See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/275585232

Mechanism for buckling of shield tunnel

linings under hydrostatic pressure
ARTICLE in TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY JUNE 2015
Impact Factor: 1.59 DOI: 10.1016/j.tust.2015.04.012

DOWNLOADS

VIEWS

25

32

5 AUTHORS, INCLUDING:
Jian hong Wang

W.J Zhang

Research and development center, Nippoo

Tianjin University

12 PUBLICATIONS 15 CITATIONS

6 PUBLICATIONS 11 CITATIONS

SEE PROFILE

SEE PROFILE

Xiang Guo
Tianjin University
28 PUBLICATIONS 167 CITATIONS
SEE PROFILE

Available from: W.J Zhang

Retrieved on: 26 June 2015

Tunnelling and Underground Space Technology 49 (2015) 144155

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

journal homepage: www.elsevier.com/locate/tust

Mechanism for buckling of shield tunnel linings under hydrostatic

pressure
J.H. Wang a, W.J. Zhang b,c,, X. Guo d, A. Koizumi e, H. Tanaka a
a

Research & Development Center, Nippon Koei Co., Ltd., Ibaraki Prefecture 300-1259, Japan
School of Civil Engineering, Tianjin University, Tianjin 300072, China
c
Key Laboratory of Transportation Tunnel Engineering, Ministry of Education, Sichuan 610031, China
d
School of Mechanical Engineering, Tianjin University, Tianjin 300072, China
e
Department of Civil and Environmental Engineering, Waseda University, Tokyo 169-8555, Japan
b

a r t i c l e

i n f o

Article history:
Received 23 February 2014
Received in revised form 30 March 2015
Accepted 13 April 2015

Keywords:
Shield tunnel
Segmental joint
Buckling
Hydrostatic pressure

a b s t r a c t
In this paper, the effects of segmental joints, dimensions of segments, and ground conditions on buckling
of the shield tunnel linings under hydrostatic pressure are studied by analytical and numerical analysis.
The results show that radial joints have signicant impacts on the buckling behavior: the shield tunnel
linings with exible joints buckles in a single wave mode in the vicinity of K joint, while those with rigid
joints buckles in a multi-wave mode around the linings. Hydrostatic buckling strength is found to
increase with the exural rigidity of the radial joint and the thickness of segment increasing. This study
shows that ground support increases the buckling strength dramatically, while earth pressure reduces
the capacity to resist hydrostatic buckling. The tunnel linings during construction are found to be easier
to buckle than that during operation. Meanwhile, the buckling of tunnel linings is studied by theoretical
analysis of buried tube buckling.
Crown Copyright 2015 Published by Elsevier Ltd. All rights reserved.

1. Introduction
In recent decades, the use of deep underground beneath seas
and rivers has rapidly increased in order to meet the civic requirements and improve the urban environment. Many new tunnel utilities, including undersea and riverbed tunnels, have been
constructed at progressively greater depths by shield tunneling
method, due to the congested uses of shallow ground and aboveground space in many large cities (Watanabe, 1990; JCRDB, 2006).
For such underwater tunnels, hydrostatic pressure should be
considered as a principal design load, as it is almost equivalent
to the acting load on the operating shield tunnel (Koyama, 2003;
Mashino and Ishimura, 2003; Yahagi et al., 2005). Under high
hydrostatic pressure, the tunnel linings are predominated by the
compressive hoop force, which is required by the design of waterproof of joint. This eventually makes the exible joint be utilized
extensively and the circular shield tunnel can be designed more
easily by the wide and thin segment (Kimura and Koizumi, 1999;
Koizumi, 2000). Therefore, the structural stability of the tunnel linings should be checked to avoid the buckling of circular tube under
hydrostatic pressure. However, the current design specications of
Corresponding author at: School of Civil Engineering, Tianjin University, Tianjin
300072, China.
http://dx.doi.org/10.1016/j.tust.2015.04.012
0886-7798/Crown Copyright 2015 Published by Elsevier Ltd. All rights reserved.

the shield tunnel (JSCE, 2007) only check the material safety by the
allowable stress method and ultimate state method. Equivalently,
the structural stability of the whole tunnel linings is ignored.
Tunneling accidents usually have complicated causes and eventually result in ground collapse, so that the structural problem of
the tunnel linings is often underestimated and even ignored in
design practice and relevant research, especially for reinforced
concrete linings. Heathrow Express Tunnel (NATM) collapse in
October 1994 (HSE, 2000), Gerrards Cross Tunnel (Tesco tunnel,
three pin arch) collapse in June 2005 (Wikipedia, 2005), and
Kurashiki Undersea Tunnel (shield tunnel) collapse in February
2012 (NBP, 2012) are some typical failure examples of the concrete
linings in recent years. These accidents have motivated many
researchers to investigate the buckling of tunnel linings (Croll,
2001; Tamura and Hayashi, 2005; Wang et al., 2014).
Particularly, the recent shield tunnel collapse in Japan caused a loss
of ve lives, which is a very severe accident throughout the construction history of the modern shield tunnel. Structural failure
of segmental linings has been analyzed to be the most likely cause
of undersea tunnel collapses (NBP, 2012). However, the failure
mechanism of the tunnel linings has not been claried yet. Since
the surrounding ground is dense and the external hydrostatic pressure is large, the buckling of segmental linings should be checked
as the necessary inuential factor in the collapse. Actually, the

145

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

Air exhausted by the

vacuum pump

Specimen
Sand

Flexible joint
(single wave)
Buckled mode

Rigid joint
(multi-wave)
(a) Free cylinder

(b) Buried cylinder

Fig. 1. Experimental prole and buckling of cylindrical shells with one exible joint.

