Sie sind auf Seite 1von 35

INTERNATIONAL JOURNAL FOR NUMERICAL AND ANALYTICAL METHODS IN GEOMECHANICS

Int. J. Numer. Anal. Meth. Geomech., 2004; 28:57–91 (DOI: 10.1002/nag.327)

Analysis of shield tunnel

W. Q. Ding1, Z. Q. Yue2,n,y, L. G. Tham2, H. H. Zhu1, C. F. Lee2 and T. Hashimoto3


1
Department of Geotechnical Engineering, Tongji University, Shanghai, China
2
Department of Civil Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China
3
Geo-Research Institute, Osaka, Japan

SUMMARY
This paper proposes a two-dimensional finite element model for the analysis of shield tunnels by taking
into account the construction process which is divided into four stages. The soil is assumed to behave as an
elasto-plastic medium whereas the shield is simulated by beam–joint discontinuous model in which curved
beam elements and joint elements are used to model the segments and joints, respectively. As grout is
usually injected to fill the gap between the lining and the soil, the property parameters of the grout are
chosen in such a way that they can reflect the state of the grout at each stage. Furthermore, the contact
condition between the soil and lining will change with the construction stage, and therefore, different stress-
releasing coefficients are used to account for the changes. To assess the accuracy that can be attained by the
method in solving practical problems, the shield tunnelling in the No. 7 Subway Line Project in Osaka,
Japan, is used as a case history for our study. The numerical results are compared with those measured in
the field. The results presented in the paper show that the proposed numerical procedure can be used to
effectively estimate the deformation, stresses and moments experienced by the surrounding soils and the
concrete lining segments. The analysis and method presented in this paper can be considered to be useful
for other subway construction projects involving shield tunnelling in soft soils. Copyright # 2004 John
Wiley & Sons, Ltd.

KEY WORDS: subway construction; shield tunnelling; soil–structure interaction; numerical procedure;
finite element method; ground settlement; lining; soft soils

1. INTRODUCTION

Since 1950s, urban developments in many cities have experienced continuous growth and
expansion. Because of the large populations and shortage of land resources, these cities have
always had a strong demand for efficient, economic and environmental friendly urban civil
infrastructure systems to accommodate the daily and routine travels of thousands and millions
of commuters. The subway system is an obvious solution to meet the demand. To minimize the

n
Correspondence to: Dr. Q. Z. Q. Yue, Department of Civil Engineering, The University of Hong Kong, Pokfulam
Road, Hong Kong, China.
y
E-mail: yueqzq@hkucc.hku.hk
Contract/grant sponsor: Research Grant Council of Hong Kong SAR Government
Contract/grant sponsor: Hong Kong Jockey Club Charities Trust

Received 23 August 2002


Revised 9 July 2003
Copyright # 2004 John Wiley & Sons, Ltd. Accepted 30 August 2003
58 W. Q. DING ET AL.

impact on the existing traffic during construction, tunnelling is usually adopted for the
construction of the subway. Because of its efficiency and safety, shield tunnelling is one of the
most popular tunnelling methods for the construction of subway tunnels in soft soil ground.
Over the last 30 years, the shield tunnelling method has experienced continuous improvement
and development. New shield tunnelling methods including the earth pressure balanced shield,
the slurry shield, the simultaneous backfill grouting as well as some improved grouting materials
have been introduced and developed in recent years [1].
Shield tunnelling will inevitably induce ground deformation and the surrounding soils will
also act on the shield lining segments. Quantitative and accurate prediction of such soil and
structure interaction will be of significant importance in many aspects including lining segment
design, construction safety, ground settlements, potential damage to existing structures and
facilities, and operation of the subway system.
Peck [2] developed an empirical method for the ground settlement associated with tunnelling
in soft soils by assuming the settlement trough to be a Gaussian distribution curve. The
actual ground settlement is predicted based on the estimation of the ground loss ratio. This
method was used and improved by many engineers and researchers such as Clough and
Schmidt [3] and Rowe et al. [4]. Recently, Schmidt [5] has proposed the semi-empirical
error function method for the estimation and prediction of ground settlement. Though these
empirical or semi-empirical methods are useful in the evaluation of the ground settlements
caused by shield tunnelling construction, they must be applied with caution as they may not be
applicable to situations significantly different from the cases on which they are based on
Wang [6].
Another important design parameter is the load acting on the lining. Such load can be
taken to be either the sum of the overburden pressure and water pressure or that determined
using Terzaghi’s formula, Schulze–Duddeck method or other empirical methods. It is noted
that such load only considers the final state that is after the completion of the shield tunnelling.
To provide a better simulation of the interaction between the lining and soil, beam–
spring models, in which the lining and soil are modelled by beam and spring, respectively,
can be adopted. As shield tunnel lining is composed of several concrete segments, a number of
researchers and engineers had also taken into consideration the effects of the segment ring
joints [7, 8] in their studies. Lee and Ge [9] suggested an equivalent method to determine
the correction factor for approximating a jointed shield-driven tunnel lining as a continuous
ring structure under a plane strain condition. The present literature review has revealed that
there are three widely accepted methods for modelling the effects of shield segment joints
[10, 11].
Furthermore, shield tunnelling usually adopts staged construction and supporting techniques.
Consequently, the responses such as soil displacements and lining forces induced by the
construction will be different at different stages. It is believed that an optimal construction
process can improve the safety and reduce the disturbance to the surrounding soils. Hence, it
becomes very important to take into account the actual construction process in the shield
tunnelling design [12, 13]. Due to the complex nature of the problems, one may have to resort to
numerical approach for analysing the problems. It is well known that the finite element method
is a powerful tool for the analysis of soil–structure interaction in geomechanics and it has been
applied to tunnel excavations by taking into account the construction process, the different soil
layers, complex geometries, various loading conditions and soil–lining interfaces [14]. Though
shield tunnelling, strictly speaking, is a three-dimensional and time-dependent soil–structure

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 59

interaction problem [15–18], many engineers and researchers have demonstrated that the two-
dimensional finite element models can still give a fairly accurate prediction on the behaviour of
the tunnel [12, 19, 20]. Benmebarek et al. [21] proposed two methods for taking into account the
three-dimensional effects in a two-dimensional model. Further assessment of two-dimensional
models in analysing shield tunnelling was reported in a recent publication by Negro and de
Queiroz [22].
In this paper, we propose a two-dimensional finite element method for analysing shield
tunnels during construction. The construction of each lining segment of a shield tunnel is
divided into four stages. To model the discontinuous displacements between segments, a beam–
joint discontinuous model is adopted for simulating the shield action. Moreover, the changes
from fluid and solid states of the backfill grout at different stages are also considered.
Furthermore, stress-releasing coefficients are used to account for the different contact conditions
between the soil and external lining surface. These coefficients are estimated based on field
measured settlements and professional experiences. The numerical results obtained by the
present model are compared with those measured during the construction of the No. 7 Subway
Line Project in Osaka, Japan. The comparison shows that the proposed model can be used to
estimate the deformation, stresses and moments experienced by the surrounding soils and the
concrete lining segments during tunnelling.

2. BACKGROUND

During shield tunnelling, the shield will advance segment by segment with a balanced soil
pressure that supports the soil. After cutting through a length of a segment, the lining will
then be installed. The lining is essentially an assembly of concrete segments linked by bolts
which joints will be covered with waterproof flexible plates. Before the shield tail is
detached, grout will be injected into the gap between the surrounding soil and the lining.
After grouting, the shield will move forward. As the grout hardens and consolidates, the
surrounding soil will deform and induce pressures on the lining. Consequently, the ground will
settle gradually as well. All movements will cease only when a state of equilibrium state is
reached.
Field observations have confirmed that the ground movements and the earth pressure on the
lining segments have developed according to the construction process of the shield tunnelling
[1, 21, 23–25]. Factors affecting ground movements due to shield tunnelling have been
summarized in recent publications by Nomoto et al. [25] and Hashimoto et al. [1, 26]. The
construction process is one of the most important factors. Closer study of the field observations
has revealed that the construction process for a segment can be divided into four stages below:
Stage 1}Balanced cutting and shield supporting. This stage includes face cutting and
shield advancing. The existing soil pressure in the cutting face is balanced by pressure
from the machine behind the cutter. Consequently, the stress changes due to the cutting and
the balancing pressure will not result in significant soil movement. Furthermore, the
surrounding soil outside the shield will not be able to release its stress due to the rigid
support of the shield. However, one has to consider the possible disturbance on the surrounding
soil due to over-cutting, snaking and friction between the shield and the soil when the shield
advances.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
60 W. Q. DING ET AL.

