Sei sulla pagina 1di 17

Tunnelling and Underground Space Technology 18 (2003) 377–393

Effect of pore water pressure on tunnel support during static and seismic
loading
A. Bobet*
Purdue University, West Lafayette, IN, USA

Received 27 September 2002; received in revised form 13 January 2003; accepted 27 January 2003

Abstract

The support of underground structures must be designed to withstand static overburden loads as well as seismic loads. New
analytical solutions for a deep tunnel in a saturated poroelastic ground have been obtained for static and seismic loading. The
static solution accounts for drainage and no-drainage conditions at the ground–liner interface. Linear elasticity of the liner and
ground, and plane strain conditions at any cross-section of the tunnel are assumed. For tunnels in which ground stresses and pore
pressures are applied far from the tunnel center, the drainage conditions at the ground–liner interface do not affect the stresses in
the liner. The analytical solution shows that the stresses in the liner are exactly the same whether there is drainage or not at the
ground–liner interface. Hence, if the drainage conditions in the tunnel are changed from full drainage to no-drainage or vice versa
the stresses in the liner are not affected. However, the stresses and displacements in the ground change significantly from drainage
to no-drainage conditions. For seismic loading a new analytical formulation is presented which provides the complete solution for
the ground and the liner system for both dry and saturated ground conditions. The formulation is based on quasi-static seismic
loading and elastic ground response; for a saturated ground, undrained conditions are assumed which indicate that the excess pore
pressures generated during the seismic event do not dissipate. The results show that the racking deformations of a liner in dry or
saturated ground are highly dependent on the flexibility of the liner.
䊚 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Drainage; Poroelastic; Saturated ground; Seismic; Tunnel; Support

1. Introduction the magnitude of seismic-induced solicitations in tunnel


liners (Merritt et al., 1985; Penzien, 2000; Hashash et
The support system of underground facilities in seis-
al., 2001). These relationships are based on the premise
mic zones must be designed to withstand static overbur-
that tunnel liners under seismic loading will tend to
den loads as well as to accommodate the additional
deform with the surrounding ground, and thus the liner
deformations imposed by the earthquake induced
must accommodate the ground deformations without
motions. Seismic-induced deformations of tunnel liners
loss of its structural integrity. Seismic-induced stresses
can be produced by direct shearing displacements of
in the liner need to be added to the stresses imposed by
active faults intersecting the tunnels, by ground failure,
the ground loads and construction process in order to
or by ground shaking around the opening.
evaluate the structural behavior of the liner.
There are two basic approaches in present seismic
Complicating all this is the fact that deep tunnels are
design. One approach is to carry out dynamic, non-
often found below the ground water table and thus the
linear soil–structure interaction analyses using numerical
surrounding ground is fully saturated. Both the static
methods. Input motions are applied to the boundaries of
and dynamic loads on the liner depend on the drainage
a ‘soil island’ to represent vertically propagating waves.
conditions at the contact between the support and the
This analysis is complex, requires a significant effort,
ground. If there is no-drainage at the contact, the liner
and thus can only be justified for very special cases.
must support the pressures generated by the water as
The second approach provides relationships to evaluate
well as the pressures generated by the ground. With full
*Tel.: q1-765-494-2166; fax: q1-765-496-1364. drainage the pore water pressures become zero at the
E-mail address: bobet@ecn.purdue.edu (A. Bobet). contact and the liner only needs to support the pressures

0886-7798/03/$ - see front matter 䊚 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0886-7798(03)00008-7
378 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

