See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/315794159

Numerical analysis of lateral movements and

strut forces in deep cement mixing walls with
top-down construction in soft...

Article in Computers and Geotechnics April 2017

DOI: 10.1016/j.compgeo.2017.03.018

CITATIONS READS

0 5

5 authors, including:

Pitthaya Jamsawang
King Mongkut's University of Technology North Bangkok
22 PUBLICATIONS 140 CITATIONS

SEE PROFILE

All content following this page was uploaded by Pitthaya Jamsawang on 06 April 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 Computers and Geotechnics 88 (2017) 174181

Contents lists available at ScienceDirect

Computers and Geotechnics

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

Technical Communication

Numerical analysis of lateral movements and strut forces in deep cement

mixing walls with top-down construction in soft clay
Pitthaya Jamsawang a,, Sittisak Jamnam a, Pornkasem Jongpradist b, Pornpot Tanseng c,
Suksun Horpibulsuk c
a
Soil Engineering Research Center, Department of Civil Engineering, King Mongkuts University of Technology North Bangkok, Thailand
b
Department of Civil Engineering, Faculty of Engineering, King Mongkuts University of Technology Thonburi, Bangkok, Thailand
c
School of Civil Engineering and Center of Excellence in Innovation for Sustainable Infrastructure Development, Suranaree University of Technology, Thailand

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

Article history: This article presents the observed and simulated lateral movements and strut forces induced in deep
Received 18 October 2016 cement mixing walls under deep excavation using top-down construction techniques in soft Bangkok
Received in revised form 22 March 2017 clay. The walls are supported laterally by permanent concrete slabs and temporary struts. A three-
Accepted 23 March 2017
dimensional numerical model is first calibrated with observed data from a case study. Then, a parametric
study is performed to compare this construction method with the bottom-up method and investigate the
influence of the DCM wall thickness on lateral movements and strut forces of the wall.
Keywords:
2017 Elsevier Ltd. All rights reserved.
Deep excavation
Deep mixing
Finite element
Simulation
Top-down construction in three dimensions
Wall

1. Introduction
Deep cement mixing (DCM) walls have been used for deep
excavation works in soft clays to protect adjacent properties in
many countries [1]. In a DCM wall, columns are formed by mixing
in situ soil with cement. The DCM wall cross section is necessarily
relatively thick due to its low tensile strength, and it is typically
excavated without struts using the bottom-up (BU) construction
technique. The top-down (TD) construction method is used for
deep excavations in urban areas when there are extremely strict
environmental protections, insufficient working spaces and extre-
mely short construction times. One advantage of this method is
that a basement excavation and a support from the critical path
of the project can be removed after the walls and pile foundations
are constructed and the first slab is cast in place. The slabs act as
permanent lateral braces for the wall, which are considerably stif-
fer than cross-lot braces that should minimize adjacent ground
movements typically encountered in BU construction [2]. Long
[3], Moormann [4] and Wang et al. [5] presented databases of a
large number of case histories of deep excavations through
soft soils. The walls constructed using the TD method included
contiguous pile walls and diaphragm walls, whereas the sheet pile
walls and DCM walls were constructed using the BU method. They
concluded that TD methods generally resulted in smaller lateral
wall movement. A large amount of research has been conducted
using three-dimensional (3D) finite element analysis to investigate
the lateral movements of diaphragm walls and contiguous pile
walls with the TD and BU methods in the context of deep excava-
tion in soft soils [1,613]. However, research on the lateral move-
ments and strut forces of DCM walls with TD construction for deep
excavations in soft clays has been limited.
This paper focuses on the numerical analysis of a field case
study of a DCM wall for a deep excavation with TD construction
in soft clay in Bangkok, Thailand. In the field study, lateral move-
ment profiles and strut forces were observed during the final
stages of excavation. The 3D finite element analysis incorporated
in the commercial software program PLAXIS 3D version 2013
was used for the numerical analysis. The numerical analysis simu-
lated the lateral movement behavior and axial forces of the struts.
In addition, the numerical analysis was used to investigate the
influence of the thickness of the DCM walls on lateral movements
and the strut forces for the DCM wall, and to compare the results of
the construction method with those of the BU method.

