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

net/publication/332833913

Finite Element Analysis of a Deep Excavation in Kenny Hill Formation with

consideration of Small Strain Stiffness

Conference Paper · September 2018

CITATIONS READS

0 3,000

1 author:

Kim Hing Law

KH Geotechnical Sdn Bhd
11 PUBLICATIONS   136 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Kim Hing Law on 03 May 2019.

The user has requested enhancement of the downloaded file.

 The IEM-CIE-HKIE Tripartite Seminar 2018, Putrajaya, Malaysia, 4 September 2018

Finite Element Analysis of a Deep Excavation in Kenny Hill Formation with

consideration of Small Strain Stiffness
Cheah, Frankie3,1, Law, K.H2 and Mohd Ashraf Mohd Ismail3
1
AECOM Perunding Sdn Bhd, Kuala Lumpur, Malaysia
2
KH Geotechnical Sdn Bhd, Kuala Lumpur, Malaysia
3
School of Civil Engineering, Universiti Sains Malaysia, Pulau Pinang, Malaysia
E-mail: Frankie.Cheah@aecom.com

ABSTRACT: Deep excavations give rise to movements in the surrounding ground with consequent potential for damage to surrounding
structures and buried services. When the excavations are in the vicinity of existing buildings and structures, it is necessary to assess the
influence of these excavations on the existing buildings such as building settlement and distortion. This paper presents a deep excavation
case history for a metro station which has been built in the Kuala Lumpur city centre. In this paper, the complexity of the excavation
sequences and challenges faced, especially on the excavation works carried out under live traffic condition will be described. The
performance of the deep excavation was also back analysed with Hardening Soil Small Strain model, an advanced constitutive model with
the consideration of small strain stiffness of the soil. The results demonstrate that reasonable agreement between measured and back-
calculated building settlement and wall displacement could be achieved when small strain stiffness of soil is taken into consideration. The
study of the effect of deep excavation of on adjacent building settlement was also conducted. It is demonstrated that to obtain a reasonable
agreement with measure building settlement, building stiffness needs to be taken into consideration in the numerical modelling on top of the
small stiffness of soil.

Keywords: Deep excavation, 2D finite element analysis, advanced constitutive model, small strain characteristic

1. INTRODUCTION station with a plan dimension of approximately 23 m width and 138

Deep excavations give rise to movements in the surrounding ground
with consequent potential for damage to surrounding structures and
buried services. When the excavations are in the vicinity of existing
buildings and structures, it is necessary to assess the influence of
these excavations on the existing buildings such as building
settlement and distortion. Hence, to ensure that the excavation-
induced deformations are within the allowable limit, comprehensive
finite element (FEM) analysis would be necessary. To produce a
comprehensive FEM analysis, several considerations are vital:-
m long (Figure 2), and 33 m deep was constructed using a top-down
construction method. Top-down construction method has been
adopted for this particular station as it provides a very rigid support
system during excavation process that would minimize adjacent
ground and buildings movements as compared to other construction
methodology.

 Detailed and comprehensive geological model of the project

with consideration of the soil layering, geological formation
and geotechnical parameters and characteristics;
 Detailed construction activities, technology, consideration of
the type of retaining structure that is most suitable for the
respective deep excavation;
 Suitable constitutive soil model that is able to predict the
ground deformation caused by excavation on top of retaining
structure during the excavation;
 Instrumentation scheme to monitor and observe the
performance of the retaining wall at each excavation stage with
consideration on necessary contingency measurement.
Figure 1: Location map of the respective underground station
In this paper, the complexity of the excavation sequences and
challenges faced, especially on the excavation works under live
traffic condition is presented. The performance of the deep
excavation was evaluated using both Hardening Soil (HS) model
(Schanz et al., 1999) and Hardening Soil with small strain stiffness
(HSS) model (Benz, 2007). The finite element analysis results
demonstrate that reasonable agreement between measured and soil
deformation and wall displacement could be achieved when small
strain stiffness of soil is taken into consideration. The study for the
building settlement behavior as compared to green field settlement
was also included in this study. The result reveals that the building
settlement could be realistically predicted with the consideration of
building stiffness.
Figure 2: Geometry for BB Station together with section indication
2. CASE HISTORY DESCRIPTION