from the ground. Although the concept is clear, quanti- 2. The relative stiffness method
fication of the loads on the support under any of the
two drainage conditions described is not trivial. If there There are a significant number of empirical, analytical
is no-drainage, the pore pressures at the contact will and numerical procedures to obtain stresses in the
induce deformations of the liner; as a consequence, the ground and in the liner for a tunnel, given the liner and
soil surrounding the tunnel will move with the liner and the ground properties, pore water pressures and drainage
thus carry additional load. Hence there is a transfer of conditions at the tunnel perimeter (see for example
the loading from the support to the ground and the Einstein and Schwartz, 1979; Aristorenas, 1992; Einstein
support may not carry the full load from the water. If et al., 1995; Einstein and Bobet, 1997; Suwansawat,
there is full drainage the pore pressures are zero at the 1998; Bobet et al., 1999; ITA, 2000; Bobet, 2001). Of
contact, but additional seepage forces are generated in the analytical procedures, the most widely used is the
the ground due to the movement of the water towards relative stiffness solution method (Einstein and
the tunnel (Lambe and Whitman, 1969). The seepage Schwartz, 1979; Bobet, 2001). In this method it is
forces produce deformations in the ground which in the assumed that: (1) the ground and the liner are elastic
end are transmitted to the support because of the materials; (2) there is no water in the ground; (3) the
excavation of the tunnel and the installation of the liner
compatibility of deformations between the ground and
are simultaneous; (4) the cross-section of the tunnel is
the support. As a result the support will carry additional
circular and (5) plane strain conditions apply at any
load due to the water draining towards the tunnel. Thus,
cross-section of the tunnel. Although some of the
the assumption that the liner will carry all the load from
assumptions may be too restrictive, the method has the
the water if there is no-drainage is conservative, and the following advantages: (1) it is simple to use; (2) it
assumption that the liner will carry only the load from models the effects of the most significant ground and
the ground if there is full drainage is unsafe. support characteristics and (3) it can be used for
During an earthquake the loading is very rapid com- preliminary design and to facilitate more advanced or
pared with the permeability of the ground; as the soil detailed modeling of the tunnel.
tries to deform, excess pore pressures are generated The axial thrust (T) and the moment (M) acting on
because the ground cannot change volume (i.e. there is the liner, obtained using the relative stiffness method
no time for the water to move in or out of the pores); for the case of no relative displacement between the
because of the rapid loading, no dissipation of pore liner and the ground, are as follows (see Fig. 1 and
pressures occur. Thus an ‘undrained’ analysis should be Appendix A for notation; for the derivation of the
performed. Once the seismic loading ends all the excess equations, see Einstein and Schwartz, 1979):
pore pressures generated dissipate with time and the
initial state is recovered (at least for an elastic ground, 1 1
Ts Žs9vqs9h.Ž1qC1.roq Žs9vys9h.Ž1q2C2.rocos2u
which is the focus of this work). 2 2
This paper concentrates primarily on the evaluation 1
Ms Žs9vys9h.Ž1y2C2q2C3.ro2cos2u (1)
of the effects of pore water pressure on the loads of the 4
support for a deep circular tunnel located below the
water table. A new analytical tool, based on the relative where
stiffness method (Einstein and Schwartz, 1979; Penzien,
2000), is presented to determine static and seismic CF(1yn)
C1sy
loading on the support. The solution can account for: CqFqCF(1yn)
(1) drainage conditions at the ground–liner interface (6qF)C(1yn)q2Fn
bs C2sbC3
and (2) effect of groundwater pressure on ground and 3Fq3Cq2CF(1yn)
support response. As with the relative stiffness method, C(1yn)
it is assumed that the ground and the liner are elastic, C 3s (2)
2 C(1yn)q4ny6by3bC(1yn)x
w
and that plane strain conditions apply at any cross-
section of the tunnel. These assumptions may limit the and C and F are the compressibility and flexibility
practical applicability of the results obtained; however, ratios, respectively, which are defined as:
the objective is not to find a general solution, which is
unrealistic given the complexity of tunnels, but rather
to develop a simple formulation for preliminary esti- EroŽ1yn2s .
mates that can be used for further analysis, or even to Cs
EsAsŽ1yn2.
acquire additional insight into a problem with a small
Er3oŽ1yn2s .
computational effort. At the same time the solution Fs (3)
includes the most relevant variables of the problem. EsIsŽ1yn2.
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 379

3. Static analysis

Two extreme conditions at the ground–liner interface


are possible: full drainage, or no-drainage (Fig. 2). For
the full drainage case the pore pressures, u, at the
interface are zero and there is water flow towards the
opening; for the no-drainage case the pore pressures are
equal to the far-field pore pressures, uf , and there is no
flow. Full drainage or no-drainage at the interface
depend on the relative permeabilities of the ground and
the liner. In some cases full drainage or no-drainage
may apply, but in other instances partial drainage may
occur. Only the two extreme conditions are considered
in the following discussion. Differences in the liner
stresses (and similarities, as will be shown) in each of
the two drainage conditions are investigated.
Fig. 2 shows a tunnel with and without drainage at
the ground–liner interface. This problem can be divided
into two new problems by the principle of superposition
in elasticity. This is shown in Fig. 3. The total far-field
external stresses in the ground are decomposed into
effective stresses (Fig. 3b), and pore pressures (Fig.
3c). The solution of the problem shown in Fig. 3b can
be obtained using the relative stiffness method; the liner
loads are given by Eq. (1).
In the problem shown in Fig. 3c, equilibrium between
the ground and the liner is established when the displace-
ments in the ground at the tunnel wall coincide with the
displacements of the liner, and when there has been a
transfer of stresses between the ground and the liner.
This requires that at rsro, (i.e. at the tunnel wall) the
radial stresses, sr, and the radial displacements, Ur, of
the ground are equal to those of the liner; note that
because of the radial symmetry of the problem, tangen-

Fig. 1. Tunnel and support.

