Вы находитесь на странице: 1из 6

HOME

Analysis of the Creep Behaviour of Tunnels in Sandstone/Shale


R Chen1 and J C Small2
ABSTRACT
Time-dependent deformation occurs when geo-materials are under an
applied load; this may lead to additional deflection of a tunnel structure
and the stress redistribution in the surrounding rock, shotcrete lining and
rock bolt support system over time. A better understanding of this creep
behaviour can improve the reliability of design, construction and
maintenance of tunnels. Hence, the present investigation was carried out
on the time-dependent properties of rock, shotcrete and the combined
tunnel support system.
Tunnels constructed in different ground conditions were modelled. The
closure changes of the tunnel with time and the redistribution of stress in
the surrounding rock, shotcrete lining and rock bolts due to the creep
deformation were analysed. Results of the present analysis were
compared with the results from the commercial program ABAQUS.

INTRODUCTION
Due to the creep behaviour of rock and the supporting shotcrete
lining, a tunnel excavated in a squeezing rock may undergo an
increasing deformation over a long time period and result in final
collapse. The Laerdal tunnel excavated in Norway broke down
nearly four years after its excavation due to the squeezing of the
rock (Grimstad, 2001). The creep performance is complex and
determined by many factors, but for many materials, a higher
stress level and a higher temperature will accelerate the creep
rate. Normally the creep curve is subdivided into three stages:
primary, steady and tertiary state creep, but the tertiary creep
only happens when the stress is relatively high and can lead to
damage of the material (Figure 1; Singh and Verma, 2005).

the lining. Shalabi (2005) used two methods to model the creep
deformation of a circular tunnel but did not consider the rock
bolts and did not focus on the stress change in the materials. In
their models, the shotcrete lining was applied to the whole
circumference of the tunnel section. In the present paper, a tunnel
with a nearly flat roof was modelled and a shotcrete lining was
only applied on the crest. The surrounding rock and lining were
considered to have different creep rates in the model as their
creep behaviour affects the deformation of the whole structure.
Steel reinforced shotcrete linings are widely used in
underground tunnelling projects. In this paper, the Novotex 0730
steel fibre reinforced shotcrete (hereafter referred to as Novotex
fibre) was investigated and employed in the tunnel model. The
creep properties of the shotcrete were determined by round
determinate panel (ASTM C-1550) tests (Bernard, 2004) and
appropriate values of parameters were chosen to fit the
experimental results. Three tunnels were modelled numerically
and rock with different creep parameters and different elastic
moduli were employed in the models to analyse the timedependent behaviour of these tunnels. The results of the present
program were compared with results obtained from the
commercial finite element software ABAQUS. A logarithmic
function in time was used to represent the bulk and shear
modulus in the present creep model. ABAQUS uses a
time-dependent power law to fit the creep behaviour that is
simple to use, but cannot represent the tertiary stage of creep
(ABAQUS Inc, 2003).

CREEP BEHAVIOUR
In the classical theory of elasticity, strain is linearly related to
the stress by Hookes Law, in which Youngs modulus E and
Poissons ratio are constants. Under simple one-dimensional
conditions, we may write:
=E

(1)

For this simple case, if the material were to creep we could


write:
( t ) = 0 R( t )

(2)

where:
(t) is the stress at any time
0
FIG 1 - Creep curve (Singh and Verma, 2005).

The surrounding rock and shotcrete lining of an excavated


tunnel will generally have different creep rates, and how the two
different materials affect each other and the deformation and
stress change of the whole tunnel system has been studied by
many researchers before. Sahli et al (2001) modelled the creep
deformation of a circular tunnel but did not consider the creep of

is the strain applied at time t = 0

R( t ) is a relaxation function
If the strain is changing with time, then we must evaluate a
convolution integral:
t

( t ) = 0 R( t ) + R( t - )( )d
0

(3)

This relationship can be simplified by applying a Laplace


transformation to it:

1.

School of Civil Engineering, Room 360, University of Sydney, NSW


2006. Email: r.chen@civil.usyd.edu.au

= s R

2.

Professor, Pro-HOD (Postgraduate and Research), School of Civil


Engineering, Room 413, University of Sydney, NSW 2006.
Email: j.small@civil.usyd.edu.au

where the superior bar denotes a Laplace transform and s is the


Laplace parameter.