2.1 General Description

The metro station for this project was built traversing through some
of the busiest areas in the Klang Valley of Kuala Lumpur. Figure 1
shows the location map of the underground metro station. The metro
 The IEM-CIE-HKIE Tripartite Seminar 2018, Putrajaya, Malaysia, 4 September 2018

2.2 Site Geology and Ground Conditions

Geological map of Kuala Lumpur, Malaysia, 1993 (Figure 3)
indicates that the site is underlain by weathered residual soils of
Kenny Hill Formation. It is also referred as meta-sedimentary rock
formation considering that the meta-sedimentary rocks have been
decomposed into residual and completely weathered soils (Grades V
and VI) as a result of intense weathering process of tropical climate.
The Kenny Hill Formation is a sequence of interbedded sandstones,
siltstones and shales/mudstones. The formation is also known to
have undergone mild and regional metamorphic event, resulting in
changes of sandstone/siltstone to quartzite and schist/phyllite
respectively (Wong & Muhinder, 1996). The geotechnical
investigation was carried out in the vicinity of the proposed metro
station.

Figure 5: Subsoil profile for section T2

Figure 3: Geological Map of Kenny Hill and Kuala Lumpur

Limestone Formation (Geological map of Selangor, Sheet 94)

Based on the completed boreholes, the subsoil condition within the

excavation site could be divided into three distinctive soil profile
based on the standard penetration test (SPT-N) result, namely T1,
T2 and T3 as shown in Figure 4, 5 and 6 respectively. Generally, the
SPT-N value is increasing with depth with a hard material (SPT-
N>100,blows/300mm) is encountered at approximately 30 m below
the existing ground level. The residual soil mainly comprises sandy
silt or clayey silt material. The depth of SPT-N>100 blows/300mm
layer is lower at section T2 as compared to section T1 and T3. The
measured groundwater table is located at a depth of about 4 m below
ground surface.
Figure 6: Subsoil profile for section T3
3. PLAXIS 2D FINITE ELEMENT ANALYSIS
3.1 Finite Mesh and Boundary Condition
The excavation geometry of the case history was carried out for a
plan area of approximately 23 m x 138 m. The ratio of primary
(long) wall length to complementary (short) wall length (L/B) is
about 6, suggesting that a plane strain 2D analysis is appropriate for
section T2 diaphragm wall (Ou et al., 1996). The numerical study of
the case history has, therefore, been conducted using 2D finite
element program PLAXIS 2D. Figure 6 shows the 2D finite element
mesh adopted in the numerical study. The left and right boundaries
were fixed horizontally but are free to move vertically, and the
bottom boundary was fully fixed. There are total 3 sections adopted
in this case history at both the design and construction stage.
However, for brevity, only section T2 results are presented and
discussed in this paper.

