IOP Conference Series: Earth and Environmental Science

PAPER • OPEN ACCESS You may also like

- Stability analysis of foundation pit against
Failure Investigation of Braced Excavation in Soft uplift based on circular sliding
Zhiguang Guo, Yongfu Liu and Fan Zhang
Clays: Case Study on the Collapse of Nicoll - Simulating the effects of braced
excavation on the adjacent tunnel using
Highway numerical method
Shun Yi, Wenhui Ke, Xiangguo Huang et
al.
To cite this article: Ari Surya Abdi and Chang-Yu Ou 2023 IOP Conf. Ser.: Earth Environ. Sci. 1184
012010 - Finite Element Analysis of Geotechnical
Excavation Based on COMSOL
Multiphysics
Jue Han and Qianqian Ma

View the article online for updates and enhancements.

This content was downloaded from IP address 188.119.119.37 on 02/06/2023 at 14:28

Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

Failure Investigation of Braced Excavation in Soft Clays: Case

Study on the Collapse of Nicoll Highway

Ari Surya Abdi1, Chang-Yu Ou2

1
Department of Civil & Construction Engineering, National Taiwan University of Science and
Technology, No. 43, Section 4, Keelung Rd., Da’an District, Taipei City 106, Taiwan
2
Department of Civil & Construction Engineering, National Taiwan University of Science and
Technology, No. 43, Section 4, Keelung Rd., Da’an District, Taipei City 106, Taiwan
Email: [email protected]

Abstract. Deep excavation in clay normally causes large soil movement, which may induce
basal heave failure. In this study, the three-dimensional finite element analysis with the strength
reduction method was employed to evaluate the typical failure mechanism of braced excavation
in soft clays. The collapse of Nicoll Highway was selected as a failure case history and further
conducted by considering the full elastoplastic structural behavior. The results showed that the
computed wall displacements were successfully verified through the observation data. It was
also found that the excavation failure was initiated by the yielding of the strut-waler connection,
which was similar to that of the field observation. Finally, the proposed finite element model
adopted in this study can be used as a design recommendation to evaluate the failure mechanism
and further prevent possible braced excavation failure in soft clays.
Keywords: basal heave, clays, deep excavation, elastoplastic structural, finite element

1. Introduction

The bracing system is commonly adopted in deep excavations to ensure the diaphragm wall can
withstand construction. In general, the bracing system consists of the center post, horizontal steel strut,
and steel waler. The horizontal steel struts are installed perpendicular to the diaphragm wall to resist the
earth pressure acting behind the wall, while the steel waler is applied to transfer the force acting on the
wall to the horizontal strut. Furthermore, the function of center posts is to support the strut to not fall
out due to their own mass [1].
As a matter of fact, excavation activities in clay often resulted in a large soil movement behind the
wall. Consequently, the stress of the structural support system may exceed its strength material and
induce excavation failure. According to previous studies [2–5], the yield of the diaphragm wall and the
structural bracing system could trigger basal heave failure. For example, the collapse of the Nicoll
Highway excavation case was initially caused by the buckling of the strut-waler connection, followed
by the failure of the diaphragm wall and construction area [6]. Hence, an appropriate design of the
structural system should be considered properly in braced excavations.
In this study, the collapse of the Nicoll Highway excavation case was investigated to evaluate the
typical braced excavation failure mechanism. A three-dimensional finite element method (3D FEM) was
used to model the actual excavation geometry accurately. The full elastoplastic structural behavior was
used to let the structural model could yield during the analyses. Furthermore, the strength reduction
method was adopted in the stability analysis to assess the factor of safety against basal heave at the final
excavation stage.

