วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.

3, 2009

Influence of Soil Models on Numerical Simulation of Geotechnical works in Bangkok subsoil

Tanapong Rukdeechuai, Pornkasem Jongpradist, Anucha Wonglert, Theerapong Kaewsri
Department of Civil Engineering, Faculty of Engineering, King Mongkut’s University of Technology Thonburi, Bangkok 10140
Thailand
Tel 0-2470-9134
E-mail: [email protected]

Abstract
In the present paper, series of analyses of geotechnical works profile and its geotechnical engineering properties, stiffness of
in Bangkok subsoil using three soil models with different structure and support system. The finite element method
levels of complexity are carried out by FEM and discussed (FEM) is often used to predict ground movements induced by
with observed data. The analyses presented include pile load such soil-structure interaction problems. The interaction
test, deep excavation and tunneling. The results obtained by between existing structures and underground activities is a
using Mohr-Coulomb model, Hardening Soil model, complex phenomenon in which the behavior of the
Hardening Soil model with small strain stiffness are surrounding ground is one of the main aspects to be taken
compared. The impacts of the chosen constitutive models on into account. Consequently, a reasonable ground model is
numerical analysis of such geotechnical works are shown. It crucial in order to estimate the magnitudes and distribution of
can be concluded that using more sophisticated constitutive the strains. The constitutive model frequently used in
models which includes non-linear pre-failure and high numerical simulation of an underground work is linear elastic
stiffness under very small strain considerably improves the perfectly plastic with a Mohr–Coulomb (MC) failure
movement prediction. criterion. The greatest advantage of MC is that only five
parameters, which includes two elastic parameters (i.e.,
1. Introduction Young’s modulus E and Poisson’s ratio ν) and three plastic
In dense urban environments where land is scarce and
‫ ﺷﺣﯾﺢ‬- ‫ﻧﺎدر‬
parameters (i.e., friction angle φ, cohesion c and dilatancy
buildings are closely spaced, tunneling and cut-and-cover angle ψ), are sufficient in describing the plastic behavior.
excavations are widely used for basement construction and Moreover, the parameters can be easily determined.
development of underground facilities. One of the main However, the model does not take into account the
design constraints in these projects is to prevent or minimize fundamental aspects of soil behavior, such as variation of
damage to adjacent structures. Since all buildings are modulus according to stress state and different modulus in
‫اﺳﺗﺑﻌد‬
supported by foundations, to eliminate or reduce the loading and unloading conditions. Therefore, in general, the
possibility of such damage, an effective method is needed to numerical results by MC are in good agreement with those of
accurately ‘‘predict” the excavation-induced movement of field observation at a certain strain range for each case. To
building foundations for such complex condition. The achieve good agreement with the measured values, the
magnitude of the settlement and lateral movement and their parameters are usually obtained from back calculation by
distributions depend on a large amount of factors, such as soil curve fitting with history records. Although this technique

RECEIVED 4 March, 2009

ACCEPTED 12 June, 2009
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

could improve the accuracy of prediction in practice, it is parameters as mentioned above. Since it is a well-known
only applicable for specific range of each work. To overcome model, the details will not be described herein.
such shortages, it is necessary to consider at least an elasto- The Mohr-Coulomb yield condition is an extension of
plastic model with isotropic hardening. In addition to Coulomb’s friction law to general states of stress. In fact, this
previously mentioned aspects, the non-linear pre-failure condition ensures that Coulomb’s friction law is obeyed in
initial stiffness of soil at small strain range is necessary for any plane within a material element. The full Mohr-Coulomb
analysis of some geotechnical earthworks [1]. Figure 1 yield condition consists of six yield function when formulated
illustrates the reduction of stiffness with increasing strain and in terms of principal stresses [3]. The yield surface in
typical strain ranges of some geotechnical works. principal stress is shown in Figure 2.
In this paper, three constitutive models with enhancing
levels of complexity are used to simulate three types of
geotechnical works (deep excavation, tunneling and pile load
test) in Bangkok subsoil condition. First, the selected models
are described and the calibration of models on the basis of the
extensive in situ test results for stiff clay and laboratory test
results for soft clay are carried out. The impact of selecting a
soil model for geotechnical work analysis on accuracy of the
predictions of soil displacements is highlighted.