Figure 4: Subsoil profile for section T1

 The IEM-CIE-HKIE Tripartite Seminar 2018, Putrajaya, Malaysia, 4 September 2018

Table 2 shows the geotechnical parameters adopted for the case

study at Section T2.
Table 2: Soil Input Parameters for the finite element analysis
Symbol Unit Backfill Sandy Sandy
(MC) SILT,N=100 SILT, N=10
c' kPa 0.5 15 5
' [o] 30 29 28
 o
[] 0 0 0
E50ref MPa 10 174 17.4
ref
Eoed MPa - 174 17.4
E ref
ur
MPa - 522 52.2
Figure 7: 2D Finite Element Analysis (Section T2) m [-] 0.5 0.5 0.5
vur [-] 0.2 0.2 0.2
p ref kPa 100 100 100
3.2 Material Models and Input Parameters
Rint er [-] 0.67 0.67 0.67
During construction stage, the stress-dependent stiffness model, the 𝑟𝑒𝑓
𝐺0 MPa - 544 54
Hardening Soil (HS) model (Schanz et al., 1999) as implemented in - 2x10-4 2x10-4
finite element program PLAXIS 2D was employed for the numerical 𝛾0.7
prediction of this particular project. In this study, the Hardening Soil Symbol Unit Sandy Sandy Sandy
with small strain stiffness model (HSS) (Benz, 2007) was adopted to SILT,N=14 SILT, N=24 SILT, N=40
evaluate the both the ground and building settlement at the retained c' kPa 5 5 10
side of the diaphragm wall.  ' [o] 28 28 28
HSS model (Benz, 2007) requires 13 parameters for determination  [o] 0 0 0
of mechanical properties of soil in which 11 parameters are the same ref MPa 24.4 41.8 69.6
E50
as the HS model (Schanz et al. 1999). The descriptions of the input
parameters of the HSS model are shown in Table 1. ref
Eoed MPa 24.4 41.8 69.6
Eurref MPa 73.1 125 209
Table 1: Input parameters of HSS model
Parameter Unit Definition HS model m [-] 0.5 0.5 0.5
[o] vur [-] 0.2 0.2 0.2
' Friction angle Yes
- Cohesion Yes p ref kPa 100 100 100
c'
Rf - Failure ratio Yes Rint er [-] 0.67 0.67 0.67
ψ [o] Dilatancy angle Yes 𝑟𝑒𝑓
𝐺0 MPa 76 130 218
ref
E50 MPa Secant stiffness in standard Yes 2x10-4 2x10-4 2x10-4
𝛾0.7
drained triaxial test
E ref MPa Tangent stiffness for Yes
oed 3.3 Structural Element Input Properties
primary oedometer loading
E ref
ur
MPa Unloading-reloading Yes For the excavation structures, 6 noded isotropic linear elastic plate
stiffness element were used to model both the diaphragm wall and floor
ν - Poisson’s ratio Yes slabs. The 1.2 m thick diaphragm wall was modelled as elastic
K0 - Yes material with concrete Young’s modulus and Poisson’s ratio of 28
𝑟𝑒𝑓 GPa and 0.2 respectively. To compensate for the overlapping of unit
𝐺0 MPa Shear modulus small strains Yes
weight and volume between soil and other material, the soil unit
𝛾0.7 Reference shear strain No
weight should be subtracted from the real unit weight of concrete
OCR Over-consolidation ratio Yes materials. Node to node anchor was used to model the steel struts to
support the diaphragm wall with axial stiffness determined by EA,
According to Law et al. (2014a), the general problem in the analysis where E is the Young’s modulus of the steel and was taken as E =
and design of deep excavations in weathered residual soils of the 205 GPa, and A is the cross-sectional area of the steel strut. Table 3
Kenny Hill Formation are the fact that soil tests data is often limited shows the structural elements input parameters adopted in the
or low-quality due to the friable nature of the weathered samples. numerical analysis.
Given that, the soil stiffness parameters are usually correlated with
the Standard Penetration Test (SPT) data for the design of deep Table 3: Structural Elements Input Parameters
excavations in Kenny Hill formation. The empirical correlation with Structures Concrete Wall E ν
SPT-N value is well reported by Law et al. (2013, 2014a, 2014b). Grade Thickness
(MPa) (m) (kN/m2)
The additional two parameters of the HSS model include Goref is a Diaphragm wall 40 1.20 28 x 106 0.2
reference initial shear stiffness corresponding to the reference Tunnel Lining 60 0.275 28 x 106 0.2
pressure pref and shear strain  0.7 at which the secant shear modulus Slab 40 Varies 28 x 106 0.2
is reduced to 70% of Go. In this study, Goref was obtained by first Roof 40 1.20 28 x 106 0.2
determining the Eo/Eur ratio based on the chart by Alpan (1970) and Concourse 40 0.70 28 x 106 0.2
assuming Eur = 3E50, where Eo is the small strain Young’s modulus, Upper Platform 40 0.95 28 x 106 0.2
and subsequently using the expression Goref = Eoref / 2(1+vur) with vur Plant 40 0.95 28 x 106 0.2
taken as 0.2. Vucetic and Dobry (1991) reported that  0.7 only varies Lower Platform 40 1.80 28 x 106 0.2
within a narrow range between 1 x 10-4 and 4 x 10-4, in this paper
 0.7 = 2 x 10-4 was assumed.
 The IEM-CIE-HKIE Tripartite Seminar 2018, Putrajaya, Malaysia, 4 September 2018