13th Australian Tunnelling Conference

Melbourne, VIC, 4 - 7 May 2008

(4)

197

R CHEN and J C SMALL

cr

A similar approach may be used for a continuum where the


stress-strain relationship takes the form:
( t ) = D( t ) 0

= Fq~ n t m

where:

(5)

cr

The time-dependent components of the stresses (t) are related


to a constant set of strains 0 by the time-dependent matrix D( t ).
For general three-dimensional stress states, there will be six
stress components and so the D matrix is a 6 6 matrix.
We may now apply a Laplace transform as in Equation 2 if the
strains are time dependent:
= s D

If we make the assumption that the bulk modulus K and shear


modulus G vary with time (or relax) we can write:

(7)

Here the bulk modulus K and shear modulus G of the materials


are represented by logarithmic functions in time:
K ( t ) = K 0 ( A + B log(1 + t ))
G( t ) = G0 ( A + B log(1 + t ))

(8)

where:
A, B,

~
q

is the uniaxial equivalent creep strain rate

is the total time

F, m, n

are the creep parameters and can be determined


from experimental results (ABAQUS Inc, 2003)

is the uniaxial equivalent deviatoric stress

Laboratory round determinate panel (ASTM C-1550) creep


tests were carried out to determine the creep behaviour of
Novotex fibre reinforced shotcrete samples (Bernard, 2004). Two
constant loads of 8100 kN (test Novo2-1) and 7120 kN (test
Novo2-2) were applied separately at the centre of the circular
panels and the deflection changes of the samples were recorded
over a few months or even longer. A set of creep parameters
F = 4.0e-37, n = 5, m = -0.1 (the parameters depend on the units
used; here days, metres and Newtons) in the time-hardening
power creep law model was then selected to fit the experimental
data as shown in Figure 2, and later applied to the tunnel model.
A uniaxial creep test with specimens of dimension 54 54
108 mm3 was modelled to determine the creep parameters of
sandstone samples. Pellet (2000) did several uniaxial tests on
sandstones samples, and a set of creep parameters F = 7.0e-20,
n = 2.19, m = -0.98 in the power creep law model can give a
good fit to the creep deflection of Pellets (2000) records
(Figure 3). A sandstone with the same creep parameters but
much smaller initial elastic modulus was also selected in the
tunnel model. Another set of creep parameters F = 3.0e-19,
n = 2.2, m = -0.9 was used to represent a sandstone which creeps
faster. From Figure 4 we can see that the rock under pressure
produces some instantaneous elastic strain which is dependent on
this elastic modulus. The creep strain is dependent on the chosen
creep parameters, which are related to many factors such as
stress level and temperature.

(6)

K + 4G / 3 K 2G / 3 K 2G / 3 0 0 0

K 2G / 3 K + 4G / 3 K 2G / 3 0 0 0
K 2G / 3 K 2G / 3 K + 4G / 3 0 0 0
D=

0
0
0
G 0 0

0
0
0
0 G 0

0
0
0
0 0 G

(9)

are the creep parameters

The transformed K and G are then put into the finite element D
matrix of Equation 7 to form the stiffness matrix for the problem.
The resultant equations can then be solved with a threedimensional finite element analysis based on the transformed
variables. Tablots inversion is finally used to invert the
numerical solution to get the result in real time.
Another creep model used was the time-hardening power
creep law model, which is implemented in the commercial
computer code ABAQUS and can be directly used in the
analysis. This power law model may be written:

NUMERICAL SIMULATION OF A TUNNEL


A typical section of the tunnel, as shown in Figure 5, was
modelled by the three-dimensional finite element program
developed here called SAFEA (semi-analytical FE analysis) to
analyse the excavation and creep behaviour. Fits of the creep
data with the logarithmic model were made and are shown in

0.8

Creep Deflection (mm)

Novo2-1
0.6

0.4

Novo2-2
Novo2-1 Load=8100 F=4e-37,n=5,m=-0.1
(A BA QUS )
Novo2-2 Load=7120 F=4e-37,n=5,m=-0.1
(A BA QUS )

0.2

0
0.0001

0.001

0.01

0.1

10

100

100

Time (Days)
FIG 2 - Creep deflection of Novotex shotcrete sample.

198

Melbourne, VIC, 4 - 7 May 2008

13th Australian Tunnelling Conference

ANALYSIS OF THE CREEP BEHAVIOUR OF TUNNELS IN SANDSTONE/SHALE

Sandstone Creep Parameter


Time (day)

0
0.01
-0.02

0.1

10

100

Deflection (mm)

-0.04
Sandstone Load = 5.2 MPa F.Pellet 2000

-0.06
Load=5.2 MPa F = 7.0e-20,n = 2.19, m = -0.98
(ABAQUS)

-0.08
-0.1
-0.12
-0.14
-0.16

FIG 3 - Creep deflection of sandstone.