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

2. Numerical Methods

2.1. Collapse of Nicoll Highway Excavation Case

The Nicoll Highway excavation case was part of the MRT Circle Line project located in Singapore. The
project was occupied a 211-m-long, while the final excavation level was planned to go up to 11th stage
(GL – 33.6 m). However, the collapse occurred at the 10th stage (GL – 30.6 m), while the failure was
extended from Type M3 area to M2 area, as illustrated in Figure 1. According to the COI report [6], the
buckling of strut-waler connection was initially found during the 7th excavation stage. Prior to collapse,
the buckling of strut-waler connection was also found at the ninth level of strut (10th excavation stage),
which further caused the excavation failure, resulting the crater with a diameter of about 100 m.

TSA

I-104

Inclinometer
Diaphragm wall
Steel strut

Figure 1. Detail of the collapsed area in the Nicoll Highway excavation case.

The soil layer dominantly consists of marine clay, while the old alluvium (defined as hard stratum)
is found approximately GL – 41.5 m below the ground surface. The ground surface was covered by
backfill with a unit weight (γt) of 19 kN/m3, while the unit weight of marine clay ranged from 16 to 16.8
kN/m3. The unit weight of the upper fluvial deposit ranged from 19 to 20 kN/m3, whereas the unit weight
of estuarine deposits was 16.8 kN/m3. The unit weight of the lower fluvial deposit and old alluvium is
found to be 20 kN/m3. In addition, the groundwater level was located at Gl -2.4 m.

The excavation was supported by a diaphragm wall with a length and thickness are 44.3 m and 0.8
m, respectively. The bracing system was adopted to restrain the excessive wall displacement, consisting
of nine strut levels arranged by an average horizontal spacing of 4 m. Two continuous jet grout piles
(upper and lower JGP parts) were implemented near the final excavation level to support the diaphragm
wall during excavation. The soil condition, detailed structural support system, and excavation profile
are detailed in Figure. 2.

2
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

10.05 m 10.05 m S u (kPa)

0 20 40 60 80 100

H-350 Excavation stage

Fill
H-400 (G50)
Diaphragm wall
2H-400 (G50) Estuarine
Cement mortar
Plate stiffner
Waler 2H-400 10
Upper
marine clay
2H-400

H-414 (G50)
Strut Fluvial deposit
2H-400 (G50) 20

Depth (m)
2H-414 (G50)
Bracket 2H-400 (G50) Lower marine
clay
Upper JGP
30

Lower JGP
tdw = 0.8 m
Estuarine

H-350
Fluvial deposit 40

Diaphragm Old Alluvium

Wall
Whittle and Davies
Center Post
50

Figure 2. Detail of structural support system, excavation profile, and soil condition in the Nicoll
Highway excavation case (su was obtained from Whittle and Davies [7]).

2.2. Determination of soil and structural properties

The three-dimensional finite element method (3D FEM) was performed by using the commercial
geotechnical software PLAXIS 3D v.21 [8]. The soil was modeled by a 10-node tetrahedral element,
while the Mohr-Coulomb soil model was used to simulate the soil strength characteristic near failure. In
the analyses, the backfill layer was performed by drained material, in which the cohesion ( c' ), internal
friction angle ( ' ), and Poisson’s ratio (ν) were assumed to be 0.1 kPa, 30 ,ͦ and 0.3, respectively.
Young’s modulus of the backfill layer (E′) was estimated based on the N-SPT value (E′ = 2000N), where
the N value of backfill was approximately 5. On the other hand, the undrained material was adopted to
simulate the soft marine clay with cu = su (see Figure. 2), ϕ = 0, and νu = 0.495. The value of Eu = 200su
was chosen for the typical marine soft clay in Singapore. The design of undrained shear strength and
Young’s modulus of the JGP layer were 300 kPa and 150 MPa, respectively [6].

The plate element was used to model the diaphragm wall, while the horizontal strut and center post
were modeled as beam elements. The nominal value for Young’s modulus of diaphragm wall concrete
was estimated by E  4700 fc' ( fc' = 40 MPa), following the suggestion by ACI 318-95 [9]. A reduction
factor of 0.8 was used for wall stiffness to consider the cracks caused by the large bending [10,11]. On
the other hand, the yield stress (σy), Poisson’s ratio (νs), and Young’s modulus of steel material were
assumed to be 250 MPa, 0.3, and 2.04 × 108 kPa, respectively. The plastic axial stress (Np) and plastic
bending moment (Mp) were estimated based on σy value, cross-section area (A), and section modulus of
material (S). A reduction of 10% in the strut and center post parameters (i.e., E and σy) was applied to
represent the improper installation during construction.