Shield machine Groundwater pressure

Groundwater and earth pressures

Stable ground
Loosing
ground

Segmental linings

Void

Load condition 1

grouting material
before complete
hardening

grouting material
before complete
hardening

Load condition 2

(a) Load condition 1 and 2 (LC1 and LC2, in construction)

(b) Load condition 3 (LC3 in operation)

Fig. 2. Load conditions considered in buckling analysis during construction and operation.

experimental study using the tunnel models with radial joints has
indicated that exible segment joints reduces the buckling load of
cylindrical shells signicantly (Wang and Koizumi, 2010).
The aim of this paper is to study the stability of deep shield
tunnel linings under hydrostatic pressure. The buckling of the
segmental linings will be analytically and numerically investigated to clarify its structural failure mechanism. The following
factors will be identied: (a) the rotational rigidity of radial
joints; (b) the number and orientation of radial joints; (c) the
thickness and width of segment; and (d) the overburden depth
and ground support. In addition, the examination of stress will
be performed to check the material safety by allowable stress
method, and the buckling of tunnel linings will also be examined
by the analytical solutions of Winkler model and the elastic
continuum model.

2. Overview of cylinder buckling theories

2.1. Relevant buckling theories
Buckling of a free cylindrical shell under external pressure was
studied by Levy (1884) and Timoshenko and Gere (1961). Buckling
of a tube encased shell in rigid cavity was investigated by Cheney
(1971), Amstutz (1970), Jacobsen (1974), Glock (1977), El-Sawy
and Moore (1998), and El-Sawy (2001). The former was usually
classied into the multi-wave buckling theory, while the latter into
the single-wave buckling theory. The free buckling strength of a
beam ring can be estimated by the following equation:

Prcr

3EI
BR3

146

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

pe1
qe1
Self-weight (g)

pe1
qe1

g
Pw

Pw
qe2

External pressure (Pw)

pe1+g
(a) load condition 1 (LC1)

Ground
support
(K)

Earth pressure

(b) load condition 2 (LC2)

qe2
pe1+g

(c) load condition 3 (LC3)

Fig. 3. Load modes.

where R, E, I, and B (commonly B = unit length) are mean radius,

Youngs modulus, second moment of inertia, and width, respectively. Considering the effect of Poisson ratio, the buckling formula
for an innite long pipe can be obtained as follows:

Winkler spring model for ground to resist both inward and outward deformation; the critical hoop force (Ncr) resulting in the
instability can be formulated by

 3
E
t
Pcr
41  v 2 R

Ncr  2
2

where t is thickness and v is Poisson ratio.

Glock (1977) presented a simple solution to evaluate the buckling strength of a pipe encased in rigid cavity, i.e.,

PGcr

 2:2
E
t
1  v 2 2R

Ncr


1=2
EIK
BR

Sea

H0

K
B

A
A

Shield Tunnel
(RC segment)
B

Do=4950mm B

Sandy
3

s=18 kN/m

v=0.495
E=50 MN/m2
Fig. 4. Prole of tunnel and ground conditions.

On the other hand, Cheney (1976) and Moore and Booker

(1985) proposed an elastic continuum theory for multi-wave buckling, using the continuum model to simulate the surrounding
ground. When the interface between structure and ground is
smooth, the critical hoop force can be formulated by

 1=3 
2=3
EI
Es
Ncr 1:2
2
B
1  ms

Hw

where K is modulus of subgrade reaction.

The single wave buckling of a ring in Winkler medium was
studied by Falter (1980) and Gumbel (1983); and the corresponding solution was expressed by Eq. (5), in which only the effects of
ground resistance to outward deformation is considered,

The works of introducing imperfection were done by Boot

(1998), El-Sawy and Moore (1997), and El-Sawy (2013).
However, for a circular tube embedded in soil medium, the theories of buried structure buckling should be applied. Stevens
(1952), Flgge (1962), Luscher (1966) and Chelapati and Allgood
(1972) presented a multi-wave buckling solution, using the


1=2
EIK
BR

where Es and vs are Youngs modulus and Poisson ratio of the

ground, respectively. Both inward and outward deformation are
assumed to be resisted by ground. However, the linear buckling
solution using the elastic continuum model is applicable only when
hoop force is generated by ground rather than by hydrostatic pressure (Moore, 1989). In addition, all theories of buried buckling can
only be applied to the thin-walled tube.
The relation between modulus of subgrade reaction K and elastic modulus of ground Es can be expressed by an empirical equation (Wood, 1975; Kimura and Koizumi, 1999)

3Es
1 v 5  6v R

As for the studies on buckling of tunnel linings, Croll (2001) proposed several models on buckling failure and indicted that both
linings pull-off and bending failure resulted in the degeneration
of tunnel rigidity. Tamura and Hayashi (2005) identied the buckling failure phenomenon by replacing the tunnel with some thinwalled aluminum pipes and also the ground with some aluminum
rods. Finally, they proposed a two-dimensional numerical solution
by a frame model of elastic beams and rotational springs. Blom
(2002) studied the local instability of segmental linings and
demonstrated the snap-through problem when linings was subjected to the uniform tangential compressive load and ovalisation
load. Meanwhile, the analytical solution using Eulers theory was

147

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155
Table 1
Material properties of RC segment.
Concrete

Reinforcement (SD345)

Youngs modulus
Ec (N/mm2)

Poisson ratio

Compressive strength

Allowable stress

rc (N/mm2)

rca (N/mm2)

Youngs modulus
Es (N/mm2)

Yield strength
fy (N/mm2)

rsa (N/mm2)

3.3  104

0.17

42.0

16

2.1  105

0.17

345

200

Table 2
Earth pressures.
Items

Ground 1
(basic)

Ground
2

Ground
3

Overburden thickness H0 (m)

Vertical earth pressure pe1 (kN/
m2)
Lateral earth pressure qe1 (kN/m2)
Lateral earth pressure qe2 (kN/m2)
Sea depth Hw (m)
Self-weight g (kN/m2)

10
80.00

5
40.00

15
120.00

40.00
20.00
60.00
60.00
40.00
80.00
Hw0 = 18 m (basic), variable
3.12 (t = 120 mm), 4.16 (t = 160 mm),
5.20 (t = 200 mm)

Notice: soilwater separated method in load calculation, and the coefcient of

lateral earth pressure k = 0.5.