Sandstone Creep Parameter


0.01

0.1

10

100

-0.02

F = 7.0e-20, n = 2.19, m = -0.98


1.2 MPa (ABAQUS)
1.2 MPa A = 1 B = -0.0066
Alpha = 100 (SAFEA)

-0.025
-0.03
Deflection (mm)

1000

F = 3.0e-19, n = 2.2, m = -0.9


1.2 MPa (ABAQUS)

-0.035
-0.04

1.2 MPa A = 1 B = -0.027


Alpha = 100 (SAFEA)

-0.045
-0.05

E = 2.227e9 F = 7.0e-20
n = 2.19, m = -0.98
1.2 MPa (ABAQUS)
E = 2.229e9 1.2 MPa A = 1
B = -0.003 Alpha = 100
(SAFEA)

-0.055
-0.06
-0.065
-0.07
Time (day)

FIG 4 - Creep deflections of three types of sandstone of block samples.

FIG 5 - Typical section of tunnel (Adams, Lechner and


Lamb, 2001).

13th Australian Tunnelling Conference

Figure 4. The parameters A, B used to achieve the fits are given


in Figure 4 and also presented in Table 1, along with values used
to fit the creep behaviour of the shotcrete. ABAQUS was also
used to simulate the same tunnel section. The tunnel was
symmetrical so only half of the tunnel was analysed. The rock
was represented by a 20-noded solid element, the shotcrete lining
was modelled by an eight-noded shell element and the rock bolts
were modelled with two-noded beam elements. The shell
elements and beam elements share nodes with the solid elements,
thus the shotcrete lining was modelled as tied to the exposed
tunnel surface and the rock bolts were modelled as embedded
into the rock elements. The dimensions and mesh for the tunnels
are shown in Figure 6 and the properties of the rock, shotcrete
lining and rock bolts are given in Table 1. The same shotcrete
lining and rock bolts were used in each of the models but
different sandstone models were applied in the analysis of the
tunnel behaviour under different kinds of underground
conditions. In Table 1, M5 represents a tunnel in rock which
creeps according to Pellets (2000) records; M5C represents a
tunnel in rock which creeps faster and M5D represents a tunnel
in soft rock which has the same creep parameters as that of case
M5 but a much smaller initial elastic modulus.

Melbourne, VIC, 4 - 7 May 2008

199

R CHEN and J C SMALL

TABLE 1
Properties of rock, lining and bolts in tunnel model.

Sandstone

Shotcrete lining
(Novotex fibre)

Rock bolts

M5

M5C

M5D

E = 4.976e9 Pa
= 0.3

E = 4.976e9 Pa
= 0.3

E = 2.227e9 Pa
= 0.3

Creep parameters
F = 7.0e-20
n = 2.19
m = -0.98 (ABAQUS)

Creep parameters
F = 3.0e-19
n = 2.2
m = -0.9 (ABAQUS)

Creep parameters
F = 7.0e-20
n = 2.19
m = -0.98 (ABAQUS)

A=1
B = -0.0066
Alpha = 100 (SAFEA)

A = 1,
B = -0.027
Alpha = 100 (SAFEA)

A=1
B = -0.003
Alpha = 100 (SAFEA)

E = 5.848e9 Pa
= 0.15
Thickness = 0.2 m

E = 5.848e9 Pa
= 0.15
Thickness = 0.2 m

E = 5.848e9 Pa
= 0.15
Thickness = 0.2 m

Creep parameters
F = 4.0e-37
n = 5.0
m = -0.1 (ABAQUS)

Creep parameters
F = 4.0e-37
n = 5.0
m = -0.1 (ABAQUS)

Creep parameters
F = 4.0e-37
n = 5.0
m = -0.1 (ABAQUS)

= 1.8 MPa
A=1
B = -0.266
Alpha = 0.1

=1.8 MPa
A=1
B = -0.266
Alpha = 0.1

= 3.6~0.8 MPa
A=1
B = -0.26* + 0.14
Alpha = 0.1

= 0.5 MPa
A=1
B = -0.032
Alpha = 0.01 (SAFEA)

= 0.5 MPa
A=1
B = -0.032
Alpha = 0.01 (SAFEA)

= 1.1~0.6 MPa
A=1
B = -0.533* + 0.2499
Alpha = 0.01 (SAFEA)

M22 3000 mm
E = 2.0e11 Pa
= 0.3

M22 3000 mm
E = 2.0e11 Pa
= 0.3

M22 3000 mm
E = 2.0e11 Pa
= 0.3

FIG 6 - Dimension and mesh of tunnels (unit: metres).

