diaphragm
wall of this shaft is composed of 12 initial panels and 12
subsequent panels, meaning that 24 steps are required to
complete the structure. In determining the installation
sequence of each panel, the area of 360 was divided into
the three areas of I, II and III as shown in Figs. 2 and 4.
0.
(a) Type A (b) Type B
1 2
2 1 3
1
4
2
Fig. 5. Types of soil excavation within the diaphragm wall.
2
5
.
6
m
C
l
a
y
(
C
)
4
0
.
8
m
S
a
n
d
(
S
)
1
9
.
1
m
B
a
s
e

m
u
d

r
o
c
k
(
B
)
8
5
.
5
m
0.3m or 0.6m or 1.2m
Initial panel
Subsequent panel
G
u
t
t
e
r
3
G
u
t
t
e
r
2
G
u
t
t
e
r
1
Layer B1(G.L.-75.95m)
Selected element
Layer S3(G.L.-39.35m)
Selected element
Layer C3(G.L.-12.75m)
Selected element
2
5
.
6
m
C
l
a
y
(
C
)
4
0
.
8
m
S
a
n
d
(
S
)
1
9
.
1
m
B
a
s
e

m
u
d

r
o
c
k
(
B
)
8
5
.
5
m
0.3m or 0.6m or 1.2m
Initial panel
Subsequent panel
G
u
t
t
e
r
3
G
u
t
t
e
r
2
G
u
t
t
e
r
1
Layer B1(G.L.-75.95m)
Selected element
Layer S3(G.L.-39.35m)
Selected element
Layer C3(G.L.-12.75m)
Selected element
Fig. 6. Geometry of the initial and subsequent panels and selected elements for lateral stress examination (not to scale).
Table 2
Cases analyzed
Diaphragm wall thickness (index) Soil excavation type
(index)
Case
index
1.2 m (t12) Type A t12-A
Type B t12-B
0.6 m (t06) Type A t06-A
Type B t06-B
0.3 m (t03) Type A t03-A
Type B t03-B
796 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
In the rst phase, the initial panels were installed in a clock-
wise direction maintaining an interval of 120 for each one.
In the second phase, subsequent panels were installed in a
similar way to avoid concentrating the installation eects
Lateral stress : kN/m
2
0
10
20
30
40
50
60
70
80
90
0 250 500 750 1000 1250 1500
D
e
p
t
h

:

m
(c) Behind Gutter 3
Cl ay
Sand
Base mud rock
Lateral stress : kN/m
2
0
10
20
30
40
50
60
70
80
90
0 250 500 750 1000 1250 1500
D
e
p
t
h

:

m
Ben1
Ben13
Ben123
Con123
Initial total horizontal stress
Ben1
Ben13
Ben123
Con123
Initial total horizontal stress
(a) Behind Gutter 2
Lateral stress : kN/m
2
0
10
20
30
40
50
60
70
80
90
0 250 500 750 1000 1250 1500
Ben1
Ben13
Ben123
Con123
Initial total horizontal stress
(b) Behind Gutter 1
Clay
Sand
Base mud rock
Clay
Sand
Base mud rock
D
e
p
t
h

:

m
Toe
Toe
Toe
Fig. 7. Lateral total stress distribution with depth for various stages of initial panel construction: eect of position of gutters.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 797
of each panel locally as in actual diaphragm wall
construction.
2.3.3. Modeling soil excavation within the diaphragm wall
Once the diaphragm wall is installed in the ground to an
embedded depth of 85.5 m, soil excavation within the wall
is carried out up to a depth of 52.7 m in nine layers as illus-
trated in Fig. 1c. The numerical simulation of the excava-
tion process is to remove the elements of each layer. The
type of soil excavation may aect the nal condition of
the surrounding soil, and two possible types were consid-
ered. Type A involves excavating a quarter of each layer
in a clockwise direction, while in Type B two opposite
quarters of each layer are excavated at the same time, as
illustrated in Fig. 5.
2.3.4. Cases analyzed
Analysis was carried out for the two major possible con-
trolling factors of wall thickness (1.2 m, 0.6 m, 0.3 m) and
the type of excavation within the wall (Types A and B).
These three dierent wall thicknesses and two dierent
excavation types (totaling six cases) were examined, as
summarized in Table 2.
3. Results and discussion
3.1. Lateral stress changes during initial panel construction
One of the issues in terms of the eects of the installation
process is its inuence on lateral stress distribution along
the wall panel depth (as previously examined by Ng and
Normalized distance from the edge of Gutter 2 by embedded depth
0.6
0.8
1.0
1.2
1.4
1.6
1.8
0.0 0.5 1.0 1.5 2.0
L
a
t
e
r
a
l

