Acta Geotechnica

DOI 10.1007/s11440-015-0416-6

RESEARCH PAPER

Efficiency of excavations with buttress walls in reducing

the deflection of the diaphragm wall
Pio-Go Hsieh1 • Chang-Yu Ou2 • Wei-Han Hsieh2

Received: 18 January 2015 / Accepted: 13 September 2015

Ó Springer-Verlag Berlin Heidelberg 2015

Abstract Installation of buttress walls against diaphragm has fully been mobilized prior to the final excavation depth,
walls has been used as an alternative measure for the the efficiency of buttress walls in reducing the wall
protection of adjacent buildings during excavation, but deflection in a deep excavation was much less than that in a
their mechanism in reducing movements has not yet been shallow excavation. Rectangular shape of buttress walls
fully understood. This study performs three-dimensional was of a better effect than T-shape in the shallow exca-
finite element analyses of two excavation case histories, vation because frictional resistance between buttress walls
one in clay with T-shape buttress walls and another in and adjacent soil played a major role in reducing the wall
dominant sand with rectangular buttress walls, to establish deflection rather than bearing resistance of the flange.
analysis model. Then, a series of parametric study were When the excavation went deeper, the difference in
performed by varying soil types, types and length of but- reducing the wall deflection between the R-shape and
tress walls based on the above-mentioned excavations. T-shape became small.
Results show that the mechanism of buttress walls in
reducing wall deflections mainly came from the frictional Keywords Buttress wall  Deep excavation  Shallow
resistance between the side surface of buttress wall and excavation  Wall deflection
adjacent soil rather than from the combined bending stiff-
ness from diaphragm and buttress walls. The buttress wall
with a length\2.0 m had a poor effect in reducing the wall 1 Introduction
deflection because the soil adjacent to the buttress wall had
almost the same amount of movement as the buttress wall, Deep excavation may induce excessive wall deflections
causing the frictional resistance little mobilized. Since the and ground settlements and thus damage adjacent build-
frictional resistance of buttress walls in a deep excavation ings. To avoid the damage of adjacent buildings during
excavation, it is necessary to adopt effective measures to
limit the wall deflection or ground settlement. Ground
& Chang-Yu Ou improvement is a common measure to reduce the excava-
[email protected] tion-induced ground movements [8, 18, 25, 29]. Recently,
Pio-Go Hsieh cross walls, constructed perpendicularly and connected to
[email protected] the two opposite diaphragm wall, have been widely used in
Wei-Han Hsieh some countries as an alternative measure. The authors have
[email protected] shown that use of cross walls in deep excavations can
1
reduce the wall deflection to a very small amount [11, 23,
Department of Assets and Property Management, Hwa Hsia
24]. However, use of cross walls in a very wide excavation
University of Technology, 111, Gongzhuan Rd., Zhonghe
Dist., New Taipei City 23568, Taiwan would be costly. Therefore, buttress walls have been
2 adopted recently as an alternative to cross walls for the
Department of Civil and Construction Engineering, National
Taiwan University of Science and Technology, 43, Sec. 4, protection of adjacent buildings during deep excavation [4,
Keelung Rd, Taipei 10672, Taiwan 9, 10, 12, 13].

123
 Acta Geotechnica

The basic configuration of a buttress wall is depicted in dimensional finite element method. The buttress walls in
Fig. 1. A buttress wall is similar to a cross wall in terms of both cases were demolished along with excavation. The
construction. It is a concrete wall perpendicular to the rationality of analysis results was validated by the com-
diaphragm wall constructed before excavation, but not parison between the monitoring data and analysis results.
connected to the opposite diaphragm wall. Conceptually, a Characteristics of wall deflection and development of lat-
buttress wall may provide frictional resistance developing eral resistance from buttress walls were further studied
on the two sides of the buttress wall and thus increase the using the three-dimensional finite element method by
overall lateral resistance against the movement of the varying the length, shape and soil properties in these two
diaphragm wall during excavation. It may also function cases.
like a T-beam as that in reinforced concrete structures,
enhancing the capability of moment resistance of the dia-
phragm wall. No matter which mechanism of buttress walls 2 Finite element analysis and constitutive model
exists in deep excavations, the wall movement is expected
to reduce to a certain extent. A three-dimensional finite element computer program,
According to the experiences of practicing engineers, PLAXIS 3D [26], was used as a basic analysis tool. The
buttress walls seem to have a good effect in reducing the Hardening Soil model [28], referred to as the HS model,
movement of diaphragm walls though their effectiveness was adopted for analysis. In addition to the strength
and mechanism are not yet verified [10]. Buttress walls are parameters (c0 , /0 ), the HS model requires the other seven
thus gradually popular in reducing excavation-induced parameters, namely w, Eref ref ref
50 , Eoed, Eur , m, Rf and mur, to
movements in some Asian countries in recent years. describe the stress–strain behavior of soil. Table 1 lists the
Moreover, although buttress walls have been applied in parameters, their definition and the way of evaluating their
many deep excavations, research on their behavior is rather values in this study. As shown in this table, the way of
deficient. Only a few studies on buttress walls have been determining the stiffness parameters of sandy and clayey
found in the literature. For example, Hsieh and Lu [9] soils was different. According to the relations listed in
introduced a preliminary design method for a buttress wall, Table 1, the unloading/reloading referential stiffness (Erefur )
and Hwang et al. [13] investigated the behavior of the of clayey soil was evaluated first based on the initial void
buttress wall. Chen et al. [4] examined the influence of the ratio (e) and the swelling index (Cs) that were little influ-
geometry of the buttress walls (shape, thickness, and enced by the possible disturbance of soil sampling process.
length) on the displacement of buttressed diaphragm wall Then estimate the referential secant stiffness (Eref 50 ) fol-
by performing three-dimensional finite element analysis. lowed by the evaluation of the referential oedometer
The mechanism of buttress walls in reducing the defor- stiffness (Eref
oed). For sandy soil, the Young’s modulus (Es)
mation of diaphragm wall remains resolved. of in situ soil was estimated first according to the rela-
In this paper, two excavation cases with buttress walls, tionship of Es & (2000 * 3000)N that was obtained by
one with T-shape buttress walls in clay and the other with back-analyzing excavation cases in sandy soil, where N is
rectangular shape buttress walls, referred to as R-shape, in the standard penetration number [15]. Then estimate the
predominating sandy soil, were analyzed using the three- Eref ref
50 followed by the evaluation of the Eoed and Eur
ref