Stage 2}Backfill grouting. The second stage is to install the lining segments and to backfill the
gap at the shield tail with grout. The lining will form a circular ring within the shield tail cover
plate. Once the shield moves forward, a gap between the soil and the lining segment will be
created. A grout must be simultaneously injected to backfill the gap space. The amount of grout
is usually equal to 1.2–1.3 times of the gap volume and the grout pressure is between 0.2 and
0:4 MPa: Such backfill grouting has three functions:
* preventing soil deformation immediately after the shield tail is detached,
* stabilizing lining segments, and
* improving tunnel water-proof performance.
The soil will start to interact with the lining although the grout is still in a fluid state. It is
noted that this backfill grouting serves a controlling measure to prevent soil deformation.
Excessive grout volume and pressure may lead to soil heave around the shield tail and must be
avoided.
Stage 3}Grout hardening. This stage is a transient stage. The grout will harden and consolidate,
and the soil deformation will also increase with time though its rate decreases. The lining
segments and ground soil interaction will increase with time until an equilibrium state is
reached. It is, therefore, necessary to assess the effects of the grout hardening and pressure
distribution on the response of the lining structure at this stage.
Stage 4}Hardened grout. In the final stage, the grout becomes hardened and gains its full
stiffness and strength. The settlement will almost cease to increase.

3. NUMERICAL PROCEDURES FOR SHIELD TUNNELLING

In this paper, a two-dimensional finite element model is developed for modelling shield
tunnelling process. As such process is a very complicated one, exact simulation is almost
impossible. Therefore, the model only tries to take into account major factors, namely the
complex properties of the soil materials, the snaking of the shield direction, the volume loss, the
grout properties and pressure, the joints of lining segment, and the construction process.
Following the standard finite element procedures, one has to obtain the initial stress before the
construction simulation can be carried out. The initial stresses at any point i is calculated by the
following equations:
P
sy0 ¼ gj Hj
ð1Þ
sx0 ¼ K0 ðsy0  Pw Þ þ Pw
where sy0 and sx0 are the vertical and horizontal initial earth stresses, respectively, gj is the unit
weight of the jth earth stratum above the point; Hj is the corresponding thickness; K0 is the
lateral earth pressure coefficient, and Pw is the water pressure at the point.
As pointed out in Section 2, the construction of a tunnel segment can be divided into
four stages. As the soils behave non-linearly during the construction, an elasto-plastic
model is adopted to model the soil. Similarly, the lining segment and the joints between
lining segment are modelled by adopting appropriate curved members. Different models
are adopted to simulate the grout as its behaviour is very much different at the various

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 61

stages. To take into account such non-linear complex problems, an incremental-


iterated technique with constant stiffness within each construction stage is used in the
analyses. A flow chart illustrating the above FEM simulation for the soil–structure interaction
analysis of shield tunnelling is presented in Figure 1. Mathematically, the solution process

Input calculating data and controlling information

Calculate initial or current earth stress and stress releasing equivalent nodal forces {∆F }rea

Construction stage (stress releasing step): j=1

Calculate newly added and stress releasing nodal forces at

current construction stage: α


j
{∆F }rea , {∆F j }add

[K ]ini + ∑ [∆K ζ ]
j
Form element and global stiffness matrix:
ζ =1

Iteration step: k=1

j=j+1
Solve FEM equation, gain: ∆δ { jk
}
k=k+1

Nonlinear equivalent
Calculate incremental strain and stress: ∆ε { jk
}{
, ∆σ },
jk

nodal forces: then gain total displacement, strain and stress.

{∆F } jk
non

No Soil and contact element nonlinear analysis


No
Iteration step:
New over-excessive stress element exists?
k≥N
Yes ∆σ ejk − ∆σ jkp
Precision satisfied ? < 0.00001
σ j (k −1) − ∆σ jkp
Yes
No Construction stage: j ≥ 4
Yes
Output results

Figure 1. Flow chart of the finite element simulation for shield tunnelling.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
62 W. Q. DING ET AL.

can be expressed as
!
X
j
½Kini þ ½DK z  fDd jk g ¼ a j fDF grea þ fDF j gadd þ fDF jk gnon ð2Þ
B¼1

where j ¼ 1–4 and k ¼ 1–N ; N is the number of non-linear iterative steps; ½Kini is the initial
stiffness matrix of soil ground and structure (if it exists) before excavation; ½DK z  is the
increment or decrement stiffness matrix of the soil ground excavated and the supporting
structure installed or removed at the construction stage z; fDF grea is the vector of the stress
releasing equivalent nodal forces due to the stress state of the previous construction process
acting along the currently excavated boundary. For one construction (excavation) process, it is
due to the initial stress state; fDF j gadd is the vector of newly added nodal load at construction
stage j; fDF jk gnon is the vector of P
current
R incremental equivalent excessive nodal forces caused
by non-linear stress. fDF jk gnon ¼ e Ve fBgT fDsa g dV where fBg is strain matrix and fDsa g is
the incremental non-linear stress vector; fDd jk g is the vector of incremental nodal displacement
at the stage j and iterative step k; a j is the stress releasing coefficient at construction stage j:
The iteration for each stage will cease after 10 iterative steps, that is N ¼ 10; or the prescribed
iterative precision is achieved. In the analysis, the iterative precision is chosen to be
 
 Dse  Dsp 
 jk jk 
 50:00001 ð3aÞ
sjðk1Þ  Dspjk 

where Dsejk and Dspjk are the elements of the vectors of incremental elastic stress matrix fDsejk g
and plastic stress matrix fDspjk g at the stage j and the iterative step k; respectively, and sjðk1Þ is
the corresponding total stress at the stage j and the iterative step k  1: Dsejk and Dspjk can be
estimated using the following equations:
fDsejk g ¼ ½D½BfDd jk g ð3bÞ

fDspjk g ¼ ½Dp ½BfDd jk g ð3cÞ


where ½D and ½Dp  are the elastic and plastic material property matrices, respectively.
After solving Equation (2), one can obtain the total displacement fdi g; strain fei g and stress
fsi g at the construction stage i ði ¼ 1; 2; 3; 4Þ using the following equations:
X i XN
fdi g ¼ fd0 g þ fDd jk g
j¼1 k¼1

X
i X
N
fei g ¼ fe0 g þ fDe jk g
j¼1 k¼1

X
i X
N
fsi g ¼ fs0 g þ fDs jk g ð4Þ
j¼1 k¼1

where fd0 g and fe0 g are the initial displacement and strain respectively, they are usually set to
zero. fs0 g is the initial earth stress; fDe jk g and fDs jk g are the vectors of the incremental strain
and stress releasing step (construction stage) j and the iterative step k; respectively.
Furthermore, the lining internal forces (hoop force, shear force and bending moment) can
then be determined from the displacements.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 63

4. A CASE STUDY: SHIELD TUNNELLING FOR OSAKA NO. 7 SUBWAY LINE

Extensive monitoring was carried out during the construction of a tunnel of the Osaka Subway
Line. The results are readily available. This case history is used to assess the accuracy that can be
achieved by the present approach. For completeness, a brief of the relevant information will be
given and more detailed information can be found in the report prepared by the Japanese
Research Society of Construction Maintenance Techniques [27].

4.1. General
Figure 2 shows the general arrangement and longitudinal profile of the No. 7 Subway Line
Project in Osaka, Japan. The subway is between the Morinomiya Station and the Osaka
Business Park Station. Its east bound tunnel line is about 970:4 m long. Its west bound tunnel
line is about 974:5 m long. The tunnels are about 16–30 m below the soil ground surface. The
lining is of reinforced concrete and each segment is 1:2 m wide and 280 mm thick, and has the
outer diameter 5:3 m: Two 5:44 m in diameter earth pressure balanced shield machines with
synchronous grouting technique were used to construct the tunnels. In this paper, we will focus
on Section A of the west bound tunnel.

4.2. Ground conditions


Figure 3 shows the soil stratum characteristics at Section A of the tunnels. There are three
different clayey and sandy soil strata, i.e. AC1, AC2 and AS1, above the proposed tunnel. The
tunnel is located at about 20 m below the ground. The tunnel passes through a sand layer, AS2,

Figure 2. General layout of the Osaka No. 7 Subway Line: (a) line plan; and (b) tunnel profile.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
64 W. Q. DING ET AL.

0 50 Standard Penetration
Soil layer: Ac1 Test (SPT) value

Ac2

As1 -10

As2
Tunnel
-20

Ac3

As3 -30

Ac4

-40
Depth (m)
Figure 3. The soil layer of Section A.

Table I. The adopted ground soil parameters.


