6th International Conference on Earthquake Geotechnical Engineering

1-4 November 2015

Christchurch, New Zealand

Transverse Seismic Analysis of Shield Tunnel Lining

in Multi-Layered Soft Ground Using Different Methods

J.H. Wang 1, H. Tanaka1, M. Nakano1, H. Sugiyama1, A. Koizumi 2 and F. Chen 3

ABSTRACT
This paper investigates the transverse seismic performance of a shield tunnel in multi-layered
ground. A three-dimensional dynamic finite element (3D FE) analysis and a simplified two-
dimensional (2D) frame analysis were performed accounting for bolt-jointed segmental lining and
multi-layered ground. Meanwhile, the seismic performance of the tunnel was analysed by a
closed-form solution. The analysis results from the three methods indicate that the ground
composition greatly impacts the seismic response of the ground. The 3D FE analysis should be
used to analyse the seismic performance of tunnel lining for a conservative design.

Introduction
The ASCE (1974) and JSCE (1995) have reported that there are many underground structures
that suffered damages during past earthquakes. The former documents the damage in the Los
Angeles area during the 1971 San Fernando Earthquake, and JSCE (1995) recorded some
underground utilities (e. g., collapse of Daikai station) damaged in the 1995 Hyogoken-Nanbu
Earthquake (HNE). Power et al. (1998) and Kaneshiro et al. (2000) present summaries of 217
case histories as an extensive database of seismic damage to underground structures. Generally,
underground structures suffer relatively less damage than surface structures, except for those
constructed in poor ground. However because the damage to underground structures is usually
due to the peak ground motion, the impact of multi-layer ground on underground structures
should be clarified in addition to the magnitude and source-to-site distance of the earthquake.

The seismic design philosophy for underground facilities has been standardized (e. g., JSCE
2006) using the two-level design criteria: the higher design level earthquake for life safety and a
lower design level earthquake for the basic function of the facility (Youssef et al. 2001).
Meanwhile, numerous seismic analysis approaches have been used for different facilities. In
general, the seismic analysis methods can be placed into three categories: the closed-form
solutions, pseudo-static analysis and dynamic analysis. In particular, because the closed-form
solutions (Newmark 1968, Wang 1993 and Penzien 2000) are based on the propagation of
earthquake waves in the idealized single layer free field, the application to multi-layered soil
deposits needs to be verified.

Therefore, this study aims to clarify the impact of multi-layered ground on the transverse seismic
performance of a shield tunnel and to examine the related seismic analysis approaches. A shield
tunnel coupled with the multi-layered soft ground is extensively investigated by the closed-form
solutions, 2D frame pseudo-static analysis and 3D FE dynamic analysis in terms of the response
of the ground and tunnel lining.
1
Dr. Wang, et al. Res. & Dev. Center, Nippon Koei Co., Tsukuba, Japan. A6956@n-koe.co.jp
2
Prof. Koizumi, Dep. Of Science and Engng., Wased Univ. Tokyo, Japan. Koizumi@waseda.jp
3
Mr. Chen, D&C Structure Analysis Co., Matsudo, Japan. Chin@dckozo.co.jp
 Seismic analysis methodology

Closed-form solutions

Solutions for the transverse seismic analysis of underground structures were provided using the
deformation model of the free field due to the propagation of a shear wave. Newmark (1968),
Wang (1993), Power et al. (1996), and Penzien (2000) proposed the closed-form solutions in
which the transverse ovalling / racking is used to compute the strains and forces in the lining.
The related equations to evaluate the free field ground and the internal forces in the linings are
summarized in Table 1 using two methods: Penzien (2000) and Wang (1993). In the equations,
ρ m is density, G m is the shear modulus, and E m is the elastic modulus of the ground; Cs is the
apparent velocity of S-wave propagation, a s is the peak particle acceleration associated with S-
wave, ν m is Poisson’s ratio, and R velo&acce is the ratio of the peak velocity to the peak acceleration
at the surface; γ max is the maximum free-field shear strain; d, t, I and El are the diameter,
thickness, inertia moment and elastic modulus of the tunnel lining, respectively; and Ts max , Vs max ,
Ms max and T max , V max and M max are the maximum thrust force, shear force and bending moment
for full-slip and no-slip conditions, respectively.