Fig. 1 Schematic diagram of the buttress wall a plan; b A–A section

123
Acta Geotechnica

Table 1 Parameters of the HS model

Physical implication Method of evaluation References

w The angle of dilatancy w=/0 - 30° Bolton [2]

Eref
50 The referential secant stiffness in primary loading Eref ref
50 = Eur /3 for clay Calvello and Finno [3]
    m 
ref
E50 ¼ Es 3 r03 pref for sand Khoiri and Ou [15]
Eref
oed The referential tangent referential stiffness in Eref
oed = 0.7Eref
50 for clay Calvello and Finno [3],
primary oedometer loading Eref
oed = 1.5Eref
50 for sand Khoiri and Ou [15]
Eref
ur The referential unloading/reloading stiffness ref
Eur ¼ 3ð1þeÞpref ð12vur Þ
for clay Lim et al. [17]
Cs =ln 10
Khoiri and Ou [15]
Eref
ur = 3Eref
50 for sand

m The power for stress-level dependency of stiffness m = 1.0 for clay Schanz et al. [28]
m = 0.5 for sand
Rf The ratio of the asymptotic strength to the failure strength 0.9 Duncan and Chang [7],
Schanz et al. [28]
mur The Poisson’s ratio in the unloading–reloading state 0.2 –
ref
The referential pressure p = 100 kPa

according to the correlations listed in Table 1. Moreover, it is subject to a large deflection [22], which is also an
the silty gravel soil exists in the Taipei Basin around 45 m overall effect of assumption of linear elastic behavior of
below the ground surface and its stiffness parameters, material. The temporary struts were with axial elements and
back-analyzed by Deng [6] through back analysis, that the also simulated as linear elastic material. The axial stiffness
Eref ref ref
ur , E50 and Eoed were about equal to 256, 85 and 85 MPa, of the struts was reduced by about 50 %, considering pos-
respectively. sible defects of alignment of struts in the field [22].
Installation of diaphragm wall and buttress wall and The interaction between the wall and soil was simulated
their construction quality may affect the stress redistribu- with interface elements, whose behavior follows the Mohr–
tion of the soil [5, 19, 27], and numerical analysis may not Coulomb model. In this study, the friction angles of
exactly simulate all the installation effects. For simplifying interface elements were set the same as soil.
analysis, the diaphragm wall was assumed to be wished in It should be noted that the parameters of soils and
place and the soil before excavation was in the at-rest state structural members in the following case studies were
but the weight of the concrete from D-wall and buttress evaluated preliminarily according to the above description.
wall over the existing soil was applied to the soil. The Except for the K0 values, the parameters were also cali-
coefficient of the at-rest earth pressure for sandy soil (K0) brated at the early stages of excavation based on the
can be obtained by the equation K0 = 1 - sin/0 [14]. The monitored wall deflections, especially those correlations
K0,OC value for clay can be obtained according to Ladd given in a range.
et al. [16] as
K0;OC ¼ K0;NC ðOCRÞa ð1Þ
3 Case studies
where K0,OC is the at-rest earth pressure coefficient at the
overconsolidated state, and K0,NC is that at the normally 3.1 The Park-2001 project
consolidated state, OCR is the overconsolidated ratio, and
a is an empirical coefficient and can be approximated by The Park-2001 project was a 44 by 42 m excavation
a = sin/0 . located in Taipei. The diaphragm wall was 21 m in depth
The structural members such as diaphragm walls, but- and 0.6 m in thickness. In order to protect a gas station near
tress walls and concrete floor slabs employed in the top- the excavation, three T-shape buttress walls, with the
down construction method were with plate elements and thickness (tbw) of 0.6 m, the flange length (Lf) of 2.5 m and
simulated as linear elastic material. The Poisson’s ratio for web length (Lw) of 5 m, were constructed from the 2.0 m
concrete is set equal to 0.15. The Young’s modulus of below the ground surface level (GL -2.0 m) to GL
pffiffiffiffi
concrete was estimated using the equation Ec ¼ 4700 fc0 , -22.0 m against the diaphragm wall as shown in Fig. 2a,
where fc0 is the compressive strength of concrete (MPa) [1]. b. Four excavation stages were conducted to reach the final
In analysis, the Ec of concrete was reduced by 20 %, con- depth of 8.6 m. Three levels of struts were employed to
sidering that cracks may appear in the diaphragm wall when support the diaphragm wall, in which the first strut level

123
 Acta Geotechnica

Fig. 2 Park-2001 excavation project a excavation geometry and instrumentation; b types of buttress wall; c profile of the subsurface soil and
construction sequence

was H330 9 330 9 10 9 15 while the second and third from 1 to 4, the water content ranges from 30 to 50 %, the
strut levels were H400 9 400 9 13 9 21. The average liquid limit ranges from 37 to 50, and the plastic limit
horizontal spacing between struts was 6.0 m in the east– ranges from 21 to 26. The silty sand locates from GL -20
west direction and 5.5 m in the north–south direction. to GL -26 m with the total unit weight ranging from 17.36
Figure 2c shows the construction sequence of excavation. to 19.52 kN/m3, and the SPT-N about 11. In addition, the
As shown in Fig. 2c, the silty clay locates from ground ground water level is at 0.5 m below the ground surface.
surface level to GL -20 m, in which the total unit weight Two inclinometers, SI4 and SI6, were installed to observe
ranges from 17.0 to 18.04 kN/m3, SPT-N value ranges wall deflection during excavation, as shown in Fig. 2a.

