1042

STUDY ON SEISMIC CALCULATION METHOD FOR SHIELD TUNNELS UNDER

STRONG GROUND MOTION
Tengyan LI1 And Shiro TAKADA2

SUMMARY
Hyogo-ken Nanbu Earthquake of January 17th 1995 caused severe damage to underground
structures, due to which the seismic problems on them were once again taken seriously. The antiearthquake code of shield tunnels and bridges was revised, in which two design levels representing
low-to-moderate and strong earthquakes are considered. However, under both design levels, the
elastic properties of the tunnel-ground system are assumed. This assumption is not realistic under
Level-2(strong) earthquakes, because the seismic forces are so strong that the system will behave
inelastically. In several relevant design guides, the seismic deformation method is recommended
as a design guide for shield tunnel anti- earthquake designing. However, on the modeling of the
shield tunnel itself, the method of replacing the lining structure with equivalent stiffness beam is
frequently adopted, which means the actually discontinuous lining structure is considered as a
continuous and uniform stiffness beam in the same axial direction. But in case of strong
earthquakes such a replacement method might ignore the effect of the interaction between the
shield tunnel and soil, and that of non-linear features of the ring joints and linings on the seismic
response of the shield tunnel. As a result, the internal forces which occur in the shield tunnels will
be evaluated excessively.
In this paper, the transfer matrix method is employed to derive a calculating method for antiearthquake designing of shield tunnels. The method is proved to be satisfactory and the above
problems are solved.

INTRODUCTION
With the development of science and technology, shield tunneling, method is now frequently applied to such
urban tunnel structures as underground railways, water supply and sewerage. With the advantages of little
vibration, few noises and no influence on the ground traffic, it is most suited to adapt to build tunnels on the
ground of retardation and poor subsoil. Therefore, shield tunnels are becoming one of the important urban
constructions.
The seismic calculating method of shield tunnels has been developed a lot through researchers' unceasing efforts.
A great deal of experimental studies have been made by Miwa on the axial rigidity features of shield tunnels,
which succeeds in clarifying the rigidity features of ring joints and segment rings[Miwa,1987]. Sakuma proposes
to simulate ring joint and segment ring with spring and bar[Sakuma, 1990]. But by this method calculation will
be rather limited, for a shield tunnel is composed of numerous ring joints and segment rings. Besides, only
modeled by concentrated springs, the interaction between shield tunnel and foundation can not be accurately
reflected. In order to solve this problem, Takada and Li establish transfer beam element to analyze twodimensional seismic response along the shield tunnel axis[Takada and Li,1992]. Shiba and Kawajima employ
equivalent stiffness beam to set up a seismic calculating formula of shield tunnel along the axial direction, which

1
2

Earthquake Disaster Mitigation Department, Japan Engineering Consultants Co., Ltd. , Japan
Department of Civil Engineering, Kobe University, Japan

is adopted by Guidelines for Seismic Design Method of Large Underground structures[Shiba and
Kawasima,1989,1992]
Hyogo-ken Nanbu Earthquake, which took place in the south of Hyogo Prefecture, Japan on January 17th, 1995
, caused severe damages to civil constructions, including underground structures. For example, the platforrn of
subway Daikai Station was almost completely destroyed. Although the shield tunnels didn't suffer great
damages, those applied to sewerage systems in the areas of Karumojima, Rokko island and Naruomikage
cracked a lot, which means that shield tunnel did not function well in reality. At this juncture, research
committees were established by such organizations and associations as the Japan Society of Civil Engineers and
so on to analyze the reasons of the damages and the suggestions' book was published on concerning antiearthquake problems.
Due to warnings taken from the Great Earthquake, the earthquake code of water supply, sewerage and bridges
was revised, dividing ground motions into Level-1 and Level-2.. Level-1 ground motion is equal to external
seismic force as termed in the previous designing code, happening once or twice during the lifetime of the
structure, while Level-2 ground motion is that of rare occurrence, including continental type like Hyogo-ken
Nanbu Earthquake and large-scale oceanic type whose location is close to the continent. Up to now seismic
deformation method is normally applied for seismic design of shield tunnel which is taken as equivalent elastic
foundation beam. By this method the non-linear feature of shield tunnel and that of the interaction between
ground and shield tunnel will not be considered. Therefore, it does not fit the seismic calculation of Level-2
ground motion.
Upon the above problems, transfer matrix method is employed in this paper to solve the above problems, and
discussion is made on how the calculation result is affected by non-linear feature of shield tunnel. Finally,
comparison is made with the concurrent seismic code.