recommended. Although the buckling failure was discussed in the

above investigations, the effect of the radial joints has not been
addressed.
Wang and Koizumi (2010) conducted experiments of a thinwalled cylindrical shell with longitudinal joints to investigate the
buckling of shield tunnel linings under hydrostatic pressure. It
was found that the free cylindrical shell with exible longitudinal
joints buckles in a single-wave mode, while that with rigid joints
buckles in a multi-wave mode, as shown in Fig. 1(a). It was found
that the buckling strength decreases as the exural rigidity of joint
decreases. Furthermore, they veried that the buried cylindrical
shell with exible joints can buckle under external air pressure,
in a similar deformed mode with the free cylindrical shell, as
shown in Fig. 1(b); but its buckling strength is increased by about
26% than the free one. The experiments have claried that the
radial joint and ground support have important inuence on the
buckling strength of shield tunnel linings, although the specimens
used the thin-walled cylindrical shells (R/t > 150) with one radial
joint, rather than a segmental linings (R/t < 20) with multi-joints.
Besides, the inuence of imperfection of inward deformation at

Poisson ratio

Allowable stress

joint have been identied by numerical analysis, and the good

agreement with experiments has been obtained (Wang and
Koizumi, 2010).
2.2. Buckling analysis considering load conditions
Shield tunnel linings in the life cycle will undergo many loading states during construction, while three typical load conditions
should be selected when examining the stability, as shown in
Fig. 2: load condition 1 and 2 corresponding to construction period, and load condition 3 corresponding to operation period. The
load condition 1 (LC1) is the loading state of a tunnel linings just
leaving the shield tail. The buckling strength under LC1 can be
studied by Eq. (2). The load condition 2 (LC2) is the loading state
of the grouting material before complete hardening; and the load
condition 3 (LC3) is the loading state of the support and loads
applied on the tunnel linings during operation. The corresponding
loading models can be dened based on the current design specications (JSCE, 2007), as shown in Fig. 3. Model 1 in Fig. 3 will be
used to simulate LC1; model 2 will be used to simulate LC2 by
considering the earth pressure, because the earth pressure acts
on the segmental linings through the soft grouting material;
and model 3 is used to simulate LC3 by adding the ground support beside the LC2, considering the completion of ground settlement. For LC3, the buckling investigation should consider the
instability of the tunnel linings under earth pressure and hydrostatic pressure.
The conventional load conditions and the corresponding load
models are dened, considering the complexity and uncertainty
of the real load conditions involving the ground and construction.
Moreover, this study mainly focuses on the hydrostatic buckling,
although the effect of the ground support and the earth pressure
on the buckling strength are investigated.
2.3. Numerical analysis model of segmental linings

No.1

No.1

=41.5 No.2

=13.8

No joint (Eq.(2))

J1(n=1)

No.1
No.2
=124.6

No.3

J3(n=3)

J2(n=2)

No.1
No.2

No.1
No.5

=152.3

No.3
No.4

J4(n=4)
Fig. 5. Number and location of joints.

Shield tunnel linings are assembled by the segments with

straight or staggered joints. The staggered conguration can
enhance the integrity of the tunnel linings, and the effect is considered by a factor to reduce the moment of the radial joints and
increase the moment of the adjacent segments in structural design.
However, each ring can be simply considered as an independent
structure in buckling analysis, because the hydrostatic pressure
mainly causes a circumferential compression. Therefore, the buckling investigation only needs to be focused on a ring of the tunnel
linings. Three-dimensional joint-shell model is employed, considering the locations of radial joints of bolts and treating the segments as the cylindrical shell.

No.2
=69.2

No.3
No.4

J5(n=5)

3. Numerical analysis
A two-dimensional beam-spring model is used to check the
safety of the tunnel linings based on material strength. Extensive
studies are performed to investigate the effects of joints, segment
dimensions and ground conditions on the buckling strength of tunnel linings, based on three-dimensional joint-shell model.

148

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

Table 3
Parameters and cases in numerical analysis.
Parameters and
cases

Tunnel lining
Outer
diameter
D0 (m)

Thickness of
segment
t (m)

Width of
segment
B (m)

Joint effective exural

rigidity factor

Basic
Extensive

4.95

0.16
0.12, 0.20

1.2
0.9, 1.5

Ground

Lateral earth pressure

factor
k

Modulus of subgrade
reaction
K (MN/m3)

Overburden
depth
H (m)

01
01

0.5
0.5

20
0200

10.0
5.0, 15.0

Nodal ties (x,y,z,Rx,Ry)

Nodal ties (x,y,z,Rx,Ry)
& Rotation spring (Rz)

Ry
Rz

Rx
Segment B

Segment A

Shell element

Segment/radial
joint

(b) Joint modeling details (LC1and LC2)

Nodal ties (x,y,z,Rx,Ry)
& Rotation spring (Rz)

B
A

Segment B

Segment A

Shell element
A

Shell element
Ground reaction spring

Ground

(c) Joint modeling details (LC3)

(a) FE model of a ring lining

Fig. 6. FE model and joint modeling details.

Table 4
Rotational rigidity of segment joint kh (kN m/rad).
Segment

Rotational rigidity of segment joint kh (kN m/rad)

t (mm)

B (m)