123
Acta Geotechnica

Figure 3 shows the finite element mesh used for analysis For evaluating the effectiveness of buttress walls in
where the fine mesh as defined by PLAXIS was used, as excavations quantitatively, the ratio of reduction in the
suggested by Ou et al. [21]. The depth of the model was set maximum wall deflection (MR) is defined as
at 26 m where hard rock exists. The horizontal boundaries dhm;nobw  dh;bw
were located at the distance two times the final excavation MR ¼  100 % ð2Þ
dhm;nobw
depth from the diaphragm wall as suggested by Ou and
Shiau [20]. where dhm,nobw is the maximum deflection of the dia-
The parameters of the sand and clay at the construction phragm wall with the assumption of no buttress walls
site were determined in a way as shown in Table 1 and as installed, dh,bw is the deflection of the diaphragm wall, with
the description in the preceding sections. Furthermore, the buttress walls, at a depth corresponding to the maximum
parameters of the soils and structure were calibrated at the deflection of diaphragm wall without buttress walls.
earlier stages of excavation, and therefore, the lower bound As shown in Fig. 4, the MR value was 45.5 % at the
of the correlation between the Es and SPT-N for sandy soil final excavation stage. Installation of buttress walls can
was adopted for analysis. Table 2 summarizes the param- reduce the maximum wall deflection significantly.
eters of the soils used in the analysis where the clayey soil
and sandy soil were assumed to be undrained and drained 3.2 The Jinshan south road project
material, respectively. The analysis followed the con-
struction procedure as shown in Fig. 2c. The Jinshan south road project was an 18-story building
Figure 4 plots a comparison of the computed and with 7-level basements locating in Taipei. As shown in
monitored wall deflections at SI4, near the location of a Fig. 6a, the excavation was 64 m in length and 43 m in
buttress wall. It can be observed that the computed wall width, and surrounded by five buildings. The diaphragm
deflections were in general close to the monitored values wall was 43 m in depth and 1.3 m in thickness, penetrating
but slightly overestimated at the last two stages. For 3.9 m into gravel layer. Three types of R-shape buttress
evaluating the effectiveness of buttress walls in reducing walls with different tbw and length (Lr) were installed as
the wall deflection, analysis of the excavation with shown in Fig. 7, and their allocations are shown in Fig. 6a.
assumption of no buttress walls was performed and the The basement was constructed with the top-down con-
results are also shown in Fig. 4. For global evaluation of struction method up to 26.45 m in depth. To monitor the
the efficiency of buttress walls, the wall deflection at dif- wall deflection, seven inclinometers were installed pene-
ferent sections along the diaphragm wall at a depth of final trating 5 m into gravel layer. Among them, inclinometer
excavation bottom, usually close to the maximum value, is SI1 was damaged at the beginning of excavation so their
shown in Fig. 5. It can be found that the wall deflections data were excluded in this study. Inclinometers SI3 and SI4
were reduced significantly due to the installation of buttress remained functional until stage 8, and the monitored values
walls. The computed maximum wall deflections with but- at all stages except for stage 8 were employed for study.
tress wall and without buttress walls were 76.1 and Figure 6b shows that the subsoil is comprised of seven
41.5 mm, respectively. alternating layers of clayey and sandy soils, among which
the second and fourth layers of soils mainly affect the wall
deflection. These layers are medium dense silty sand (SM)
with friction angle about 30°. The SPT-N value, on the
other hand, ranges from 5 to 17 for the second layer and
from 22 to 27 for the fourth layer.
As shown in Fig. 6b, the soil was excavated, along with
the demolishment of buttress walls, to GL -2.5 m, referred
to as stage 1. The concrete floor slab at the ground level
(1FL) was then installed, and the soil and buttress walls
were excavated and demolished, respectively, down to GL
-5.45 m, referred to as stage 2. The procedure repeated
until B5FL was installed, and soil and buttress walls were
excavated and demolished, respectively, down to GL
-20.2 m (stage 7). After completion of stage 7, the con-
crete floor slab was constructed at GL -18.60 m (B6FL)
and the soil was excavated to the final depth, GL
Fig. 3 Finite element mesh used for the Park-2001 excavation -26.45 m, but the buttress walls stayed at the same depth,
project no demolishment, at this stage, referred to as stage 8. At

123
 Acta Geotechnica

Table 2 Parameters of the HS model used in the Park-2001 project

Depth (m) Soil type Drainage type ct (kN/m3) c0 (kPa) /0 (°) w (°) Eref
ur (kPa) Eref
50 (kPa) Eref
oed (kPa) m mur

0–20 CL Undrained 17.00–18.04 0 24 0 17,832–23,518 5944–7839 4161–5488 1.0 0.2

20–26 SM Drained 17.36–19.52 0 30 0 22,012–24,101 7337–8034 11,006–12,050 0.5 0.2