Figure 2 The Mohr-Coulomb yield surface in principal stress

space (c=0) [4]

2.2 Hardening soil model (HS)

The hardening soil (HS) model is derived from the hyperbolic
model of [5], with some improvement on the hyperbolic
formulations. The model can be adapted to all types of soil.
Figure 1 Characteristic stiffness-strain behavior of soil with The input parameters are as follows:
typical strain ranges for laboratory tests and structure [2]. • resistance parameters including friction angle, φ, the
cohesion, c, and the dilatancy angle, ψ
2. Soil models • stiffness of the soil defined by parameters E50, Eoed, which
2.1 Mohr-Coulomb model (MC) governs the volumetric behavior (where E50 denotes for
The linear elastic perfectly plastic behavior, so-called the secant modulus at 50 percent of maximum stress, Eoed is the
Mohr-Coulomb Model (MC) was used with its five slope of stress-strain curve obtained from oedometer test) and
Eur the unloading–reloading modulus, and a Janbu [6] type

18
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

parameter m that governs stress dependent stiffness according model incorporates a formulation of small-strain stiffness.
to a power function. Many authors have studied the behavior of soils using high
• yield function, f which has the following formulation precision triaxial tests. They obtained a reversible behavior
and high stiffness for strains less than 10-5 and showed that
f = f − yp (1) the shear modulus was constant under very small-strain
(strains between 10-6 and 10-5). This behavior is described in
where f is a function of stress and hardening parameters the HSsmall model using an additional strain-history
calculated by parameter and two additional material parameters, i.e., G0 and
γ0.7. G0 is the small-strain shear modulus and γ0.7 is the strain
1 c ⋅ cot ϕ + σ 3 m q 2q c ⋅ cot ϕ + σ 3 m
(2) level at which the shear modulus has reduced to 70% of the
f = ⋅( ) ⋅ − ⋅( )
E50ref c ⋅ cot ϕ + p ref 1−
q Eurref c ⋅ cot ϕ + p ref small-strain shear modulus.
qa

and y p is a function of plastic strains. The parameter q 2.4 Small Strain Stiffness
denotes deviatoric stress, Eurref is the unload-reloading Soil stiffness is an important soil parameter when ground
modulus at reference stress and qa is the ratio of ultimate deformations in engineering earthworks are analyzed. Soil
deviatoric stress to failure ratio. stiffness exhibits strong non-linearity. The strain dependency
of the stiffness is an important parameter to predict the
y p = ε1p − ε 2p − ε 3p = 2ε1p − ε vp (3) precise ground deformation at small strain level. Jardine [8]
reviewed the overall ground movement with typical working
•The cap-type yield function, f cap which has the formulation load from many case histories such as shallow foundations,
piles, excavations and tunnels, in various type of soils. It was
q2 shown that the ground strains mostly range between 0.001 to
f cap = + ( p + c ⋅ cot ϕ ) 2 − ( p p + c ⋅ cot ϕ ) 2 (4)
M2 0.5 percent. According to case histories reported, the strains
in the ground of hard soil are mostly less than 0.1 percent and
where the largest strain is 0.5 percent.

6 ⋅ sin ϕ (5)
M=
3 − sin ϕ
3. Bangkok subsoil conditions and calibration of soil
parameters
and p p is isotropic pre-consolidation stress. Bangkok Soft Clay was deposited in marine conditions at the
delta of the rivers in the Chao Phraya Plain. The typical
2.3 Hardening soil with small strain stiffness model Bangkok subsoil is shown in Figure 3. The soil properties
(HSsmall) used in the analyses are mainly determined from local
‫ﯾﺷﻛل‬
The HSsmall model constitutes an extension of the HS investigated data correlation from comprehensive in-situ tests
model. All the features of the HS model are included in the [9] and previous laboratory tests [10-13]. Figures 4 and 5
HSsmall model [7]. In addition to the HS model, the HSsmall show the examples of in situ data for strength and coefficient

19
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

of lateral earth pressure of Bangkok soils from previous mass

transit projects [9] adopted in this study.