CALCULATION METHOD

GL

Ring-joint

Segment-ring

Shield tunnel

Bace Rock
Ring-joint

In this paper, seismic deformation method is used to analyze seismic response of shield tunnels. As the
modeling method for a shield tunnel, springs and elastic foundation beams are employed respectively to simulate
ring joints and segment rings, as shown in Figure 1. To solve the problem of infiniteness of ring joints
and segment rings that a shield tunnel has, transfer matrix method is introduced.

Ring-joint

Subsurface

-1
@

Segment-ring
@

Interaction
springs

Gruund

Seismic wave

Segment-ring
Interaction
springs

Figure 1: Modeling method for a shield tunnel

Figure 2: Model of Segment ring

According to Figure 2, the following equations of equilibrium can be set up for segment ring No.i.

d 2U
X cos
)
+ K sxU = K sxU sx 0 sin (t
2
C
dX
d 4V
X cos
)
EI
+ K syV = K syVsy 0 sin (t
4
C
dX

EA

(1)

(2)

in which U is longitudinal displacement of shield tunnel, V is transversal displacement, I is inertia moment and
A is acreage. Ksx is a longitudinal ground spring coefficient while Ksy is a transversal one. Usx0 is longitudinal
amplitude of seismic wave and Usy0 is transversal one. C is propagation velocity of seismic wave. By solving
2

1042

Equation (1) and (2), the following relation of state vector on the two shoulders of Segment ring No.i can be
obtained,

Vi = FiVi
R

(3)

in which F is called field transfer matrix, determined by geometry and external force conditions of segment
rings, while ViR and ViL indicate the state vector on the right and left shoulders of Segment No.i, composed of
displacement and internal force of the segment ring.
On the other hand, as shown in Figure 3, according to the equilibrium condition of displacement and force on the
two shoulders of the ring joint, where the following relation of state vector can be obtained,
lk

lk+1
MkR

lk+1

lk

L
Mk+1

Qk +1 = Qk
L
R
Vk +1 = Vk
L

M k +1 = M k = k R (k +1 k )
L

N kR

L
Qk+1

QkR

lk+1

lk

Nk+1

N k +1 = N k = k T (U k +1 U k )
L

Figure 3: Model of ring joint

Vi +1 = PiVi
L

(4)

Here Pk is called Point Transfer Matrix. According to Equation (3) and Equation (4), the state vector of the left
shoulder of the shield tunnel as shown in the analytical model Figure 1 can be transferred to its right shoulder
through the following equation,

V N = FN PN 1 P2 F2 P1 F1V1
R

(5)

With the boundary condition of the analytical model substituted into Equation (5), the displacement and internal
force occurred in the shield tunnel can be obtained.

MODEL AND RESULT OF CALCULATION

Model of Calculation
The ground condition of calculation model is as shown in Table 1. The shield tunnel is buried in the layer of
alluvial clay, at a distance of 16.1 m from the ground surface, Its characteristics are shown in Table 2. In line
with the design code for sewerage systems, ground displacement and seismic wave length can be respectively
obtained by the following equations,

U sx 0 ( z ) =
L=

2
z
S v Ts cos
2
2H

(6)

VDS VBS
Ts
V DS + VBS

(7)

Here Sv means the design velocity response spectrum. Sv in this paper is supposed to be 24 m/s for Level-1
ground motion, and for Level-2 seismic motion Sv to be 80 m/s. Ts is the fundamental period of vibration of the
subsurface ground. H is the thickness of the subsurface ground. Vds is the shear wave velocity of the subsurface
ground, while Vbs is that of the base rock. Usx0(z) is the horizontal displacement at a distance of z meters from
the ground surface. The spring coefficient, which indicates the interaction between the shield tunnel and the
ground, is supposed to be equivalent to the shear modulus of the ground, and the yield shear stress of the ground
3

1042

equivalent to the standard penetration coefficient of the ground. Figure 4 shows non-linear features of the ring
joints and the ground. The non-linear features of the ring joints are defined by the result of the experiment.