Fig. 5 Wall deflection at different sections along the diaphragm wall

at a depth of the excavation bottom for the Park-2001 excavation
project

calibrated in a way similar to the preceding case. Table 3

summarizes the soil parameters used in this study.
Figure 8 shows the finite element mesh for the Jinshan
south road project. The depth of mesh was set at GL
-54.9 m, considering that hard rock would not deform
during excavation. The horizontal boundaries were located
at the distance two times the final excavation depth from
Fig. 4 Comparison of monitored and computed wall deflections at the diaphragm wall. The finite element model was there-
SI4 at all stages for the Park-2001 excavation project fore 169.8 m in length and 148.8 m in width. The analysis
followed the construction procedure as shown in Fig. 6b,
stage 9, the final stage, a 0.9-m-thick mat foundation was including the weight of surrounding buildings and the
cast at the final depth, i.e., GL -26.45 m, and then the depth of foundation, up to stage 8 because of no monitored
buttress walls were demolished down to the final depth and data afterward.
the concrete floor slab (B7FL) was built. It should be noted Figure 9 shows a comparison of monitored and com-
that the concrete floor slabs 1FL, B1FL and B7FL were puted wall deflection at stage 8. For evaluation of the
0.25 m in thickness while the rest of slabs 0.4 m. effectiveness of installation of buttress walls, analysis of
In finite element analysis, sandy and clayey soils were the excavation with assumption of no buttress walls
treated as drained and undrained material, respectively. installed was also performed and the results are shown in
The void ratio and swelling index of the soils were the same figure. Moreover, the wall deflections at different
obtained directly from laboratory tests. Similar to the Park- sections along the diaphragm wall at a depth of final
2001 project, stiffness parameters of the gravel soil, excavation bottom, usually close to the maximum value,
locating below GL -39.6 m, were: Eref ur = 256 MPa, are shown in Fig. 10. Despite no monitored wall deflec-
Eref ref
50 = 85 MPa and Eoed = 85 MPa. The parameters of the tions at SI3 and SI4 at stage 8, the monitored data before
soils were determined in a way as shown in Table 1 and stage 8 were still used for validation and the computed wall

123
Acta Geotechnica

Fig. 6 Jinshan south road project a excavation geometry and instrumentation; b profile of the subsurface soil and construction sequence

The agreement for SI3 and SI4 was also quite well from
beginning up to stage 7 though their comparison was not
shown here. The analysis validates the constitutive model
adopted and analysis procedure in this study.
Figure 9 also shows that the wall deflections at SI2, SI3,
SI4 and SI6 were slightly smaller than those without but-
tress walls. Their MR values were equal to 17.5, 17.4, 18.2
and 17.0 %, respectively. Installation of buttress walls had
some effects in reducing the wall deflection. The wall
deflections at SI5 and SI7 with buttress walls were moder-
ately smaller than that without buttress walls. Their MR
values were 26.9 and 23.9 %, respectively. Installation of
buttress walls had a moderate effect in reducing the wall
deflection.

Fig. 7 Types of R-shape buttress walls used in the Jinshan south road 4 Factors affecting the deflection of diaphragm
project a type A; b type B; c type C wall with buttress walls

deflections at stage 8 were also used for studying the The above-mentioned two case histories were with differ-
effectiveness of installation of buttress walls. ent excavation depths, soil types, types and length of but-
The computed wall deflections generally agreed with the tress walls, as summarized in Table 4. As indexed by the
monitored data for SI2, SI5, SI6 and SI7 at all excavation MR values, the Park-2001 project obviously had a better
stages, and the comparison at stage 8 is shown in Fig. 9. effect in reducing the wall deflection. Does it imply that

123
 Acta Geotechnica

Table 3 Parameters of the HS model used in the Jinshan south road project
Depth (m) Soil type Drainage type ct (kN/m3) c0 (kPa) /0 (°) w (°) Eref
ur (kPa) Eref
50 (kPa) Eref
oed (kPa) m mur

0–5 CL Undrained 18.44 0 28 0 19,794–21,660 6598–7220 4619–5054 1.0 0.2

5–15.9 SM Drained 19.42 0 30 0 15,543–40,573 5181–13,524 7771–20,287 0.5 0.2
15.9–19.5 CL/ML Undrained 18.74 0 30 0 26,116–26,807 8705–8936 6094–6255 1.0 0.2
19.5–27.7 SM Drained 19.03 0 31 1 43,880–48,621 14,627–16,207 21,940–24,311 0.5 0.2
27.7–36.8 CL/ML Undrained 18.93 0 30 0 36,480–44,356 12,160–14,785 8512–10,350 1.0 0.2
36.8–39.6 SM Drained 19.23 0 32 2 43,864 14,621 21,932 0.5 0.2
39.6–54.9 GM Drained 22.07 0 38 8 256,000 85,000 85,000 0.5 0.2

Fig. 8 Finite element mesh used for the Jinshan south road excava-
tion project

buttress walls in the shallow excavations, and/or with

dominating clay layer and/or with T-shape had a better
restraining effect than those in the deep excavation, and/or
with dominating sandy soil layer and/or with R-shape?
Fig. 9 Comparison of monitored and computed wall deflections at
To clarify the queries, a series of three-dimensional
stage 8 for the Jinshan south road excavation project
finite element analyses were performed where the exca-
vation geometry and construction sequence of the Park-
2001 project and the Jinshan south road project, denoted as
‘‘shallow excavation’’ and ‘‘deep excavation’’, respec-
tively, were used as a basis, combined with different soil
types, length and types of buttress walls. The soil profile as
shown in Fig. 2b was treated as the ‘‘clay’’ type of the
subsoil in the shallow excavation, while the soil profile as
illustrated in Fig. 6b was regarded as the ‘‘sand’’ type of
subsoil in the deep excavation. In addition, Fig. 11a, b
shows the ‘‘sand’’ types of the subsurface soil profile in the
shallow excavation and the ‘‘clay’’ type in the deep exca-
vation, respectively. The construction of the shallow
excavation was exactly the same as shown in Fig. 2b, while Fig. 10 Wall deflection at different sections along the diaphragm
the construction sequence of the deep excavation was wall at a depth of the excavation bottom for the Jinshan south road
similar to that shown in Fig. 6b except for stage 8 where excavation project