Figure 4 Shear Strength of Bangkok soft clay and first stiff

clay [9]

Figure 3 Description of typical Bangkok subsoil soil profile

[14]

Figure 5 Coefficient of earth pressure for Bangkok clay, first

stiff clay, and Bangkok aquifer sand [9]

20
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

The numerical models of drained triaxial tests corresponding to 500Su, where Su is undrained shear strength. Based on
to different stress paths are formulated to calibrate the soft Bender element test [11], the maximum shear modulus ranges
clay parameters of each model. The clay samples referred in from 400Su to 570Su. In this study, the value of 570Su is used.
this analysis are undisturbed soft Bangkok clay taken from a In stiff clay layers, it is difficult to collect an
site within the Asian Institute of Technology campus. undisturbed sample for laboratory test. Thus, it is more
Samples from a depth of 3 to 4 m are selected using the convenient to relate the low strain shear modulus of clay to
existing triaxial test results [12-13]. The stress paths followed the shear wave velocity by using the empirical formula [16]
in this calibration are compression loading for pile problem, as
compression unloading for deep excavation and extension
unloading for tunneling. Using the Ko value from Figure 5, G0 = ρVs2 (6)
the results of triaxial tests for stress path of 72, 135 and 252
degree are chosen. Figures 6 to 8 show examples of the where ρ is density and Vs is the shear wave velocity. The
calibration results in terms of stress-strain and volumetric shear wave velocity can be calculated by using the empirical
strain-shear strain, relationships corresponding to the equation correlated to SPT number (N) [16] as
experimental data from Uchaipichat [12] and Navaneethan
[13]. The calibrated as well as correlated parameters for the Vs = 80.2 N 0.292 (7)
three models of five relevant layers are summarized in Table
1 and 2. By using the above formulas, the calculated G values of stiff
0

Based on a study by Teparaksa [15], it was found that soils are found to be in the range of 1500Su to 3000Su as
the elastic modulus of Bangkok clay can be calculated with suggested by Seed and Idriss [17].
investigated parameter, Su from field vane shear test. The strain level where the shear modulus reaches 70% of
Teparaksa suggested that the ratio of elastic modulus to initial value (γ0.7) is calculated by using Hardin-Drnevich
undrained shear strength of soft Bangkok clay and stiff clay [18] relationship as
were Eu / S u = 500 and 2000 for E (MC) and E50ref (HS and
HSsmall), respectively. Additionally, the E oedref is assumed as γ 0.7 =
1
[2c′(1 + cos(2ϕ ′)) − σ 1′(1 + K 0 ) sin(2ϕ ′)] (8)
9G0
E50ref [4]. The stress dependent Young’s modulus Eurref for
unloading/reloading has been calculated from the secant
Table 1 MC soil model parameters
modulus E50ref by assuming a ratio of Eurref / E50ref equal to 3 [4].
Soil layer Wea. Soft Med. Stiff Sand
Since the empirical formula from the above was proposed for
Crust Clay Clay Clay
total analysis, the Poisson’s ratio of 0.5 is used to convert
γsat[kN/m ] 17
3
16 18 18 20
total parameters to effective parameters. The low strain
ν ′ [-] 0.32 0.33 0.33 0.33 0.3
stiffness of Bangkok clay is determined from previously
ϕ ′ [°] 22 22 22 22 36
investigated data correlation. For soft clay, by using the
E’ [kPa] 570Su 570Su 1000Su 2000Su 2600N
laboratory test and in-situ field vane shear test [10], it was
found that the maximum shear modulus, G is around 300Su
0

21
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

Table 2 HS and HSsmall soil model parameters 60.0

Soil layer Soft Med. Clay Stiff Clay 50.0

Deviator stress, q (kPa)

Clay 40.0

E50ref [kPa] 500 Su 1000 Su 2000 Su 30.0

E oed [kPa] E50

ref ref ref
E50 E50ref 20.0
Test 135 Degree(Uchaipichat, 1998)
HSsmall
10.0 HS
Eurref [kPa] 3 E50ref 3 E50ref 3 E50ref MC