g = 1.0E6

g = 0.05

g = 0.1

g = 0.2

g = 0.5

g = 0.7

g = 0.9

g = 1.0

120

0.9
1.2
1.5

8.67E+01
1.16E+02
1.44E+02

4.00E+02
5.33E+02
6.66E+02

5.20E+02
6.94E+02
8.67E+02

7.44E+02
9.92E+02
1.24E+03

1.73E+03
2.30E+03
2.88E+03

3.36E+03
4.48E+03
5.60E+03

1.14E+04
1.52E+04
1.89E+04

1.00E+08
1.00E+08
1.00E+08

160

0.9
1.2
1.5

2.06E+02
2.74E+02
3.43E+02

9.48E+02
1.26E+03
1.58E+03

1.23E+03
1.64E+03
2.06E+03

1.76E+03
2.35E+03
2.94E+03

4.10E+03
5.46E+03
6.83E+03

7.96E+03
1.06E+04
1.33E+04

2.69E+04
3.59E+04
4.49E+04

1.00E+08
1.00E+08
1.00E+08

200

0.9
1.2
1.5

4.01E+02
5.35E+02
6.69E+02

2.47E+03
2.47E+03
3.09E+03

3.21E+03
3.21E+03
4.01E+03

4.59E+03
4.59E+03
5.74E+03

1.07E+04
1.07E+04
1.33E+04

2.07E+04
2.07E+04
2.59E+04

7.02E+04
7.02E+04
8.77E+04

1.00E+08
1.00E+08
1.00E+08

3.1. Tunnel condition and study cases

A 4.95 m outer diameter (O.D.) underwater shield tunnel
assembled by staggered conguration is selected, as shown in
Fig. 4. The shield tunnel ring is composed of ve reinforced concrete at segments with two A-segments (aA = 83.1), two B-segments (aB = 83.1) and a K-segment (ak = 27.7). The basic ground
conditions are overburden H0 = 10 m, sea water level Hw = 18 m,
density cs = 18 kN/m3, Youngs modulus Es = 50 MN/m2, and
Poisson ratio v = 0.495. The coefcient of lateral earth pressure
k = 0.5 and modulus of subgrade reaction K = 20 MN/m3 are taken,
based on in-situ ground condition and shield tunnel design specication for the dense sand (JSCE, 2007). The dimensions of the segment are t = 160 mm in thickness and B = 1200 mm in width.
Segmental joints are assumed to have a variable exural rigidity.
The material properties of concrete and reinforcement are shown
in Table 1.

The above three load conditions are used to study the buckling behavior of the tunnel linings. The earth pressures shown
in Table 2 are used to check the stability of the tunnel linings
under variable hydrostatic pressure and calculated based on
the assumption of soilwater separated. Extensive studies are
performed to investigate the effects of joints, segment dimensions and ground conditions on the buckling strength of tunnel
linings. Four joint patterns arranged by adding each joint in
the clockwise direction are used to clarify the interaction of
joints, as shown in Fig. 5. The effect of segment dimension is
investigated in cases of three widths (B = 900, 1200 and
1500 mm) and three thickness (t = 120, 160 and 200 mm).
Additionally, the inuences of earth pressure are studied by
comparing three overburden thickness of H0 = 5, 10 and 15 m,
and that of ground supports using the modulus of subgrade reaction K = 0200 MN/m3. The dimensions of tunnel linings, ground
parameters are summarized in Table 3.

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

Maximum

19.53 (kNm)

Hoop force
Location
Minimum
Hoop force
Location

822.4 (kN)
0 ( )
-15.17 (kNm)
875.2 (kN)
289 ( )

Maximum

882.5 (kN)

Location

180 (

Bending moment

149

Hoop force

(a) Internal forces calculated using equation method for basic ground conditions
Compressive stress

250

Allowable compressive strength of steel bar sa=200 N/mm2

Hoop stress (N/mm2)

200

Steel bar

150

Concrete

100

Basic ground condition

50

ca=16 N/mm2
Allowable compressive strength of concrete

0
0
-50

500

28m 50m

1000

100m

1500

150m

2000

200m

2500

Hw+H0(m)

Tensile stress

Hydrostatic pressure at tunnel crown Pw ( kN/m2)

(b) Maximum and minimum stress in concrete lining

Fig. 7. Stress examining results in the basic design condition under variable hydrostatic pressure.

3.2. Numerical model

Numerical model of the segments uses thick-shell elements,
which are four-node elements with global displacements and rotations of degrees of freedom. All meshes are ensured to be sufciently small and have a good aspect ratio of about 1 in order to
achieve better convergence. Generally, segment joints are modeled
by springs to describe the relation between force and displacement, with each node six degrees of freedom. Considering that
buckling is mainly determined by exural rigidity, the segment
joints are modeled by rotational springs and nodal ties. The nite
element analysis and details of the joint modeling are illustrated
in Fig. 6. The rigidity kh of the rotational spring is estimated by
(Wang and Koizumi, 2010)

kh 

0:30 EI
log g R

so that joints with an effective exural rigidity factor (non-dimensional parameter g) within the range 01 can be investigated. The
calculated results of rotational rigidity are shown in Table 4.
For LC1 and LC2, four nodes in the middle of the model along
the longitudinal and circumferential directions are xed.

Considering the Winkler method has been widely used in current

tunnel design practice and the target load is the dominant groundwater, the Winkler method is applied in LC3, by modeling the
ground support as a number of independent springs to resist the
radial deformation. In this model, the rigidity of spring is taken
as zero for the inward deformation and as K for the outward
deformation.

3.3. Buckling analysis approach

The analysis is performed by a nite element analysis software
package of Marc (MSC, 2005). Analysis can be done by an elastic
buckling analysis as an eigenvalue problem or a nonlinear buckling
analysis by performing eigenvalue analysis in each increment as a
nonlinear problem. For both geometrical (large deformation) and
material (elasto-plastic material) nonlinear problems, the critical
load should be estimated by an auto-incremental analysis based
on arc-length method (Riks, 1979). In this study, although the tunnel linings can be easily considered elastic, the auto-incremental
analysis is applied to consider the effects of joints, earth pressure
and nonlinear ground support. Criselds arc-length method
(Criseld, 1981, 1983) is employed and described by a governing

150

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

3500

LC1

LC2

Q n

LC3

Buckling strength Pcr (kPa)

3000
2500

=0.9

2000
1500

where {DO} and kO are the n  1 step-displacement vector and load

factor, and {P(n)} and Q(n) are residual errors. However, when buckling occurs with the sudden development of nonlinearity, it should
be ensured that the arc-length remains sufciently small prior to
the occurrence of buckling. The buckling strength is estimated
when the tangent rigidity approaches zero.

=1.0

Pcr=2295 kN/m (Eq.(2))

=0.7

=0,0.05, =0.5

i
2
1h 2
n T
n
r  fd g fd g  kn
2

0.1,0.2

1000

4. Results and discussions of numerical analysis

500
0
0

10000

20000

30000

40000

50000

60000

Rotational rigidity of joint K (kNm/rad)

(a) Relation of buckling critical pressure and

rotation rigidities of the joints
3500

LC1

LC2

Numerical results are presented to investigate the safety of tunnel linings based on material strength, and the buckling of tunnel
linings under three load conditions, where the maximum hydrostatic pressure acting on the crown of tunnel is used as the critical
pressure of the bucking strength. The effects of joints, as well as the
segment dimension, on buckling are studied. Meanwhile, ground
conditions including the overburden depth and modulus of subgrade reaction are discussed.