123
Acta Geotechnica

Table 4 Comparison of excavation condition, buttress wall type and 4.1 Mechanism of buttress walls in the reduction
dimension, and wall deflection between the Park-2001 and Jinshan in wall deflection
south road excavation projects
Case Park-2001 Jinshan Two scenarios were assumed. One was exactly the same as
the Park-2001 project that was a shallow excavation in clay
Excavation depth (m) 8.6 26.45
with T-shape buttress walls with Lw = 5 m and
Excavation type Shallow excavation Deep excavation
Lf = 2.5 m. The other was similar to the Jinshan south
Soil layer Clayey soil Sandy soil
road project but with R-shape buttress walls with
Buttress wall type T-shape R-shape
Lr = 5.6 m. In each scenario, an analysis was performed
Lr (m) – 2.5–7
with the assumption of no frictional resistance between the
Lw, Lf (m) 5, 2.5 –
side surface of web of buttress walls and the adjacent soil.
tbw (m) 0.6 0.6
Figure 12a shows the computed wall deflections for the
dhm;bw (mm) 41.5–40.6 13.8–38.5
deep excavation in sand with R-shape buttress walls.
dhm;nobw (mm) 76.1–79.0 16.7–52.7 Results show that with consideration of frictional resis-
MR (%) 45.5–48.6 17.0–26.9 tance between the buttress walls and the adjacent soil, the
computed wall deflection was less than that without but-
tress walls. The computed wall deflection for frictionless
case was almost the same as that without buttress walls.
This implies that the main mechanism of R-shape buttress
walls in reducing the wall deflection was due to the fric-
tional resistance between buttress walls and the adjacent
soil. The combined bending stiffness of the diaphragm wall
from the diaphragm wall and the R-shape buttress wall
plays insignificant role in the reduction in the wall
deflection. This is because when the buttress walls were
demolished along with the removal of soil, the buttress
walls below the excavation bottom were mainly ‘‘pushed’’
by the diaphragm wall rather than providing the bending
resistance against the deformation of the diaphragm wall.
Figure 12b shows the computed wall deflections for the
shallow excavation in clay with T-shape buttress walls.

Fig. 11 Profile of subsurface soil layers a the sandy soil in the

shallow excavation; b the clayey soil layer in the deep excavation

the buttress walls from GL -20.20 to GL -26.45 m were

demolished with excavation of soil.
As referred to the dimension in Fig. 1, the T-shape but-
tress walls were with different dimensions: Lw = 5–7.5 m,
Lf = 2.5–5 m while R-shape: Lr = 5–20 m. Analysis of
each case with assumption of no buttress walls was also
performed, and the resulting wall deflection at the same
location was used for comparison.
The wall deflection and shear stress along the side sur-
face of the buttress wall at point A in Fig. 2a and SI5 in
Fig. 12 Comparison of computed wall deflections for the cases of
Fig. 6a, both locating at about central section of the
with buttress walls, without buttress walls, and with buttress walls but
excavation, were employed for studying the reduction in no friction a R-shape buttress wall in the deep excavation in sand;
wall deflection. b T-shape buttress walls in the shallow in clay

123
 Acta Geotechnica

Results show that the excavation with buttress walls was of

less wall deflections than that without. The wall deflection
for the scenario of no frictional resistance between the web
surface of the buttress walls and the adjacent soil was much
smaller than that of no buttress walls but slightly higher
than that with frictional resistance. Since Fig. 12a has
demonstrated that the combined bending stiffness from
both diaphragm wall and buttress walls plays insignificant
role in the reduction in the wall deflection, the reduction in
wall deflection for the T-shape buttress walls mainly comes
from the bearing resistance of the flange against the soil in
front of the flange. For both with and without frictional
resistance, the upper diaphragm wall, above the excavation
bottom, deformed with curvature, but the lower diaphragm
wall, below the excavation bottom, deflected with more
close to the straight type. This is because the combined
bending stiffness below the excavation bottom was suffi-
ciently large that the wall moved laterally in a straight type
rather than bending. Compared with the lower diaphragm
wall, the bending stiffness of the upper diaphragm wall was
relatively small so that it deformed with some curvature.
Similar phenomenon has been found in the deep exca-
vation in clay and shallow excavation in sand.

4.2 Efficiency of R-shape buttress walls

As studied in the preceding section, the main restraining

effect of R-shape buttress walls comes from the frictional
resistance between buttress walls and adjacent soil when
the buttress wall was demolished along with the removal of
soil. This section will further study the efficiency of
R-shape buttress walls with different lengths, in sand/clay,
in the shallow/deep excavation.
Figure 13 shows the variation of the computed wall Fig. 13 Comparison of computed wall deflections for different
deflections for the scenarios with buttress walls where lengths of R-shape buttress wall in the excavations
Lr = 5, 10, 20 m and without buttress walls, respectively.
The amount of wall deflection certainly decreased with the shear stress and relative shear stress ratio, respectively, on
increasing length of buttress walls in both shallow and deep the surface of the buttress wall in the shallow excavation in
excavations in sand or clay. Figure 14 shows the amount of clay and in sand, with the buttress wall length. Note that
reduction in terms of the MR value due to the installation due to the boundary between different soil layers, the
of buttress walls, indicating that the longer Lr, the more fluctuation of the shear stress and relative shear stress ratio
reduction in the wall deflection. The reduction rate distribution are found in Figs. 15 and 16 and the following
decreased as the buttress wall length increased. The MR similar figures. As shown in Fig. 15, the induced shear
value for the cases of buttress walls in sand had a better stress occurred at the diaphragm wall with the smallest
efficiency in reducing the wall deflection in both shallow value and increased with the increasing distance from the
and deep excavations. diaphragm wall. The srel was also smallest near the dia-
To investigate the mobilization of shear strength, the phragm wall and generally increased with the increasing
relative shear stress ratio, srel, is defined as the ratio of distance from the diaphragm wall, up to a value of 1.0. The
shear stress to the shear strength of a soil for a given srel was mostly close to 1.0 for the short length of the
effective normal stress. The srel = 1.0 implies that the soil buttress wall, e.g., Lr = 5 m, especially in clay and majorly
adjacent to the buttress wall was at the failure state, and its much \1.0 for the longer length of the buttress wall, e.g.,
shear strength or frictional resistance was fully mobilized. Lr = 20 m, especially in sand (Fig. 16). Therefore, to
Figures 15 and 16 illustrate the variation of the maximum further restrain the wall deflection, the length of buttress