Table 1. Equations used in closed-form solution

Gm = ρ mCs2 Em = 2Gm (1 + ν m )
Δd free − field
Vs = Rvelo & acce ⋅ a s γ max = Vs / Cs =
2d (1 − v )
Full –slip- condition No –slip- condition
12 E I (5 − 6ν m ) 24 E I (3 − 4v m )
α = 3l
s
α = 3l
d G m (1 − ν l2 ) d G m (1 − ν l2 )
Parameters

4(1 − ν m ) 4(1 − ν m )
Rs = R =
αs +1 α +1
Penzien (2000)

γ max d γ max d
∆d lining
s
= ± R s ∆d free − field = ± R s ∆d lining = ± R ∆d free − field = ± R
2 2
12 E l I∆d s
π 24 El I∆d lining π
T s max = cos 2(θ + )(θ = 45°) Tmax = cos 2(θ + )(θ = 45°)
lining

d 3 (1 − ν l2 ) 4 d 3 (1 − ν l2 ) 4
Internal forces

6 E l I∆d lining
s
π 6 El I∆d lining π
M s max = cos 2(θ + )(θ = 45°) M max = cos 2(θ + )(θ = 45°)
d 2 (1 − ν l2 ) 4 d 2 (1 − ν l2 ) 4

24 E l I∆d lining
s
π 24 El I∆d lining π
V s max = sin 2(θ + )(θ = 0°) Vmax = sin 2(θ + )(θ = 0°)
d 3 (1 − ν l2 ) 4 d 3 (1 − ν l2 ) 4

Em (1 − ν l2 )d Em (1 − ν l2 )d 3
C= F=
2 El t (1 + ν m )(1 − 2ν m ) 48El I (1 + ν m )
Parameters

F [(1 − 2ν m ) − (1 − 2ν m )C ] − (1 − 2ν m ) 2 + 2
1
12(1 − ν m ) 2
Ks = K =
2 F + 5 − 6ν m [
F [(3 − 2ν m ) − (1 − 2ν m )C ] + C 5 / 2 − 8ν m + 6ν 2 m ) + 6 − 8ν m ]
Wang (1993)
Internal forces

Em Em
T s max = ± K s γ max d Tmax = ± K γ max d
12(1 + ν m ) 4(1 + ν m )
1 Em 1 Em
M s max =± Ks γ max d 2 M max =± K γ max d 2
24 (1 + ν m ) 24 (1 + ν m )
Simplified frame analysis method

The simplified frame analysis method uses beam-spring and ground-spring support models to
calculate the resultant internal forces, according to the specifications of the shield tunnel (JSCE
2006, 2011). In addition, the deformation method is recommended for the ground response
analysis based on the scientific knowledge that the movement of a flexible tunnel structure
always mimics the surrounding ground during an earthquake. The procedure for seismic analysis
includes two steps: 1) performing the ground response analysis using dynamic analysis software
such as Shake (Schnabel et al., 1972) and determining the maximum relative displacement at the
crown and invert sites and the corresponding shear stress around the tunnel at same moment; and
2) imposing the displacement and shear stress on a beam-spring model supported by ground-
springs and calculating the internal forces.