4. FINITE ELEMENT MODEL STAGE CONSTRUCTION 5. RESULTS AND DISCUSSION

The metro station is located along Jalan Bukit Bintang that suffers The diaphragm wall at section T2 was supported by four levels of
heavy traffic flows everyday. To minimize the disruption to the basement slab namely roof slab, concourse slab, under upper
traffic flow, the construction of the diaphragm wall and the roof slab platform (UUP) slab and plant room (PR) slab and one level of
has to be carried out in two phases. Table 4 describes the horizontal steel strut installed between PR slab and final excavation
construction sequences adopted in this particular project. The level (FEL). In this paper, only the displacement of the left hand
sectional detail of the excavation is shown in Figure 8. side diaphragm wall is presented.
Table 4: Construction Sequences
5.1 Lateral Wall Displacement
Stage Description
1 Divert traffic to Facilitate the Installation of Diaphragm The predicted lateral wall displacement at section T2 from the finite
wall, Barrette piles for king posts and sheet pile wall in element analysis is presented for both HS and HSS model, see
Phase 1 Figure 9. Data from inclinometer which was installed in the
2 Excavate 3m below Existing Ground level for Phase 1 diaphragm wall at section T2 (INW11) was used to evaluate the
3 Installation of the Temporary Strut S1a performance of two constitutive models.
4 Further Excavation to the Soffit of the Roof Slab
5 Construct Roof Slab and then Backfill Until Existing Road
level after Removal of Temporary strut S1a
6 Divert the Traffic and start the Phase 2 Construction of Roof
Diaphragm wall, Barrette Piles for Kingposts and Complete
Roof slab.
7 Lower Tunnel & Upper Tunnel Bore-through Concourse
8 Further Excavate to Concourse Slab & Cast Concourse
Slab.
9 Repeat Step 8 (Exc. & Casting) for Upper Platform Slab. Upper Platform
10 Repeat Step 8 (Exc. & Casting) for Plant Room Slab.
11 Further Excavate to S1level
12 Install S1 Strut and Preload Plant
13 Excavate to Final Excavation Level and Cast the Base Slab Room
14 Remove S1 and cast the Permanent Column
15 Remove the Temporary Kingpost
FEL
Backfill to Existing Ground Level and Activate the Traffic
above

Figure 9: Predicted Wall Lateral Displacement at Formation

The predicted maximum lateral wall displacement (δx) at the final

excavation level is about 108 mm and 76 mm for HS model and
HSS model respectively, see Table 4.

Table 4: Displacement at Formation with Different Soil Model

Max. Ratio Max.
Displacement Displacement
(δx) mm (δx)
Soil Model
to Maximum
Excavation Depth
H
Hardening Soil (HS) 108 0.33%H < 0.5%H
Hardening Soil Small 76 0.26%H < 0.5%H
Strain (HSS)

As can be observed from Figure 9, both the HS and HSS models are
considered reasonable and able to simulate realistic wall
displacement profile until the final excavation level.
Figure 8: Cross section of the excavation stage of the case history
It should be noted that during construction stage, to accommodate
with the construction sequence and the changes of the probe wheels
on the inclinometer (INW11) due to measurement error, calibration
of the inclinometer data needs to be carried out. Hence, in order to
continue with the monitoring works for the subsequent excavation
 The IEM-CIE-HKIE Tripartite Seminar 2018, Putrajaya, Malaysia, 4 September 2018

stage, the monitoring data is then reset from the concourse floor 5.2 Green Field Settlement Prediction and Building Settlement
excavation stage. In view of the above, the predicted lateral wall Marker Measurement
displacement at final excavation stage as presented in Figure 10 was
incremental displacement from the concourse floor excavation stage. Hsieh and Ou (1998) reported that there are two types of the ground
settlement profiles, namely spandrel and concave type. Figure 12
shows the predicted ground settlement profile and measured
50 Year 2018 (HS) buildings settlement behind the diaphragm wall caused by the deep
Year 2018 (HSS) excavation works. As expected the ground settlement profile
45 INW11-11/6/2015 (concave shape) is closely correlated to the wall deflection mode
(deep seated bulging mode). The excavation for the metro station
40 are located at a very close distance to the existing shop houses,
therefore an extensive number of building settlement markers were
installed along the rows of shop houses which are supported by
35 shallow foundation. A total of 34 nos. of the building settlement
markers were placed along the shop houses as shown in Figure 11.
30 The highest building settlement measured is about 27 mm at marker
S148 near edge of the excavation while the building settlement
measured at about 30 m away from the edge of the excavation is
RL (m)