123
Acta Geotechnica

wall needs to be increased in order to provide larger fric- Figure 14 also shows that the MR value in the shallow
tional resistance. Similar phenomena were also found in the excavation was larger than that in the deep excavation in
deep excavation cases. both sand and clay. The R-shape buttress wall in the
shallow excavation seems to have a better effect than that
in the deep excavation. This can be attributed to the fact
that the srel value in the deep excavation was generally
higher than that in the shallow excavation under the same
condition, the identical buttress wall length and same type
of soil. The frictional resistance between the buttress wall
and the adjacent soil was almost fully mobilized in the
deep excavation due to the large shear stress induced at the
early stage of excavation, and therefore, its srel was mostly
close to 1.0. Therefore, it was necessary to have a larger
length and greater depth of the buttress walls in the deep
excavation to obtain a better effect in reducing the wall
deflection.
Figures 15 and 16 also show a relative small driving
stress and relative shear stress ratio within the first 2.0 m
from the diaphragm wall, no matter how long the buttress
was. This phenomenon can be explained by the fact that the
Fig. 14 Variation of the MR values with the length of R-shape movement at any location of the buttress wall was all the
buttress wall same, almost the same as the movement of the diaphragm

Fig. 15 Distribution of shear stress for the cases with different lengths of R-shape buttress wall a shallow excavation in clay; b shallow
excavation in sand

123
 Acta Geotechnica

Fig. 16 Distribution of relative shear stress ratio for the cases with different lengths of R-shape buttress wall a shallow excavation in clay;
b shallow excavation in sand

wall because the buttress wall had a very high axial stiff- bearing resistance from the flange. This section will further
ness. The soil in front of the diaphragm wall, say, 2.0 m study the efficiency of T-shape buttress walls with different
from the diaphragm wall, directly pushed by the diaphragm lengths of the web and flange in sand/clay, in the shallow/
wall, should have almost the same amount of movement as deep excavation. Three combinations of web length (Lw)
the diaphragm wall or buttress wall. Therefore, the relative and flange length (Lf) were adopted for analysis: Lw = 5 m
displacement between the buttress wall and the soil within and Lf = 2.5 m, Lw = 5 m and Lf = 5 m (i.e., Lf increases
the first 2.0 m from the diaphragm wall was very small, but 2.5 m), Lw = 7.5 m (i.e., Lw increases 2.5 m) and
it increased gradually with the increasing distance from the Lf = 2.5 m.
diaphragm wall. The relative shear stress ratio was there- Figure 17 shows the computed wall deflections for the
fore very small near the diaphragm wall, and it increased above three combinations and case of without buttress
with the increasing distance from the diaphragm wall. It walls in shallow/deep excavations in sand/clay. The MR
was clear that if the buttress wall length was \2.0 m, the values corresponding to Fig. 17 are summarized as listed in
buttress wall was unable to restrain the wall deflection Table 5. Compared with the wall deflection for Lw = 5 m
although the combined bending stiffness from the contri- and Lf = 2.5 m, either increase in the web length up to
bution of the diaphragm wall and buttress wall seems 7.5 m or flange length up to 5.0 m can reduce the wall
increased (Fig. 14). deflection. Similar to the R-shape buttress wall, either the
T-shape buttress wall in sand or in the shallow excavation
4.3 Efficiency of T-shape buttress walls has a better reduction in the wall deflection than that in clay
or deep excavation.
As studied in the preceding section, the restraining effect Figure 18 shows the srel on the web surface of the
for T-shape buttress walls comes from the frictional T-shape buttress wall in the shallow excavation in clay and
resistance between the buttress walls and adjacent soil and in sand, with three combinations of web and flange lengths.

123
Acta Geotechnica

As shown in this figure, the srel was very small both near soil adjacent to the web was constrained by both the dia-
the diaphragm wall and near the flange. The former was phragm wall and the flange, and thus, the srel on the
similar to the R-shape buttress wall, and the latter was due T-shape buttress wall was generally smaller than that on
to the restraint by the flange because the movement of the the R-shape, comparing Figs. 16 and 18. The phenomenon
seems to be more obvious as the flange length increased by
comparing the srel in the case of Lw = 5 m and Lf = 2.5 m
with that of Lw = 5 m and Lf = 5 m. Though the srel or
frictional resistance on the web in the case of Lw = 5 m
and Lf = 5 m decreased, the bearing resistance of the
flange increased and the resulting overall efficiency
increased, as demonstrated in Fig. 17 and Table 5.
As observed in Fig. 18, the srel on the web surface for
the case of Lw = 7.5 m and Lf = 2.5 m was larger than
that of the case of Lw = 5 m and Lf = 2.5 m. The srel was
less influenced by the diaphragm wall and flange when the
web length increased. Therefore, increase in the web length
can provide more frictional resistance and the wall
deflection reduced as a result.
It is clear that either extension of the web length or
increase in the flange length can reduce the wall deflection.
Under a condition of identical length/area of buttress wall,
increase in the web length has a slightly better effect in
reducing the wall deflection than extension of the flange
length as shown in Fig. 17 and Table 5 by comparing the
case of Lw = 7.5 m and Lf = 2.5 m with the case of
Lw = 5 m and Lf = 5 m. The phenomenon can be
explained by the fact that the frictional resistance can be
mobilized at a relatively small displacement while the
bearing resistance required a relatively large displacement
to be mobilized. Extension of the flange length can increase
the bearing resistance that may not be fully mobilized,
while it would reduce the frictional resistance of the web.
The overall efficiency of increase in the flange length was
therefore less than increase in the web length.
Similarly, the characteristics of T-shape buttress walls in
deep excavations in sand/clay also have a similar behavior
to the shallow excavation. Figure 17 also shows the vari-
ation of wall deflections of the deep excavation in sand and
clay with different combinations of T-shape buttress walls,
Fig. 17 Comparison of the wall deflections for the cases with and their MR values are also summarized in Table 5.
T-shape buttress wall with different web and flange lengths