0.0

G0ref [kPa] 570 Su * 0.0 1.0 2.0 3.0 4.0 5.0

ρ (80.2 N 0.292 2
) ρ (80.2 N 0.292 2
) Shear strain, εs (%)

γ 0.7 [-] Eq(8) Eq(8) Eq(8) (a)

[18] [18] [18] 0
Test 135 Degree (Uchaipichat, 1998)

m [-] 1[4] 1[4] 0.5[4] -1 HSsmall

HS
MC

pref [kPa] 100 65 95 -2

Volumetric Strain, εv (%)

-3

-4

-5

120.0 -6

100.0 -7
0 1 2 3 4 5
Deviator stress,q (kPa)

Shear Strain, εs (%)

80.0

60.0
(b)
40.0
Test 72 Degree (Uchaipichat, 1998)
Figure 7 Deviator stress-shear strain (q-εs) (Figure a) and
HSsmall
20.0 HS
MC Volumetric strain-shear strain (εv-εs) (Figure b) for
compression unloading-stress path of 135 degree (comparison
0.0
0.0 2.0 4.0 6.0 8.0
Shear strain, εs (%)

between model predictions and drained test data)

(a)
3
Test 72 Degree (Uchaipichat, 1998)
HSsmall 0.0 1.0 2.0 3.0
HS
0.0
MC Test 252 Degree(Navaneethan, 1998)
-10.0 HSsmall
Volumetric Strain, εv (%)

2 HS
-20.0 MC
Deviator stress, q (kPa)

-30.0

1 -40.0

-50.0

-60.0
0
0 2 4 6 8
-70.0
Shear Strain, εs (%)
-80.0
Shear strain, εs (%)
(b)
Figure 6 Deviator stress-shear strain (q-εs) (Figure a) and (a)
Volumetric strain–shear strain (εv-εs) (Figure b) for
compression loading-stress path of 72 degree (comparison
between model predictions and drained test data)

22
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

0
ground is set up, the installation of pile is simulated. The load
Test 252 Degree (Navaneethan, 1998)
HSsmall
HS increment is gradually applied at the pile head following the
MC

actual values and the settlement at pile head is recorded. In

Volumetric Strain, εv (%)

-1

the analysis, the value of 0.9 for interface of soil-structure is

-2 used. The results in terms of pile load-settlement curve from
the simulation and measurement data are compared. Special
-3
0 1 2 3
attention is paid to the maximum pile settlement at the
Shear Strain, εs (%)
working and maximum loads.
(b)
Figure 8 Deviator stress-shear strain (q-εs) (Figure a) and Table 3 location of bored piles data of Bangkok soft clay
Volumetric strain-shear strain (εv-εs) (Figure b) for extension No Project name Process Diameter Depth of
unloading-stress path of 252 degree (comparison between of pile of pile (m) pile (m)
model predictions and drained test data) 1 8th Tower Dry 0.6 24
Ladpraow rd. process
4. Finite element models and analysis conditions 2 Krungdhon Wet 0.80 44.99
In this section, the finite element Program, PLAXIS 2D, is Hospital process
used to analyze three types of problem. In the analyses, soil 3 6th Tower Dry 0.8 27
layers are divided and the average value parameters are Rama6 rd. process
assumed for each layer. The ground water table is assumed to 4 Bangplee over Wet 0.8 47.2
be 1.5 m below the ground surface. The sand layer is fly bridge process
modeled with Mohr-Coulomb model only. The extension of 5 Popular Wet 0.8 45.7
the finite element mesh is wide enough for each case to Tower process
ensure the accuracy of the analysis. The model is fixed in
6 Becthai Wet 0.8 50.31
horizontal direction on both left and right sides. The bottom
process
boundary is fixed in vertical direction. The well-documented
case histories of each type of problem are collected. The
4.2 Deep excavation
measurement data of each case are compared with simulated
Plane strain models with suitable boundary conditions and FE
results to evaluate the impact of soil model chosen for
mesh are constructed for 4 cases including Green Tower,
surrounding soils in geotechnical analyses.
Thammasat University, Orifleme Building and BMAH
project. All projects were excavated using diaphragm wall. In
4.1 Pile problem
the analyses, the actual excavation depth, wall thickness and
Axis-symmetric models of six cases listed in Table 3, are
excavation sequence are employed. Only one half excavation
created to simulate the pile load test. The concrete pile is
width is modeled. The diaphragm wall and basement slab are
treated as linear elastic material using solid elements with E
modeled with plate elements. The joint connection between
of 2 x 107 kPa and ν of 0.3. After the initial condition of the
basement slabs and diaphragm wall are modeled as semi-rigid