25
about 20 mm.
20

15

10

0
-0.020 0.000 0.020 0.040 0.060 0.080 0.100 0.120
-5 Deflection(m)
Figure 10: Comparison of Predicted and Measured Deformation at
Formation Level Figure 11: Building Settlement Marker placed for the monitoring

Figure 10 compares the measured and predicted lateral wall
displacement at final excavation level. As can be seen from Figure
10, the computed lateral wall displacement using HSS model closely
matched the measured wall displacement profile, whereas HSS
model grossly over-predicted the lateral wall displacement.
However, both HSS and HS models predict similar wall
displacement profile. This implies that both the HSS and HS models
can capture well the unloading behaviour of soil during the
excavation process. Table 5 compares the predicted and measured
maximum lateral wall displacement at final excavation level.
Figure 12 compares of the measured building settlement against the
predicted ground settlement profile. It is shown that for both models
grossly over predict the building settlement. The over prediction is
due to the fact that predicted settlement represents the ground
settlement at green field condition without taking into consideration
soil-structure interaction of existing buildings foundation.

Table 5: Comparison of wall displacement at final excavation level

after recalibration of Inclinometer INW11
Incremental Measured (%) from Compariso
Displaceme Incremental measured n
Soil nt Displacement (δx) with of
Model (mm) (mm) compariso (δx) against
n with H
INW11 (%)
0.32%H < Figure 12: Settlement Profile Behind the wall – Predicted versus
(HS) 106 70 34% Instrumented Value at the Final Excavation Level
0.5%H
0.22%H<
74 70 5% 5.3 Settlement Prediction Inclusive Building Stiffness and
(HSS) 0.5%H
comparison with monitoring value of Building Settlement
Table 5 shows that the predicted lateral wall displacement by Markers
adopting HSS model agrees well with the measured lateral wall As highlighted in Section 5.2, the over prediction of measured
displacement as compared to HS model. The ratio of the maximum buildings settlement is mainly due to the soil-structure interaction
displacement (δx) to excavation depth, H is 0.32%H and 0.22%H for effect. Hence, to further quantify and improve the predicted
HS and HSS soil model respectively. The computed maximum wall buildings settlement as compared to numerical result, the effect of
displacement for both models is within an allowable deformation building stiffness was included in the numerical model as suggested
limit by as specified by BS8002 that displacements should be by Maleki et.al (2010). The adjacent building was modeled as an
limited to 0.5% of total excavation depth, 0.5%H. The discrepancy equivalent beam in which each floor of building is considered as
between HS and HSS models could be due to the higher passive slab and considering building foundation footing as one level of
resistance developed within the diaphragm wall shaft resistance and floor slab. For building with m storey height, it could be modelled as
also due to the shaft resistance of the adjacent installed barrette pile. m+1 slabs with the consideration of the distance between slabs as H
 The IEM-CIE-HKIE Tripartite Seminar 2018, Putrajaya, Malaysia, 4 September 2018