As is the cross-section area of the liner per unit length


of tunnel (i.e. Asst), and Is is the moment of inertia of
the liner per unit length of tunnel (i.e. Iss1y12t 3; Fig.
1).
Tangential stresses and strains in the liner are given,
from Eq. (1), by:

T Mt
ssus "
As 2Is
1yns2 s
´us f su (4)
Es

The relative stiffness method is applicable to a ground


medium without water (i.e. dry ground). That is, the
method cannot take into account the drainage conditions
at the ground–liner interface. Fig. 2. Drainage and no-drainage at the liner-ground interface.
380 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

Fig. 3. Decomposition of cases with drainage and with no-drainage.

tial displacements and shear stresses are zero. Hence, r2o B R2 E


the two boundary conditions at the tunnel wall are: s9rs C1y Fs9r
R2yr2o D r2 G o
ŽUr.groundsŽUr.liner at rsro; or UrZrsrosUrs r2o B
C
R2 E
Fs9r
s9us 1q
Žsr.groundsŽsr.liner at rsro; or srZrsrossrs (5) R2yr2o D r2 G o
usuf
Due to the symmetry of the problem of Fig. 3c the
(1qn) r2o B R2 E
stresses at the interface are constant and independent on Ur s rC1y2nq Fs9r
the angle u. Note that in Eq. (5) the stresses are total E R2yr2o D r2 G o
stresses since this is required for equilibrium. Uus0 (6)

3.1. No-drainage conditions where u is the pore pressure and R is the distance from
the center of the tunnel to the exterior boundary (Fig.
For no-drainage conditions, the full water pressure is 3c).
acting at the interface. The liner must deform inwards; The radial displacements in a circular liner with
because of compatibility of deformations between the exterior radial stresses ssr are given by:
ground and the liner at the interface, the ground must
follow the deformations of the liner and thus take some 1yns2 srsro2
of the loading from the water pressure. Hence the water Uss (7)
Es t
pressure at the interface is distributed between the
ground and the liner. At the tunnel wall, the radial stress
applied to the ground is sgro (tension), and to the liner From Eqs. (5) and (6), with s9rosufyssr, and taking
ssr (compression), such that sgroqssrsuf, or sgros R4ro,
s9rosufysrs.
The problem to be solved is that of a circular opening
subjected to an internal tension s9ro. This problem has Es(1qn)t
srss uf (8)
the following solution (Timoshenko and Goodier, 1970): Es(1qn)tqEŽ1yns2.ro
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 381

Fig. 4. Pore pressure decomposition for a tunnel with full drainage at the ground–liner interface.

The stresses in Eq. (8) create a uniform constant internal pressure uf, and full drainage at the tunnel wall
compressive thrust in the liner given by: (Fig. 4a); (2) a tunnel subjected to an inward internal
pressure uf, and full drainage at the tunnel wall (Fig.
Es(1qn) ro 4b).
Ts uft (9) The problem shown in Fig. 4a is solved first. The
Es(1qn)tqEŽ1yns2.ro
solution can be found in Appendix B1, where it has
Eq. (9) shows that for a perfectly incompressible been assumed that the tunnel has no liner. It will be
liner (i.e. EyEss0), the axial force in the liner is Ts shown that this assumption is correct since the liner
ufro; that is, the liner must support the full water does not carry any load. The stresses and displacements
pressure. If this is not the case (the liner is compressi- in the ground are given by:
ble), the axial force is reduced due to the load transfer
from the liner to the ground which is a function of their R
relative stiffness. The final stresses in the liner are wB E2 z
ln
1y2n r2o R uf r
2 fx
determined from Eqs. (9) and (4), together with the s9rs 2
u C F y1|q
2(1yn) R yro yD r G ~ 2(1yn) R
stresses obtained from the relative stiffness solution ln
applied to the problem shown in Fig. 3b; that is by ro
superposition of the stresses produced by the ground 1y2n ro2 wB E2
R z