Corresponding author.
E-mail address: pitthaya_kmutnb@hotmail.com (P. Jamsawang).

http://dx.doi.org/10.1016/j.compgeo.2017.03.018
0266-352X/ 2017 Elsevier Ltd. All rights reserved.
 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181
2. Subsoil conditions and project details
The project construction site was located on Sukumvit Road in
central Bangkok, Thailand. The soil profile at this site was a
2-m-thick weathered crust underlain by an 11-m-thick soft clay,
a 3-m-thick medium stiff clay, a 1-m-thick stiff clay, and an under-
lying thick dense sand layer, as shown in Fig. 1a. The average water
table was approximately 4 m below the ground surface. The basic
and geotechnical engineering properties of the soils are summa-
rized in the soil profile presented in Fig. 1bk, including the wet
unit weight, natural water content, liquid limit and plastic limit,
soil specific gravity, initial void ratio, and undrained shear strength
obtained from unconfined compression tests. Conventional
oedometer tests and conventional triaxial tests based on consoli-
dated undrained tests were performed on foundation soil speci-
mens taken from the project site at depths of
1.5,
7,
14 and

18 m for the weathered crust, soft clay, medium stiff clay and stiff
clay, respectively, to determine the soil parameters for the numer-
ical simulations. The results of the triaxial tests indicate that the
effective friction angle varied from 23 to 28, whereas the effec-
tive cohesion varied from 2 to 30 kPa. These shear strength param-
eters are consistent with the values for the numerical modeling of
Bangkok clays reported by Jamsawang et al. [1]. The ratios of swel-
ling index to compression index obtained from the oedometer tests
were 0.130.25 for the stiff and soft clays, respectively, which falls
within the range of ratios of swelling index to compression index
for the Bangkok sub soils reported by Bergado et al. [14]. The over
consolidation ratio profiles determined by the oedometer tests
show that the weathered crust, medium stiff clay and stiff clay
were heavily over-consolidated and that the soft clay was slightly
over-consolidated.
The project was an 18-story condominium project with two
underground car parks and a maximum excavation depth of
7.90 m. The excavation was performed in the soft clay layer only.
With insufficient space to construct a thick gravity DCM wall, a
DCM wall with temporary bracing systems was designed to reduce
the wall thickness. The TD construction technique was used to
minimize the construction time, employing a permanent basement
slab for lateral support. Fig. 2a shows the layout of the DCM wall
and pile foundation and the location of the temporary struts.
Fig. 2b presents the cross section of the DCM wall. The maximum
Fig. 1. Soil profiles and soil properties.
175
excavation depth was 7.90 m on the western side of the excavation
area, which was close to a public road. On the eastern side of the
excavation area, the excavation depth was 6.30 m. There were
two basements, B1 and B2. The levels of the two basement slabs
were different; the B1-A slab was at a level of
2.90 m, and the
B1-B slab was at a level of
4.50 m. The excavation for B2 was
unequal; for the B2-A side, the excavation level was
6.3 m,
whereas the excavation level for the B2-B side was
7.90 m.
Because the construction employed the TD method, the B1-A and
B1-B basement slabs were used as the lateral supports for the
DCM wall, and they were installed before the construction of the
mat foundation. Therefore, the temporary stanchions, which were
embedded into the bored piles prior to the excavation work, were
required to support the basement slabs. The DCM wall used at this
site comprised four rows of 0.7 m diameter DCM columns with
0.1 m of overlap. The entire thickness of the DCM walls was
2.5 m, and the tip was
14 m from the ground surface. The tip of
the DCM was embedded 1 m into the medium stiff clay layer.
The DCM walls were installed using a low-pressure mechanical
mixing method. The excavation construction sequences versus
elapsed time are listed in Table 1. The construction project was
started on June 16, 2012, and completed on November 21, 2012.
Four inclinometer casings were installed up to the stiff clay
layer at a depth of 19 m. They were located at the middle of the
walls on four sides of the excavation boundary to monitor the wall
lateral movement, as shown in Fig. 2a. Electrical strain gauges were
attached to the neutral axes of the struts to avoid a bending stress
component to measure the forces in the temporary struts installed
between B1-A and B1-B. A dummy strain gauge was used to elim-
inate any effect of temperature.
3. Laboratory tests on cement-admixed clay samples
The required unconfined compressive strength (qu) of the DCM
columns used for excavation work in this project was 1 MPa. Thus,
a trial mix was prepared to determine the optimum cement con-
tent to be used in in situ mixing before construction initiated. Port-
land cement type I was used as an admixture according to the
standard of the Department of Highways of Thailand. The chemical
compositions and classifications of cement used in this study are
176 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181

Table 1
Excavation construction sequences versus elapsed time.