Soil Thickness Average Elastic Internal Unit Lateral
layer of per layer value of modulus Poisson’s friction Cohesion weight pressure
name Hj ðmÞ SPT N E ðkPaÞ ratio m angle j ð8Þ C ðkPaÞ gj ðkN=m3 Þ coefficient K0
AC1 2.0 2.50 8500 0.43 0.0 25.0 16.0 0.75
AC2 5.5 6.11 20774 0.43 0.0 61.1 16.7 0.75
AS1 3.0 34.00 39500 0.35 37.6 0.0 16.6 0.55
AS2 11.0 43.05 49003 0.35 40.4 0.0 16.5 0.55
AC3 5.0 11.80 40120 0.43 0.0 118.0 17.0 0.75
AS3 7.0 42.64 48572 0.35 40.3 0.0 16.5 0.55
AC4 26.0 13.80 46920 0.43 10.0 138.0 17.4 0.75

and a clayey layer, AC3. The thicknesses of these layers are about 11 and 5 m; respectively.
Below the AC3 stratum, there are a sandy stratum, AS4, and a clayey stratum, AC4. The
permanent ground water level is at about 3 m below the ground surface. The standard
penetration test results revealed that the three upper clayey soil strata (AC1–AC3) had a SPT N -
values from 2.5 to 11.8 whereas the three sandy soil strata (AS1–AS3) had SPT N -values ranging
from 34 to 43.05 (Figure 3 and Table I).

4.3. Instrumentation system for monitoring


In order to verify the shield tunnelling design and the settlement prediction, an instrumentation
system was installed before and during the tunnel construction to monitor the ground settlement

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 65

depth (m) 0 y
S1 -0.25

S1 ~ S5 Measuring points

S2 -10.8

S3 -14.8
S4 -15.8
S5 -16.3 180°

270° o' 90°


θ

East Bound West Bound
to be constructed being constructed and
examed in this paper

Section A-A
Figure 4. The locations of the settlement gauges.

and lining structural performance. Figure 4 shows the arrangement of the five settlement
monitoring points in the soil right above the west bound tunnel. Figure 5(a) shows the
arrangement of eight earth pressure cells installed to measure the soil pressure acting on the
outer surface of the concrete lining segments. Gauges are installed to measure moment as well as
hoop forces. Their locations are shown in Figure 5(b) and 5(c), respectively.

4.4. Constitutive model for the surrounding soil


An elasto-plastic constitutive model was used in this paper for the soils surrounding the tunnel.
The model employs the classical Drucker–Prager yield criterion and the Reyes elasto-plastic
matrix [28] after a careful examination of the soil properties [12]. The following empirical
equations are used to determine the Young’s modulus E ðkN=m2 Þ the internal friction angle j,
the lateral pressure coefficient K0 and Poisson’s ratio m for the sandy soil:
E ¼ 3800 þ 1050N ð5aÞ
pffiffiffiffiffiffiffiffiffi
j ¼ 15 þ 15N ð5bÞ

K0 ¼ 1  sin j0 ð5cÞ

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
66 W. Q. DING ET AL.

EP3(180°) 180°

EP4(224°) EP2(131°)

270° 90°
EP5(272°) EP1(90°)

EP8(58°)
EP6(310°)

EP7(355°) 0°
(a)

M04(188°) 180°
M05(214°) M03(151°)

M06(261°) M02(116°)

270° 90°

M07(288°) M01(75°)

M08(323°) M10(37°)
0° M09(5°)

(b)

AF04(188˚) 180˚
AF05(214˚) AF03(151˚)

AF02(116˚)
AF06(261˚)
270˚ 90˚

AF07(288˚) AF01(75˚)

AF08(323˚) AF10(37˚)
0° AF09(5˚)
(c)
Figure 5. The locations of: (a) earth pressure cells; (b) moment gauges; and (c) hoop force gauges.

K0
m¼ ð5dÞ
1 þ K0

where j is the internal friction angle, j0 is effective internal friction angle, K0 is the lateral
pressure coefficient, N is the number of standard penetration test (SPT) value.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 67

For the clayey soils, we used the following empirical relations:


E ¼ 170qu ð6aÞ

K0 ¼ OCR0:3  0:5 ð6bÞ


qu
c¼ ð6cÞ
2
N
qu ¼ ð6dÞ
50
K0
m¼ ð6eÞ
1 þ K0
where c is the cohesion ðkN=m2 Þ; qu is the unconfined compression strength (MPa), and OCR is
the over-consolidation ratio.
The parameters of the seven soil strata are given in Table I.

4.5. Modelling of the shield lining


Figure 6 shows a typical reinforced concrete segment with a circular-arc-shape. For such
segment, one can adopt curved beam elements to simulate its action (Figure 7). As flexible plates
were used to ensure flexible joint contact between the concrete segments, the joints will
smoothen the stress concentration but they have some inherent discontinuous characteristics in
transfering deformation. In this study, we have proposed a special joint element to represent the
flexible contact between any two adjacent segments. A detailed description of the model is given
in Appendix A.
Based on the test results, the following lining parameters are adopted: the Young’s modulus
E ¼ 3:5  107 kN=m2 ; the rectangular cross-section area A ¼ 0:28 m2 ; the bending moment of
inertia I ¼ 1:83  103 m4 ; the joint parameters: kW ¼ 2:5  108 kN  m=rad=m2 ; kn ¼ 6:55 
107 kN=m3 ; ks ¼ 2:5  107 kN=m3 :

4.6. Modelling of the grout


In the analysis, both the grout stiffness and grout pressure have to be modelled adequately and
the details of the models are outlined as follows:
(a) Grout stiffness. In this study, we adopted the Goodman joint model to represent such grout
contact property [28]. In this model, the element thickness is assumed to be zero and the
stiffnesses are defined in terms of the normal and tangential stiffness. The normal stiffness Kn

joint

segment

Figure 6. Shield tunnel lining segment.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
68 W. Q. DING ET AL.

Joint
n Beam

2 1

s
Figure 7. Beam–joint discontinuous model.

depends on the grout deformation modulus whereas the tangential stiffness Ks is related to the
sliding feature of the contact.
If the normal contact stress sn is compressive, i.e. it has positive value, one can assume that
the yield criterion of the contact element will follow the Mohr–Coulomb function, that is
f ¼ jts j  ðcj þ sn tgjj Þ ð7Þ
where cj and jj are, respectively, the cohesion and internal friction angle of the contact element,
ts is the total shear stress on the contact element.
One can show readily that the elastic property matrix ½D and plastic property matrix ½Dp  for
the ideal plastic and plane strain situation can be expressed as
" #
Ks 0
½D ¼ ð8aÞ
0 Kn
" 2 #
1 Ks Ks S1
½Dp  ¼ ð8bÞ
S0 Ks S1 S12
where S0 ¼ Ks þ Kn tg2 jj ; S1 ¼ Kn tgjj :
The grout initially is in fluid state and it hardens gradually. Therefore, different parameters
should be used to simulate the behaviour of the grout. A brief description is given below to
explain how to adopt different grout parameters at the different construction stages.
In Stage 2, the grout is in the fluid state, and therefore, the tangential contact stiffness Ks is
close to zero. Although the liquid is incompressible, the trapped air voids can be compressed.
Assuming 15% of the total grout volume is occupied by air voids, the relation between the
grouting pressure and volumetric strain can be expressed as
  
Pa
ev ¼ 0:15 1  ð9Þ
Pa þ Pg
where Pa and Pg are the atmospheric pressure and the grouting pressure, respectively. In the
numerical calculations, we assumed Pa ¼ 0:1 MPa and Pg ¼ 0:15 MPa: The value of the normal
contact stiffness Kn can be estimated using the equation below
Pg
Kn ¼ ð10Þ
ev  t 0
where t0 is the thickness of shield tail void and it equals to 0:07 m for the present shield tunnel.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 69

In Stages 3 and 4, the grout hardens and its stiffness increases gradually. The tangential
contact stiffness is given as:
Kn
Ks ¼ ð11Þ
2ð1 þ nÞ
where n ¼ 0:3: The normal stiffness Kn can be estimated by using the following empirical
relation [12]:
E 150qu
Kn ¼ ¼ ð12Þ
t t
where qu is the unconfined compressive strength of grout, t is the thickness of contact element.
In this study, qu are taken to be 0.2 and 3:0 MPa for Stages 3 and 4, respectively. The normal
and tangential stiffnesses adopted at various stages are given in Table II.
(b) Distribution of grout pressure. Two models of grout pressure distribution (Figure 8) are
considered in the analysis. The first model (GP-A) assumes that the grout pressure is uniform
and such model has the advantage of being simplicity, but, it may not be able to model the
decrease in pressure as the grout flows away from the grout hole. To overcome the above

Table II. The adopted parameters of the contact elements.