s
t
r
e
s
s

r
a
t
i
o
Layer C3-Ben123
Layer C3-Con123
Layer S3-Ben123
Layer S3-Con123
Layer B1-Con123
Layer B1-Ben123
Fig. 9. Lateral stress at Layers C3, S3 and B1 normalized by the initial total horizontal stress (lateral stress ratio) with distance from behind Gutter 2
towards the outside: eect of construction sequence.
Earth pressure coefficient
0
10
20
30
40
50
60
70
80
90
100
0.4 0.6 0.8
D
e
p
t
h

:

m
t03-Ben123
t03-Con123
t06-Ben123
t06-Con123
t12-Ben123
t12-Con123
Initial total
horizontal stress
Clay
Sand
Base mud rock
1
Toe
Fig. 8. Eect of initial panel construction on lateral stress in terms of earth pressure coecient behind Gutter 2.
798 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
Yan [8] and Gourvenec and Powrie [9]). The soil stresses
and those of the diaphragm wall are the mean values of
the stresses at each integration point in the selected ele-
ments. Fig. 6 shows detailed element meshes around an ini-
tial panel. Lateral stress was examined on three selected
elements behind gutters at each ground layer (Layers C3,
S3 and B1). Fig. 6 shows the geometry of the initial and
subsequent panels in question, and Fig. 7 shows the total
lateral stress distribution with depth at various stages of
initial panel construction behind each gutter. The gure
shows that along Gutter 2 the lateral stress at Layers C3
and B1 is less than the initial total horizontal stress, while
the lateral stress at Layer S3 is greater than the initial total
horizontal stress. Incidentally, before executing the analysis
of the diaphragm wall installation, vertical stress r
V
corre-
sponding to the force of gravity was set at each integration
point of the nite element of the soil, and horizontal stress
r
h
corresponding to the Poisson ratio was set at the same
integration point. In this research, the horizontal stress
was assumed to be the initial lateral stress, and this analyt-
ical process was dened as the initial stress FEM analysis.
When the stress distribution patterns along the three
gutters are compared, it is clear that those along Gutters
1 and 3 are more or less identical, while the pattern along
Horizontal displacement : mm
0
10
20
30
40
50
60
70
80
90
-30 -25 -20 -15 -10 -5 5 10
D
e
p
t
h

:

m
Lv2-t03-Ben123
Lv2-t03-Con123
Lv2-t12-Ben123
Lv2-t12-Con123
Clay
Sand
Base mud rock
(+)
(-)
Lv2
0
Toe
Fig. 11. Change in horizontal displacement distribution with depth during installation of the rst initial panel: eect of wall thickness and construction
sequence.
Horizontal displacement : mm
0
10
20
30
40
50
60
70
80
90
-30 -25 -20 -15 -10 -5 5 10
D
e
p
t
h