Table 5 Comparison of the MR values for R-shape and T-shape buttress walls
Lw, Lf (m) Lr (m) MR (%)
T-shape R-shape
Shallow-Clay Shallow-Sand Deep-Clay Deep-Sand Shallow-Clay Shallow-Sand Deep-Clay Deep-Sand

5, 2.5 – 48.6 55.2 19.9 29.2 – – – –

5, 5 – 55.1 62.8 21.8 38.4 – – – –
7.5, 2.5 – 66.5 71.3 23.1 45.0 – – – –
– 5.6 – – – – 37.0 43.9 14.0 20.1
– 7.5 – – – – 51.4 58.8 18.4 30.0

123
 Acta Geotechnica

Fig. 18 Distribution of the relative shear stress ratio of the web of Fig. 19 Comparison of the relative shear stress ratio between the web
T-shape buttress walls in the shallow excavation a in clay; b in sand of T-shape and the R-shape buttress walls for shallow excavation a in
clay; b in sand

5 Discussion
resistance though the mobilized frictional resistance of the
As studied in the previous sections, both R-shape and web was smaller than the R-shape, as shown in Fig. 19. A
T-shape buttress walls can provide additional resistance similar phenomenon has been found in the deep excavation
against the movement of the diaphragm wall. Assuming in sand/clay.
that buttress walls were all 0.6 m in thickness, the com- Table 5 also exhibits that the case of R-shape with
puted wall deflections for the cases of T-shape with Lr = 7.5 m was of the larger MR value than the case of
Lw = 5 m and Lf = 2.5 m, R-shape with Lr = 5.6 m and T-shape with Lw = 5 m and Lf = 2.5 m in the shallow
R-shape with Lr = 7.5 m are adopted for study. Among excavation in sand/clay where these two cases had the
these cases, the case of T-shape with Lw = 5 m and same length/area of buttress walls. However, in the deep
Lf = 2.5 m was of the same distance from the end of the excavation, the MR values were almost the same for both
buttress wall to the diaphragm wall as the case of R-shape R-shape and T-shape, and their values were much smaller
with Lr = 5.6 m. The case of T-shape with Lw = 5 m and than those in the shallow excavation. As shown in Fig. 19,
Lf = 2.5 m has the same total length/area as the case of the srel for the case of R-shape with Lr = 7.5 m was much
R-shape with Lr = 7.5 m. higher than the case of T-shape with Lw = 5 m and
As listed in Table 5, the case of T-shape with Lw = 5 m Lf = 2.5 m in the shallow excavation in sand/clay. With
and Lf = 2.5 m had a larger MR value than the case of the same length/area of the buttress walls, the overall
R-shape with Lr = 5.6 m for both deep and shallow mobilized frictional resistance of the R-shape buttress wall
excavations in sand/clay. This implies that with the same was higher than the sum of the mobilized frictional resis-
distance from the end of the buttress walls to the diaphragm tance of the web and bearing resistance, less mobilized, of
wall, the T-shape buttress wall had a better effect in the flange because the wall deflection was not sufficiently
reducing the wall deflection than the R-shape. This is large. The similar trend was also found in the deep exca-
because the flange also provides an additional bearing vation, but relatively large wall deflections caused the full

123
Acta Geotechnica

mobilization of the bearing resistance of the flange as well reduction in wall deflection increased with the increas-
as the full mobilization of the frictional resistance of the ing length of the buttress wall.
web, and therefore, their MR values were much smaller. 4. Though the T-shape buttress wall was with the bearing
The difference of the overall resistance between the resistance from the flange, the movement of the soil
R-shape and T-shape buttress walls thus became very adjacent to the web was constrained by both the
small, resulting in almost the same MR values. diaphragm wall and the flange, causing a smaller
Moreover, the case studies demonstrated that the Park- mobilized frictional resistance. Such a phenomenon
2001 project, a shallow excavation in clay with the T-shape increased with the increasing flange length. Increase in
buttress wall, had a larger MR value than the Jinshan south the web length had a slightly better effect in reducing
road project, a deep excavation in sand with the R-shape the wall deflection than that of the flange length.
buttress wall. It seems that the T-shape buttress wall had a 5. Under a condition of the same amount of length/area
better efficiency in reducing the wall deflection. As listed of buttress walls, the R-shape buttress wall had a better
in Table 5 and discussed before, buttress walls in the effect than the T-shape buttress wall. When the
shallow excavation had a better efficiency than that in the excavation went deeper, the difference between the
deep excavation though buttress walls in clay provide a less R-shape and T-shape became smaller.
restraining effect than that in sand. Moreover, the T-shape 6. Buttress walls in sand, no matter in shallow or deep
buttress walls in the Park-2001 project were adopted with excavations, can provide more resistance against the
Lw = 5 m and Lf = 2.5 m. Such a dimension of the movement of the diaphragm wall than that in clay.
T-shape buttress wall was equivalent to the R-shape with 7. In the deep excavation, the frictional resistance from
Lr = 7.5 m in terms of the restraining effect, as studied the buttress wall was almost fully mobilized at early or
previously. The R-shape buttress walls with Lr = 2.5 m in intermediate stages of excavation. The buttress walls
the Jinshan south road project were constructed between thus provide less resistance at the final stage. There-
GL ?0 and GL -20.2 m and with Lr = 5 m between GL fore, to effectively reduce the wall deflection in deep
-20.2 and GL -35 m. Their effects should be much excavations, longer and deeper buttress walls were
smaller than the equivalent T-shape dimension, generally recommended.
Lr = 7.5 m. Moreover, Lr = 2.5 m would cause the soil
adjacent to the buttress wall moving together with the Acknowledgments The authors acknowledge the support provided
buttress wall. The frictional resistance was not sufficiently by the Ministry of Science and Technology in Taiwan via Grant No.
mobilized, and therefore, its effect was not as good as that MOST103-2221-E-146 -004.
in the Park 2001 project.