23
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

connection by rotational stiffness. At the level of final 5. Numerical results and comparison with measured
excavation depth, the existing bore piles are considered. The data
bore piles are modeled by elastic model. The elastic modulus 5.1 Pile load test
of bore pile is transformed to plane strain analysis using Lee From the results of pile load test, comparisons between the
et al. [19] approach. The measurement data of few analysis results of the FEM and the observed values for 6
construction stages for each case with the total number of 12 bored pile cases from 6 locations of Bangkok are employed
sets are collected. The selection of inclinometer location is for evaluation purposes. As an example, the boundary
done with consideration of plane strain condition. Only the condition and mesh of FEM analysis case Krungdhon
measured points that are far away from edge of construction Hospital project is shown in Figure 9. Figure 10 shows
are selected. comparisons between the analysis results from FEM and the
observed values. It shows that the tendencies of the analytical
4.3 Tunneling results of HSsmall models are satisfactory in terms of the
The constructions of tunnel in MRTA project are chosen as magnitude and the shape of the wall compared with other
well-document case studies. The observation points are models. The simulated settlement from MC model is highly
CS-9A, CS-8E and CS-8C in section C Rama IX - over-estimated whereas slightly over-estimated settlement is
Phetchaburi zone located at the depths of 17.1, 19 and 20.5 obtained from HS model. Figure 11 shows the maximum pile
m, respectively [20]. The outer diameter of tunnel is 6.3 m settlement at the working and maximum loads. Pile
and the tunnel lining is 0.3 m thick. Plane strain models are settlement from analyses compared with the observation
constructed to analyse tunnel excavation by shield tunneling results of all case studies. The result from FEM by using
method. The excavation is modeled by deactivation of soil HSsmall model gives highest accuracy for the entire range of
elements situed in excavation zone after the initial conditions the observed settlement. The HS model can predict the pile
have been applied. Immediately after the soil elements are settlement fairly well but slightly over-predict the values. The
removed with application of volume loss, the tunnel lining results from MC model seem to over-predict the settlement
elements which are treated as solid elements having liner for all observed data. The difference between the observed
elastic properties with E of 3.1 x 107 kPa and ν of 0.2, are and predicted loads seems to increase with magnitude of the
activated. In the analyses, the volume loss was varied to get settlement.
the appropriate distance of surface settlement and then the
values of surface settlement at tunnel center (maximum), 5.5
and 11 m from tunnel center are discussed with measurement
data.

Figure 9 Finite element mesh of Krungdhon Hospital project

24
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

0 100 200
Load (ton)
300 400 500 600
the magnitude and the shape of the wall while those of MC
model are too high.
0.0
Field Test
HSsmall
HS
-2.0 MC
Settlement (mm.)

-4.0

-6.0

-8.0

Figure 10 Pile settlement curves from analysis using Figure 12 Finite element mesh of BMAH case study
different soil models compared to measurement data of
Wall displacement (mm.) Wall displacement (mm.)

Krungdhon Hospital project 0

0 5 10 15 20
0
0 5 10 15 20 25

15 -5 -5

-10 -10
Depth (m.)

Depth (m.)
10
-15 -15
Prediction (mm.)

Stage 1 Stage 2

Observed Observed
-20 -20
HSsmall HSsmall
HS HS
MC MC
5
-25 -25

HSsmall (a) (b)

HS
MC Wall displacement (mm.)
0 5 10 15 20 25
0 0
0 5 10 15
Observation (mm.)