3D FE dynamic analysis procedure

The procedure of 3D dynamic analysis was proposed by Wang et al (2013) and is applied in
three steps. First, evaluate the equivalent shear modulus and damping coefficient of each soil
layer using the ground analysis software (Shake) for ground response analysis and the
equivalent-linear model to account for the soil non-linearity with respect to the shear strain under
cyclic loads. Second, evaluate the α and β for Raleigh damping by conducting the eigenvalue
analysis of ground to obtain the vibration frequency and damping coefficients of the double
predominant deformation modes. Finally, perform dynamic analysis for a soil-structure system
with equivalent shear modulus and the Rayleigh damping coefficients. The related dynamic
equation and Rayleigh formula are given below,

[M ]{U(t )}+ [C ]{U (t )}+ [K ]{U (t )} = {E (t )} (1)

[C ] = α [M ] + β [K ] (2)

where α = 4πf1 f 2 (2h1 f 2 −2 h2 f1 ) ; β = 2( h2 2f 2 − h21 f1 ) ; U(t), Ů(t) and Ü(t) are the displacement,
f 2 − f1 f 2 − f1
velocity and acceleration of the soil-structure system, respectively; [M], [C] and [K] are the mass
matrix, damping matrix and stiffness matrix of the soil-structure system, respectively; E(t) is
earthquake force; α and β are the coefficients of Rayleigh damping; and f 1 , f 2 and h 1 , h 2 are the
vibration frequency and damping coefficient of the two predominant modes, respectively.

Target shield tunnel and modelling

Conditions

A 4.3 m diameter shield tunnel is selected, and the structure and ground conditions are shown in
Figure 1. As shown in Figure 1 (a), one ring of linings is assembled by connecting seven steel
segments (A1 to A5, B and K ) with steel bolts, and the segment has the height of h = 150 mm,
the width of B = 900 mm, and the different perimeter from A, B and K types. The tunnel lining is
assembled by the staggered joint arrangement method, and the rings are connected with steel
bolts in the longitudinal direction.
 Target site

Landfill (Bs)

Silty sand(As)

Shield tunnel
Silt (Ac)

3D FE model
( L=50 m )
Soft rock (Ka)

(a) Lining and segment (b) Ground in 3D FE analysis (c) 2D frame analysis (d) Closed form analysis

Figure 1. Structure and ground condition

Table 2. Structural, geotechnical and earthquake parameters

Composite lining Geotechnical Earthquake ( HNE_NS)

- Diameter, d =3.4 m - Soil deposit thickness H =17.75 m -Moment magnitude Mw =7.3
-Equivalent thickness, t eq =0.167 m - Unit weight γ eq =16.35 kN/m3 - Peak ground particle
-Equivalent Young’s Modulus, -Equivalent period of the soil acceleration in soil, a max =0.818 g
E eq = 7808.8 kN/m2 deposit T =2.062 sec -Apparent velocity of S-wave
- Moment of inertia, I eq = 349.3 cm4 - Soil Poisson’s ratio, v m =0.45 propagation Cs =96.5 m/sec
HNE denotes the N-S wave recorded in 1995 Hyogoken -Nanbu Earthquake

Figure 1(b) shows the longitudinal profile of the ground, which is multi-layered deposits mainly
consisting of landfill (Bs), silty sand and silt clay. The ground condition of site boring No. 3 is
used for the seismic analysis by the simplified frame method and closed-form solution. The
geotechnical and structural parameters are given in Table 2, along with the earthquake
parameters. The equivalent thickness t eq and the Young’s modulus E eq were evaluated based on
the principle of equivalence of axial and flexural stiffnesses of composite steel segment and the
plate model.

Modelling and analysis procedure

The numerical analysis models are shown in Figure 2. A 50 m long, 100 m wide and 21.4 m high
cubic ground was employed for the 3D FE dynamic analysis. As shown in Figure 2 (a), the target
of 27 m long linings (30 rings) in the middle were modelled in detail. Transverse and rotational
springs are used to simulate the segment and ring joints, and the equivalent plate elements are
used for the segments. The remaining linings at the two ends were modelled by the constitutive
 Nodes
2D Beam
Ground
spring
(tangential

(a) 3D FE model (b) 2D frame model

Figure 2. Modelling for numerical analysis

Table 3. Material properties of ground

Density Cohesive force Friction angle Elastic modulus Poison’s ratio

3
γt (kN/m ) C (kN/m2) Φ (Deg.) E (MN/m )
2
ν
Landfill (Bs) 18 15 0 1.69 0.45
Silty sand(As) 17 0 20.5 1.4 0.495
Silt (Ac) 16.1 35.1 0 3.74 0.495
Mudstone(Ka) 19.7 263.9 21.1 65.01 0.4