During the analysis, the initial ground stress was created by


gravity forces and initial deformation was set to zero since the
deformation was already finished. The tunnel was divided into
ten sections longitudinally and the excavation of the tunnel was
stimulated by removing the element in ten steps. During each
step, rock elements were excavated and the corresponding
supporting shotcrete lining elements and bolt elements were
activated in the follow step. The total excavation length in the
analysis was 40 metres so that the tunnel deformation in the
middle section would not be affected much by the face boundary
restraints. The deviator stress caused by excavation around the
tunnels is a maximum in a surrounding region of the tunnel. This
area is set as the creep zone since it is assumed that it is mainly
the change of the deviator stress that will make the materials
creep (Figure 7). After excavation, the deviator stress around the
tunnel and on the shotcrete lining can been obtained. In the
present analysis (using program SAFEA), the elastic modulus of
the sandstone in the creep zone and shotcrete lining will
decrease according to a logarithmic law, while the elastic
modulus of the rock bolts was taken as constant. The following

200

FIG 7 - Creep zone (unit: metres).

creep analysis was based on the deviator stress level and the
creep parameters were selected based on the stress level
(Table 1). Since stress on the linings curved corners is higher
than that on the flat part, two sets of creep parameters for the
lining were chosen to fit the creep in the higher and lower stress
zone. The stress in the lining changes a lot in the soft rock
(M5D), and so the creep parameter for this case also changes
based on stress level. The deviator stresses were applied to the
numerical uniaxial test and the creep parameters of the present
model were chosen to fit the numerical curve of ABAQUS with
the power law creep model.
The displacement diagram (Figure 8) shows the vertical
closure displacements of the tunnel. It can be seen that tunnels
built in soft and hard rock (M5D versus M5) that have the same
creep rates will close at nearly the same rate. Also, the initial
closure in soft rock is bigger than in the hard rock, while a tunnel
in the rock that creeps faster will also close at a faster rate (M5D

Melbourne, VIC, 4 - 7 May 2008

13th Australian Tunnelling Conference

ANALYSIS OF THE CREEP BEHAVIOUR OF TUNNELS IN SANDSTONE/SHALE

Tunnel displacement
10

Displacement (mm)

9
8

M5 (ABAQUS)

M5 (SAFEA)

6
M5C (ABAQUS)

5
4

M5C (SAFEA)

M5D (ABAQUS)

M5D (SAFEA)

1
0
0.01

0.1

10

100

1000

Time (day)

FIG 8 - Creep closures of tunnels.

Lining stress on top shell element (MPa) versus time (day)


0.01

0.1

10

100

1000

0
-0.2
-0.4

Stress (MPa)

-0.6
-0.8
-1
-1.2
-1.4

M5 (ABAQUS)

M5 (SAFEA)

-1.6

M5C (ABAQUS)

M5C (SAFEA)

-1.8

M5D (ABAQUS)

M5D (SAFEA)

-2
Time (day)

FIG 9 - Stress behaviour in lining of tunnels.

versus M5C). Figure 9 shows the change of vertical stress in the


shell element on the crown of the tunnels. It can be seen that in
the long term the stress in the lining will decrease. In soft rock,
the stress in the lining will drop at a faster rate that in the hard
rock since in soft rock, the shotcrete lining will carry more load
and thus the initial stress in the lining will be relatively higher
than for the hard rock and the shotcrete material will creep faster
at a higher stress level. In some cases (Figure 9, M5C), if the
rock creeps at a relatively faster rate, the stress in the lining may
increase in the initial period and then decrease since the stress in
the surrounding rock and lining redistribute because of the
different creep rates of the two materials. Figure 10 shows the
change of vertical stress in the rock at the top of the tunnel. The
stresses in the hard and soft rock (M5D versus M5) are close and
decrease at a similar rate, while the stress in the rock which
creeps at a faster rate (M5C) decreases at a faster rate. Figure 11
shows the tension stress in the rock bolts. The stress in all cases
increases, although initial stresses are different for bolts in soft
and hard rock (M5D versus M5) since in soft rock, the rock bolts
will carry more load, the stress increases at a similar rate in these
two cases. The stress in bolts in rock which creeps faster (M5C)
will increase at a faster rate.