Furthermore, because the buckling could occur in the connection between the horizontal strut and
waler, the strut-waler connection was considered in the analyses. According to Abdi and Ou [2], the
strut-waler connection could be represented as a short beam element, which is typically in the range of
0.3 m to 0.4 m (depending on the steel waler dimension). Under such a case, it is expected that the yield
of the strut-waler connection is mainly due to the axial force, while the bending moment is relatively

3
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

small. The reduction of σy value for the strut–wall connection parameter (α = 0.7) was considered to
represent the possible buckling failure caused by the large compression.

Moreover, a 12-node interface element was employed to model the interaction behavior between the
wall (including the diaphragm and buttress wall) and the surrounding soil. The roughness of the interface
element is controlled by the strength reduction factor (Rinter), which is related to the friction ratio between
the wall adhesion and the soil cohesion (α = su/cw). Rinter value is set to be 0.67, following the typical
roughness of concrete surface [1].

The 3D FEM modeling was focused on the Type M3 area (largest subsided area). The horizontal
boundary was extended far from the excavation area to minimize the effects of the boundary restraints.
The nodes along the side and bottom boundaries were constrained by the roller and fixed boundary,
respectively. A very fine mesh with local refinement was adopted in the analyses to provide an accurate
numerical solution. Figures. 3(a) and (b) show the typical finite element mesh and detail of structural
modeling in the analyses.

28 m Steel strut
89.95 m
B = 20.1 m Strut-waler
89.95 m connection
50 m

He

Hp
Diaphragm
wall

200 m Center post

(a) (b)

Figure 3. Typical excavation geometry used in the 3D FEM analyses: (a) finite element mesh; (b)
detail of structural modeling.

2.3. The Concept of Strength Reduction Method

In general, the safety factor is usually defined as the ratio of the collapse load over the working load.
Indeed, this concept could be adopted to assess the safety factor for foundation engineerings, such as
footings or piles. However, for the slope and excavation case, the failure was initially induced by the
load from the overburden pressure. In such a case, the failure surface is highly affected by the soil
strength. Thus, other methods to determine the safety factor could be adopted, such as the strength
reduction method. The strength reduction method is generally performed by reducing the soil strength
until the failure occurs (i.e., divergence condition) [2,12–17]. In this analysis, the corresponding strength
reduction is defined as the factor of safety against basal heave (FSfem = SRmax). Figures. 4(a) and (b)
illustrate the reduction process of the Mohr-Coulomb (MC) failure envelope. Thus, the concept of the
strength reduction method in relation to the MC failure criterion can be expressed as:

  c   n tan  (MC failure criterion) (1)

c  tan  
cr  ; r  tan 1   (during strength reduction) (2)
SR  SR 

4
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

c  tan  
cf  ;  f  tan 1   (at failure [maximum strength reduction]) (3)
FSfem  FSfem 

where  is the shear strength;  n is the normal stress along the shear plane; c is the cohesion;  is the
internal friction angle; SR is the strength reduction value; FSfem is the factor of safety at maximum
strength reduction; and parameters with the subscript “f” refer to the reduced strength at failure.

 σ1

σ1
Intial MC yield
surface
60 ͦ MC yield surface after
r strength reduction
f σ3 σ2
cr cf σ σ2
σ3 σ1
(a) σ3 (b)

Figure 4. The strength reduction concept for the MC soil model: (a) linear envelope in the Mohr
diagram; (b) pyramidal surface in principal stress space and cross-section in the equipressure plane.