Stage Detail Elapsed

time (day)
1 Construct 0.7- and 0.8-m-diameter bored piles to
support the structural load
2 Install a temporary column (stanchion) usingH-
400  197 kg/m steel embedded into the bored piles
to transfer the weight from the basement floor during
construction to the bored piles
3 Install DCM column walls by deep wet mixing around 0
the excavated area to be used as a temporary retaining
structure for the construction of the basement
4 Excavate to
3.20 m for floor zone B1-A and install 10
concrete slab zone B1-A
5 Excavate to
4.80 m for floor zone B1-B and install 30
concrete slab zone B1-A
6 Install temporary struts between slab zones B1-A and 37
B1-B using H 300  94 kg/m steel to transfer the
lateral load due to the excavation
7 Excavate to
6.30 m for space between floor zones B1-
B and B1-A
8 Excavate to
6.30 m for floor zone B2-A 65
9 Excavate to
7.90 m for floor zone B2-B 88

Table 2
Summary of the chemical composition of cement used in this study.

Composition Content (%)

Cao (Lime) 62.81
SiO2 21.20
Al2O3 (Alumina) 4.95
Fe2O3 (Iron) 2.82
MgO (Magnesia) 4.00
Na2O and K2O (Alkali) 0.30
SO3 (Sulfuric anhydrite) 2.63
Specific gravity 3.15
Loss of ignition (%) 1.23
Classification Ordinary Portland cement
Type1-Grade 53

ranged from 15 to 16.5 kN/m3, and the water contents varied from
35% to 70%. Unconfined compression tests were performed on
samples 50 mm in diameter and 100 mm in height to determine
the field qu. The values of the field qu ranged from 1.4 to 2.1 MPa,
with an average value of 2.0 MPa, which was twice as high as the
required qu. The undrained Youngs modulus (Eu) ranged from
120 to 290 MPa, with an average value of 200 MPa, indicating an
empirical relationship Eu = 100qu, which corresponded to the test
results of Jongpradist et al. [18] and Jamsawang et al. [1,19,2022].

Fig. 2. (a) Plan view of the excavation area and (b) cross-sectional view A-A of the 4. Numerical analysis of the field case study
excavation.

4.1. Finite element mesh and boundary condition

presented in Table 2. A series of unconfined compression tests was
performed according to ASTM D2166-00 [15]. Flexural strength
and splitting tensile tests in the laboratory followed the standard
procedures of ASTM D1635-00 [16] and ASTM D3967-95a [17],
respectively. The cement contained between 140 and 260 kg/m3
of wet soil; the water-cement ratio was fixed at 1.1. The results
show that the required cement content was 200 kg/m3. However,
the cement content used for field mixing was increased to
250 kg/m3 due to the non-uniformity of in situ mixing. The empir-
ical relationships indicate that the flexural and the splitting tensile
strengths were estimated as 0.3 and 0.16 time the qu, respectively.
After the DCM wall construction was completed, core samples
were extracted from the DCM columns at various depths from
three locations, BHC-1, BHC-2 and BHC-3, for laboratory tests.
Fig. 3 presents the test results of the core samples. The unit weights
A finite element simulation using the PLAXIS 3D version 2013
software was used to describe the performance of the DCM wall
constructed using the TD method. The 3D finite element model
comprised the DCM columns, bored piles and foundation soils.
The soil volume was modeled using ten-node tetrahedral volume
elements. The stanchions and struts (Fig. 2a and b) were modeled
using beam elements, whereas the basement slabs (Fig. 2a and b)
were modeled using plate elements. Fig. 4a and b illustrates the
140,000-element 3D finite element mesh used in the analysis. At
the bottom of the finite element, the displacements were set to
zero in the three directions, x, y, and z. The vertical model bound-
aries were fixed in the x- and y-directions and free in the z-
direction. To avoid boundary effects, the length and width of the
model were 160 and 140 m, respectively, and its depth was 30 m.
 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181 177

Fig. 3. Properties of the cored DCM columns.

Fig. 4. (a) 3D finite element mesh of the wall and (b) enlargement of the modeling excavation area.

