Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT
Deep excavation for basement construction under restricted setback conditions is common these days
because of increased underground utilisation in an urban environment. A diaphragm wall with active
soil/rock anchors at two levels is adopted for a 12.5m deep basement excavation in mixed soil
conditions. The diaphragm wall is resting in weathered rock available at about 19.50m. The first level
anchor could be accommodated only at 6.0m depth from the top because of the subsoil conditions and
the elevation of basement floor slabs. The performance of the retention system was analysed using 2D
finite element program PLAXIS 2D, and the estimated wall deflection was not of cantilever type and
the maximum bending moment was on the excavation side. The actual deflection of the wall was
measured at different locations during different excavation stages. Cantilever deflection pattern was
recorded for the first 6.50m excavation, much different from the estimated deflection profile. The wall
deflection increased suddenly in two steps after the site received heavy rains. This paper discusses the
estimated & measured wall deflections along with back-analysis and certain limitations with Plaxis 2D
analysis faced in the present case.
Keywords: Diaphragm wall, Plaxis 2D, deep excavation, wall deflection, inclinometer, undrained B
1
This paper was originally presented at 16th ARC 2019 Taipei
-1-
8TH MADHAV LECTURE, GEOPRACTICES 2019- IGS HYDERABAD, 23 NOV 2019
A realistic analysis requires to input failure proposed up to 6.50m depth for installing the
modes manually for capturing all the possible anchors at 6.0m depth. Dewatering from the
failure conditions. This paper is discussing the excavation side was planned.
performance of the wall under excavation up Table 1. Sub-soil stratification at project site
to 6.50m from the ground level only.
Sl Elevation, cu φ’ E’
2. SUBSOIL CONDITIONS AND No
Type
m
N
kPa (°) kPa
ANALYSIS OF RETENTION
1 Firm S clay 15.8 - 11.8 15 90 - 28800
SYSTEM
2 Firm clay 11.8 - 6.8 16 96 - 30720
The construction of a commercial building
3 Firm clay 6.8 - 4.3 10 60 - 16800
with three basement floors requiring roughly
12.5m excavation is underway. Figure 1 is 4 Soft clay 4.3 - 1.3 8 48 - 12000
presenting the overall plan of the site where 5 Sand ����
1.3 - 0.2 65 - 42 90000
the deep excavation is to be made. �����-�����
6 Sand 0.2 1.7 85 - 42 180000
7 Sandy clay �����- 4.0
1.7 ���� 90 400 - 200000
8 w.j. rock �����- 9.2
4.0 ���� 300 1000 - 250000
-2-
8TH MADHAV LECTURE, GEOPRACTICES 2019- IGS HYDERABAD, 23 NOV 2019
Initial excavation
Final excavation
GWL 2.0m
Firm clay
Firm clay
Soft to
firm clay
(a) (b)
Firm clay
Fig. 4 (a) Wall deflection (b) BM distribution
Sand after 6.5m excavation.
Sand
full depth of excavation with two level
Weathered rock anchors and 20% increase on the estimated
BM was recommended. The main steel
Fig. 3 Plaxis 2D model of excavation
corresponding to the maximum BM was
The estimated wall deflection was 25mm at provided for the full wall depth. Undrained
wall top and 30mm at the excavation level analysis was conducted as the first level
after 6.50m excavation. The reversal of BM excavation and anchor installation was
distribution could not be understood, and the expected to be completed in 45 days.
analysis was repeated using the HS model for 3. CONSTRUCTION AND
the top clay layers using undrained B MONITORING
parameters. The results were very similar with
only some differences in the deflection and The diaphragm wall was constructed in
BM values. Review of the literature did not panels of 6m wide using the trench cutter and
provide a suitable answer for the BM bentonite mud stabilisation. Inclinometers
distribution on the excavation side for the were installed for the full depth of the wall at
cantilever wall. The analysis was made for the different locations, mostly towards the centre
of the walls. The monitoring of the walls
-3-
8TH MADHAV LECTURE, GEOPRACTICES 2019- IGS HYDERABAD, 23 NOV 2019
initiated just at the time of excavation. The wall moved to roughly 85mm
Counter measurements were made using total suddenly after 15 days and then again
station observations. stabilised at about 90mm at the top. The site
observed 20mm to 25mm gap between the
The excavation up to 6.50m depth was
wall and the soil during the initial
initially done for two wall lengths of 65m and
observations of 30mm deflection of the wall.