Figure 11 Comparison between predicted maximum pile

-5

settlement by different soil models and observation of all case -10

Depth (m.)

studies.
-15

Stage 3

5.2 Deep excavation -20

Observed
HSsmall
HS

For deep excavation work, the measurement is commonly -25

MC

expressed by inclinometer readings which are the horizontal

0 5 10 15 20 25

(c)
movements of retaining wall during and after excavation. As Figure 13 Horizontal wall displacements from analyses with
an example, the boundary condition and mesh of FEM different soil model compared to observed data; (a) stage 1 (-
analysis case BMAH project is shown in Figure 12. Figure 13 2.0 m.), (b) stage (-5.7 m.), (c) stage 3 (-11.0 m.)
compares between the analysis results of the FEM and
observed values. It shows that the tendencies of the analytical Figure 14 shows the maximum lateral wall movement from
results of HS and HSsmall models are satisfactory in terms of analyses compared with the observation results of all cases.

25
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

The predicted results by HSsmall give high accuracy for the shown in Figure 17. From FE analyses, HSsmall and HS
entire range of observed displacement. Both HS and MC model give high accurate prediction for the entire range of
model could closely predict the wall movement when the observed settlement. The results from MC model seem to be
wall movement is greater than 30 mm. It can be obviously under-predict for all observed data.
seen that for soil models which do not take the small-strain
stiffness into account, the lateral wall movement is over-
predicted.

60

50

40
Prediction (mm.)

30 Figure 15 Finite element mesh of MRTA point CS-8E case

study
20 Distance (m)
0 20 40 60 80 100
0.002
HSsmall
10
HS 0.001
MC
0.000
0 -0.001
0 10 20 30 40 50 60
Settlement (m.)

-0.002
Observation (mm.)
-0.003

Figure 14 Comparison between predicted maximum -0.004

horizontal displacement by different soil models and

-0.005

-0.006 Observed
Hssmall vl.0.75%

observation of all case studies. -0.007

-0.008
Hs vl.0.75%
MC vl.0.75%

Figure 16 Ground surface settlement in CS-8E point

5.3 Tunneling
(prediction by different soil models and observed data)
For tunneling work, the ground surface settlements were
measured by settlement marker. As an example, the boundary 14

condition and mesh of FEM analysis case MRTA point CS- 12

8E is shown in Figure 15. The results of ground surface 10

settlement are shown in Figure 16 for comparisons of the FE

Prediction (mm.)

analysis results and the observed values for case of MRTA 6

point CS-8E in section C Rama IX - Phetchaburi. It shows 4

that the analytical results from HSsmall models give 2 HSsmall

HS
satisfactory tendencies in terms of the magnitude and the 0
MC

shape of the wall compared to other models.

0 2 4 6 8 10 12 14
Observation (mm.)

The results of surface settlement at tunnel center (maximum), Figure17 Comparison between maximum predicted Ground
5.5 and 11 m from tunnel center are discussed with surface settlement by different soil models and observation of
measurement data at point CS-8E, CS-8C and CS-9A as all case studies.