In addition, the strength reduction is also applied to the interface strength element during the stability
analyses. The tolerated error was set to be 0.01, following the default from the PLAXIS for the
convergence criteria of global error. The arc-length control procedure was used in the analysis to obtain
reliable collapse loads during strength reduction. Furthermore, a very fine mesh with local refinement
in the interesting area was applied to provide accurate numerical solutions.

3. Results and Discussions

3.1. Excessive Wall Displacement

The inclinometer data located in the Type M3 area (I-104, see Figure. 1) was used to validate the 3D
FEM results. The design of maximum wall displacement is supposed to be around 14.5 cm. However,
excessive wall displacement occurred during construction. According to the COI report [6], the
maximum wall displacement at the 10th excavation stage reached 35 cm, which was located near the
excavation level. This indicates that the demolishing of the upper JGP induced excessive wall movement
and further caused the failure of the diaphragm wall.

Figure. 5 shows the comparison between the computed wall displacement (FEM results) and field
measurement (I-104). For the field measurement in stage 2, the wall displacement is like a cantilever
shape (i.e., the largest wall displacement occurred near the ground surface). As the excavation activity
goes deeper (stage 10), the trend of wall displacement along the depth is more like a curvilinear shape,
in which the largest wall displacement was found near the final excavation level. This is due to the
horizontal struts that restrain the upper part of the diaphragm wall. Furthermore, the results showed a
close agreement between the FEM results and field measurement for the initial stage (2nd excavation
stage) and final stage (10th excavation stage). This clearly implies that the FEM results were validated

5
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

through the observation data. Hence, the 3D FEM analysis adopted in this study could be reasonably
used to further assess the failure mechanism of braced excavation.

Wall displacement (cm)

0 10 20 30 40 50 60
0
Field measurement (Stage 2)
5 Field measurement (Stage 10)
FEM results

10

15
Depth (m)

20

25
Stage 2
30
Stage 10
35

40

45

Figure 5. Comparison of computed wall displacement and field measurements.

3.2. Basal Heave Failure

The stability analysis was performed at the final excavation stage (10 th excavation stage). The soil
strength was successively reduced until the convergence solutions no longer remained (i.e., divergence
condition), as mentioned in the preceding section. According to the stability results, it is found that the
maximum strength reduction is 1.120 at the divergence condition, which was defined as the factor of
safety against basal heave. This indicates that the final excavation stage in the FEM analysis was prior
to the collapse, which was similar to the field observation.

Figure. 6(a) shows the displacement vector at divergence condition. As shown in the figure, the soil
behind the wall was moving upward, passing through the lower part of the diaphragm wall, causing the
large upward soil movement inside the excavation. The soil movement further caused the upward
movement of the center post, resulting in the large bending of horizontal struts. Indeed, this phenomenon
is similar to that of the field observation, where the center post was moving upward when the excavation
failure occurred.

In addition to the stability analysis, the generation of the plastic point was further investigated, as
shown in Figure 6(b). It is found that the plastic points were concentrated below the final excavation
surface and extended far behind the diaphragm wall. It is due to the large soil movement that caused the
shear failure and formed the typical failure surface. Besides, the range of the extended failure area from
the excavation center was found to be around 49 m, which was similar to the actual failure condition
(the failure area was extended with a radius of 50 m). Hence, it is concluded that the 3D FEM modeling
adopted in this study should be considered for the proper design of braced excavation in soft clays.

6
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

Steel strut
Diaphragm wall

Soil behind the wall

moving downward

Soil below the excavation

moving upward

(a) Displacement vector at divergence condition

Extended failure area from Steel strut

excavation center = 49 m
Diaphragm wall

Possible failure surface

(FSfem = 1.120)

Plastic point

(b) Plastic point generation at divergence condition

Figure 6. The stability results at divergence condition: (a) displacement vector; (b) plastic point
generation.

3.3. Internal Force of Structural Support System

The internal force curve of the structural support system (horizontal steel struts and strut-waler
connection) was presented to evaluate the plastic behavior bracing system from the initial excavation
stage until the failure occurred, as plotted in Figure. 7. The boundary line (BL) herein has an implication
of the ultimate combination between the plastic axial force (Np) and the plastic bending moment (Mp).
Under such conditions, when the stress acting on the structural element exceeds the BL value, the plastic
condition of the structural material has been reached.