LC3

4.1. Check of safety of tunnel linings

Buckling strength Pcr (kPa)

3000
2

2500

Pcr=2295 kN/m (Eq.(2))

2000
1500
1000
500
0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

Effective flexural rigidity factor of joint

(b) Relation of buckling critical pressure and

flexural rigidity factors of the joints
Fig. 8. Buckling strength versus rigidity of joint for ground conditions
(B = 1200 mm, t = 160 mm, n = 5, K = 20 MN/m3).

equation, Eq. (9), and increment control equation, Eq. (10). The former governs the forcedisplacement relation, while the latter
implicitly denes the load increment size.

Kfdg  kff g fOg

fDdg fDdg Dk2 Dr 2

10

where k is the incremental load factor, K the tangent-stiffness

matrix, f the load vector, d the displacement vector, and r the arclength control parameter. After written into the load increment
matrix form by adding step n, Eq. (10) becomes:

"

K n DO d  ff g
n T

fd g

kn

#(

Dd
Dkn

Pn

Q n

where
n

fd g

n1
X
i
f Dd g
i1

kn

n1
X
fDki g
i1



n
n
fP n g kO kn ff g  KDO d fDO d g

11

The internal forces of tunnel linings under design loads are

examined using the allowable stress method. Here, the stress
changes with the variable hydrostatic pressure. The internal forces
are calculated by the equations in the structural design guideline
(JSCE, 2007).
For the basic design condition, the bending moment and the
hoop force are shown in Fig. 7(a). The relations between stress
and hydrostatic pressure are plotted in Fig. 7(b), in which the stress
examination for design condition is also presented.
Fig. 7 indicates that stresses in both concrete and reinforcement
steel bars are far smaller than the allowable strength. Furthermore,
the entire tunnel linings in compression can be observed under the
design condition, as well as the compressive stress increasing with
the hydrostatic pressure. The design results verify that the linings
is safe in material strength, and the hoop compressive force is more
predominant than the bending moment. Therefore, considering
that the buckling is mainly induced by compressive hoop force,
the structural stability of tunnel linings should be checked under
the external pressure.
4.2. Effect of load conditions
The effects of load conditions on buckling of the shield tunnel
linings are veried, by comparing the results of three load conditions. The relation between the buckling strength and rotational
rigidity kh and effective exural rigidity factor g of the joint shown
in Fig. 8. The buckling mode congurations are plotted in Fig. 9,
with the effective exural rigidity factor g of 0, 0.7, and 1.0, corresponding to the hinge joint, the general bolt joint, and non-joint.
Fig. 8 indicates that tunnel linings most easily buckle under LC2.
By comparing LC2 and LC1, it can be found that the rigidity of joint
signicantly affects the buckling strength and the earth pressure
decreases the buckling strength of tunnel linings. This can be
explained by the initial imperfections such ovalisation, induced
by earth pressure. On the other hand, LC3 corresponds to the highest buckling strength, although earth pressure is taken into
account. In addition, the variation of exural rigidity of the joint
has a slight effect on the buckling strength when the effective exural rigidity factor g is larger than 0.2. Accordingly, the surrounding ground support dramatically enhances the linings stability
because the tunnel linings supported by hard ground (K = 20 MN/
m3) becomes a statically-indeterminate structure.

151

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

= 0.7

Non joint ( = 1.0)

LC3

LC2

LC1

=0

Fig. 9. Buckling deformation (t = 160 mm, B = 1.2 m, K = 20 MN/m3).

On comparing Fig. 8(a) and (b), the relation between buckling

strength and rotational rigidity kh is more complex than that of
the effective exural rigidity factor g. The buckling strength drops
dramatically when kh varies from 3.59  104 to 1.06  104 kN m/
rad, corresponding to g = 0.9 and 0.7, respectively. It can be found
that the use of an effective exural rigidity factor is a valid means
to simply and clearly describe the relation between the buckling
strength and the exural rigidity of joints. Therefore, the above
numerical analysis is reasonable since the numerical results under
LC1 agree well with the calculated results of Eq. (2), which was
derived for cylindrical shells without joints (g = 1) under the external pressure.
From Fig. 9, buckling deformation presents a good explanation
for the buckling strength changing with the exural rigidity of
the joints and load conditions. For the buckling of tunnel linings
with hinge joints, the buckling mode is quite different from the
general ovalisation (two-wave) mode of cylindrical shell. The buckled deformation mainly occurs close to joints of K-segment, while a
single-wave buckling mode occurs under all load conditions.
Actually, this phenomenon (shown in Fig. 1) has been veried in
experimental studies (Wang and Koizumi, 2010; Wang et al.,

2014). However, as the joint becomes more rigid, the buckling

mode changes from a single-wave mode to two-wave mode under
LC1 and LC2, but the buckling of linings under LC3 shows a highorder buckling mode having four waves. This can explain why
the critical load of the linings under LC3 is always larger than that
under LC1 and LC2.
The above results show that shield tunnel linings is most likely
to buckle during construction when the hydrostatic/grouting pressure and earth pressure (LC2) act on the tunnel linings.
4.3. Effect of joints on buckling
The interactional effect of joints on buckling should be claried,
since the shield tunnel linings is assembled by segments and joints.
The relations between buckling strength and joint number are
plotted in Fig. 10 in the conditions of different effective exural
rigidity.
From Fig. 10, it is shown that the buckling strength of tunnel
linings with rigid joints (g = 1) has no variation, while the buckling
strength of other linings declines as the number of joints increases
under both LC1 and LC2. The decreasing of buckling strength due to

152

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

3000

=0

=0.1

=0.2

=0.5

=0.7

=0.9

=1.0

Eq.(2)

Pcr=2295 kN/m (Eq.(2))

2287

2000

1500

1111
1000

463.2

500

4.4. Effect of segment dimension

463.2

362.8
121.2

0
1

Number of joints (n)

(a) LC1
3000

Buckling strength Pcr (kPa)

2500

=0

=0.1

=0.2

=0.5

=0.7

=0.9

=1.0

Eq.(2)

Pcr=2295 kN/m (Eq.(2))

2000

1500

In elastic buckling of a ring without joint, the dimension and

Youngs modulus can determine the buckling strength. As for
shield tunnel linings, the effects of thickness and width of segment
on the buckling strength should be investigated. The buckling
strengths of designed linings under LC1 and LC2 are plotted in
Figs. 11 and 12, respectively, considering variation of exural rigidity of joint. Furthermore, to clarify the relation between buckling
strength and thickness described in Eq. (2), the buckling strengths
of tunnel linings with thickness t1 = 120 and t2 = 200 mm are calculated by equations Pcr1 = Pcr2(t1/t2)3 and Pcr3 = Pcr2(t3/t2)3 and plotted in Fig. 11. Here, Pcr2 is the buckling strength for t2 = 160 mm.
5000

1265

LC1_t120
Pcr1_Cal.