:

m
Lv1-t12-Ben123
Lv1-t12-Con123
Lv2-t12-Ben123
Lv2-t12-Con123
Lv3-t12-Ben123
Lv3-t12-Con123
Lv4-t12-Ben123
Lv4-t12-Con123
Clay
Sand
Base mud rock
(+)
(-)
Lv4
Lv3
Lv2
Lv1
0
Toe
Fig. 10. Change in horizontal displacement distribution with depth during installation of the rst initial panel: eect of construction sequence.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 799
Gutter 2 (located in the middle) diers, particularly above
the toe of the wall panel. This implies that stress redistribu-
tion and concentration occurs behind Gutter 2 during the
construction process from Ben13 to Ben123. Specically,
lateral stress decreases slightly in the construction process
from Ben1 to Con123 in the upper clay layer, and suddenly
increases in the construction stage from Ben13 to Ben123 in
the sand layer. Fig. 8 summarizes the eects of initial panel
construction on lateral stress in terms of the earth pressure
coecient (lateral total stress/vertical total stress) along
Gutter 2. The reason for choosing Gutter 2 is that the
major inuence of the initial panel construction process
appears behind the middle gutter of the wall panel. Initial
panel construction causes a decrease in lateral stress in
the clay layer and an increase in the sand layer, and these
trends become more marked when the panel wall is thicker,
as seen in Fig. 8. It should be noted here that the lateral
stress distribution in multi-layered ground is dierent from
that in single-layer ground.
It is also important to identify the zone that is aected
by wall panel installation. Fig. 9 plots the lateral stress at
Layers C3, S3 and B1 normalized by the initial total hori-
zontal stress (i.e. the lateral stress ratio) with distance from
the outer surface of Gutter 2. The lateral stress ratio varies
with the soil layer; the change is less marked during con-
creting (from Ben123 to Con123), as shown in Fig. 9.
The lateral stress ratios at Layers C3 and B1 decrease from
1.0 to 0.65 and 0.80 respectively, and the ratio at Layer S3
increases from 1.0 to 1.7 towards the outer surface of Gut-
ter 2 at the time of wall panel installation. These lateral
stress ratios change rapidly with distance at about 0.2D
(where D is the depth of the wall panel) from the outer
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
L
a
t
e
r
a
l

s
t
r
e
s
s

r
a
t
i
o
(t12)
Behind Gutter 16 in Layer C3
Behind Gutter 17 in Layer C3
Behind Gutter 18 in Layer C3
Behind Gutter 19 in Layer C3
Behind Gutter 20 in Layer C3
In i A B C D E F G H I J K L
Installation of panels at selected steps
Fig. 13. Increase and decrease of lateral stress behind existing panel at Layer C3 during the whole process of panel construction.
Normalized distance from the outer surface of gutter by embedded depth
-10
-8
-6
-4
-2
0
2
4
0.0 0.2 0.4 0.6 0.8 1.0 1.2
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

:

m
m
Lh1-t03-Ben123
Lh1-t03-Con123
Lh1-t12-Ben123
Lh1-t12-Con123
Lh2-t03-Ben123
Lh2-t03-Con123
Lh2-t12-Ben123
Lh2-t12-Con123
(+)
(-)
Lh1
Lh2
(+)
(-)
(+)
(-)
Lh1
Lh2
(+)
(-)
Lh1
Lh2
(-) (-) (-)
Fig. 12. Dierence in the vertical displacement prole on the ground surface along Lh1 and Lh2: eect of wall thickness and construction sequence.
800 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
surface of Gutter 2 for Layers C3, B1 and S3. This is con-
sistent with the previous results obtained by Ng and Yan
[8] showing that the zone of rapid change is within 0.2D.
The distance required to recover the initial stress condition
in the present analysis is about 0.5D for Layers C3 and B1,
and 0.2D for S3. These results are not always the same as
those found by Ng and Yan [8], which show that the dis-
tance is about 1.0D. This nding shows that the previous
results of two-dimensional analysis may not be wholly
applicable to circular diaphragm walls in multi-layered
ground.
3.2. Horizontal and vertical displacement proles during
initial panel installation
There must be horizontal and vertical displacement in
the ground associated with changes in stress near the
panels. The selected positions of horizontal and vertical
Horizontal displacement : mm
0
10
20
30
40
50
60
70
80
90
-35 -30 -25 -20 -15 -10 -5 0 10
D
e
p
t
h