References
6 Conclusions
1. ACI Committee 318 (1995) Building code requirements for
Based on the studies in this paper, the following conclusion structural concrete (ACI 318-95) and commentary (ACI 318R-
95). American Concrete Institute (ACI), Farmington Hills
can be drawn: 2. Bolton MD (1986) The strength and dilatancy of sands.
Geotechnique 36(1):65–78
1. When buttress walls in excavations were demolished
3. Calvello M, Finno R (2004) Selecting parameters to optimize in
along with excavation of soil, the effect of reduction in model calibration by inverse analysis. Comput Geotech
the wall deflection mainly came from the frictional 31(5):410–424
resistance between the side surface of buttress walls 4. Chen SL, Ho CT, Li CD, Gui MW (2011) Efficiency of buttress
walls in deep excavations. J GeoEng 6(3):145–156
and adjacent soil rather than from the combined
5. Comodromos EM, Papadopoulou MC, Konstantinidis GK (2013)
bending stiffness contributing from the diaphragm wall Effects from diaphragm wall installation to surrounding soil and
and buttress walls. adjacent buildings. Comput Geotech 53:106–121
2. When the short length of buttress walls was adopted, 6. Deng WB (2013) The effect of buttress walls on wall deflection
in deep excavation. National Taiwan University of Science and
for example, 2.0 m, the soil adjacent to the buttress
Technology, Master’s Thesis (in Chinese)
walls will almost have the same amount of movement 7. Duncan JM, Chang CY (1970) Nonlinear analysis of stress and
as the buttress walls, causing the frictional resistance strain in soils. J Soil Mech Found Div 96(5):637–659
between the buttress walls and adjacent soil being little 8. Gaba AR (1990) Jet grouting at Newton station. In: Proceedings
of the 10th Southeast Asia geotechnical conference, Taipei,
mobilized. Therefore, it would be unable to reduce the
pp 77–79
wall deflection effectively. 9. Hsieh HS, Lu LC (1999) A note on the analysis and design of
3. The main mechanism of the R-shape buttress wall was diaphragm wall with buttresses. Sino-Geotechnics 76:39–50 (in
due to the frictional resistance. The amount of Chinese)

123
 Acta Geotechnica

10. Hsieh HS, Wu LH, Lin TM, Cherng JC, Hsu WT (2011) Per- 20. Ou CY, Shiau BY (1998) Analysis of the corner effect on the
formance of T-shaped diaphragm wall in a large scale excavation. excavation behavior. Can Geotech J 35(3):532–540
J GeoEng 6(3):135–144 21. Ou CY, Chiou DC, Wu TS (1996) Three-dimensional finite ele-
11. Hsieh PG, Ou CY, Lin YL (2013) Three dimensional numerical ment analysis of deep excavation. J Geotech Eng 122(5):337–345
analysis of deep excavations with cross walls. Acta Geotech 22. Ou CY, Liao JT, Lin HD (1998) Performance of diaphragm wall
8(1):33–48 constructed using top-down method. J Geotech Geoenviron Eng
12. Hsieh PG, Ou CY, Lin YK, Lu FC (2015) Lessons learned in 124(9):798–808
design of an excavation with the installation of buttress walls. 23. Ou CY, Hsieh PG, Lin YL (2011) Performance of excavations
J GeoEng 10(2):25–35 with cross walls. J Geotech Geoenviron Eng 137(1):94–104
13. Hwang RN, Moh ZC, Wang CH (2007) Performance of wall 24. Ou CY, Hsieh PG, Lin YL (2013) A parametric study of lateral
systems during excavation for core pacific city. J GeoEng wall deflections in deep excavations by the installation of cross
2(2):53–60 walls. Comput Geotech 50:55–65
14. Jaky J (1944) The coefficient of earth pressure at rest. J Soc Hung 25. Parashar S, Mitchell R, Hee MW, Sanmugnathan D, Sloan E,
Archit Eng 78(22):355–358 Nicholson G (2007) Performance monitoring of deep excavation
15. Khoiri M, Ou CY (2013) Evaluation of deformation parameter at Changi WRP project. In: Proceedings of the 7th international
for deep excavation in sand through case histories. Comput symposium on field measurements in geomechanics. Singapore,
Geotech 47(1):57–67 pp 1–12
16. Ladd CC, Foott R, Ishihara K, Schlosser F, Poulos HG (1977) 26. PLAXIS 3D (2013) Computer software. PLAXIS, Delft
Stress-deformation and strength characteristics. In: Proceedings 27. Poh TY, Wong IH (1998) Effects of construction of diaphragm
of the 9th international conference on soil mechanics and foun- wall panels on adjacent ground: field trial. J Geotech Geoenviron
dation engineering, vol 2. Tokyo, pp 421–494 Eng 124(8):749–756
17. Lim A, Ou CY, Hsieh PG (2010) Evaluation of clay constitutive 28. Schanz T, Vermeer PA, Bonnier PG (1999) The hardening soil
models for analysis of deep excavation under undrained condi- model—formulation and verification. In: Brinkgreve RBJ (ed)
tions. J GeoEng 5(1):9–20 Beyond 2000 in computational geotechnics. Balkema, Rotterdam,
18. Liu GB, Ng CWW, Wang ZW (2005) Observed performance of a pp 281–296
deep multistrutted excavation in Shanghai soft clays. J Geotech 29. Wong KS, Wong IH, Broms BB (1987) Methods of improving
Geoenviron Eng 131(8):1004–1013 the stability of deep excavations in soft clay. In: Proceedings of
19. Ng CWW, Yan WM (1999) Three-dimensional modeling of a the 8th Asian regional conference on soil mechanics and foun-
diaphragm wall construction sequence. Geotechnique dation engineering. Kyoto, pp 321–324
49(6):825–834

123