Table 4. Ground response analysis in closed-form solution

Mw D L (km) a max (g) γ m (kg/m3) Cs (m/s) νm

Earthquake
7.3 16.5 0.818 1635 96.5 0.45
a s (g)
G m (kPa) E m (kPa) R velo&acce Vs (m/sec) γ max
Ground (αs=0.9)
15226 44155 0.736 194.4 1.43 0.00477

plate elements disregarding the joints, which is used to relax the influence of boundaries on the
targeted 30 ring linings. The geometry and material conditions of the plate element uses the
parameters shown in Table 2, and the nonlinear properties of the springs are given in Figure 2.
Moreover, a cubic solid element was used to simulate the surrounding ground, and the material
properties of each deposit are shown in Table 3. For the simplified frame analysis, two-
dimensional beams were employed to model the segment ring, and the ground springs were used
to simulate the soil-structure interaction. The FE software package of Soil plus (CTC 2009) was
employed for 3D dynamic analysis, and the general analysis code 2D frame and Shake were used
for 2D analysis. The ground response analysis in the closed-form solution used the equations
shown in Table 1, and the calculated results are shown in Table 4.
 Analysis results and discussion

Results of seismic analysis

The analysis results from the three methods are presented in terms of the peak ground
displacements and the maximum internal forces. To compare the ground response results, the
peak ground displacements and the relative shear deformation due to ground distortion are
shown in Figure 3, where the analytical results from the closed-form solution were estimated by
the simple equation δ =h i γ max (where h i is the thickness from the bed ground) based on the
maximum shear strain γ max . The maximum thrust, bending moments and shear force of the
tunnel lining due to the seismic motion are shown in Figure 4 in terms of the 2D simplified frame
analysis and 3D FE dynamic analysis. To clearly show the effects of joint allocation in different
segments and rings, the average value of each element is displayed for the contour in the 3D FE
analysis. Moreover, the maximum internal forces are summarized in Table 5, in which the
analytical results were estimated by Wang (1993) and Penzien (2000) for the no-slip condition
and the full-slip-condition.

Peak ground displacement δ (cm) Peak ground displacement δ (cm) Peak ground displacement
0 5 10 15 20 25 30 0 5 10 15 20 25 30
0.0 0.0
δmax=26.3cm
2.0 2.0

4.0 4.0 δmax=15.3cm

6.0 6.0
δU=14.6cm δU=13.3cm
8.0 8.0

10.0 10.0 Δd_3D= 5.57cm

Relative shear
12.0 δL=8.2cm 12.0 δL=7.2 cm deformation

14.0 Δd_A=6.38cm 14.0 Δd_2d=6.08cm

+23
+20
+18
+16
+13
+11
+9
+6
+4
+2
+0
Reletive shear Reletive shear
16.0 deformation 16.0 deformation

18.0 Depth H (m) 18.0 Depth H (m)

(a) Closed-form solution (b) 2D frame analysis (c) 3D FE analysis

Figure 3. Peak ground displacement and relative shear deformation Δd (cm)