:

m
(t03)
Lv19-Initial panels complete
Lv19-Diaphragm wall complete
Lv20-Initial panels complete
Lv20-Diaphragm wall complete
(t12)
Lv19-Initial panels complete
Lv19-Diaphragm wall complete
Lv20-Initial panels complete
Lv20-Diaphragm wall completet
Clay
Sand
Base mud rock
(+)
(-)
Lv20
Lv19
5
Toe
Fig. 15. Horizontal displacement proles with depth along Lv19 and Lv20 during the construction of initial and subsequent panels: eect of wall
thickness.
I
-
5
1 3
1 5
1 4
I - 4
1
6
S
- 4
I
-
6
2
4
2
3
2
2
2
1
2
0
S
-
5
S
-
6
I
-
5
1 3
1 5
1 4
I - 4
1
6
S
- 4
I
-
6
2
4
S
-
5
S
-
6
2
0
I
-
5
I - 4
1
6
S
- 4
I
-
6
2
4
2
0
S
-
5
S
-
6
Step A Step B Step D
I
-
5
I - 4
1
6
S
- 4
I
-
6
2
4
S
-
5
S
-
6
I
-
5
I - 4
S
- 4
I
-
6
2
4
S
-
5
S
-
6
(Legends)
I
S
: Newly installed panel
: Already installed panel
Step G Step J
: Subsequent Panel
: Initial Panel
Fig. 14. Increase and decrease of lateral stress behind existing panel at Layer C3 during adjacent panel construction.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 801
displacement for the initial panel are indicated as explan-
atory notes on each related gure (see Figs. 1012). Lv1,
Lv2, Lv3 and Lv4 are the lines for the horizontal dis-
placement prole behind the panel, while Lh1 and Lh2
are for the vertical displacement proles. Figs. 10 and
11 show the change in horizontal displacement distribu-
tion with depth during installation of the initial panel.
Fig. 11 shows the eect of wall panel thickness on hori-
zontal displacement distribution with depth, particularly
in the clay layer. Corresponding to the lateral stress dis-
tribution shown in Figs. 7 and 8, the soil in the clay
layer moves inward, whereas that in the sand layer
moves outward.
The magnitude of horizontal displacement is signi-
cantly greater along Lv2 and Lv3 and smaller along Lv1
and Lv4, regardless of the layers. Increased wall thickness
requires wider excavation, causing larger horizontal dis-
placement in the clay layer as seen in Fig. 11.
A clear dierence in the vertical displacement prole
on the ground surface along Lh1 and Lh2 is also seen
in Fig. 12, implying that settlement may be observed
behind the middle of the wall panel, and a slight heave
may occur behind the outer surface of the wall panel.
In particular, Fig. 12 indicates the eects of wall panel
thickness on the vertical displacement distribution with
distance.
Normalized distance from the outer surface of panel by embedded depth
-30
-25
-20
-15
-10
-5
0
0.0 0.5 1.0 1.5 2.0
H
o
r
i
z
o
n
t
a
l

d
i
s
p
l
a
c
e
m
e
n
t

:

m
m
(t12)
Lh19-Initial panels complete
Lh19-Diaphragm wall complete
Lh20-Initial panels complete
Lh20-Diaphragm wall complete
(+)
(-)
Lh19
Lh20
Fig. 17. Horizontal displacement prole on the ground surface during the construction of initial and subsequent panels.
Normalized distance from the outer surface of panel by embedded depth
-10
-8
-6
-4
-2
0
2
4
6
0.0 0.5 1.0 1.5 2.0
V
e
r
t
i
c
a
l

d
i
s
p
l
a
c
e
m
e
n
t

:

m
m
(t03)
Lh19-Initial panels complete
Lh19-Diaphragm wall complete
Lh20-Initial panels complete
Lh20-Diaphragm wall complete
(t12)
Lh19-Initial panels complete
Lh19-Diaphragm wall complete
Lh20-Initial panels complete
Lh20-Diaphragmwallcomplete
(+)
(-)
Lh19
Lh20
+
(-)
(+)
(-)
Lh19
Lh20
(-) (-)
( )
(-)
Fig. 16. Vertical displacement proles on the ground surface along Lh19 and Lh20 during the construction of initial and subsequent panels: eect of wall
thickness.
802 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
It is of practical importance to know the zone aected
by the installation of the wall panel. In this analysis, the
inuence of diaphragm wall construction ceases at a dis-
tance approximately equal to one wall-panel depth
(1.0D), which is consistent with the results of the centrifuge
model tests reported by Powrie and Kantartzi [19]. The
results obtained by Gourvenec and Powrie [9] indicated
that avoiding the inuence of wall installation requires a
distance of up to 40 m from the wall face. This corresponds
to more than 2.0D, which is much less than the results
obtained from two-dimensional plane strain nite element
analysis.
Table 3
Ratios of inward lateral stress to outward lateral stress
Layer index t03-B t06-B t12-B
Maximum Minimum Maximum Minimum Maximum Minimum
C3 1.13 0.90 1.19 0.83 1.26 0.56
S3 1.03 0.98 1.01 0.95 1.00 0.93
Lateral stress : kN/m
2
0
10
20
30
40
50
60
70
80
90
-1500 -1250 -1000 -750 -500 -250 0
D
e
p
t
h

:

m
t03 - Minimum
t03 - Maximum
t06 - Minimum
t06 - Maximum
t12 - Minimum
t12 - Maximum
Initial total horizontal stress
(a) Inward
Lateral stress : kN/m
2
0
10
20
30
40
50
60
70
80
90
0 250 500 750 1000 1250 1500
D
e
p
t
h

:

m
t03 - Minimum
t03 - Maximum
t06 - Minimum
t06 -Maximum
t12 - Minimum
t12 - Maximum
Initial total horizontal stress
(b) Outward
Clay
Sand
Base mud rock
Clay
Sand
Base mud rock
Toe
Toe
Fig. 18. Maximum and minimum lateral stress distribution inward and outward from the wall with depth after completion of diaphragm wall: eect of
wall thickness.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 803
3.3. Lateral stress distribution and horizontal and vertical
displacement proles during the installation of initial and
subsequent panels
Fig. 13 demonstrates how the lateral stress ratios behind
Gutters 1620 uctuate in the upper clay layer (Layer C3)
during the whole process of wall panel construction. The
reason for this uctuation is considered to be the local for-
mation of arching and stress transfer during the process of
construction, as is schematically illustrated in Fig. 14. In
the construction of initial panel No. 5, Gutter 17 is exca-
vated rst, followed by Gutters 19 and 18. At Step A indi-
cated in Fig. 14, the lateral stress ratio behind Gutter 18
decreases from 1 to 0.6, while the ratio behind Gutters 17
and 19 increases from 1 to 1.1. At this stage there is no
inuence from lateral stress in neighboring areas (the lat-
eral stress ratio is 1 at the location of Gutters 16 and 20).
At Step B, initial panel No. 6 is constructed, resulting in
an increase in the lateral stress ratio only behind Gutter
19, the nearest gutter. Similarly, at Step G, subsequent
panel No. 5 is constructed, resulting in an increase of the
lateral stress ratio only behind Gutter 20, the nearest gut-
ter. In the nal step (Step L), the lateral stress ratio nally
becomes 1.2 behind Gutter 17, 1.1 behind Gutters 19 and
20, 0.9 behind Gutter 16 and 0.6 behind Gutter 18.
The displacement at initial panel No. 5 shown in Figs. 4
and 14 is to be examined. The locations of the line of inter-
est in this consideration are indicated as explanatory notes
on each related gure (see Figs. 1517). Fig. 15 shows hor-
izontal displacement proles with depth along Lv19 and
Lateral stress : k/m
2
0
10
20
30
40
50
60
70
80
90
0 250 500 750 1000 1250 1500
D
e
p
t
h

:

m
t12-A-Maximum
t12-B-Maximum
t12-A-Minimum
t12-B-Minimum
Initial total horizontal stress
before soil excavation
Clay
Sand
Base mud rock
Fig. 20. Maximum and minimum lateral stress acting on the outside wall face during the soil excavation stage: eect of soil excavation type.
Lateral stress : kN/m
2
0
10
20
30
40
50
60
70
80
90
0 250 500 750 1000 1250 1500
D
e
p
t
h