only (considering the ground buoyant unit weight), and s9usy u fxC
2(1yn) R2yro2 yD r G
F q1 |
~
by the water pressure only. In fact, close inspection of R
Eq. (9) shows that this is the solution for a tunnel with ln q1y2n
uf r
a uniform external stress of magnitude uf, using the q
2(1yn) R
relative stiffness method. Hence, the total stresses in the ln
liner can be obtained directly using Eq. (1) with total ro
stresses (svss9vquf, shss9hquf). (1qn)(1y2n) w ro2 R2q(1y2n)r2

3.2. Drainage conditions


Urs
2(1yn)E x
y 2 2 uf
R yro
y
r

ufr B R Ez
The problem of a tunnel with a liner and full drainage
conditions at the liner–ground interface can also be
q
ln
ro
Cln q1ynF
RD r G
~
|
decomposed into the ‘effective stress’ problem (Fig. 3b)
and the ‘pore pressure’ problem (Fig. 3c). The effective Uus0 (10)
stress problem can be solved with the relative stiffness
method, as before (i.e. Eq. (1)). The tunnel with only
pore pressures (Fig. 3c) can be further decomposed into With the condition R4ro which implies that the pore
the following: (1) a tunnel subjected to an outward pressure boundary conditions are far from the tunnel
382 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

Table 1

Case 噛 1 2 3 4 5 6 7 8 9
E (MPa) 5000 1000 1000 100 1000 1000 1000 1000 1000
n 0.15 0.15 0.15 0.15 0.15 0.30 0.15 0.15 0.15
Es (MPa) 25 000 25 000 50 000 25 000 5000 25 000 25 000 25 000 25 000
ns 0.25 0.25 0.25 0.25 0.25 0.35 0.25 0.25 0.25
t (m) 0.10 0.10 0.1 0.10 0.1 0.10 0.10 0.20 0.10
Ko 0.5 0.5 0.5 0.5 0.5 0.5 1.0 2.0 0.5
ro (m) 2.5
s9v (kPa) 659.5
uf (kPa) 490.5 981.0

excavation, Eq. (10) can be written as: the cases analyzed cover EyEs ratios that change over
two orders of magnitude, which are thought to be
uf w B r E 2z
representative of tunnels excavated in stiff soils or in
s9rs x1q(1y2n)C o F | rocks where there is limited yielding of the ground. The
2(1-n) y D r G ~

uf w B r E2z tunnel radius and the effective vertical stress are kept
s9us x1y(1y2n)C o F | the same in all cases (Table 1). The effective vertical
2(1-n) y D r G ~
stress corresponds to the stress at the center of a tunnel
(1qn) (1y2n) uf r2yro2 located 50 m below the surface with a saturated ground
Urs
2(1yn) E r unit weight gs23 kNym3, and water table at the surface.
Uus0 (11) It is assumed in all cases that there are no relative
displacements between the ground and the liner.
From Eq. (11) one gets UrZrsros0. This is a very
The tunnel shown in Fig. 2 is discretized with 8-node
important conclusion since it indicates that the problem
isoparametric plane strain elements with pore pressure
shown in Fig. 4a does not affect the displacements of
at the corner nodes for the ground and 8-node isopara-
the ground at the tunnel wall, and therefore has no effect
metric elements without pore pressure for the liner. This
on the liner stresses. In other words, the initial assump-
allows for the simultaneous solution of the equations of
tion that there is no liner is correct. Therefore, the liner
equilibrium and groundwater flow. Because of the sym-
stresses from the problem shown in Fig. 3c are those of
metry of the problem, only one quarter of the tunnel is
the problem shown in Fig. 4b. However, the solution to
discretized. Along the vertical and horizontal axes of
this problem (Fig. 4b) has already been obtained in
symmetry no horizontal or vertical displacements,
Section 3.1 and is given by Eq. (9), which is the
respectively, are allowed. The far-field stresses are
solution of a tunnel with no-drainage. Thus the stresses
applied to the top and right lateral boundaries of the
in the liner due to the water pressure are independent
mesh with magnitude equal to the effective stress plus
of the drainage conditions at the contact between the
pore pressure; the boundaries are located at a distance
liner and the ground, and can be obtained in either case
R from the center of the tunnel such that Ryros100.
from Eq. (8) or Eq. (9).
The stresses in the liner are obtained at the end of the
This important conclusion is also reached by solving
following steps:
the same problem but using the concept of seepage
forces (Lambe and Whitman, 1969). The solution is 3.2.1. Step 1: geostatic
presented in Appendix B2, and shows that indeed the The entire continuum is discretized. The elements that
stresses in the liner are the same irrespective of the are within the excavation of the tunnel are active, and
drainage conditions at the interface. the elements of the liner are not active (i.e. the tunnel
In addition, the analytical solution is compared with has not been excavated yet). At the end of the step all
a numerical model. The finite element method (FEM) the elements have a constant effective vertical stress,
program ABAQUS, (ABAQUS, 1998) is used for this effective horizontal stress, and pore pressure equal to
purpose. Nine different cases are studied in which the the far-field stresses applied. An advantage of the FEM
ground and the liner stiffness are varied over a wide ABAQUS is that this can be accomplished with zero
range of values. The material properties, liner thickness displacements of the ground. This is a convenient feature
and loading conditions are shown in Table 1. The liner because the liner should deform only because of the
thicknesses shown in the table are not intended to give displacements caused by the excavation and not because
the best possible design, but rather to cover a range of of the initial conditions.
typical dimensions. Since it is the ratio of the ground’s
Young’s modulus to the liner’s Young’s modulus (Ey 3.2.2. Step 2: excavation
Es) which determines the stresses in the liner rather than In this step the tunnel is excavated and the liner is
either the ground’s or the liner’s Young’s modulus alone, installed. This is accomplished by deactivating the ele-
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 383