13th Australian Tunnelling Conference

CONCLUSIONS
Conclusions are drawn based on the specific material moduli and
creep parameters selected in the numerical model.
The creep parameters of the Novotex fibre shotcrete material
are sensitive to the stress level.
Tunnels built in different underground situations will undergo
different amounts of internal closure, while a higher initial
closure does not necessarily mean a faster closure rate. The
initial closure displacement, initial lining stress and initial bolt
stress of tunnels built in soft rock are higher than in hard rock,
while the initial rock stress in all three cases is similar. The
closure displacement, bolt stress and rock stress of tunnels built
in rock which creeps faster will also change at a faster rate while
the lining stress of a tunnel built in soft rock decreases faster
since the Novotex fibre shotcrete lining creeps much faster at a
high stress level.
The stress in the shotcrete lining and surrounding rock were
predicted to decrease while the stress in the rock bolts was
predicted to increase. This shows that during the creep process,
the load carried by the lining and surrounding rock will transfer
to the rock bolts.

Melbourne, VIC, 4 - 7 May 2008

201

R CHEN and J C SMALL

Stress on top rock element (MPa) verus time (day)


1.5
1.45
1.4

Stress (MPa)

1.35
1.3
1.25
1.2
M5 (ABAQUS)

M5 (SAFEA)

1.1

M5C (ABAQUS)

M5C (SAFEA)

1.05

M5D (ABAQUS)

M5D (SAFEA)

1.15

1
0.01

0.1

Time (day)

10

100

1000

FIG 10 - Stress behaviour in surrounding rock of tunnels.

Stress on top bolt element (MPa) versus Time (day)


90
80

Stress (MPa)

70
60
50
40
30
20
10
0
0.01

0.1

M5 (ABAQUS)

M5 (SAFEA)

M5C (ABAQUS)

M5C (SAFEA)

M5D (ABAQUS)

M5D (SAFEA)

10

100

1000

Time (day)

FIG 11 - Stress behaviour in rock bolts.

For the Novotex fibre shotcrete, these three analyses show


that the 1000 day creep closure might reach about 1.04 (M5),
1.05 (M5D) or 1.22 (M5C) times the original closure. The creep
deformation is acceptable in these cases.

ACKNOWLEDGEMENTS
The authors would like to thanks their industry partners
Readymix Holdings, Elasto-Plastic Concrete Pty Ltd (EPC), the
Roads and Traffic Authority of New South Wales (RTA), BOSFA
and the Australian Research Council (ARC) for their support of
this research.

REFERENCES
ABAQUS Inc, 2003. Analysis Users Manual, version 6.4 (ABAQUS,
Inc, USA).
Adams, D N, Lechner, M K and Lamb, I, 2001. M5 east tunnels: A flat
roofed, bolt and shotcrete-lined highway, in RETC Proceedings,
pp 501-512.

202

Bernard, E S, 2004. Creep of cracked fibre reinforced shotcrete panels, in


Shotcrete: More Engineering Developments, pp 47-57 (Taylor and
Francis Group: London).
Grimstad, E, 2001. Behaviour of steel fibre reinforced shotcrete during
large deformations in squeezing rock, in Shotcrete: Engineering
Developments, pp 119-122 (Swets and Zeitlinger: Lisse).
Pellet, F, Sahli, M, Boidy, E and Boulon, M, 2000. Modelling of
time-dependent behaviour of sandstones for deep underground
openings, in Proceedings International Symposium on Civil
Engineering in the 21st Century, Beijing, pp 431-438.
Sahli, M, Pellet, F, Boidy, E and Fabre, G, 2001. Modeling of viscous
behaviour of rocks for deep tunnel, presented to ISRM Regional
Symposium, Eurock.
Shalabi, F I, 2005. FE analysis of time-dependent behaviour of tunneling
in squeezing ground using two different creep models, in Tunnelling
and Underground Space Technology, 20:271-279.
Singh, T N and Verma, A K, 2005. Prediction of creep characteristic of
rock under varying environment, Environ Geol, 48:559-568.

Melbourne, VIC, 4 - 7 May 2008

13th Australian Tunnelling Conference

Вам также может понравиться