45m. The weather was dry during this
This detachment was for about 3m from the
excavation. The actual deflection of the wall
surface. The ground settlement was not
was measured during different excavation
observed initially. Figure 6a is presenting the
stages by continuous monitoring of the
site observations after the wall moved around
inclinometers. Cantilever deflection pattern
30mm and Figure 6b is when the wall further
was registered for the first 6.50m excavation,
moved immediately after the rains.
much different from the estimated deflection
profile and about 32mm deflection at the top
and close to 12mm at the excavation level
were recorded. For a comparison, the Figure 5
presents the predicted and measured wall
deflections.
6
4. BACK ANALYSIS AND
DISCUSSION
4
2
The wall deflection was more than twice
the predicted deflection that warranted a
0 Estimated -70% interaction
Measured I-04 stability study. The difficulty in capturing the
-2
Measured I-05 cantilever wall movement by Plaxis 2D
-4 analysis could have caused structural
Fig. 5 Predicted and measured wall instability when the excavation depth without
deflection @ 6.5m excavation lateral support is more as in the present case.
It is, hence, necessary to look into the reasons
for such anomaly in the Plaxis 2D analysis.
The construction of anchors and pre- Ideally, the excavation without lateral support
stressing did not progress as planned, and the should have been limited to about 4.0m.
anchors installed were not pre-stressed for
almost a month during which the deflection Literatures on back analyses of the
remained more or less stable at about 32mm. performance of the instrumented diaphragm
The groundwater table stood at 2.00m 2.50m walls with struts and anchored tiebacks
depth. The site received heavy rains at this showed that Hardening Soil (HS) model
stage, and the wall started moving and reached predicts the wall deflection more realistically
a stabilised maximum deflection of about 65 than the Mohr-Coulomb (MC) model [Suched
mm at the wall top. The groundwater table Likitlersuang et al. (2013) and Bin-Chen
was at 0.50m below ground level after the Benson Hsiung and Sy-Dan Dao (2014)]. HS
rains. Even though the rains stopped, further model with undrained B parameters was tried
work could not commence for another 15 in the present case, but the results were not
days. significantly different.
-4-
8TH MADHAV LECTURE, GEOPRACTICES 2019- IGS HYDERABAD, 23 NOV 2019
Elevation, m
Elevation, m
8 8
4 4
negative pore pressure. The bending of the 2 2
wall towards the soil side during the initial 0 Estimated - 0.1 interaction
and Wedge failure
0
to each other during the computation. The new Fig. 8 (a) Estimated and observed wall
study was carried out with zero interaction deflection @ 6.5m excavation and long after
between the wall and the top soil layer and the the rains (b) Estimated BM distribution
estimated wall deflection along with the In this case also, an appreciable match was
observed one is presented in Figure 7a. The observed between the estimated and measured
corresponding BM distribution is shown in deflection and the BM distribution resulted
Figure 7b. from the analysis was used to confirm the
BENDING MOMENT
structural safety. The differential bending
deflection at this stage is 1/1030, well within
WALL DEFLECTION AT 6.60M EXCAVATION
(AFTER RAINS) BM, kNm/m
Displacement, mm
0 10 20 30 40 50 60 70 80 90 100
16
-400 -300 -200 -100 0 100
the structural acceptability limit of 1/300.
16
14
14
The analysis was further done from this
12 12
stage for the remaining excavation steps so
that the stability was ensured. Since the
10 10
Elevation, m
8
further excavations were continued after
Elevation, m
8
6
providing the inclined anchors as designed,
6
4 4
2 2
the wall deflections and bending moments
0 Estimated - No interaction 0 remained within the estimated values. Figures
-2
Measured I-04
-2 (b) 9a and 9b illustrate measured wall deflections
Measured I-05 (a)
-4 -4 comparable with the estimated ones.