Fig. 5. Comparison between FEM and analytical solutions for drained and undrained conditions.

ments that constitute the excavation and by activating the liner for each case, regardless of where it is produced
the elements of the liner. Compatibility of deformations (i.e. interior or exterior fibers, springline, crown, etc.).
between the liner and the ground is ensured by requiring The figure shows an excellent agreement between the
that the nodes of the elements of the ground in contact numerical and analytical results with errors below 5%
with the liner have the same radial and tangential in all cases. The results support the conclusion that
displacements as the corresponding nodes of the liner. drainage at the boundary does not change the stresses
In this step the drainage condition at the interface in the liner.
between the ground and the liner is imposed; for the The results shown in Fig. 5 also indicate the relative
drainage case, the pore pressures at the interface are importance of some of the input parameters. Cases 1
forced to be zero (i.e. at atmospheric pressure); for the through 4 illustrate the significance of the relative
no-drainage case, the pore pressures are not constrained. stiffness of the ground and the liner; as the relative
stiffness (i.e. the ratio of EyEs ) decreases, that is as the
3.2.3. Step 3: steady-state ground becomes comparatively softer, the stresses in the
The ground around the tunnel consolidates and the liner increase. Case 5 has the same liner stresses as case
pore pressures reach equilibrium. For the drainage case 1, although the ground in case 5 is softer than in case
this means that, at the end of the step, a steady-state 1; this is so because the two cases have the same relative
flow is established from the exterior boundary of the stiffness. Case 6 is similar to case 2, except for the
mesh, where the pore pressures are equal to the far-field Poisson’s ratio, and it is used to illustrate that the
pore pressures, towards the tunnel where the pore pres- importance of the Poisson’s ratio is small. Cases 7 and
sures are zero. For the no-drainage case, the pore 8 compare the numerical and analytical solutions with
pressures everywhere in the ground are equal to the different coefficients of earth pressure at rest (Ko), and
ambient pore pressures at the end of the step. case 9 compares the two solutions with a larger column
The tangential stresses in the liner are obtained at the of water on top of the tunnel.
end of Step 3 (i.e. when final equilibrium has been In conclusion, if the ambient pore pressures are
reached) and compared with the analytical solution. Fig. prescribed far enough from the tunnel, the stresses in
5 shows the maximum compressive tangential stress in the liner are independent of the drainage conditions.
384 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