4.2. Constitutive model and model parameters Table 3

Parameters used in MohrCoulomb model.
The stiffness of the DCM column was much higher than that of DCM column
the foundation soils, and the final stage of the excavation was per-
Unit weight (kN/m3) 15
formed under working stress. Therefore, a bilinear elastic-perfectly Undrained Youngs modulus (MPa) 200
plastic MohrCoulomb model was used to model the DCM wall Undrained Poissons ratio 0.49
[1,2027]. The tensile strength of the DCM columns was Undrained cohesion (MPa) 1
0.160.23qu based on the measured correlation, which was consid- Friction angle () 0
Material behavior Undrained type C
ered using the tension cutoff in the model. The undrained type C
was employed to simulate the undrained behavior of the DCM
columns using a total stress analysis. The parameters of the
MohrCoulomb model are listed in Table 3. Since the elapsed time [3032]. The hardening soil model was used to model the behavior
for each stage of construction was relatively short (Table 1), the of the weathered crust, soft clay, medium stiff clay and stiff clay in
creep effect was much smaller than the stress relaxation from this study. The soft soil creep model was not used to model the
the excavation. The lateral movement induced due to excavation behavior of the soft clay because the elapsed time for each stage
was much larger than the movement arising from the creep of of construction was shorter than the stress relaxation time for
the soft clay layer. The hardening soil model is a non-linear the excavation. Thus, lateral movement induced due to excavation
advanced model that is frequently used for simulating the behavior was much more pronounced than that due to the creep effect of the
of soft soils and stiff soils under deep excavation work soft clay layer. The linear elastic model was employed to model the
[1,2022,2830]. A value of v ur 0:2 is typically used in this model behaviors of the concrete slabs, stanchions, struts and bored piles
178 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181
[3]. The parameters of the hardening soil model and linear elastic
model are listed in Tables 4 and 5, respectively.
4.3. Soil parameter calibration for the foundation soils
The laboratory test results and hardening soil model were cali-
brated using the PLAXIS soil test facility to obtain reasonable soil
parameters for simulating the field behaviors of the foundation
soils. The test results of the triaxial tests are presented in the form
of deviator stress versus axial strain, whereas those of the oedome-
ter tests are shown as plots of the logarithm of the vertical effective
stress versus vertical strain curves, as shown in Fig. 5a and b,
respectively, for soft clay. The input shear strength parameters /0
and c0 for the foundation soils were obtained from the triaxial test
results, as shown in Fig. 1j and k, respectively. Eref ref ref
50 , Eoed , Eur and m
are independent input parameters in the hardening soil model.
These parameters were adjusted to obtain suitable values to pro-
vide the best-fit results of the stress-strain curves. The example
of calibration results for soft clay shown in Fig. 5a and b reveals
good agreements for the stress-strain curves. The similar fitting
curves were obtained for other soil layers. Therefore, suitable
parameters of the foundation soils used for the 3D finite element
analysis for this study are presented in Table 4.
4.4. Numerical results and comparisons with measurements
4.4.1. Lateral movement profile
The observed lateral movement profiles of the DCM wall at the
middle of each side of the excavation area were obtained from the
four inclinometers (shown in Fig. 2a), I1, I2, I3 and I4, at the west,
east, south and north parts of the excavation, respectively. The lat-
eral wall movement profiles at the final stage of the excavation
(stage 10) are shown in Fig. 6ad. In the figures, the computed lat-
eral movement profiles from the 3D finite element analysis are also
included for comparison purposes. The trends of the lateral move-
ment profiles were reasonably well captured, and the computed
magnitudes were generally in good agreement with the observed
data for I1, I2, I3 and I4. The 20% maximum overestimation of
the computed dhm at a depth of 8 m (20 mm) for I4 was 5 mm,
whereas the 20% maximum underestimation of the calculated
dhm at a depth of 9 m (15 mm) for I3 was 5 mm.
For the I1 side, the wall movement showed a small curvature,
which means that the wall was tilted like a block. The wall was
permitted to deflect as a cantilever beam. The maximum lateral
wall movement (dhm) located at the top of the wall (near the
ground surface) was 58 mm, and the movement at the tip of the
wall (at a depth of 14 m) was 15 mm, which implies that the move-
ment pattern was a combination of slight overturning and sliding.
For inclinometer I2, the amount of wall movement was less than I1