:

m
t03-B
t06-B
t12-B
Initial total horizontal stress
Clay
Sand
Base mud rock
before soil excavation
Toe
Fig. 19. Lateral stress distribution behind the wall with depth at the center of initial panel No. 5 after soil excavation: eect of wall thickness.
804 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
Lv20 during initial and subsequent panel construction for
various wall thicknesses.
The general trends at Lv20 on the outer surface of wall
panel No. 5 are similar to those of Lv4 (see Fig. 10), and
the trends at Lv19 in the middle are similar to those of
Lv3 (see Fig. 10). The main dierences in the displacement
prole are seen in the clay layer and in the increased hori-
zontal displacement with thicker walls. Fig. 16 shows the
vertical displacement prole with various wall thicknesses
at the ground surface level. The prole is largely dependent
on the relative location of the wall panel. Settlement is
observed at Lh19, and heave is seen at Lh20. It is interest-
ing to note that the magnitude of the dierential displace-
ment between Lh19 and Lh20 remains the same for
dierent wall thicknesses, suggesting that wider wall exca-
vation leads to larger heave, compensating for the poten-
tially large settlement.
Muramatsu et al. [12] reported the results of eld mea-
surements for displacement during circular shaft construc-
tion. Although the ground and construction conditions
were dierent from those used in this study, comparison
of the order of magnitude may be useful for verication
of the present study. The thickness of the soft ground
was 8 m from the surface in the eld, while the present
study assumed a thickness of 25 m from the surface. The
predicted vertical displacement was 2.0 mm (see Fig. 16
at Lh20 (t12)) compared to the corresponding eld value
of 3.0 mm.
The prediction in this study for horizontal displacement
at 5.5 m from the outer surface of the wall at the time of
diaphragm wall installation was 15.0 mm (see Fig. 17 at
Lh20), which is comparable to 5.0 mm of horizontal dis-
placement at the corresponding location in the eld
measurement.
3.4. Lateral stress distribution and horizontal displacement
distribution after the completion of diaphragm walls
The values of maximum and minimum lateral total
stress distribution with depth inward and outward for the
circular diaphragm wall are presented for various wall
thicknesses in Fig. 18. The value of the initial total horizon-
tal stress with depth is also plotted in the gure, and is
found to be almost the average of the maximum and min-
imum lateral stress levels both inward and outward. The
Horizontal displacement : mm
0
10
20
30
40
50
60
70
80
90
-20 -15 -10 -5 5
D
e
p
t
h

:

m
(t12-Diaphragm wall complete)
x1
y1
x2
y2
(t12-B-Soil excavation complete)
x1
y1
x2
y2
Clay
Sand
Base mud rock
x2 x1
y1
y2
0
Toe
Fig. 22. Change in horizontal displacement distribution with depth along x1, x2, y1 and y2 at the time when diaphragm wall construction stage and soil
excavation stage were completed.
Normalized internal displacement by shaft diameter
0
10
20
30
40
50
60
70
80
90
-0.0006 -0.0004 -0.0002 0.0000
D
e
p
t
h

:

m
Lx-t03-B
Lx-t06-B
Lx-t12-B
Ly-t03-B
Ly-t06-B
Ly-t12-B
Clay
Sand
Base mud rock
Ly
Lx
Toe
Fig. 21. Internal displacement divided by shaft diameter with depth at Lx
and Ly after soil excavation: eect of wall thickness.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 805
inuence of the wall thickness is noticed in the clay layer.
The thicker the wall is, the greater the dierence between
the maximum and minimum lateral stress values.
Table 3 summarizes the range of ratios of lateral stress
on the internal wall face to the lateral stress on the outer
wall face for various wall thicknesses at the clay and sand
layer. It can be seen that the thinner the wall, the smaller
the range of the ratio.
3.5. Lateral stress changes during and after the soil
excavation stage
No literature is available relating to the numerical study
of installation eects, including the installation stage and
the subsequent soil excavation stage. Fig. 19 shows the lat-
eral stress distribution with depth for various wall thick-
nesses behind the center of initial panel No. 5, indicating
that the value of earth pressure diers considerably from
that of the initial total horizontal stress, and that the eect
of the wall thickness is as small as 20 kN/m
2
. The maxi-
mum and minimum lateral stresses acting on the outer wall
face during the soil excavation stage for a wall thickness of
1.2 m are plotted with depth for the two types of soil exca-
vation in Fig. 20. It is noted that the type of soil excavation
has no eect on the lateral stress distribution for either the
maximum or minimum values. The initial total horizontal
stress of each soil layer is also close to the maximum value
for the soft clay layer, and gives approximately the mean
value of the maximum and minimum values for sti-layer
ground. The type of excavation is also predicted to have
no eect on the internal horizontal displacement prole
with depth.
Fig. 21 shows the internal horizontal wall displacement
divided by the shaft diameter with depth at the lines of Lx
and Ly indicated as explanatory notes on the gure after
soil excavation for various wall thicknesses. The magnitude
of the maximum inward movement is in approximate
inverse proportion to the wall thickness.
Fig. 22 shows the changes in horizontal displacement
distribution with depth at x1, x2, y1 and y2 indicated as
explanatory notes on the gure at the times when the dia-
phragm wall construction stage and soil excavation stage
were completed. In these cases, during the construction
stage the horizontal displacement gradually accumulates
towards the inside of the wall in the clay layer and the base
mud rock layer, whereas horizontal displacement in the
sand layer occurs in the opposite direction. However, once
the excavation stage starts, horizontal displacement only
occurs toward the inside of the wall at all depths. The mag-
nitude of the accumulated displacement in the clay layer is
in the order of y2, x2, y1 and x1, clearly indicating that the
magnitude of such displacement depends on the construc-
tion sequence. The ratios of the displacement after dia-
phragm wall completion to the accumulated displacement
in clay are 0.15 at x1 and 0.1 at x2, x3 and x4.
Corresponding to the wall deection, the vertical bend-
ing moment changes with depth. Fig. 23 shows the vertical
bending moment (M
v
) distribution with depth for various
wall thicknesses, and indicates that thinner walls have a
smaller bending moment, although the overall pattern of
the bending moment distribution remains similar.
4. Conclusions
The processes of installing a circular diaphragm wall
and excavating soil within the wall were analyzed as
three-dimensional events. The main ndings on lateral
stress in the ground and ground movements, examined by
three-dimensional FEM analysis, are as follows:
(1) Numerical evaluation using FEM analysis was suc-
cessfully carried out to investigate the eects of
M
v
: kNm/m
0
10
20
30
40
50
60
70
80
90
-50 50 100 150
D
e
p
t
h

:

m
(t03)
Gutter 5
Gutter 6
Gutter 7
(t12)
Gutter 5
Gutter 6
Gutter 7
Clay
Sand
Base mud rock
0
Fig. 23. Vertical bending moment distribution with depth at initial panels No. 2: eect of wall thickness.
806 Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807
installing circular diaphragm walls, including the
detailed installation stage and the subsequent soil
excavation stage. The realistic multi-layered ground
used has a more complicated lateral stress distribu-
tion. Results from single-layer ground may not have
provided an overall picture of this complexity.
(2) The results of the analysis showed that the situation
even prior to excavation within the wall is no longer
axisymmetric. This non-uniform characteristic with
periodic uctuations was formed at an early stage
of the initial panel construction.
(3) The inuence of the type of soil excavation within the
wall may be negligible.
(4) The maximum value of lateral stress after excavation
within the wall was greater than the initial total hor-
izontal stress, and the minimum was smaller than the
initial total horizontal stress. The mean of the maxi-
mum and minimum values was found to be close to
the initial total horizontal stress.
(5) Reducing the wall thickness resulted in a decrease in
the vertical section forces of the wall after excavation
within it.
Acknowledgements
The authors are grateful to Dr. Shikai Du for his help
with nite element analysis using MARC, and to Dr. Mo-
toi Iwanami and Keiji Oishi for their help with the outline
of the construction sequence for the circular shaft.
References
[1] De Moor EK, Stevenson MC. Evaluation of the performance of a
multi-propped diaphragm wall during construction. In: Mair, Taylor,
editors. Proc. int. symposium on geotechnical aspects of underground
construction in soft ground. Balkema; 1996. p. 1116.
[2] Gunn MJ, Clayton CRI. Installation eects and their importance in
the design of earth-retaining structures. Geotechnique
1992;42(1):13741.
[3] Kutmen G. The inuence of the construction process on bored piles
and diaphragm walls: a numerical study, MPhil thesis, University of
Surrey; 1986.
[4] Gunn MJ, Satkunananthan A, Clayton CRI. Finite element modeling
of installation eects. Retaining structures. London: Thomas Telford;
1993. p. 4655.
[5] De Moor EK. An analysis of bored pile/diaphragm wall installation
eects. Geotechnique 1994;44(2):3417.
[6] Ng CWW, Lings ML, Simpson B, Nash DFT. An approximate
analysis of the three-dimensional eects of diaphragm wall installa-
tion. Geotechnique 1995;45(3):497507.
[7] Ng CWW, Yan RWM. Stress transfer and deformation mechanisms
around a diaphragm wall panel. J Geotech Geoenviron Eng ASCE
1998;124(7):63848.
[8] Ng CWW, Yan RWM. Three-dimensional modeling of a diaphragm
wall construction sequence. Geotechnique 1999;49(6):82534.
[9] Gourvenec SM, Powrie W. Three-dimensional nite-element analysis
of diaphragm wall installation. Geotechnique 1999;49(6):80123.
[10] Goto S, Muramatsu M, Sueoka T, Saka F, Yabe H, Watanabe H,
et al. Ground movement earth and water pressures due to shaft
excavations. In: Fujita, Kusakabe, editors. Proc int symposium on
underground construction in soft ground. Balkema; 1995. p. 1514.
[11] Ariizumi K, Kumagai T, Kashiwagi A. Behaviour of large-scale
cylindrical earth retaining wall. In: Kusakabe, Fujita, Miyazaki,
editors. Proc int symposium on geotechnical aspects of underground
construction in soft ground. Balkema; 2000. p. 4816.
[12] Muramatsu M, Abe Y. Considerations in shaft excavation and
peripheral ground deformation. In: Mair, Taylor, editors. Proc int
symposium on geotechnical aspects of underground construction in
soft ground. Balkema; 1996. p. 1738.
[13] Japan Society of Civil Engineers Tunnel Committee, Standard
Specications for Tunnel [Cut and Cover Tunnel Version], JSCE;
2006. p. 1179 [In Japanese].
[14] Advanced Research Institute for Science and Engineering, Waseda
University, specic proposals for deep-level infrastructure for Central
Tokyo 1995; October [In Japanese].
[15] Lings ML, Ng CWW, Nash DFT. The lateral pressure of wet
concrete in diaphragm wall panels cast under bentonite. In: Proc
institution of civil engineers, geotechnical engineering, no. 107, July;
1994. p. 16372.
[16] Harrison TA, Clear CA. Concrete pressure on formwork, Construc-
tion Industry Research and Information Association, Report 108,
London; 1985.
[17] Arii T. Experimental research of lateral pressure of concrete acting on
the diaphragm wall. In: The 18th Japan national conference on soil
mechanics and foundation engineering; 1983. p. 12256 [In Japanese].
[18] Japan Society of Civil Engineers Concrete Committee, Standard
Specications for Concrete [Structures Materials and Construction
Version]; 1998. p. 1113 [In Japanese].
[19] Powrie W, Kantartzi C. Ground response during diaphragm wall
installation in clay: centrifuge model test. Geotechnique 1996;46(4):
72539.
Y. Arai et al. / Computers and Geotechnics 35 (2008) 791807 807