and thickness of each slab is tslab and then based on the parallel axes REFERENCES
theorem (Timoshenko, 1995), second moment of area of equivalent
beam can be computed as input into the finite element model. The Alpan I. The geotechnical properties of soils. Earth Sci Rev
second moment of area, I and cross section area, A of the slab are 1970;6:5–49.
defined by equation (1) to (4) below. Benz T. Small-strain stiffness of soil and its numerical
3
𝑡𝑠𝑙𝑎𝑏 𝐿 (1) consequences. PhD Thesis, University of Stuttgart; 2007.
Islab = Law, K.H., Hashim, R., and Ismail Z. (2013) “Performance of multi-
12
𝐴𝑠𝑙𝑎𝑏 = 𝑡𝑠𝑙𝑎𝑏 𝐿 (2) propped deep excavation in Kenny Hill formation”, Proc. 18th
𝑚+1 (3) Southeast Asian Geotechnical Conference, Singapore, pp705-
(𝐸𝑐 𝐼)𝑏𝑒𝑎𝑚 = 𝐸𝑐 ∑ (𝐼𝑠𝑙𝑎𝑏 + 𝐴𝑠𝑙𝑎𝑏 ℎ𝑚 2 ) 712.
1 Law, K.H., Othman, S.Z., Hashim, R., and Ismail, Z. (2014a)
(𝐸𝑐 𝐴)𝑏𝑒𝑎𝑚 = (𝑚 + 1)(𝐸𝑐 𝐴)𝑠𝑙𝑎𝑏 (4) “Determination of soil stiffness parameters at a deep excavation
construction site in Kenny Hill Formation”, Measurement 47,
where pp645-650.
L is out of plane dimension of the slab Law, K.H., Hashim, R., and Ismail, Z. (2014b) “3D numerical
Ec is Young’s modulus of concrete analysis and performance of deep excavations in Kenny Hill
formation”, Proc. 8th European Conference Numerical Methods
In this study, m is equivalent to 3 for the existing 2 storey shop in Geotechnical Engineering (NUMGE), Delft, pp759-764.
houses behind the diaphragm wall. Figure 13 compares the predicted Ou, C.Y., Chiou, D.C., and Wu, T.S. (1996) “Three-dimensional
and measured building settlement. As can be seen from Figure 3, finite-element analysis of deep excavations”, Journal of
reasonable agreement between the predicted building settlement and Geotechnical and Geoenvironmental Engineering, 122(5),
the field measurement data could be achieved with the inclusion of pp337-345.
building stiffness in the numerical model. Schanz, T., Vermeer, P. A., and Bonnier, P. G. (1999) “The
hardening soil model: formulation and verification”, Beyond
2000 in Computational Geotechnics, Balkema, Rotterdam,
pp281-296.
Frankie, C., Mak, W.K and Law, K.H. (2016) Diaphragm Wall
Design Challenges in 33m Deep Excavation at Congested Urban
Site in Kuala Lumpur. Proc. Of 19th SEAGC, Kuala Lumpur,
pp703-707.
Linlong Mu, Maosong Huang. (2016) Small strain based method for
predicting three dimensional soil displacements induced by
braced excavation. Tunnelling and Underground Space
Technology 52 (2016)
Figure 13: Settlement Profile Behind the wall (Inclusive Building L. Sebastian Bryson, David G. Zapata. (2012) Method for
Stiffness) – Predicted vs Instrumented Value at the Final Excavation Estimating System Stiffness for Excavation Support Walls.
Level Journal of Geotechnical and Geoenvironmental Engineering.
M. Maleki, H. Sereshteh, M. Mousivand, M. Bayat (2011) An
It is observed that consideration of building stiffness leads to the equivalent beam model for the analysis of tunnel-building
decreased in soil vertical deformation and the building stiffness interaction.
contributes significantly to the soil-structure interaction to resist Vucetic M, Dobry R. Effect of soil plasticity on cyclic response. J
against vertical deformation of soil caused by the deep excavation Geotech Eng ASCE 1991;117(1):89–107.
works. Wong, J., and Muhinder, S. (1996) Some engineering properties of
weathered Kenny Hill Formation in Kuala Lumpur”, Proc. 12th
South Asian Geotechnical Conference, Kuala Lumpur, pp179-
6. CONCLUSION 187

Lessons have been learnt from this completed deep excavation

works for this particular underground metro station construction.
Based on the numerical study in this paper, the following conclusion
can be drawn:-
 For excavation works that involve complex construction
sequences together with many adjacent buildings and services,
an advanced constitutive model such as HS and HSS models
should be adopted.
 Comparison of predicted and measured wall deformation and
building settlement indicates that the adoption of HSS model
with consideration of small strain stiffness is more appropriate
as compared to HS model.
 To realistically predict the building settlement, the building
stiffness should be included in the numerical model. The
assessment of building settlement based on predicted green
field settlement profile may lead to unnecessary building
strengthening works and uneconomic design.

View publication stats