because the excavation depth was smaller. The dhm was 32 mm at
the top of the wall, and the movement at the tip of the wall was
10 mm. The lateral movement profiles for the I3 and I4 sides are
presented in Fig 6c and d, respectively. The magnitudes of the lat-
eral movements were considerably smaller than those of the I1 and
I2 sides because of the smaller wall length and sufficient lateral
support from the concrete slab bracings B1-B and B1-A. Because
the final excavation depths were the same (He), there were no sig-
nificantly different lateral wall deflections on the two sides of I3
and I4. The dhm values were 0.73, 0.51, 0.23 and 0.33% He for walls
located at I1, I2, I3 and I4, respectively. The lateral movement pro-
files developed into a bulged profile inward of the excavation area,
indicating that the walls of the two sides were well propped near
the surface. Thus, the location of the maximum lateral movement
occurred at a deeper depth.dhm was 22 and 26 mm at distances
4.5 and 7 m below the ground surface for the walls located at I3
and I4, respectively. The tip movements of the retaining wall were
only 5 mm for I3 and I4.
The location of the maximum lateral movement was at the
ground surface for I1 and I2, whereas it was 0.63 and 1.0 He below
the ground surface for I3 and I4, respectively. Ou et al. [33] found
that the location of the maximum lateral movement of eight case
histories in Taipei soft soil were often observed near the excavation
surface. The analysis of Moormann [4] showed that the dhm for
most deep excavations in a soft soil was observed at depths of
0.51.5He under the ground surface. Wang et al. [5] reported that
the location of the maximum lateral movement was observed at
depths of 0.51.0 He under the ground surface in 53% of the case
histories. For 43% of the case histories, the location of the maxi-
mum lateral movement was observed at depths of 1.01.4He under
the ground surface. The location of the maximum lateral move-
ment was observed at the top of the wall for only approximately
4% of the case histories. The dhm for the DCM walls without internal
struts occurred at the top of the walls. Deep-seated wall displace-
ments were observed when internal struts were used to support
the DCM walls. The results of Ou et al. [33], Moormann [4] and
Wang et al. [5] have been broadly confirmed by this study.
The tip movement of the retaining wall was found to occur at I1
and I2. Wang et al. [34] reported that the embedded depth ratio of
the wall, which is defined as the embedded length of the wall to
final excavation depth, may contribute to the tip movement. Here,
the embedded depth ratio was 0.77 for the DCM wall mainly varied
between 0.8 and 1.4 and was 1.08 on average [34]. Thus, embedded
depth ratio was the smallest in this study. A larger embedded
depth ratio could suppress toe movement because more soils
under the excavation surface are strengthened.
4.4.2. Strut force
Fig. 7 shows comparisons of the measured and computed strut
forces induced by the 7.9 m deep excavation for all struts, as shown
in Fig. 2. The measured values of the strut forces were 310, 150,
330, 370, 440, 350 and 450 kN, whereas the computed values were
280, 100, 350, 360, 390, 320 and 510 kN for struts S1S7, respec-
tively. Strut S7 was farthest from the edge of the slab and was
exposed to a larger axial force than the other struts. The yield
strength of the steel used in the struts was 250 MPa, resulting in
a yield axial force of 3000 kN. Thus, a minimum factor of safety
against structural failure of 5.8 was obtained for this project.
A comparison of the observed and computed data indicated a
maximum underestimation of 33% and an overestimation of 13%
for struts S2 and S7, respectively. However, the average error in
the comparison was only 12%. Therefore, the computed magni-
tudes of the strut forces were generally in good agreement with
the observed data. The calculated force in strut S7 was approxi-
mately twice the measured and calculated forces in strut S1
because the spacing of S7 was twice the spacing of S1. This shows
that the strut forces determined by computational aid provide reli-
able results.
4.5. Parametric study on lateral movements and strut forces
To investigate the effectiveness of the DCM wall with TD con-
struction (DCM-TD) used for deep excavation work, the perfor-
mances of DCM walls without a permanent concrete slab bracing
or built with a BU construction method (DCM-BU) were character-
ized. The effect of the construction method was simulated by omit-
ting the concrete slabs and the temporary struts from the
excavation area. As shown in Fig. 8, the DCM wall simulation with
BU construction resulted in considerably larger lateral movements.
The dhm/He ratios were 1.28, 1.42, 0.83 and 0.85% He for DCM-BU
 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181 179