Shear stiffness Normal stiffness Internal friction Cohesion
Construction stage Ks ðkN=m3 Þ Kn ðkN=m3 Þ angle j ð8Þ c ðkPaÞ
No. 1 } } } }
No. 2 100 2:381  104 0.0 0.0
No. 3 1:648  105 4:286  105 20.0 100.0
No. 4 2:473  106 6:429  106 50.0 3000.0

180°
Grouting hole
217°
170° GP-A GP-B
232° 143°
242°
270° 95°
90°
303° θ
19°

0° Segment joint
Grouting pressure (kPa)

180
160
140
120
100
80
60
40
20
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 8. Grout pressure models.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
70 W. Q. DING ET AL.

limitation, a non-uniform pressure distribution is assumed in the second model (GP-B).


Mathematically, the grout pressure at the hoop angle y is defined as
yb
P ðyÞ ¼ P1 þ P2 cos ðp4y  b4pÞ ð13Þ
2
where b is the grout hole angle; P1 is the pressure at y ¼ b; P2 is the pressure at y ¼ p þ b:
As two grout holes are used in the present case and they are symmetric about the vertical axis,
the non-uniform model of the grouting pressure distribution can be expressed as follows:
X2
y  bi
P ðyÞ ¼ 2P1 þ P2 cos ðp4y  b1 ; y  b2 4pÞ ð14Þ
i¼1
2
where b1 and b2 are the grout hole angles. In the present study, P1 ¼ 5:47 kPa; P2 ¼ 77:32 kPa;
b1 ¼ 1438; b2 ¼ 2178:

4.7. Modelling of the stress release: stress releasing coefficient


The releasing coefficient for each construction stage is determined according to the field
displacements and experience.
Depending on the degree of snaking and volume loss, the stress release coefficient is taken to
be 0.0 to 0.1 in the first stage. In the second stage, the lining structure is installed and the grout is
injected at prescribed pressure to fill up the gap between the lining and the ground. The pressure
applied will not only restrain some of the ground movements but it may cause heaving, if the
pressure is high. Consequently, the coefficient depends on the grout pressure. In the third stage,
the stress releasing is mainly caused by the gap of the shield tail, and therefore, the coefficient
will depend on the gap. In the fourth stage, it depends on the stresses that have not yet released.
The releasing coefficients for the first to the fourth construction stages are chosen to be 0.10,
0.45, 0.3 and 0.15, respectively.

4.8. FEM mesh and boundary conditions


The initial finite element mesh adopted for the analysis is shown in Figure 9. In the finite element
simulation, the soil is modelled by 512 elements containing 545 nodes. Lining segments and their
joints are simulated with 30 beam elements and six joint elements, respectively. Thirty contact
elements are used to model the interface between the ground and lining segments. The left and
right boundaries of the mesh are laterally restricted. The bottom boundary is vertically restricted
whereas the top surface is assumed to be a free boundary.

5. RESULTS AND ANALYSIS

Section A of the Osaka No. 7 subway line is analysed in this paper and the results are compared
with the field measurement data documented in the report prepared by the Japanese Research
Society of Construction Maintenance Technique [27].

5.1. Soil displacements and settlements


Figure 10 shows the original and deformed meshes around the shield tunnel before grouting and
after completion of Stage 4. In Table III, the displacements at four points around the tunnel are

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 71

Figure 9. Initial finite element mesh.

tabulated. They are the results before grouting and after the completion of the Stage 4 using the
uniform grout pressure models. The settlement at the crown point A is about 7 mm: The
horizontal displacement at the points C and D are about 2–3 mm: The heave at the bottom
point B is about 11 mm: These results have shown that the soil displacements can gain
substantial increases with the progress of the construction stages.
The computed settlements above the tunnel after completion of Stage 4 are compared to the
measured values in Figure 11. The results show that the computed settlements agree fairly well
with the measured ones.

5.2. Lining bending moment, hoop force and earth pressure


Figure 12 shows the measured and computed bending moments of shield lining at the Stages 2, 3
and 4. In Figure 13, the measured and computed hoop forces of the shield lining segments are
compared for the Stages 2, 3 and 4. Furthermore, the earth pressure at the Stage 4 only
is depicted in Figure 14. These results were obtained using the uniform grout pressure model
GP-A. In those figures, the corresponding predicted values using the conventional semi-
analytical design method at the Stage 4 are also presented. The conventional method is
summarized in Appendix B for an ease of reference. Details of this conventional method can be
found in [10, 11, 29].
From Figure 12, it can be observed that the variations of the measured and the present FEM
predicted bending moments with respect to the hoop angle y and the construction stages are
basically similar although there are some significant discrepancies at the measurement locations
M10 ðy ¼ 758Þ and M08 ðy ¼ 3238Þ: At those two locations, the measured moments have
irregular tends with the construction stages 2–4, which indicates the measured moments might
not be correct. The absolute values of the FEM predicated moments are generally greater than
those of the measured moments. The moments predicted using the conventional method after

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
72 W. Q. DING ET AL.

C D

Scale of total displacement from an


original point to its corresponding point: 0 20(mm)
(a)

C D

Scale of total displacement from an


original point to its corresponding point: 0 20(mm)

(b)
Figure 10. Deformed meshes of around the tunnel: (a) before grouting; and (b) after completion of Stage 4.

the completion of Stage 4 are generally less than those measured values, which could cause an
unsafe design.
From Figure 13, it can be observed that the variations of the measured and the present FEM
predicted hoop forces with respect to the hoop angle y and the construction stages are basically
similar although there are some significant discrepancies at the measurement locations AF01
ðy ¼ 758Þ: Basically, the hoop forces increased as the construction stage progressed, which
cannot be predicted using the conventional method. Furthermore, the measured hoop forces are
not symmetric about the vertical centre (y ¼ 0 or 1808), while the predicted results are symmetric
about the vertical axis. The percentage relative differences between each set of the measured and
FEM predicted hoop forces are between 113 and 35%. A majority of the relative differences
are within 20%: Furthermore, the predicted results using the conventional method at the
Stage 4 have the lowest variations with respect to the hoop angle y:

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 73

Table III. The displacements of the typical points around tunnel.

Point A Point B Point C Point D


Construction
stage Horizontal Vertical Horizontal Vertical Horizontal Vertical Horizontal Vertical
1 0 2.24 0 2.95 0.92 0.12 0.92 0.12
2 0 5.32 0.03 6.06 1.39 0.38 1.45 0.38
3 0.08 6.28 0.05 9.55 1.86 0.93 2.34 0.69
4 0.11 6.67 0.11 11.21 2.14 1.64 2.85 1.19
Note: The positive values are horizontal rightward and vertical upward; unit: mm.

Settlement (mm)
0
0 1 2 3 4 5 6 7
-2

-4

Measured value
-6
GP-A
-8

-10

-12

-14
Depth (m)

-16

-18
Figure 11. Variation of settlement with depth.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
74 W. Q. DING ET AL.

110
Stage 2 & measured Stage 2 & GP model A
Stage 3 & measured Stage 3 & GP model A
90 Stage 4 & measured Stage 4 & GP model A
Conventional design method Location of joints
70
Bending moment (kN.m)
50

30

10

-10

-30

-50

-70
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 12. Variation of measured and predicted bending moments with hoop angle at the
completion of construction stages 2, 3 or 4.

900
Stage 2 & measured Stage 2 & GP model A
Stage 3 & measured Stage 3 & GP model A
800 Stage 4 & measured Stage 4 & GP model A
Conventional design method Location of joints

700
Lining ring hoop force (kN)

600

500

400

300

200

100
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 13. Variation of measured and predicted hoop forces with hoop angle at the
completion of construction stages 2, 3 or 4.

From Figure 14, it is clear that the present FEM predicted earth pressure values are very close
to those measured results at the cell Nos. EP8 ðy ¼ 588Þ; EP1 ðy ¼ 908Þ and EP7 ðy ¼ 3558Þ and
are about two times greater than those measured results at the other cell Nos. EP2 to EP6.
Furthermore, the present FEM predicted result generally decreases as the hoop angle y increases

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 75

340

310

280

Normal earth pressure (kPa) 250

220

190

160

130
Measured Present method
100 Conventional Location of joints
70

40
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 14. Variation of earth pressure after completion of Stage 4.

from 0 to 1808 and then increases as the hoop angle y increases from 180 to 3608; which is
considered to be reasonable. However, the results predicted with the conventional method have
four cycles of increasing and decreasing. The four cycles correspond to the hoop angle intervals
04y5908; 9084y51808; 18084y52708 and 27084y53608; which are evidently caused by the
basic assumptions associated with the semi-analytical formation of the conventional method
(see Appendix B). Moreover, the predicted results using the conventional method are close to
those predicted using the present FEM method. The predicted results do not have significant
differences.
From the above analysis and discussions, one can have the following general findings:
(a) The present method can be used to make a best prediction on the hoop forces and a
good prediction on the earth pressures and a reasonable prediction on the bending
moment. The predicted results can be used for a safe design for each of the three
construction stages.
(b) The in situ measurements have a best performance for the hoop forces, a good
performance for the earth pressure and a reasonable performance for the bending
moments. There seems a need to develop more reliable senses for in situ measurement of
the bending moment in the lining.
(c) The conventional method could give a reasonably good prediction on the hoop forces,
an adequate estimation on the earth pressure and an under-estimation on the bending
moment after the completion of the Stage 4. The use of this conventional method in
design estimation of the bending moment needs a great care.
(d) These results have also shown that the mechanical responses of the shield tunnel
construction at the construction stages can be predicted by the numerical procedure
presented above. It is noted that it is very important to select correct soil parameters in
the numerical prediction.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
76 W. Q. DING ET AL.