Figure 7(a) shows the detailed schematics of the structural level and excavation stage. It should be
noted that the number inside the circle indicates the excavation stage. In the internal force curve (see
Figure 7[b]), the sequence number of the excavation stage was started after installing the strut and
connection. For example, because the 1st level of strut and connection were installed after completing
the 1st excavation stage, the internal force development was started from the 2nd excavation stage. On
the other hand, the 2nd level of strut and connection was initially developed from the 3rd excavation stage.

As shown in Figure 7(b), the strut-waler connection is mainly subjected to the axial force, while the
bending moment is relatively small. It is because the joint between the diaphragm walls and the strut-
waler connection is assumed to be hinged, while its length is typically short. Hence, the internal force is
only developed in the Np axis. The axial force in the strut-waler connection was generally decreased
during the initial excavation stage and relatively stable at a certain stage. For the horizontal steel strut,
the internal force developed in the Np and Mp axis, which was different from that of the strut-waler
connection. It is due to the large upward movement of the center post, causing the bending in the
horizontal steel strut, as mentioned earlier. The bending moment gradually increased at the certain stage,

7
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

particularly at the divergence condition. Nevertheless, the internal force of 1st and 2nd level of strut and
strut-waler connections did not reach the plastic condition.

The internal force curve of 4th, 5th, and 6th are shown in Figure 7(c). It is found that the 6th level of
connection reached the boundary line, implying that the strut-waler connection was subjected to
buckling failure. Furthermore, the 9th level of connection also achieved the limiting value of axial force,
as indicated in Figure 7(d). This further caused the rapid numerical solution to reach the divergence
condition, resulting in a lower factor of safety (FSfem = 1.120). Besides, these findings are similar to that
of the field observation (the buckling occurred at 6th and 9th levels of the strut-waler connection). Hence,
the proper design of a structural support system should be highlighted when designing braced
excavation.

1st level of strut

and connection Excavation N (kN)
stage -7000
1st connection
1 -6000 2nd connection
Location to
be observed -5000 1st strut
-4000
2nd strut
3
Diaphragm 4 -3000
wall
-2000 10 DC

-1000 2
M (kNm)
7 0
-1200 -1000 -800 -600 -400 -200 200 400 600 800 1000 1200

9th level of strut 1000 DC

and connection 10
2000

10 3000 BL of 2nd strut

4000 BL of 1st strut

5000
BL of 2nd connection
DC = divergence condition
BL of 1st connection
(a) (b) 6000

7000

N (kN) N (kN) 3rd connection

Location to
-
12000
4th connection -
20000 7th connection
be observed 5th connection Location to 8th connection
-
10000 6th connection be observed -
9th connection
15000
-8000 4th strut 3rd strut
7 5th strut - 10 7th strut
DC
-6000
6th strut 10000
8th strut
8
-4000 10
DC
9th strut
-5000
DC
-2000 10
10
M (kNm) M (kNm)
0 0
-2000 -1600 -1200 -800 -400 400 800 1200 1600 2000 -3500 -3000 -2500 -2000 -1500 -1000 -500 500 1000 1500 2000 2500 3000 3500

2000
5000
4000
th th
BL of 8th strut
BL of 4 & 5 strut
6000 10000
BL of 3rd, 7th, 9th strut
BL of 6th strut
8000 BL of 8th connection
BL of 4th & 5th connection
(c) 10000
BL of 6th connection
(d) 15000

BL of 3rd, 7th, 9th connection

12000 20000

Figure 7. Internal force curve of the horizontal steel strut and strut-waler connection: (a) detail of
structural level and excavation stage; (b) 1st and 2nd level; (c) 4th, 5th and 6th level; (d) 3rd, 7th, 8th, and
9th level.