Table 4
Parameters used in hardening soil model.

Weathered crust Soft clay Medium stiff clay Stiff clay

3
Unit weight (kN/m ) 16 14 16 20
Secant stiffness, Eref 10,000 1200 10,000 20,000
50 (kPa)
Tangential stiffness, Eref (kPa) 12,000 960 12,000 25,000
oed
Unloading and reloading stiffness, Eref
(kPa) 35,000 4000 45,000 95,000
ur
Power of the stress-level dependency of the stiffness, m 1.0 1.0 0.9 0.9
Poissons ratio for unloading-reloading, v ur 0.2 0.2 0.2 0.2
Effective cohesion (kPa) 15 2 9 30
Effective friction angle () 27 23 26 28
Angle of dilatancy () 0 0 0 0
Over consolidation ratio 2.0 1.5 2.0 2.4
Material behavior Undrained Undrained Undrained Undrained

Table 5
Parameters used in linear elastic model.

Concrete slab Bored pile Temporary strut Stanchion

Unit weight (kN/m3) 24 24 78 78
Moment of inertia (m4) 1.3  10
3 m4/m 0.0120.020 2.0  10
4 7.1  10
4
Cross-sectional area (m2) 0.25 m2/m 0.380.50 0.012 0.025
Youngs modulus (kPa) 2.8  107 2.0  107 2.1  108 2.1  108
Poissons ratio 0.15 0.15
Material behavior Plate element Non-porous Beam element Beam element

tively, yielding an average improvement ratio of 2.4. The high

Fig. 5. An example of soil parameter calibrations with (a) consolidated undrained
triaxial and (b) oedometer test results for soft clay layer.
and 0.81, 0.63, 0.28 and 0.30% He for DCM-TD for walls located at
I1, I2, I3 and I4, respectively. These results indicate that the
DCM-TD can reduce lateral movement with improvement ratios
of 1.6, 2.4, 3.0 and 2.8 for walls located at I1, I2, I3 and I4, respec-
improvement ratio reflects the strong influence of concrete slab
bracing. The shapes of the computed lateral movement profiles
for the walls located at I3 and I4 in Fig. 8 tended to move farther
toward the excavation, unlike those for the case study. This finding
confirms that the presence of a slab affected the type of wall lateral
movement profiles for walls located I3 and I4, as noted in
Section 4.4.1.
The DCM wall thickness for this project was limited to 2.5 m, or
four rows of a DCM wall, due to insufficient space. Thus, the effect
of DCM wall thickness on lateral movement and strut forces was
investigated by varying the number of rows, using 2, 4 (case study),
6, 8 and 10, which correspond to wall thicknesses of 1.3, 2.5, 3.7,
4.9 and 6.1 m, respectively. Fig. 9a presents the relationship
between wall thickness and mean values of dhm/He for walls
located at I1, I2, I3 and I4. The numerical results show that the
mean dhm/He decreases with increasing wall thickness due to
increasing wall rigidity [1] for both DCM-TD and DCM-BU. In addi-
tion, the DCM-TD for the field case study requires half the wall
thickness of the DCM-BU to obtain same mean value of dhm/He.
In addition, the difference in the mean value of dhm/He between
DCM-TD and DCM-BU decreases with increasing wall thickness,
which reflects a decrease in the improvement ratio. The improve-
ment ratios were 2.4, 2.2, 2.0, 1.8 and 1.5 for wall thicknesses of
1.3, 2.5, 3.7, 4.9 and 6.1 m, respectively. Based on past experiences
in engineering practice, the improvement ratio should be greater
than 2. Thus, the use of a wall thickness greater than 3.7 m or 6
rows in this study was economically undesirable for DCM-TD.
Fig. 9b shows the effect of wall thickness on magnitude of all
strut forces S1S7. The figure shows that the wall thickness had
a substantial effect on the strut forces. As expected, the results
show that the overall tendency is the reduction in strut forces for
all locations with increasing wall rigidity, resulting in the possibil-
ity of reducing the number of struts. However, this reduction is
only significant when the wall thickness varied from 1.3 to
2.5 m. The influence of the wall thickness was not easily detectable
when wall thickness exceeded 2.5 m, indicating that the influence
of wall thickness is only significant when the wall has a low
rigidity.
180 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181

Fig. 6. Comparison of measured and computed lateral movements.

1200
1100 Slab B1-B

1000
900 8.0 2.5 2.5 3.1 5.1 3.3 4.8 16.0
Strut force (kN)

800 S1 S2 S3 S4 S5 S6 S7

700
Measured
600 Simulated
500
400
300
200
100
0
S1 S2 S3 S4 S5 S6 S7

Fig. 7. Comparison of measured and calculated strut loads.