Table IV. The relative displacements at joints of lining segment at the completion
of construction stage No. 4.
y 198 958 1708 2328 2428 3038
2 2 2 2 2
GP-A Du ðmmÞ 1.91  10 2.19  10 1.60  10 1.79  10 1.91  10 2.13  102
4 4 4 5 5
Dv ðmmÞ 1.62  10 9.60  10 8.54  10 1.22  10 6.15  10 1.35  103
DW ð8Þ 8.83  106 1.12  105 9.51  106 3.05  107 3.87  106 5.08  106

GP-B Du ðmmÞ 1.80  102 2.12  102 1.53  102 1.72  102 1.84  102 2.03  102
4 4 4 5 5
Dv ðmmÞ 1.46  10 9.01  10 8.55  10 1.66  10 5.09  10 1.42  103
DW ð8Þ 9.31  106 1.23  105 1.06  105 1.23  107 4.62  106 4.81  106
Note: Du; Dv and DW are relative normal displacement, relative shearing displacement and relative rotational angular
displacement between two nodes of joint element, respectively.

340

310
GP-A GP-B
Normal earth pressure (kPa)

280

250

220

190

160

130

100
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 15. Effect of grout pressure models: earth pressure.

5.3. Joint displacement and rotation


Table IV gives the relative normal displacement, shearing displacements and rotational angular
displacements between two nodes of the joint elements after the completion of Stage 4. From
Table IV, it can be observed that the relative normal displacements vary between 0.015 and
0:022 mm: The relative shearing displacements vary between 0.000012 and 0:0014 mm: The
relative rotational angular displacements vary between 1:23  107 and 1:23  105 : Such small
relative displacements at the joints can be justified by the combined factors that the joint
stiffness values (kW ¼ 2:8  108 kN m=rad=m2 ; kn ¼ 6:55  107 kN=m3 ; ks ¼ 2:5  107 kN=m3 )
are much greater than those of the grout stiffness values between 100 and 6:429  106 (see
Table II), these values are close to the lining concrete bending stiffness (say, Young’s modulus/
thickness E=h ¼ 3:5  107 =0:28 ¼ 1:25  108 kN=m3 ) in magnitude, as well as there is an
arching effect along the circular lining.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 77

100
80
GP-A GP-B
60

Bending moment (kN.m)


40
20
0
-20
-40
-60
-80
-100
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 16. Effect of grout pressure models: bending moment.

900
800 GP-A GP-B
Lining ring hoop force (kN)

700
600
500
400
300
200
100
0 30 60 90 120 150 180 210 240 270 300 330 360
θ (degree)
Figure 17. Effect of grout pressure models: hoop force.

5.4. Effects of grout pressure models


The effects of the distribution of the grout pressure were also studied by adopting both uniform
(GP-A) and non-uniform (GP-B) models. The results are plotted in Figures 15–17. They have
shown the following phenomena:
(a) The GP-A model can result in an up to 24% higher earth pressure value than those
associated with the GP-B model since the grout pressure in GP-A model is assumed much
higher than that in GP-B model (Figures 8 and 15). By comparing the corresponding
results in Figures 14 and 15, it is evident that the earth pressure values predicted from the
conventional method is well covered by the earth pressure ranges predicted using the GP-
A and GP-B over the entire hoop angles.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
78 W. Q. DING ET AL.

(b) The two grout pressure models can produce very close results in the beam bending
moments although it is interesting to note that the GP-B model has slightly higher earth
pressure values around y ¼ 908; y ¼ 1808 or y ¼ 2708 (see Figure 16).
(c) The lining ring hoop forces predicted with the GP-A model is uniformly slightly (about
7%) greater than those predicted with the GP-B model over the entire hoop angle ð04
y53608Þ (see Figure 17).

6. CONCLUSIONS

Based on the main characteristics of the tunnel construction process, the properties of the grout
and the stiffness of the lining and its joints, we have presented a two-dimensional finite element
model for shield tunnels. In the analysing, we pay special attentions to the contact factors
between the soil and the lining structure. These factors include the gap closing, the grout
pressure distribution and hardening. Based on the analysis, we may have the following
conclusions:

* The results obtained by the present FEM method, in particular, the settlements, the earth
pressures and the hoop forces, are very close to the values measured on site. The measured
values are well within the ranges of the numerical results. Furthermore, although the
present predicted lining bending moment values are generally greater than those measured
on site, they can be used for a safe design. These findings have shown that the numerical
procedure presented in this paper is practical and reliable for evaluation, design and
displacement prediction for shield tunnel construction.
* The hardening of the grout materials with the process of shield tunnelling can be simulated
by varying parameters such as the stiffness, the internal friction angle and cohesion at
different construction stages.
* The two grouting pressure models give similar results for the displacements, moments and
forces. It may be concluded that the grouting pressure models do not have a significant
effect on the behaviour of the tunnelling in soil ground.
* Although the conventional method have been used for prediction of the lining bending
moment, hoop forces and earth pressures after the completion of the construction stages,
the predicted values could be lower and the method could not take into account many
complex factors and construction stages that the present FEM method can accommodate
in the design calculations.
* The above FEM analysis has further shown that numerical procedures can make adequate
predictions on the mechanical behaviour of complex soil–structure interaction system such
as shield tunnelling during and after construction. Such numerical predictions can enhance
the design and construction measures of geotechnical works in soft soils. It is noted that it
is also critical to select correct values for the mechanical parameters of soils in the
theoretical predictions. Good knowledge and experience of the local soft soil conditions
and properties are always assets in selection of the parameter values for prediction.
Additional experience on the selection of the soil parameters can be gained from the
comparisons between the predicted and measured mechanical responses that have been
presented in this paper.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 79

APPENDIX A: BEAM–JOINT DISCONTINUOUS MODEL

As shown in Figure 7, the beam–joint discontinuous model uses several curved beam elements
for a segments and a joint element for a joint. Figure A1 depicts a curved beam element and
Figure A2 shows a joint element in the forms of a double node.
By using the Castigliano’s second theorem [12] we can establish the relation of the nodal
forces Fi ¼ fNi Qi Mi gT with the nodal displacements di ¼ fui vi Wi gT ði ¼ 1; 2Þ in a local co-
ordinate system ðn; sÞ: It can be shown readily that:
( ) " #( ) ( )
F1 K11 BK11 d1 d1
¼ ¼ ½K ðA1Þ
F2 BK11 K22 d2 d2

where ½K is the stiffness matrix and ½B is the correlative matrix between the two nodal forces of
the beam element.

Q2 v2
M2  2

N2 u2 2 P

M1  1
1
Q1 v1
β ϕ
N1 u1
z o′

o x
Figure A1. Nodal forces and displacements of curved beam model.

α2

γ
2
z 1
s

o x
α1

Figure A2. Double nodes of joint and nodal direction angles in global co-ordinates.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
80 W. Q. DING ET AL.