Shear stress

force

K=G/1000

kT2
f N tf/m2
Tension

K=G

kT1

strain

displasement
Compression

kc

kc=3.39 1012 tf/m2

kT1=1.52 105 tf/m2

G=677 tf/m2

kT2=7.5 103 tf/m2

(Ring-joint)

(Ground springs)

Figure 4: Non-linear features of the ring joints and the ground

Table 1: Ground condition
Depth
(m)

Thickness
of layer(m)

Bankfill

-2.3

2.3

Bankfill

-9

6.7

-13.6

Alluvial
sandy
soil
Alluvial
clay
Diluvial
sandy
soil
Diluvial
clay
Diluvial
gravel

Table2: Characteristics of tunnel

Density
(tf/m3)

Velocity
(m/s)

1.8

150

15

1.9

200

4.6

15

1.8

200

Tunnel elastic
modulus(kgf/cm2)

360000

-18.3

4.7

1.6

150

Bolt type

M22

-22.8

4.5

15

1.9

200

Bolt acreage(m2)

3.03

-30.0

7.2

20

1.7

280

-43.0

13.0

50

2.1

350

N-valve

Item

Value

Tunnel
diameter(m)
Tunnel
thickness(m)

2.15
0.9

Bolt elastic
modulus(kgf/cm2)
Bolt yield
stress(kgf/cm2)

2100000
2450

Result of calculation
The seismic response of shield tunnels is analyzed in this paper in the light of Level-1 ground motion and Level2 ground motion. Seismic deformation method is applied for seismic calculation. Seismic wave is supposed by P
wave in sine distribution. Discussions are made separately on the 5 cases as shown in Table 3 in order to find
Table 3 : Calculation cases
Linear

Case1
Case2
Case3
Case4
Case5

Non-linear

Segment stiffness Equivalent stiffness Beam(uniform beam)

(uniform beam)
spring
#
#
#
#
#
#:means being considered.

Ring
joint
#
#

Ground
spring
#

1042

out how seismic response of shield tunnels is affected by various modeling methods.
Level-1 ground motion
Table 4 shows the stress and the displacement of the ring joints that occurs in the shield tunnel under Level-1
ground motion in different cases. It can be known from their displacement that the ring joints are not in the
plastic region in any of the cases. The reason is that the external seismic force of Level-1 is comparatively weak.
Although for tension stress the errors that might occur in various cases is under the rate of 2%, for compression
stress the error rate in Case 1 and Case 3 is 12% and that in Case 3 and Case 5 is 41%. Therefore, equivalent
stiffness beam values at a low rate the compression stress occurring in the shield tunnel, on which the non-linear
features of the ground have a great effect. Shown in Figure 5 is the displacement of ring joints, and in Figure 6
the stress distribution. As the ring joints are within the linear region, the result of Case 3 is almost the same as
that of Case 4. The result of Case 5 is rather small than that of any other case due to the influence of the nonlinear features of the ground.

Table 4: Result of Calculation

Relative displacement(cm)

Case1
Case2
Case3
Case4
Case5

Compression
stress(tf/m2)
1617
125
1820
1820
1282

Tension stress
(tf/m2)
1617
125
123
123
120

Displacement of
ring joint(m)

0.00076
0.00076
0.00074

Tension:+
Compression:-

0.06

CASE3
CASE4
CASE5

0.04
0.02
0
0

100

200
Stretch(m)

300

Axial stress(tf/m )

Figure 5: Relative displacement of ring joint

1000

Tension:Compression:+

CASE1
CASE2
CASE3
CASE4
CASE5

-1000
0

100

200
Stretch(m)

300

Figure 6: Axial stress distribution of shield tunnel

1042

Level-2 ground motion

Table 5 shows in various cases the stress and the displacement of the ring joints that occurs the shield tunnel
Table 5: Result of Calculation

Case1
Case2
Case3
Case4
Case5

compression
stress(tf/m2)
5389
417
6067
6077
1552

tension stress
(tf/m2)
5389
417
412
178
175

Displacement of
ring joint(m)

0.00254
0.00275
0.0026

CASE1
CASE2
CASE3
CASE4
CASE5

0.05
0
-0.05
0

100

200
Stretch(m)

300

Figure 7: Relative displacement of ring joint

Relative displacement(cm)