Ground deformation

From Figure 3, the nonlinearity of peak ground displacement can be observed in the 2D frame
analysis and the 3D FE analysis, in which the multi-layer ground was considered. The maximum
shear displacement evaluated by the closed-form solution was the largest, whereas that estimated
by the 2D frame analysis is the smallest. The maximum shear displacements were 26.3 cm, 15.3
cm and 23 cm for the closed-form solution, 2D frame analysis and 3D analysis, respectively.
However, the relative shear displacements between the bottom and the top of the tunnel lining
were Δd_3D = 5.57 cm, Δd_2D = 6.08 cm and Δd_A = 6.38 cm for the 3D analysis, 2D frame
analysis and the closed-form solution, respectively. Accordingly, the composition of deposit
layers affects the seismic response of the ground. In addition, the little difference of the relative
shear deformation between the three methods agrees well with the relation of the ground
condition used for the three analysis. That is, the condition of the tunnel-buried deposit layer
with the thickness of 8.75 m and the initial shear-wave velocity of 89 m/s are similar to that of an
idealized single layer ground with a thickness of 17.75 m and an initial shear-wave velocity of 94
m/s (see Figure 1). This may imply that the application of the closed-form solution is valid if the
tunnel-buried deposit is thick enough. Furthermore, the difference of the ground response
between 2D frame and 3D FE analysis can be explained by their different conditions. The
former used two-dimensional ground, whereas the latter employed three-dimensional ground,
accounting for the change of thickness of each soil layer along the tunnel. Therefore, the 3D FE
analysis method should be used to achieve a more reasonable ground response for multi-layered
ground.

Figure 4. Maximum internal forces per ring tunnel by numerical analysis

 Table 5. Comparison of maximum internal forces

Analysis Soil Bending moment Hoop thrust Shear force

method Structure M (kN・m/ring) N (kN/ring) V (kN /ring)
Full-slip ±54.6 ±25.5 ±50.9
Penzien
No-slip ±54.5 ±50.8 +50.8
Full-slip ±54.6 ±25.5 －
Wang
No-slip ±54.6 ±478.4 －
2D No-slip ±39.3 ±54.4 ±37.3
53.1 61.7 143.1
3D No-slip
-52.5 -63.5 -157.1

Internal forces of tunnel linings and comparison

Figure 4 shows that the circumferential distributions of bending moment, hoop thrust and shear
force are similar between 2D frame analysis and 3D FE analysis. The peak value happens at the
site of ±π/4 radial angle with respect to the springline for the bending moment and hoop thrust
force and at 0 and π radial angles for shear force. This is consistent with the theory of the closed-
form solution and the damage observation of tunnels due to earthquakes (Lanzano, et al. 2008).
In addition, the discontinuity of internal forces around the ring illustrates well the structural
feature of a shield tunnel composed of segments and radial joints. In addition, the staggered
configuration of the adjacent rings results in the variation distribution of internal forces along the
lining in the 3D FE analysis.

The results comparison shown in Table 5 indicates that the analytical peak bending moment
agrees well with the result of the 3D FE analysis but is approximately 30% higher than the 2D
frame analysis result. Conversely, the hoop thrust forces of the three methods coincide with each
other if the results of Penzien’s solution for the no-slip condition is considered. However, the
solution from Wang for the no-slip condition yields a rather higher thrust force than both of the
two numerical analyses and the Penzien solution, although giving agreeable analytical thrusts by
two solutions under the full-slip condition. Furthermore, the shear force computed by 3D FE
dynamic analysis is larger than those from the 2D frame model analysis and the closed-form
solution. This may imply that the three-dimensional model of the tunnel can take into account the
spatial effects of the tunnel by providing more conservative results than those of the other two
methods.

The above results clearly illustrate that the circular tunnel lining is vulnerable to the transverse
ovalling due to the ground shear deformation under earthquake events. The soil-tunnel system
must be employed in seismic analysis considering that the seismic loading to the tunnel lining is
mainly produced by the ground motion. Moreover, because the shield tunnel is a flexible
structure, the slip condition may be disregarded for a conservative design. The closed-form
solution from Penzien can be used as a simple approach to evaluate the internal forces for a
tunnel completely buried in a thick deposit layer, while the 3D FE dynamic analysis method may
be applied to evaluate the internal forces for any tunnel because of its high capacity to reproduce
the soil layer composition of the ground.
 Summary and remarks

The seismic response of shield tunnel lining was extensively investigated by three methods: the
closed-form solution for an idealized single layer ground, 2D frame analysis for horizontally
multi-layered soil deposits and the 3D FE dynamic analysis accounting for the relatively flexible
joints and multi-layered ground. The peak ground displacement and the tunnel lining’s internal
force analysed by the three analysis methods were discussed, and the application of each method
was identified. From this study, the following remarks can be given:

1) For multi-layered ground, the soil layer composition greatly impacts the seismic response of
the ground. However, if a tunnel is completely buried in a relative thick deposit layer, the
relative shear deformation of the tunnel lining shows little difference for the idealized single
layered ground and the multi-layered ground. Therefore, the single layered ground model
can be acceptable as a simple engineering solution for a tunnel completely buried in a thick
deposit layer; otherwise, the multi-layered ground model must be employed for a more
reasonable ground response.
2) The closed-form solution may be used as a simple approach to evaluate the internal forces
for a tunnel completely buried in a single thick deposit layer. The 2D frame analysis is
prone to provide the smaller internal forces, whereas 3D FE dynamic analysis always
produces conservative results in terms of the internal forces. Therefore, the 3D FE dynamic
analysis method should be applied for the safe design of tunnel lining under earthquake
events.

Although in this study the 2D frame analysis and 3D FE dynamic analysis were performed with
the soil-tunnel system under the no-slip condition, the seismic analysis of a soil-tunnel system
under full-slip condition or the real condition based on the shear strength of deposits should be
investigated extensively. In addition, more studies should be carried out for different grounds.

References

ASCE. Earthquake damage evaluation and design considerations for underground structures, February. American
Society of Civil Engineers, Los Angeles Section, 1974.
CTC. Manual of Soil Plus 2009. ITOCHU Techno-Solutions Corporation.
JSCE (1995). Secondary reports on evaluation of earthquake damages due to Hyogoken-Nanbu Earthquake.
Japanese Society of Civil Engineers: Tokyo, 1995.
JSCE (2007). Standard specification for tunneling--2006, shield tunnels. Japanese Society of Civil Engineers:
Tokyo, 2007.
JSCE (2011). Guideline for seismic performance evaluation and earthquake resistance measures of underground
structure. Japanese Society of Civil Engineers: Tokyo, 2011.
Schnabel, P.B., Lysmer, J. and Seed, H. Bolton.. SHAKE: A computer program for earthquake response analysis of
horizontally layered sites. Report No. UCB/EERC-72/12, Earthquake Engineering Research Center, University of
California, Berkeley, Dec. 1972, 102.
Power, M., Rosidi, D., Kaneshiro, J.. Seismic vulnerability of tunnels-revisited. In: Ozedimir, L., Ed. Proceedings of
the North American Tunneling Conference. Elsevier: Long Beach, CA, USA, 1998.
Kaneshiro, J.Y., Power, M., Rosidi, D.. Empirical correlations of tunnel performance during earthquakes and a
seismic aspects of tunnel design. Proceedings of the Conference on Lessons Learned From Recent Earthquakes _
On Earthquakes in Turkey 1999, Nov. 8_11, 2000.
Newmark, N.M.. Problems in wave propagation in soil and rock. Proceedings of the International Symposium on
Wave Propagation and Dynamic Properties of Earth Materials, 1968.
Wang, J.-N.. Seismic Design of Tunnels: A State-of-the-Art Approach, Monograph, monograph 7. Parsons,
Brinckerhoff, Quade and Douglas Inc: New York, 1993.
Wang, J.H.,et al. A procedure for assessing the seismic capacity of aging shield sewage tunnel using 3D-FEM
Seismic analysis. Proceedings of the seventh China-Japan conference on shield tunnelling, Xian, China, On
September 21-24, 168-177, 2013.
Penzien, J., Wu, C.. Stresses in linings of bored tunnels. Int. J. Earthquake Eng. Struct. Dyn. 1998; 27, 283_300.
Youssef M.A. Hashasha, Jeffrey J. Hooka, Birger Schmidtb,John I-Chiang Yao. Seismic design and analysis of
underground structures, J. Tunnelling and Underground Space Technology 2001; 16:247_293.
Lanzano G., Bilotta, E. andRusso G.. Tunnels under seismic loading: a review of damage case histories and
protection methods. Strategies for reduction of the seismic risk, 2008.