8
Second International Symposium on Civil Engineering and Environmental Research IOP Publishing
IOP Conf. Series: Earth and Environmental Science 1184 (2023) 012010 doi:10.1088/1755-1315/1184/1/012010

4. Conclusion

The three-dimensional finite element analysis with the strength reduction method was employed to
evaluate the excessive wall displacement and failure mechanism in the collapse of the Nicoll Highway
excavation case. The structural support systems, such as the diaphragm wall, center post, and horizontal
steel strut, were modeled as elastoplastic structural behavior to represent the actual failure condition. In
addition, the strut-waler connection model was also proposed to consider the possible buckling failure
in the connection between strut and waler. Furthermore, the following conclusions can be drawn:
1. The close agreement between the computed wall displacement and field measurement can be
obtained by considering the proposed FEM model.
2. A similar failure mechanism was obtained between the FEM model and the actual failure condition,
where the excavation failure was initiated by the yielding of the strut-waler connection, followed
by the failure of the horizontal strut and surrounding soil.
3. The proposed FEM model can be used as a design recommendation to evaluate the failure
mechanism and further prevent possible basal heave failure in future construction.

References
[1] Ou C-Y 2006 Deep Excavation Theory and Practice Taylor Fr.
[2] Abdi A S and Ou C-Y 2022 A Study of the Failure Mechanism of Braced Excavations Using
3D Finite-Element Analysis Int. J. Geomech. 22 1–14
[3] Do T N, Ou C Y and Chen R P 2016 A study of failure mechanisms of deep excavations in
soft clay using the finite element method Comput. Geotech. 73 153–63
[4] Ukritchon B, Whittle A J and Sloan S W 2003 Undrained Stability of Braced Excavations in
Clay J. Geotech. Geoenvironmental Eng. 129 738–55
[5] Choosrithong K and Schweiger H F 2020 Numerical Investigation of Sequential Strut Failure
on Performance of Deep Excavations in Soft Soil Int. J. Geomech. 20 04020063
[6] COI C of I 2004 Report of the Committee of Inquiry into the incident at the MRT circle line
worksite that led t the collapse of Nicoll Highway
[7] Whittle A J and Davies R V 2006 Nicoll Highway collapse: evaluation of geotechnical factors
affecting design of excavation support system Int. Conf. Deep Excav. 28 30
[8] Brinkgreve R, Engin E and Swolfs W 2013 PLAXIS 3D 2019 reference manual 470
[9] ACI A C I 2014 Building Code Requirements for Structural Concrete Am. Concr. Inst. 524
[10] Lim A, Hsieh P G and Ou C Y 2016 Evaluation of buttress wall shapes to limit movements
induced by deep excavation Comput. Geotech. 78 155–70
[11] Lim A, Ou C Y and Hsieh P G 2019 An innovative earth retaining supported system for deep
excavation Comput. Geotech. 114 103135
[12] Goh A T C, Zhang W G and Wong K S 2019 Deterministic and reliability analysis of basal
heave stability for excavation in spatial variable soils Comput. Geotech. 108 152–60
[13] Goh A T C 2017 Deterministic and reliability assessment of basal heave stability for braced
excavations with jet grout base slab Eng. Geol. 218 63–9
[14] Faheem H, Cai F, Ugai K and Hagiwara T 2003 Two-dimensional base stability of excavations
in soft soils using FEM Comput. Geotech. 30 141–63
[15] Faheem H, Cai F and Ugai K 2004 Three-dimensional base stability of rectangular excavations
in soft soils using FEM Comput. Geotech. 31 67–74
[16] Do T N and Ou C Y 2020 Factors affecting the stability of deep excavations in clay with
consideration of a full elastoplastic support system Acta Geotech. 15 1707–22
[17] Do T-N, Ou C-Y and Lim A 2013 Evaluation of Factors of Safety against Basal Heave for
Deep Excavations in Soft Clay Using the Finite-Element Method J. Geotech.
Geoenvironmental Eng. 139 2125–35