5. Summary and conclusions

A field case study of a DCM wall applied using a TD construction

method for a deep excavation in soft Bangkok clay is reported. The
lateral movements of the wall system and the strut forces at the
proposed excavation depth were observed. A calibration of labora-
tory test results with the HSM was performed to obtain the optimal
parameters for simulating the behavior of the foundation soils. The
Fig. 8. Effect of BU construction on computed lateral movement profiles.
software program PLAXIS 3D version 2013 was used to simulate
the behavior of the wall. Finally, the results of the construction
method in comparison to the BU method were investigated, and
the influence of wall thickness on wall lateral movement and strut 2. The observed strut forces varied from 150 to 450 kN, with an
forces was determined to evaluate the effectiveness of TD con- average of 340 kN. The minimum factor of safety against struc-
struction method and to identify an appropriate wall thickness. tural failure of 5.8 was obtained for this project, and it has been
The following conclusions can be drawn based on the observed confirmed by a performance-based design.
and simulated results: 3. The numerical results obtained from the 3D model were consis-
tent with the field data for the wall lateral movements and strut
1. The observed data show that dhm were 0.73 and 0.51, 0.23 and forces. Comparisons of the observed and computed data
0.33% He for the long and short sides of the walls, respectively. revealed 20% maximum and 12% average errors for the com-
The long sides of the walls were permitted to deflect as a can- puted dhm and strut forces, respectively.
tilever beam. However, the short sides of the walls were well 4. Based on the numerical results, the concrete slab used in the TD
supported near the surface due to sufficient lateral support of construction had a significant effect on the lateral movement of
the concrete slab. the wall. When the concrete slab was introduced, the lateral
 P. Jamsawang et al. / Computers and Geotechnics 88 (2017) 174181
Fig. 9. Effect of wall thickness on (a) mean value of dh/He and (b) strut force.
movements were sharply reduced, with an average improve-
ment ratio of 2.4. The DCM-TD in the field case study required
half the wall thickness of the DCM-BU to obtain the same mean
value of dhm/He. A wall thickness greater than 3.7 m was consid-
ered an economically impractical design for DCM-TD based on
the improvement ratio of 2.0.
5. From the numerical analysis, all strut forces decreased with
increasing wall thickness. Thus, the number of struts can be
reduced for the thick wall. The effect of the DCM wall thickness
appears to be insignificant when the wall thickness is greater
than 2.5 m.
Acknowledgments
The authors gratefully acknowledge the financial support of the
Faculty of Engineering, King Mongkuts University of Technology
North Bangkok Contract no. ENG-59-01 and the Thailand Research
Fund under the TRF Senior Research Scholar program Grant No.
RTA5980005.
References
[1] Jamsawang P, Voottipruex P, Jongpradist P, Bergado DT. Parameters affecting
the lateral movements of compound deep cement mixing walls by numerical
simulations and parametric analyses. Acta Geotech 2015;10:797812.
[2] Arboleda-Monsalve LG, Finno RJ. Influence of concrete time-dependent effects
on the performance of top-down construction. J Geotech Geoenviron Eng
2015;141:98594.
[3] Long M. Database for retaining wall and ground movements due to deep
excavations. J Geotech Geoenviron Eng 2001;127:20324.
[4] Moormann C. Analysis of wall and ground movements due to deep excavations
in soft soil based on a new worldwide database. Soils Found 2004;44:8798.
View publication stats
181
[5] Wang JH, Xu ZH, Wang WD. Wall and ground movements due to deep
excavations in Shanghai soft soils. J Geotech Geoenviron Eng
2010;136:98594.
[6] Hou YM, Wang JH, Zhang LL. Finite-element modeling of a complex deep
excavation in Shanghai. Acta Geotech 2009;4:716.
[7] Hsieh P, Ou C, Lin Y. Three-dimensional numerical analysis of deep excavations
with cross walls. Acta Geotech 2013;8:3348.
[8] Ignat R, Baker S, Larsson S, Liedberg S. Two- and three-dimensional analyses of
excavation support with rows of dry deep mixing columns. Comput Geotech
2015;66:1630.
[9] Ou CY, Hsieh PG, Lin YL. A parametric study of wall deflections in deep