Response displacement(m)

under Level-2 ground motion, which is different from those under Level-1 seismic motion. In the cases of Level2 seismic motion, the joints are all in the plastic region. Thus the tension stress in Case 2 is 2.4 times stronger
than that in Case 4. The non-linear features of the ground have few effect on tension stress but great effect on
compression stress. The compression stress in Case 4 is about 4 times of that in Case 5. Therefore, the non-linear
features of the shield tunnel and the ground should be considered when seismic response of the shield tunnel is
evaluated in case of Level-2 ground motion. Shown in Figure 7, Figure 8 and Figure 9 are the displacements of
the shield tunnel and the ring joints, and the stress occurring in the shield tunnel. Due to the effect of the nonlinear features of the ground, the range of the maximum relative displacement occurring in the joints in Case 5 is
smaller than in any other case, and the stress distribution is also far less than in any other case. In addition, for
the sake of the discontinuous properties of the shield tunnel, the ranges of displacements which cause tension
stress occurring in the shield tunnel in Case 3, Case 4 and Case 5 are smaller than those which cause
compression stress. This is also because compression stress is stronger than tension stress.

Tension:+
Compression:-

0.2

CASE3
CASE4
CASE5

0.1

0
0

100

200
Stretch(m)

300

Figure 8 : Relative displacement of ring joints

1042

Axial stress(tf/m2)

5000

CASE1
CASE2
CASE3
CASE4
CASE5

Tension:Compression:+

-5000
0

100

200

300

Stretch(m)
Figure 9 : Axial stress distribution of shield tunnel
CONCLUSION
In this paper, transfer matrix method is employed to analyze seismic response of shield tunnels and is compared
with equivalent stiffness beam method. For Level-1 ground motion, the seismic force is comparatively weak. As
a result the discontinuity of shield tunnels and the inelastic properties of the ground has a small effect on the
seismic response of the shield tunnel. By the above-mentioned two methods, the error rate of tension stress
occurs in the shield tunnel is under 2%, and that of compression stress is only 12%. Therefore, equivalent
stiffness beam method is feasible to calculate seismic response of shield tunnels in case of Level-1 ground
motion.
But in case of Level-2 ground motion, the discontinuity of shield tunnel and the inelastic properties of the ground
has a great effect on the seismic response of shield tunnel. The stress obtained by equivalent stiffness beam
method is 2.4 times stronger than that obtained by the method in which the effect of the inelastic properties are
well considered. Therefore the effect of the inelastic properties can not be neglected when Level-2 ground
motion is applied for seismic design of shield tunnels, and equivalent stiffness beam is not fit for seismic
calculation in case of Level-2 ground motion.
In this paper, the problem of the discontinuity of shield tunnels is solved by transfer matrix method, which well
reflects the inelastic properties of the ring joints and the ground. It can be concluded that transfer matrix method
is a good method for seismic calculation in case of Level-2 ground motion. But in this paper the analysis is only
made on seismic response of shield tunnels in the axial direction. Further study should be made on that in the
transversal direction.
REFERENCES
1.

2.

3.
4.

5.

6.
7.
8.

Miwa, Z.(1987),"Evaluating Experiment on the stiffness of Shield Tunnels in Axial Direction",

Proceedings of the 42th Annual Conference of the Japan Society of Civil Engineers, pp596-600.(in
Japanese)
Sakuma, S.(1990),"Effects of the Axial Stiffness of a Shield Tunnel Including Effects of Shearing
Interaction between Primary and Secondary Linings", Proceedings of the Japan Society of Civil
Engineers,No.424/ -14,pp.251-259.(in Japanese)
Takada, S. and Li,T.Y.(1992),"Non-Liner Seismic Response Analysis for Large Scale Buried Structures,
ASME, PVP-Vol.227, pp65-72.
Shiba, Y. and Kawasima, K.(1989),"Evaluation Procedure for Seismic Stress Developed in Shield Tunnels
based on Seismic Deformation method", Proceedings of the Japan Society of Civil Engineers, No.404/ 11,pp.385-394. (in Japanese)
Asakura, H. and Kawashima, K.(1992),"Guidelines for Seismic Design Methods of Large Underground
Structures", 24th Joint Meeting of U.S.-Japan Cooperative Program in Natural Resources, Panel on Wind
and Seismic Effects, Gaitherburg, MD, U.S.A., pp39-53.
Takada, S. and Li,T.Y.(1995),"Seismic Motion and Damage Characteristics", Journal of the Japan Society
of Civil Engineers,Vol.80, pp50-57. (in Japanese)
Takada, S. and Li, T. Y.(1996),"Damage and Measures to Port Facilities", Journal of Japan Society
Engineering, Vol.35, No.1, pp8-14. (in Japanese)
Takada, S(1991), Lifeline Earthquake Engineering, Kyoritsu Publishing House. (in Japanese)

1042