LC1_t160
Pcr2_Base

LC1_t200
Pcr2_Cal.

1000

629.6
500

287.2

274.7
114.1

0
1

0
5

Number of joints (n)

(b) LC2

Buckling strength Pcr (kPa)

Buckling strength Pcr (kPa)

2500

is smaller due to the second joint, while the reduction of the buckling strength becomes larger with the joint number increasing. The
reason lies in the buckling mechanism that tunnel linings buckles
locally near the K-segment for hinge joints, but buckles in a twowave mode for moderately-rigid joint. Generally, the effect of the
joints under LC1 and LC2 have the same tendency. Therefore, considering the shield tunnel linings are assembled by segments and
joints, the analysis of buckling strength must take the interaction
effects of the number, the location, and exural rigidity of joints
into consideration.

Pcr=4483 kN/m (Eq.(2))

4000

3000
2

Pcr=2295 kN/m (Eq.(2))

2000
2

Pcr=968 kN/m (Eq.(2))

1000

Fig. 10. Buckling strength versus effective exural rigidity factor with the number
of variations of joint.

0
0.0

0.4

0.6

0.8

1.0

1.2

Effective flexural rigidity factor of joint

(a) LC1
3000

Buckling strength Pcr (kPa)

the increasing of joint number vary with the exural rigidity of the
joint. The more exible the joint is, the smaller buckling strength
is. In addition, it is found that the joints near the crown, spring-line
and invert of the tunnel dramatically reduce the buckling strength,
compared with exible joints, and the buckling strength decreases
with the joint number nonlinearly. As shown in Fig. 10(a), the
buckling strength of tunnel linings with hinge joints under LC1
decreases from 2287 kPa to 1111 kPa and 463.2 kPa when the
number of joints of K-segment increases from zero to one and
two, from 463.2 kPa to 362.8 kPa when the number of joints of
K-segment increases from three to four and the fourth joint is
located in the invert, and from 362.8 kPa to 121.2 kPa when the
number of joints of K-segment increases from four to ve and
the fth joint is located in the spring-line. Similarly, for the tunnel
linings with hinge joints under LC2, Fig. 10(b) shows that the buckling strength decreases from 1265 kPa to 629.6 kPa and 287.2 kPa
when the number of joints of K-segment increases from zero to
one and two, from 274.7 kPa to 114.1 kPa when the number of
joints of K-segment increases from three to four, and from
114.1 kPa to 0 when the number of joints of K-segment increases
from four to ve. However, if the joint is moderately rigid, the
buckling strength decreases with the increase of the joint number
rather than the joint location. For instance, in the case of a joint
with exural rigidity g = 0.7, the reduction of the buckling strength

0.2

LC2_t120
Pcr1_Cal.

LC2_t160
Pcr2_Base

LC2_t200
Pcr3_Cal.

2400

1800

1200

600

0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

Effective flexural rigidity factor of joint

(b) LC2
Fig. 11. Buckling strength versus effective exural rigidity factor with the variations of the segment thickness (B = 1200 mm, H0 = 10 m).

153

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

Load condition 1
Load condition 2

LC1_B900
LC2_B900

LC1_B1200
LC2_B1200

LC1_B1500
LC2_B1500

2000

Buckling strength Pcr (kPa)

Buckling strength Pcr (kPa)

2500

1500

1000

500

4500

=0

=0.2

=0.5

4000

=0.7

=0.9

=1.0

3500
3000
2500
2000
1500
1000
500
0

0
0.0

0.2

0.4

0.6

0.8

1.0

0.1

1.2

10

100

1000

Modulus of subgrade reaction K (MN/m3)

Effective flexural rigidity factor of joint

Fig. 12. Buckling strength versus effective exural rigidity factor with the variations of the segment width (H0 = 10 m, t = 160 mm).

Fig. 14. Buckling strength versus modulus of subgrade reaction with the variations
of g (t = 160 mm, B = 1200 mm, H0 = 10 m, LC3).

Fig. 11 shows that the segment thickness substantially affects

the buckling resistance and that the buckling strength increases
as the thickness of segment increases under LC1 and LC2. In addition, the calculated buckling strengths of tunnel linings with thickness t1 = 120 mm and t3 = 200 mm agree well with the numerical
results under LC1, while they are different from the numerical
results under LC2. From Fig. 11(b), it is shown that the numerical
results become smaller for thin segment and larger for thick segment, comparing with the calculated results under LC2. This
implies that the imperfection induced by the earth pressure
increases as the thickness of segment decreases, leading to
decreasing of buckling strength. However, it can be found in
Fig. 12 that the width of the segment has only a slight effect on
the buckling strength of tunnel linings under LC1 and LC2.
Generally, the relation between buckling strength and thickness
can be described by Eq. (2), as well as the width of the segment, if
the tunnel linings is mainly subjected to the hydrostatic pressure.

overburden depth under LC2 and the modulus of subgrade reaction

under LC3, respectively.
Fig. 13 shows that the earth pressure induced by overburden
can reduce the buckling strength of tunnel linings. This can be
explained by the fact that imperfections in the tunnel linings
become numerous as the earth pressure increases. On the other
hand, from Fig. 14 it can be found that the buckling strength
increases as the modulus of subgrade reaction becomes larger.
The effect of ground support on buckling strength changes with
the joint rigidity and modulus of subgrade reaction when the
ground reaction K is less than 50 MN/m3. If the ground is moderately rigid with modulus of subgrade reaction above 50 MN/m3,
the exural rigidity of joint has a slight effect on the buckling
strength.