Fig. 7 (a) Estimated and observed wall
deflection @ 6.5m excavation and rains (b) WALL DEFLECTION AT 10.20M EXCAVATION WALL DEFLECTION AT 12.5M EXCAVATION
Displacement, mm Displacement, mm
Estimated BM distribution 16
0 10 20 30 40 50 60 70 80 90 100
16
0 10 20 30 40 50 60 70 80 90 100
14 14
The estimated and observed wall deflection 12 12
8 8
2 2
30mm after maintaining a stable position for 0 Estimated +5.60m exc 0 Estimated -+3.40mExc
about 15 days given suspicion to a wedge -2
I05-+5.60m Exc
I04- +5.6mExc (a) -2 I05-+3.4m EXc
(b)
failure of the soil mass behind the wall. There -4 -4
was minor subsidence behind the wall Fig. 9 (a) Estimated and observed wall
partially closing the gap at the surface. The deflection @ 10.2m excavation and first level
anchors (b) @ 12.5m excavation and 2nd level
analysis was performed manually inducing
anchors
45° wedges. This analysis was necessary for
ascertaining the structural stability of the wall. The assumption of no interaction with the
The results of the analysis along with the site wall may be the right approach but may not be
measurements are presented in Figure 8a. The suited to all site conditions. Figures 10a and
-5-
8TH MADHAV LECTURE, GEOPRACTICES 2019- IGS HYDERABAD, 23 NOV 2019
10b illustrate the shear stress distribution on reverses as the interface is almost zero. Since
the D wall in the design stage estimation with this analysis did not provide any insight into
0.7 interface factor and zero interface factor the reasons for the anomaly in the moment
for the top layer. distribution, further study is done modelling
the clay as undrained A type with c’ and φ’. It
is seen that as the c’ increases for
accommodating high undrained strength, the
bending moments are on the excavation side
and the reversal occurs when c’ is as small as
10 kPa. The apparent prop action comes into
play with large cohesion intercept and always
when undrained B model is used.
(a) (b)
The shear stress in the D wall top portion 0.7 0.5 0.3 0.1 0.01
suggests a propping action forcing the wall Fig. 12 BM distribution on a 600mm thick wall
bend towards soil when interface reduction is @ 6.0m excavation for different soil wall
0.7 that is not happening in reality. interface constants
To study further, analyses of a 15m deep
wall in a single layer soil profile comprising 5. CONCLUDING REMARKS
sand and firm clay with 6.0m excavation were The Plaxis 2D analysis modelling firm clay
performed assuming 0.7 interface reduction with undrained B procedure did not capture
factor. The clay was modelled as undrained B the cantilever behaviour of an unstrutted wall.
with cu as the input parameter. The results are Analyses with zero interface and manually
presented in Figures 11a and 11b. Even induced wedge failure showed agreement
though the wall is cantilevered for 6.0m in with the actual performance. Present case
both the cases, the BM distributions are on study reveals the limitations when using
different sides of the wall. undrained B model and the necessity of
additional manual input during the design
stage for predicting more realistic field
performance.
Even though 6.0m cantilever wall is stable
in form to stiff clay, the cantilever length of
the wall should have been limited to 4.0m for
stability under unforeseen environment as
experienced in the present case.
6. REFERENCES
Bin-Chen Benson Hsiung and Sy-Dan Dao (2014) Evaluation
Fig. 11 BM distribution on a 600mm thick wall of Constitutive Soil Models for Predicting Movements
@ 6.0m excavation (a) in sand layer and Caused by a Deep Excavation in Sands, EJGE, Vol. 19,
pp 17325-17344.
(b) in clay layer
Likitlersuang S, Surarak C, Wanatowski D, Erwin Oh, and A.
Analysis was performed with different Balasubramaniam (2013), Finite element analysis of a
interface reduction factors for clay, and the deep excavation: A case study from the Bangkok MRT,
results are shown in Figure 12. The analyses Journal Soils and Foundations by The Japanese
Geotechnical Society. Elsevier B.V., Volume 53, Issue 5,
show that the BM distribution remains more October 2013, Pages 756-773.
or less same on the excavation side for
interface constants more than 0.01 and it
-6-