excavations with the installation of crosses walls. Comput Geotech
2013;50:5565.
[10] Hsiung BB, Yang K, Aila W, Hung C. Three-dimensional effects of a deep
excavation on wall deflections in loose to medium dense sands. Comput
Geotech 2016;80:13851.
[11] Zhang W, Goh ATC, Xuan F. A simple prediction model for wall deflection
caused by braced excavation in clays. Comput Geotech 2015;63:6772.
[12] Likitlersuang S, Surarak C, Wanatowski D, Oh E, Balasubramaniam A. Finite
element analysis of a deep excavation: a case study from the Bangkok MRT.
Soils Found 2013;53:75673.
[13] Lim A, Hsieh P, Ou C. Evaluation of buttress wall shapes to limit movements
induced by deep excavation. Comput Geotech 2016;78:15570.
[14] Bergado DT, Teerawattanasuk C, Youwai S, Voottipruex P. Finite element
modeling of hexagonal wire reinforced embankment on soft clay. Can Geotech
J 2000;37:120926.
[15] ASTM D221610. Standard test methods for laboratory determination of water
(moisture) content of soil and rock by mass. West Conshohocken (PA): ASTM
International; 2010.
[16] ASTM D163500. Standard test method for flexural strength of soil-cement
using simple beam with third-point loading. West Conshohocken (PA): ASTM
International; 2010.
[17] ASTM D396795a. Standard test method for splitting tensile strength of intact
rock core specimens. West Conshohocken (PA): ASTM International; 1995.
[18] Jongpradist P, Jumlongrach N, Youwai S, Chucheepsakul S. Influence of fly ash
on unconfined compressive strength of cement-admixed clay at high water
content. J Mater Civ Eng 2010;22:4958.
[19] Jamsawang P, Bergado DT, Voottipruex P. Field behavior of stiffened deep
cement mixing piles. Proc Inst Civil Eng Ground Improve 2011;164:3349.
[20] Jamsawang P, Voottipruex P, Boathong P, Mairaing W, Horpibulsuk S. Three-
dimensional numerical investigation on lateral movement and factor of safety
of slopes stabilized with deep cement mixing column rows. Eng Geol
2015;188:15967.
[21] Jamsawang P, Boathong P, Mairaing W, Jongpradist P. Undrained creep failure
of a drainage canal slope stabilized with deep cement mixing columns.
Landslides 2016;13:93955.
[22] Jamsawang P, Yoobanpot N, Thanasisathit N, Voottipruex P, Jongpradist P.
Three-dimensional numerical analysis of a DCM column-supported highway
embankment. Comput Geotech 2016;72:4256.
[23] Voottipruex P, Suksawat T, Bergado DT, Jamsawang P. Numerical simulations
and parametric study of SDCM and DCM piles under full scale axial and lateral
loads. Comput Geotech 2011;38:31829.
[24] Voottipruex P, Bergado DT, Suksawat T, Jamsawang P, Cheang W. Behavior and
simulation of deep cement mixing (DCM) and stiffened deep cement mixing
(SDCM) piles under full scale loading. Soils Found 2011;51:30720.
[25] Huang J, Han J. 3D coupled mechanical and hydraulic modeling of a
geosynthetic-reinforced deep mixed column-supported embankment.
Geotext Geomemb 2009;27:27280.
[26] Mun B, Kim T, Moon T, Oh J. SCM wall in sand: numerical simulation and
design implications. Eng Geol 2012;151:1523.
[27] Wonglert A, Jongpradist P. Impact of reinforced core on performance and
failure behavior of stiffened deep cement mixing piles. Comput Geotech
2015;69:93104.
[28] Surarak C, Likitlersuang S, Wanatowski D, Balasubramaniam A, Oh E, Guan H.
Stiffness and strength parameters for hardening soil model of soft and stiff
Bangkok clays. Soils Found 2012;52:68297.
[29] Sexton BG, McCabe BA. Numerical modelling of the improvements to primary
and creep settlements offered by granular columns. Acta Geotech
2013;8:44764.
[30] Zhao C, Lavasan AA, Barciaga T, Zarev V, Datcheva M, Schanz T. Model
validation and calibration via back analysis for mechanized tunnel
simulationsthe western Scheldt tunnel case. Comput Geotech
2015;69:6014.
[31] Schanz T, Vermeer PA, Bonnier PG. The hardening-Soil model: formulation and
verification. In: Brinkgreve RBJ, editor. Beyond 2000 in computational
geotechnics. Rotterdam: Balkema; 1999. p. 28190.
[32] Jongpradist P, Kaewsri T, Sawatparnich A, Suwansawat S, Youwai S, Kongkitkul
W, et al. Development of tunneling influence zones for adjacent pile
foundations by numerical analyses. Tunn Undergr Space Technol
2013;34:96109.
[33] Ou C, Hsieh P, Chiou D. Characteristics of ground surface settlement during
excavation. Can Geotech J 1993;30:75867.
[34] Wang ZW, Ng CW, Liu GB. Characteristics of wall deflections and ground
surface settlements in Shanghai. Can Geotech J 2005;42:124354.