4.5. Effect of ground condition

The effect of the ground condition on the buckling strength is
investigated. Figs. 13 and 14 show the buckling loads change with

5. Theoretical verication of buckling of tunnel linings

The buckling of tunnel linings under design condition is investigated by the buried tube buckling theories. Buckling strength is
calculated by Eqs. (4)(6), corresponding to the linear multi-wave
and single-wave theories based on the Winkler model, and the
multi-wave theory based on the elastic continuum model, respectively. The elastic modulus E is evaluated from the assumed

106

1800
Load condition 2

H0=5

H0=10

H0=15

Max. hoop force N (Hw=18 m)

Max. hoop force N (Hw=190 m)
Eq.(4)_Winkler_M.W
Eq.(5)_Winkle_S.W.
Eq.(6)_Continuum_M.W.

Critical hoop force Ncr (kN)

Buckling strength Pcr (kPa)

1500

1200

900

600

105

104

103

300

102

0
0.0

0.2

0.4

0.6

0.8

1.0

1.2

Effective flexural rigidity factor of joint

Fig. 13. Buckling strength versus effective exural rigidity factor with the variations of the overburden H0 (t = 160 mm, B = 1200 mm, LC2).

0.1

10

100

1000

Modulus of subgrade reaction K (MN/m3)

Fig. 15. Critical hoop force versus modulus of subgrade reaction based on the
buried tube buckling theories (t = 160 mm, B = 1200 mm, H0 = 10 m, Hw = 18 m).

154

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155

subgrade reaction factor by Eq. (7). Meanwhile, the maximum

design hoop forces in various modulus of subgrade reaction are calculated by the equation in design specication of tunnel linings.
The calculated results are presented in Fig.15.
In Fig. 15, it can be found that the buckling under effective
ground stress does not occur, except that the modulus of subgrade
reaction is smaller than 0.1 MN/m3, which means the ground has
Youngs modulus E less than 250 KN/m2. Such ground is quite soft
so that it is not proper to construct a tunnel. Besides, the critical
hoop force in elastic continuum theory is larger than that calculated by the single-wave Winkler model, while it is smaller than
that of multi-wave Winkler model for soft ground and it is inversed
for hard ground. The comparison results between the above theories are consistent with the analytical results in Gumbel (1983) and
Moore (1989). Moreover, the buckling will not occur for the
groundwater level Hw = 190 m. From the above discussion, it can
be concluded that the operating tunnel linings will not buckle
under earth pressure and hydrostatic pressure, even though the
maximum hoop forces are over-evaluated by considering all
weight of overburden.

However, since the buckling of shield tunnels is an important

but rare structural problem, external experimental studies should
be conducted, especially when considering that the buckling of
tunnel linings always results in a disaster during construction of
shield tunnel. Validation by experiments should provide more
insight in this phenomenon of tunnel engineering. The publication
of this paper will be helpful in stimulating research interest on the
buckling of this concrete linings of shield tunnels.

6. Conclusions and recommendations

Amstutz, E., 1970. Buckling of Pressure Shaft and Tunnel Linings. Water and Power
(November), 391399 (based on the original work published in German in
1950).
Blom, C.B.M., 2002. Design Philosophy of Concrete Linings for Tunnels in Soft Soils.
Delft University Press.
Boot, J.C., 1998. Elastic buckling of cylindrical pipe linings with small imperfections
subject to external pressure. Trenchless Technol. Res. 12 (12), 315.
Chelapati, C.V., Allgood, J.R., 1972. Buckling of cylinders in a conning medium.
Highway Res. Rec. 413, 7788.
Cheney, J.A., 1971. Buckling of soil-surrounded tubes. J. Eng. Mech. Div. 97 (4),
11211132.
Cheney, J.A., 1976. Buckling of Thin-walled Cylindrical Shells in Soil. Supplementary
Report 204, Transp. Res. Lab., Crowthorne, Berkshire, England.
Criseld, M.A., 1981. A fast incremental-iterative solution procedure that handles
snap-through. Comput. Struct. 13, 5562.
Criseld, M.A., 1983. An arc-length method including line searches and
accelerations. Int. J. Numer. Meth. Eng. 19 (9), 12691289.
Croll, J.G.A., 2001. Buckling of cylindrical tunnel liners. J. Eng. Mech. 127 (4), 333
341.
El-Sawy, K., 2001. Inelastic stability of tightly tted cylindrical liners subjected to
external uniform pressure. Thin-walled Struct. 39 (9), 731744.
El-Sawy, K., 2013. Inelastic stability of liners of cylindrical conduits with local
imperfection under external pressure. Tunn. Undergr. Space Technol. 33, 98
110.
El-Sawy, K., Moore, I.D., 1997. Parametric study for buckling of liners: effect of liner
geometry and imperfections. Trenchless Pipeline Projects Practical Applications.
ASCE, Boston, Massachusetts, June 1518, pp. 416423.
El-Sawy, K., Moore, I.D., 1998. Stability of loosely tted liners used to rehabilitate
rigid pipes. J. Struct. Eng. 124 (11), 13501357.
Falter, B., 1980. Grenzlasten von einseitig elastisch gebetteten kreiszylindrischen
Konstruktionen (Ultimate loads of elastically bedded circular cylindrical
constructions bedded at the outside). Bauingenieur 55, 381390.
Flgge, W., 1962. Stress in Shells. Springer, Berlin (corrected reprint 1962).
Glock, D., 1977. berkritisches verhalten eines starr ummantelten kreisrohres bei
wasserdruck von aussen und temperaturerhhung (Post-critical Behavior of a
rigidly encased circular pipe subject to external water pressure and
temperature rise). Der Stahlbau 46 (7), 212217.
Gumbel, J.E., 1983. Analysis and Design of Buried Flexible Pipes. PhD Thesis. Univ. of
Surrey, UK.
Health and Safety Executive (HSE), 2000. Rep. the Collapse of NATM Tunnel at
Heathrow Airport. Health and Safety Executive, London, U.K.
Jacobsen, S., 1974. Buckling of circular rings and cylindrical tubes under external
pressure. Water Power 26, 400407.
Japanese City and Regional Development Bureau (JCRDB), 2006. The Utilization of
Deep Underground Space.
Japan society of civil engineer (JSCE), 2007. Standard Specications for Design and
Construction Tunnel: Shield Tunnel, 2006 ed. Tunnel Engineering Committee,
JSCE, Tokyo.
Kimura, S., Koizumi, A., 1999. A design method of shield tunnel linings taking into
account of the interaction between the linings and the ground. Proc. JSCE 624,
123134 (In Japanese).
Koizumi, A., 2000. New Technology of Segment. Dobokukogakusya Ltd, Tokyo,
Japan, ISBN 4-88624-083-6. (In Japanese).
Koyama, Y., 2003. Present status and technology of shield tunneling method in
Japan. Tunneling Undergr. Space Technol. 18, 145159.
Levy, M., 1884, Mememoire sur un nouveau cas integrable du problem de
lelastique et lune de ses applications (Memoir on a New Integrable Case of the
Problem of Elasticity and One of its Applications). J. Math. Pure et Appl.
(Lioville), 10(3), 542.