½K and ½B can be expressed as


Kii ¼ ½dik‘ 1
33 ðA2Þ
2 3
cos b sin b 0
6 7
½B ¼ 6
4 sin b cos b 07
5 ðA3Þ
Rðcos b  1Þ R sin b 1
where R and b are the radius and central angle of the curved beam, respectively; dik‘ is element of
the flexibility matrix, and
dikl ¼ dilk ðA4Þ

R R3
d111 ¼ ð0:5b þ 0:25 sin 2bÞ þ ð1:5b  2 sin b þ 0:25 sin 2bÞ ¼ d211 ðA5Þ
EA EI
R R3
d112 ¼ sin2 b þ ðcos b þ 0:5 sin2 b  1Þ ¼ d212 ðA6Þ
2EA EI
R2
d113 ¼ ðb  sin bÞ ¼ d213 ðA7Þ
EI
R R3
d122 ¼ ð0:5b  0:25 sin 2bÞ þ ð0:5b  0:25 sin 2bÞ ¼ d222 ðA8Þ
EA EI
R2
d123 ¼ ðcos b  1Þ ¼ d223 ðA9Þ
EI
R
d133 ¼
b ¼ d233 ðA10Þ
EI
In Equations (A4)–(A10), E; A and I are the elastic modulus, cross-sectional area and moment
of inertia of the curved beam, respectively.
The local stiffness ½K can be assembled into the global stiffness ½KG  in a global co-ordinate
system as follows:
½KG  ¼ ½T  ½K ½T T ðA11Þ
where ½T  is a transformation matrix between the local co-ordinate system and the global
co-ordinate system; and ½T T is the transpose of the matrix ½T :
The discontinuity of the two adjacent segments is described by the following relative
displacements with three variables between the two nodes 1 and 2 in the local co-ordinate
system ðn; sÞ
Du ¼ u1  u2
Dv ¼ v1  v2 ðA12Þ
DW ¼ W1  W2
where s is defined as the direction of the bisecting angle between the two beams with a positive
direction inward to opening; n is the direction perpendicular to s with a positive direction as
counterclockwise. Accordingly, Du is the relative normal displacement of the joint along the n

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 81

k
kn

ks
Figure A3. The stiffness kn ; ks ; kW in normal, shearing and rotational directions of joint.

direction, Dv is the relative shearing displacement along the s direction, and DW is the relative
rotational angular displacement; ðui ; vi ; Wi Þ; where i ¼ 1; 2; are the displacement components at
the node i along the local co-ordinate directions respectively.
Furthermore, the force–deformation relationship of the joint element can be expressed as
follows:

fF g ¼ ½KfDdg ðA13Þ

where
2 3
kn 0 0
6 7
fF g ¼ fN Q MgT ; fDdg ¼ fDu Dv DWgT ; ½K ¼ 6
40 ks 07 5
0 0 kW

(Figure A3), and N ; Q; M are the two forces along the n and s directions and the moment of the
joint element, respectively.
Expressing ðui ; vi ; Wi Þ as fdg ¼ fu1 v1 W1 u2 v2 W2 gT ; we can have the following equation:

fDdg ¼ ½C fdg
2 3
1 0 0 1 0 0
6 7
½C  ¼ 6
40 1 0 0 1 0 7
5 ðA14Þ
0 0 1 0 0 1

As a result, we obtain the following equation:


(
½KJ  ¼ ½CT ½K ½C
ðA15Þ
½KG  ¼ ½T  ½KJ  ½T T

where ½KJ  is the stiffness matrix of the joint element in the local co-ordinates, ½T  is a co-ordinate
transformation matrix, ½KG  is the stiffness matrix of the joint element in the global co-ordinates

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
82 W. Q. DING ET AL.

ðx; zÞ; and


2 3
A1 A3 0 A1 A3 0
6 7
6 A2 0 A3 A2 0 7
6 7
6 7
6 kW 0 0 kW 7
6 7
½KG  ¼ 6 7 ðA16Þ
6 A1 A3 0 7
6 7
6 7
6 A2 0 7
4 5
sym: kW

where A1 ¼ kn sin2 g þ ks cos2 g; A2 ¼ kn cos2 g þ ks sin2 g; A3 ¼ ðks  kn Þ cos g sin g; g is intersec-


tion angle between the opposite direction of local co-ordinate s and the direction of global co-
ordinate x (Figure A2).
It is noted that the flexible plate has a large effect on the bending behaviour of the tunnel
lining [30]. When the bending moment exceeds a certain value at the joint, the joint will open.
The rotational stiffness coefficient of the segment joint is, therefore, a non-linear function of the
relative rotation angle DW as follows:

kW ¼ ðkW1  kW2 Þ elDW þ kW2 ðA17Þ

where kW1 ; kW2 ; l are constants that can be determined from the bending test for the segment
joint.

APPENDIX B: THE SEMI-ANALYTICAL METHOD FOR SHIELD TUNNELLING

For comparisons, we also present an analysis using the semi-analytical method [10, 11, 29] for
shielding tunnelling. As the details are already document in the reference, we only provide a
highlight of the key points of the method.
The loads acting on the shield lining is shown in Figure B1. In the figure, P0 is overload, and
usually equal to 10 kN=m2 ; R0 is external radius of shield lining; Rc is radius of middle line of
shield lining; g is gravity of lining; Pe1 and Pw1 are, respectively, the vertical earth pressure and
water pressure, respectively, acted on the up side of shield lining. The lateral earth pressure and
water pressure vary linearly and act on both sides of the shield lining. They are equal to qe1 and
qw1 at the top of shield lining and qe2 and qw2 at the bottom of shield lining; Pe2 and Pw2 are
respectively the vertical earth pressure and water pressure respectively acted on the bottom side
of shield lining; Pg is the vertical resistance of lining weight acted on the bottom side of shield
lining.
For sandy clay, the earth pressure and water pressure are assumed to act on the lining
separately. If the overburden thickness is two times larger than the external diameter D of shield
lining, an effective overburden thickness h0 ðh0 52DÞ should be used and it is determined by
following Terzaghi’s formula (Figure B2) below:

B1 ð1  ðc=B1 gÞÞ P0
h0 ¼ ð1  eK0 tan fðH =B1 Þ Þ þ ðeK0 tan fðH =B1 Þ Þ ðB1Þ
K0 tan f g

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 83

Figure B1. Loads acting on the shield.

 
ðp=4Þ þ ðf=2Þ
B1 ¼ R0 cot ðB2Þ
2
where c; f and g are cohesion, internal friction angle and unit weight of soil layer respectively;
K0 is lateral pressure coefficient; H is overburden thickness. For many soil layers condition, the
average parameter values with a weighting factor of the layer thickness will be used in Equations
(B1) and (B2).

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
84 W. Q. DING ET AL.

Figure B2. The overburden thickness h0 determined by Terzaghi’s formula.

The distribution of earth resistance is triangular in shape and its acting range is shown in
Figure B1. The magnitude of the earth resistance qk is determined by following equation:
qk ¼ k  d ðB3Þ
where k is the soil resistance coefficient, d is the deformation at the horizontal quadrant point of
shield lining, and d can be calculated by following formula:
½2ðpe1 þ pw1 Þ  ðqe1 þ qw1 Þ  ðqe2 þ qw2 ÞR4c
d¼ ðB4Þ
24ðEI þ 0:0454  k  R4c Þ
Having these parameters determined, the internal forces of the shield lining can be easily
computed. The formulae for the calculation of the beam bending moment and lining ring hoop
force are given in Table B1.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 85

Table B1. The internal forces caused by the different loads.


Loads Beam bending moment Lining ring hoop force
2
Vertical earth pressure and M¼ 1
4 ð1 2 sin yÞ  N ¼ ðPe1 þ Pw1 ÞRc sin2 y
water pressure: Pe1 þ Pw1 ðPe1 þ Pw1 ÞR2c
Lateral earth pressure and M ¼ 14 ð1  2 cos2 yÞ  N ¼ ðqe1 þ qw1 ÞRc cos2 y
water pressure (rectangle part): ðqe1 þ qw1 ÞR2c
qe1 þ qw1
1 1
Lateral earth pressure and M ¼ 48 ð6 þ 3 cos y  N ¼ 16 ðcos y þ 8 cos2 y þ
water pressure (triangle part): 12 cos2 y  4 cos3 yÞ  3
4 cos yÞðqe2 þ qw2  qel 
qe2 þ qw2  qe1  qw1 ðqe2 þ qw2  qel  qwl ÞR2c qwl ÞRc
Horizontal earth resistance: qk For 04y4p4; For 04y4p4;
¼ kd M ¼ ð0:2346  0:3536 cos yÞk  N ¼ 0:3536 cos y  k  d  Rc
d  R2c For p44y4p2;
For p44y4p2; N ¼ ð0:7071 cos y þ cos2 y þ
M ¼ ð0:3487 þ 0:5 sin2 y þ 0:7071 sin2 y cos yÞk  d  Rc
0:2357 cos3 yÞ  k  d  R2c
Lining unit weight: g For 04y4p2; For 04y4p2;
M ¼ ð18 p þ y sin y þ 56 cos y  N ¼ ðp sin y þ ðp  yÞ sin y þ
2
1 2
2p sin yÞ  g  Rc p sin2 y þ 16 cos yÞg  Rc
For p24y4p; For p24y4p;
M ¼ ð38 p  ðp  yÞ sin y þ N ¼ ððp  yÞ sin y þ 16 cos yÞg  R2c
5 2
6 cos yÞg  Rc

Note: The values beyond the expressed range can be easily obtained as the loads and lining are symmetric.

Table B2. Adopted values for the design parameters.