26
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

6. Conclusion [4] Brinkgreve, R.B.J., “PLAXIS Version 8 Material Models

The analyses of three kinds of geotechnical works carried out Manual”, Delft University of Technology & PLAXIS b.v.,
in this paper shows the impact of soil model on the The Netherlands, 2002.
simulations. Using more sophisticated soil models [5] Duncan JM., and Chang CY., “Nonlinear analysis of
considerably improves the prediction of movements. The stress and strain in soil”, J Soil Mech Found Div ASCE96,
HSsmall model which includes non-linearity prefailure and 1970, pp. 1629–1693.
high stiffness under very small strain gives high accuracy of [6] Janbu, N., “Soil compressibility as determined by
movement prediction for all three types of work covering the oedometer and triaxial tests”, In: European conference on soil
entire range of observed data. The HS model gives acceptable mechanics and foundation engineering, Wiesbaden,
movement prediction for analysis of pile and tunnel whereas Germany, Vol. 1, 1963, pp. 19–25.
the movements in excavation work are over-predicted. MC [7] Benz, T., “Small-strain stiffness of soils and its numerical
model over-predicts the movement for analysis of pile and consequences”, Ph.D. thesis, Universitat Stuttgart, 2007.
excavation work whereas the movements in tunnel work are [8] Jardine, R.J., “One Perspective of the Pre-failure
under-predicted, particularly, at range of small movement. Deformation Characteristics of Some Geomaterials”,
Proceedings of International Symposium Pre-failure
Acknowledgement Deformation of Geomaterials, (ed. by Shibuya et al.),
The authors would like to thank the following; Seafco Public Balkema, Vol. 2, 1994, pp. 855-885.
Co. Ltd, STS instruments Co. Ltd, Mr. Kittisak Daungpan [9] Prust, R.E., Davies, J., and Hu, S., “Pressuremeter
from Department of Highways for their support on observed investigate for mass rapid transit in Bangkok”, Thailand,
data. The permission to use V.I.Plaxis Service Program by journal of the transportation research board, Transportation
PLAXIS ASIA (PLAXIS BV) is gratefully acknowledged. research of the national academies, Washington D.C., No.
1928, 2005, pp. 207 -217.
References [10] Shibuya, S., Hanh, L.T., Wilailak K., Lohani T.N., and
[1] Burland, J.B., “Small is beautiful - the stiffness of soils at Tanaka H., “Characterizing stiffness and strength of soft
small strains”, Can. Geotech. J., Vol. 26 (4), 1989, pp. 499- Bangkok clay from in-situ and laboratory tests”, First Int.
516. Conf. on Site Characteristics, 1997.
[2] Atkinson, J.H., and Sallfors, G., “Experimental [11] Theramast N, “Characteristic of pseudo-elstic shear
determination of stress-strain-time characteristics in modulus and shear strength of Bangkok clay”, M. Eng.
laboratory and in situ tests”, Proceedings, 10th European Thesis, AIT, 1998.
Conference on Soil Mechanics and Foundation Engineering, [12] Uchaipichat, A., “Triaxial Tests on Soft Bangkok Clay
Florence, Vol. III, 1991, pp. 915-956. with Different Applied stress Paths”, M. Eng. Thesis, AIT,
[3] Smith, I.M., Griffith, D.V., “Programming the Finite Bangkok, 1998.
Element Method”, Second Edition, U.K., 1982. [13] Navaneethan, T., “Extension Behavior of Soft Bangkok
Clay with Selected Applied Stress Paths”, M. Eng. Thesis,
AIT, Bangkok, 1999.

27
วิศวกรรมสาร ฉบับวิจัยและพัฒนา ปที่ 20 ฉบับที่ 3 พ.ศ. 2552 RESEARCH AND DEVELOPMENT JOURNAL VOLUME 20 NO.3, 2009

[14] Wonglert, A., Jongpradist, P., Kalasin, T., “Wall [17] Seed, H.B. and Idriss, I.M., “Soil moduli and damping
Movement Analysis of Deep Excavation in Bangkok Subsoil factor for dynamic response analysis”, Report EEC 70-10,
considering Small Strain Stiffness”, Journal of Research in College of Engineering, University of California, Berkeley,
Engineering and Technology, Vol.5, No.4, October- CA, 1970.
December. 2008, pp. 393-405. [18] Hardin, B.O. and Drnevich, V.P., “Shear modulus and
[15] Teparaksa, W., Thasnanipan, N., and Tanseng, P., damping in soil: Design equation and curves”. Proc. ASCE,
“Analysis of lateral wall movement for deep braced Journal of the Soil Mechanics and Foundations Divisions,
excavation in Bangkok subsoils”, Proceeding of the Civil and 98(SM7), 1972, pp. 667-692
Environmental Engineering Conference, Bangkok, Thailand, [19] Lee, F. H., Yong, K. Y., Quan, K.C.N., and Chee, K.T.,
1999, pp.II-67 – II-76. “Effect of corners in strutted excavation: Field mornitoring
[16] Imai T., “P- and S-wave velocities of the ground in and case histories”, J. of Geotech. Eng., ASCE, Vol. 124,
Japan”, In:Proceedings of the IX international conference on No.4, 1998. pp. 339-349.
soil mechanics and foundation engineering, Vol. 2, 1977, pp [20] Suwansawat, S., “Earth Pressure Balanced (EPB) Shield
127–132. Tunneling in Bangkok: Ground Response and Prediction of
Surface Settlements Using Artificial Neural Network”, MIT’s
Dissertation, 2002.

28