Buckling of tunnel linings under external pressure has been

investigated by analytical and numerical analysis. The buckling
behavior under different load conditions has been claried, as well
as the effects of joint, dimensions of segment, and the ground condition. The following conclusions can be drawn based on the above
analysis:
(1) Buckling of tunnel linings can be affected by the thickness of
the segment, the joint and ground condition. Buckling occurs
in the vicinity of the K-segment joint under low hydrostatic
pressure in a single-wave mode, when the tunnel linings is
assembled by thin segments and exible joints. The buckling
failure will occur in the tunnel linings during construction,
particularly for the segmental rings just leaving the shield
tail.
(2) The buckling of tunnel linings occurs more easily as the segmental joints becomes moderately exible. Segment width
has a slight effect while the thickness can dramatically
reduce the buckling resistance. The ground condition substantially affects the buckling strength of tunnel linings:
the softer ground is, the larger earth pressure acting on the
tunnel linings, the tunnel linings buckle easily.
Conclusively, radial joint as the structural imperfection of circular tube, greatly affects the buckling behavior; the effect of ground
condition can be considered due to the ovalisation deection
induced by earth pressure and subgrade reaction.
As recommendations, the following issues should be paid more
attention to during the design and construction of a shield tunnel:
(1) For the design of a shield tunnel subjected to high hydrostatic pressure, the buckling failure should be checked rather
than only the material failure, especially for the case of the
loading state of the grouting material before complete hardening during construction stage.
(2) For the analysis of buckling, numerical solution should take
the effects of joints, segment dimensions, and ground conditions into account.
(3) The use of exible segment joint should be avoided when a
tunnel is subjected to high hydrostatic pressure, and the
application of circular shape retainer is recommended, as
well as the rapidly-hardening grouting material during tunnel construction.

Acknowledgments
Dr. J.H. Wang acknowledges the support from Deep Tunnel
Technical Research Committee (Japan). Dr. W.J. Zhang acknowledges the supports from National Natural Science Foundation of
China (Grant no. 51378342) and Ph.D. Program Foundation of the
Ministry of Education of China (Grant no. 20120032120050). Dr.
X. Guo acknowledges the support from National Natural Science
Foundation of China (Grant nos. 11102128 and 11372214).
References

J.H. Wang et al. / Tunnelling and Underground Space Technology 49 (2015) 144155
Luscher, U., 1966. Buckling of soil-surrounded tubes. J. Soil Mech. Found. Div., ASCE
92 (SM6), 211228.
Mashino, H., Ishimura, T., 2003. Evaluation of the load on shield tunnel linings in
gravel. Tunneling Undergr. Space Technol. 18, 233241.
Moore, I.D., 1989. Elastic buckling of buried exible tubes a review of theory and
experiment. J. Geotech. Eng. 115 (3), 340358.
Moore, I.D., Booker, J.R., 1985. Simplied theory for the behavior of buried exible
cylinders under the inuence of uniform hoop compression. Int. J. Solids Struct.
21 (9), 929941.
MSC. Marc, 2005. Theory and User Information, MSC. Marc2005. Marc Analysis
Research Corp., vol. A, 53031.
Wood, A.M. Muir, 1975. The circular tunnel in elastic ground. Gotechnique 25 (1),
115127.
Nikkei Business Publications (NBP), 2012. Shield Machine Submerged Due to
Segmental Linings Collapse. NIKKEI Construction, 27, pp. 1823 (in Japanese).
Watanabe, Y., 1990. Deep underground space the new frontier. Tunneling
Undergr. Space Technol. 5 (1/2), 912.
Riks, E., 1979. An incremental approach to the solution of snapping and buckling
problems. Int. J. Solids Struct. 15 (7), 529551.

155

Stevens, G.W.H., 1952. The stability of a compressed elastic ring and of a exible
heavy structure spread by a system of elastic rings. Quart. J. Mech. Appl. Math. 5
(2), 221236.
Tamura, T., Hayashi, Y., 2005. Buckling analysis of tunnel linings considering
interaction with ground. J. JSCE 792 (III-71), 199210.
Timoshenko, S.P., Gere, J.M., 1961. Theory of Elastic Stability, second ed. McGraw
Hill, New York.
Wang, J.H., Koizumi, A., 2010. Buckling of cylindrical shells with longitudinal joints
under external pressure. Thin-Walled Struct. 48 (12), 897904.
Wang, J.H., Koizumi, A, Tanaka H., Liu C., Zhong X., 2014. Structural strength
investigation for concrete shield tunnel linings in construction material
strength vs. structural stability. In: Second International Conference on
Advances in Civil, Structural and Environmental Engineering-ACSEE 2014.
Wikipedia, 2005. Gerrards Cross Tunnel Collapse <http://en.wikipedia.org/wiki/
Gerrards_Cross_Tunnel>.
Yahagi, S, Fujiki, I., Oishi, K, Saitou, M., Arai T., 2005. Long term in-situ
measurement of railroad shield tunnel in diluvial deposits. In: Proceeding of
the 3rd Japan-China Technological Exchange of Shield-driven Tunneling, Tokyo.
ISBN 4-9902645-0-9 (in Japanese).