No. Parameters Design values
1 Lining segment: concrete plate
2 External diameter of segment D ¼ 5:3 m
3 External radius of shield lining R0 ¼ 2:65 m
4 Radius of middle line of shield lining Rc ¼ 2:51 m
5 Width of segment B ¼ 1:2 m (Calculating width:/m)
6 Thickness of segment h ¼ 0:28 m
7 Elastic modulus of segment E ¼ 3:5  107 kN=m2
8 The bending moment of inertia I ¼ 1:83  103 m4
9 Overburden thickness H ¼ 16:35 m
10 Water head at the top of lining Hw ¼ 13:35 m
11 Overload P0 ¼ 10 kN=m2
12 Soil resistance coefficient k ¼ 20000 kN=m3

Parameter of earth layer: mainly


sandy clay, see Table I and Figure 3

13 The soil average unit weight g ¼ 16:52 kN=m3


14 The soil cohesion c ¼ 17:96 kPa
15 The soil internal friction angle f ¼ 25:928
16 The lateral pressure coefficient K0 ¼ 0:378
17 The unit weight of water gw ¼ 10 kN=m3

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
86 W. Q. DING ET AL.

Table B3. Results of the semi-analytical design method.


No. Parameters Design values
!
p f
1 Then B1 ¼ R0 cot 4þ 2 B1 ¼ 4:785 m
2

 
B1 1B g
c  H

1 K tan fB
2 h0 ¼ K0 tan f 1e 0 1 þ h0 ¼ 2D ¼ 10:6 m
 H

P0 K0 tan fB
g e
1 ¼ 9:708 m52D

3 Unit weight of lining g ¼ 25  h ¼ 0:7 ðkN=m2 Þ


4 Vertical earth pressure acted on the up side Pe1 ¼ ðg  gw Þ  h0 ¼ 69:097 kPa
of shield lining
5 Water pressure acted on the up side of Pw1 ¼ gw  Hw ¼ 133:5 kPa
shield lining
6 Lateral earth pressure and water pressure qe1 ¼ K0  ðg  gw Þ  ðh0 þ h=2Þ ¼ 26:46 kPa
qe1
7 Lateral earth pressure and water pressure qw1 ¼ gw  ðh0 þ h=2Þ ¼ 134:9 kPa
qw1
8 Lateral earth pressure and water pressure qe2 ¼ K0  ðg  gw Þ  ðh0 þ R0 þ Rc Þ ¼ 38:82 kPa
qe2
9 Lateral earth pressure and water pressure qw2 ¼ gw  ðh0 þ R0 þ Rc Þ ¼ 185:1 kPa
qw2
½2ðpe1 þpw1 Þðqe1 þqw1 Þðqe2 þqw2 ÞR4c
10 d¼ 24ðEIþ0:0454kR4c Þ
¼ 0:329 mm
11 qk ¼ k  d ¼ 6:582 kPa

Table B2 gives the values of the design parameters adopted in the present study and
Table B3 tabulate the parameters computed by using these equations. These parameters are
then used to determine the beam bending moment and lining ring hoop force in the tunnel
lining.

APPENDIX C: NOMENCLATURE
Symbols in Main Text

sy0 vertical initial earth stress


sx0 horizontal initial earth stress
gj unit weight of the jth earth stratum above the point
Hj thickness of the jth earth stratum above the point
K0 lateral earth pressure coefficient
Pw water pressure at the point
N number of non-linear iterative steps
½Kini initial stiffness matrix of soil ground and structure (if it exists) before excavation
½DK z  increment or decrement stiffness matrix of the soil ground excavated and the
supporting structure installed or removed at the construction stage z
DFrea vector of the stress releasing equivalent nodal forces due to the stress state of the
previous construction process acting along the currently excavated boundary

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 87

DF j add vector of newly added nodal load at construction stage j


DF jk non vector of current incremental equivalent excessive nodal forces caused by non-
linear stress
B strain matrix
Dsa vector of incremental non-linear stress
Ddjk vector of incremental nodal displacement at the stage j and iterative step k
aj stress releasing coefficient at construction stage j
Dsejk vector of incremental elastic stress at the stage j and iterative step k
Dspjk vector of incremental plastic stress at the stage j and iterative step k
sjðk1Þ total stress at the stage j and iterative step k  1
½D elastic material property matrix
½Dp  plastic material property matrix
d0 vector of initial displacement
e0 vector of initial strain
s0 vector of initial earth stress
di vector of total displacement at stage i
ei vector of total strain at stage i
si vector of total earth stress at stage i
Dejk vector of incremental strain at stress releasing step j and iterative step k
Dsjk vector of incremental stress at stress releasing step j and iterative step k
E Young’s modulus
m Poisson’s ratio
j internal friction angle
j0 effective internal friction angle
K0 lateral pressure coefficient
N number of standard penetration test (SPT) value
c cohesion
qu unconfined compression strength
OCR over-consolidation ratio
A rectangular cross-sectional area of lining
I bending moment of inertia of lining
kW rotational stiffness of lining joint
kn normal stiffness of lining joint
ks tangential stiffness of lining joint
Kn normal stiffness of the contact element
Ks tangential stiffness of the contact element
f yield criterion value of the contact element
cj cohesion of contact element
jj internal friction angle of the contact element
sn normal stress of the contact element
ts shear stress of the contact element
ev volumetric strain
Pa atmospheric pressure
Pg grouting pressure
t0 thickness of shield tail void
qu compressive strength

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
88 W. Q. DING ET AL.

t thickness of contact element


b; b1 ; b2 grout hole angle
P1 grouting pressure at y ¼ b
P2 grouting pressure at y ¼ p þ b

Symbols in Appendix A

Fi nodal forces of nodal i; Fi ¼ Ni Qi Mi T


Ni ; Qi ; Mi hoop force, shearing force and bending moment of nodal i
di nodal displacements of nodal i; di ¼ ui vi Wi T
u i ; v i ; Wi displacement components at the node i along the local co-ordinate directions
½K local stiffness matrix of the beam element and joint element
[B] correlative matrix between the two nodal forces of the beam element
R radius of the curved beam
b central angle of the curved beam
Kii block matrix of stiffness matrix
dik‘ element of the flexibility matrix
E elastic modulus of the curved beam
A cross-sectional area of the curved beam
I moment of inertia of the curved beam
½KG  global stiffness matrix
½T  transformation matrix between the local co-ordinate system and the global co-
ordinate system
½T T transpose of the matrix ½T 
s direction of the bisecting angle between the two beams with a positive direction
inward to opening
n direction perpendicular to s with a positive direction as counterclockwise
Du relative normal displacement of the joint along the n direction
Dv relative shearing displacement along the s direction
DW relative rotational angular displacement
F vector of joint nodal forces of hoop force, shearing force and bending moment,
F ¼ N Q MT
Dd vector of joint relative displacements in normal, shearing and rotational direction,
Dd ¼ Du Dv DWT
½KJ  stiffness matrix of the joint element in the local co-ordinates
½KG  stiffness matrix of the joint element in the global co-ordinates ðx; zÞ
g intersection angle between the opposite direction of local co-ordinate s and the
direction of global co-ordinate x (Figure A2)
DW relative rotation angle
k W1 constant that can be determined from the bending test for the segment joint
k W2 constant that can be determined from the bending test for the segment joint
l constant that can be determined from the bending test for the segment joint

Symbols in Appendix B

P0 overload

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 89

R0 external radius of shield lining


Rc radius of middle line of shield lining
g unit weight of lining
Pe1 ; Pw1 vertical earth pressure and water pressure, respectively, acted on the up side of
shield lining, respectively
qe1 ; qw1 lateral earth pressure and water pressure at the top of shield lining act on both
sides of the shield lining, respectively
qe2 ; qw2 lateral earth pressure and water pressure at the bottom of shield lining act on both
sides of the shield lining, respectively
Pe2 ; Pw2 vertical earth pressure and water pressure, respectively, acted on the bottom side
of shield lining, respectively
Pg vertical resistance of lining weight acted on the bottom side of shield lining
D external diameter of segment
h0 effective overburden thickness
c cohesion of soil layer
f internal friction angle of soil layer
g average unit weight of soil layer
K0 lateral pressure coefficient
H overburden thickness
qk earth resistance
k soil resistance coefficient
d deformation at the horizontal quadrant point of shield lining
B width of segment
h thickness of segment
E elastic modulus of segment
I bending moment of inertia of segment
Hw water head at the top of lining
gw unit weight of water

ACKNOWLEDGEMENTS

The authors are indebted to Mr. Junichi Nagaya in Geo Research Institute, Osaka Soil Test Laboratory,
Osaka, Japan for his kind assistance during this investigation. The work presented in this paper is
supported by the Research Grants Council of Hong Kong SAR Government and the Hong Kong Jockey
Club Charities Trust. The authors would also like to thank the three reviewers and Editor, Professor Stein
Sture, for their invaluable comments that have enhanced the paper presentation.

REFERENCES
1. Hashimoto T, Nagaya J, Konda T. Geotechnical aspects of ground movement during shield tunneling in Japan.
One-day Seminar Geotechnical Aspect of Underground Structure, vol. 1/III–6/III. Indonesia, Jakarta. Himpunan
Ahliteknik Tanah: Indonesia, 1996 (Hatti).
2. Peck RB. Deep excavation and tunneling in soft ground. In Proceedings of 7th International Conference Soil
Mechanics and Foundation Engineering. Mexico, State of the Art Volume, 1969; 225–290.
3. Clough GW, Schmidt B. Design and Performance of Excavations and Tunnels in Soft Clay. Soft Clay Engineering.
Elsevier: Amsterdam, 1981; 569–634.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
90 W. Q. DING ET AL.

4. Rowe RK, Lo KY, Kack GJ. A method of estimating surface settlement above tunnels constructed in soft ground.
Canadian Geotechnical Journal 1983; 20(8):11–22.
5. Schmidt B. Tunneling in soft ground in the United States}National report. In Proceedings of the International
Symposium on Underground Construction in Soft Ground, Underground Construction in Soft Ground, Fujita K,
Kusakabe O (eds), New Delhi, India. A.A. Balkema: Rotterdam, 1995; 119–122.
6. Wang JY. Tunneling and technological progress in tunneling in China. In Proceedings of the International
Conference on Tunnels and Underground Structures, Tunnels and Underground Structures, Zhao J, Shirlaw JN,
Krishnan R (eds), Singapore. A.A. Balkema: Rotterdam, 2000; 97–106.
7. Zhu HH, Yang LD, Chen QJ. Two kinds of design model for joint of segment and lining in shield tunnel.
Engineering Mechanics 1994; (Suppl.):395–399 (in Chinese).
8. Hudoba I. Contribution to static analysis of load-bearing concrete tunnel lining built by shield-driven technology.
Tunneling and Underground Space Technology 1997; 12(1):55–58.
9. Lee KM, Ge XW. The equivalence of a jointed shield-driven tunnel lining to a continuous ring structure. Canadian
Geotechnical Journal 2001; 38(3):461–476.
10. Koyama Y, Kishio T, Kobayashi T. Design of linings for shield driven tunnels}a survey on Japanese shield
tunneling. In Proceedings of the International Symposium on Underground Construction in Soft Ground, Underground
Construction in Soft Ground, Fujita K, Kusakabe O (eds), New Delhi, India. A.A. Balkema: Rotterdam, 1995;
359–366.
11. Kurihara K. Report on current shield tunneling methods in Japan. In Proceedings of the International Symposium on
Underground Construction in Soft Ground, Underground Construction in Soft Ground, Fujita K, Kusakabe O (eds),
New Delhi, India. A.A. Balkema: Rotterdam, 1995; 111–114.
12. Zhu HH, Ding WQ. The analysis of the segment internal forces in the construction process of shield tunnel. Recent
Development of Theory and Practice in Geotechnology, China–Japan Joint Symposium, October 29–30, 1997; 257–266.
13. Bernat S, Cambou B. Soil–structure interaction in shield tunneling in soft soil. Computers and Geotechnics 1998;
22(3/4):221–242.
14. Oreste PP, Peila D, Poma A. Numerical study of low depth tunnel behaviour. In Proceedings of the World Tunnel
Congress’99, Challenges for the 21st Century, Alten T et al. (eds), Oslo, Norway. A.A. Balkema: Rotterdam, 1999;
155–162.
15. Soliman E, Duddeck H, Ahrens H. Effects of development of shotcrete stiffness on stresses and displacements
of single and double tunnels. In Proceedings of the International Congress on Tunneling and Ground
Conditions, Tunneling and Ground Conditions, Abdel Salam ME (ed.), Cairo, Egypt. A.A. Balkema: Rotterdam,
1994; 549–556.
16. Swoboda G, Abu-krisha A. Three-dimensional numerical modeling for TBM tunneling in consolidated clay.
Tunneling and Underground Space Technology 1999; 14(3):327–333.
17. Dias D, Kastner R, Maghazi M. Three reducing settlement caused by shield dimensional simulation of slurry shield
tunneling. In Geotechnical Aspects of Underground Construction in Soft Ground}Is}Tokyo’99, Kusakabe O, Fujita
K, Miyazaki Y (eds), Tokyo, Japan. A.A. Balkema: Rotterdam, 2000; 351–356.
18. Zhou XJ, Gao B, Hu S, Li DC. Three dimensional finite element analysis on construction process of large span
tunnel of Guangzhou Metro. In Proceedings of the International Conference on Tunnels and Underground Structures,
Tunnels and Underground Structures, Zhao J, Shirlaw JN, Krishnan R (eds), Singapore. A.A. Balkema: Rotterdam,
2000; 335–342.
19. Negro Jr AN. Design of shallow tunnels in soft ground. Ph.D. Thesis, University of Alberta: Edmonton, 1988.
20. Katano S. Behavior of cohesive soil ground during shield tunneling. In Proceedings of the International Symposium
on Underground Construction in Soft Ground, Underground Construction in Soft Ground, Fujita K, Kusakabe O (eds),
New Delhi, India. A.A. Balkema: Rotterdam, 1995; 269–272.
21. Benmebarek S, Kastner R, Ollier C. Reducing settlement caused by shield tunneling in alluvial soils. In Geotechnical
Aspects of Underground Construction in Soft Ground}Is}Tokyo’99, Kusakabe O, Fujita K, Miyazaki Y (eds),
Tokyo, Japan. A.A. Balkema: Rotterdam, 2000; 203–208.
22. Negro A, de Queiroz PIB. Prediction and performance: a review of numerical analysis for tunnels. In Geotechnical
Aspects of Underground Construction in Soft Ground}Is}Tokyo’99, Kusakabe O, Fujita K, Miyazaki Y (eds),
Tokyo, Japan. A.A. Balkema: Rotterdam, 2000; 409–418.
23. Lee KM, Ji HW, Shen CK, Liu JH, Bai TH. A case study of ground control mechanisms of EPB shield tunneling in
soft clay. In Geotechnical Aspects of Underground Construction in Soft Ground}Is}Tokyo’99, Kusakabe O, Fujita
K, Miyazaki Y (eds), Tokyo, Japan. A.A. Balkema: Rotterdam, 2000; 251–256.
24. Mair RJ. Geotechnical aspects of design criteria for bored tunneling in soft ground. In Proceedings of the World
Tunnel Congress’98 on Tunnels and Metropolises, Tunnels and Metropolises, Negro Jr A, Ferreira AA (eds), Brazil.
A.A. Balkema: Rotterdam, 1998; 183–199.
25. Nomoto T, Mori H, Matsumoto M. Overview on ground movements during shield tunneling}a survey on Japanese
shield tunneling. In Proceedings of the International Symposium on Underground Construction in Soft Ground,
Underground Construction in Soft Ground, Fujita K, Kusakabe O (eds), New Delhi, India. A.A. Balkema:
Rotterdam, 1995; 345–351.

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91
ANALYSIS OF SHIELD TUNNEL 91

26. Hashimoto T, Hayakawa K, Mizuhara K, Konda T. Investigation on successive settlement due to shield tunneling.
In Geotechnical Aspects of Underground Construction in Soft Ground}Is}Tokyo’99, Kusakabe O, Fujita K,
Miyazaki Y (eds), Tokyo, Japan. A.A. Balkema: Rotterdam, 2000; 233–238.
27. Japanese Research Society of Construction Maintenance Technique (Shield Branch). The in-situ measuring results
of the soil pressure etc. on the segment of the Osaka No. 7 subway line from Morinonomiya to Osaka Business Park.
The Sixth Conference of Shield Branch SC-6-1, 1995 (in Japanese).
28. Zheng YR, Dong YF, Xu ZY et al. The design instruction for rock bolt and shotcrete of underground structure. China
Railway Press: Beijing, China 1988 (in Chinese).
29. Nakajima S. New technology of shield tunnel (8). III. design and construction section. Tunnel and Underground
1991; 22(1):69–78 (in Japanese).
30. van Empel WHNC, de Waal RGA, van der Veen C. Segmental tunnel lining behavior in axial direction. In
Geotechnical Aspects of Underground Construction in Soft Ground}Is}Tokyo’99, Kusakabe O, Fujita K, Miyazaki
Y (eds), Tokyo, Japan. A.A. Balkema: Rotterdam, 2000; 357–362.
31. Ding WQ, Yang LD, Zhu HH. Simulation of the material behavior in shield tunnel construction. Journal of Tongji
University 1999; 27(4):297–302 (in Chinese).
32. Zhu HH, Tao LB. Study on two beam–spring models for the numerical analysis of segments in shield tunnel. Rock
and Soil Mechanics, 1998; 18(2):26–32 (in Chinese).

Copyright # 2004 John Wiley & Sons, Ltd. Int. J. Numer. Anal. Meth. Geomech. 2004; 28:57–91

Das könnte Ihnen auch gefallen