This leads to the following observation: if drainage at loading can be considered as a pseudo-static load. The
the interface changes from full drainage to no-drainage wavelength, l, can be estimated as lsVy f, where V is
or vice versa, no additional stresses in the liner occur. the shear wave or pressure wave velocity of the ground
Note, however, that from the ground point of view, there and f is the frequency of vibration of peak ground
is a difference, since the stresses in the ground given by motions. In most underground openings the quasi-static
Eq. (10) for drainage conditions and by Eq. (6) for no- conditions are usually satisfied. This assumption is used
drainage conditions are quite different. The new solution for the solution of the tunnel shown in Fig. 1, where
shows that the current practice of applying the full water the static far-field stresses sv and sh approximate the
pressure to an impermeable liner is conservative, and earthquake loading.
that the assumption that the liner will carry only stresses Pressure waves, P-waves, produce a vertical far-field
due to the buoyant weight of the ground (i.e. no seepage strain, ´vsVmax yVP, where Vmax is the peak ground
forces) when there is drainage may be unsafe. velocity, and VP the ground pressure wave velocity.
This conclusion holds only for deep tunnels with a Because the free-field vertical displacement is constant
depth much larger than the radius of the tunnel; if this everywhere in the continuum, the horizontal strains must
is not the case, the drainage conditions at the boundary be zero (i.e. in the free-field any vertical plane is a
affect the stresses in the liner (Bobet, 2001; Lee and plane of symmetry). A mechanically equivalent expres-
Nam, 2001). If extensive yielding in the ground occurs, sion in terms of far-field stresses is:
the solutions presented here may not be appropriate;
however, a limited number of analyses with the ground (1yn)E Vmax
modeled as an anisotropic elasto-plastic material (Ein- s vs
(1qn)(1y2n) VP
stein et al., 1995; Bobet et al., 1999) showed very n
similar liner stresses with drained and undrained condi- s hs sv (12)
tions, even when yielding occurred in a large volume of (1yn)
the ground. Although this appears to indicate that the
effects of drainage on the liner stresses are somewhat where E and n are the Young’s modulus and the
insensitive to the extent of yielding of the ground, Poisson’s ratio of the ground, respectively.
further verification is still necessary. Shear waves, S-waves, produce a far-field shear strain,
The effect of the tunnel depth and the distance from gsVmax yVS, where Vmax is the peak ground velocity,
the tunnel to the boundary, R, on the liner stresses is and VS the ground shear wave velocity. In terms of far-
discussed later in the paper. field stresses, this is mechanically equivalent to a shear
stress field given by tsGg, where G is the shear
4. Seismic analysis modulus of the ground (Gs1y2Ey(1qn)), which in
turn is mechanically equivalent to a compressive and a
Shear and pressure waves propagating in the plane of tensile far-field principal stress at 458 with the direction
the cross-section of the tunnel generate ground distor- of the pure shear. In this case the principal stresses are:
tions which tend to cause oval deformations of the
lining. The resulting change of the shape of the tunnel E Vmax
s vs
section generates circumferential strains in the lining 2(1qn) VS
which can cause cracking andyor crushing of the con- shsysv (13)
crete and reduce the carrying capacity of the lining. For
concrete linings, this deformation is usually the most Hence, maximum racking in Fig. 1 will occur at 458
critical and generally controls the seismic design. Thus, with the horizontal. The advantage of using horizontal
special emphasis is usually given to the evaluation of and vertical stresses is that all the equations derived in
the structural behavior of the lining distorted by racking. the appendices readily apply. Note also that Eqs. (12)
There are a number of empirical (Merritt et al., 1985), and (13) give the far-field stresses produced by the
and analytical tools (St. John and Zahrah, 1985; Penzien earthquake loading only and they should be added to
and Wu, 1998; Penzien, 2000; Fotieva et al., 2001) that the static stresses; although Eq. (13) indicates tensile
can be used to determine the response of the liner due stresses, the actual final stresses would be compressive
to racking. For the following discussion, it is assumed with the addition of the static far-field stresses. In this
that the dynamic amplification of stresses associated section a solution is presented for the earthquake loading
with a stress wave impinging on the opening is negli- only; these results should be added to the results from
gible. This assumption is correct if the wave length of the static tunnel presented in the previous section.
peak velocities, l, is at least 8 times larger than the The problem of seismic loading defined by the quasi-
width of the opening (Hendron and Fernandez, 1983). static far-field stresses sv and sh is solved in Appendix
Under these conditions the free-field stress gradient B3 for dry ground, and in Appendix B4 for saturated
across the opening is relatively small and the seismic ground with the assumption that no excess pore pres-
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 385

sures dissipate during the earthquake (i.e. undrained or This expression is less than one (i.e. free-field defor-
short term analysis). Note that this analysis is quite mations yield a conservative design) when F-(12(3y
different from the static case, where solutions are 4n))y((1yn)(1y2n)), which limits the flexibility of
obtained for drainage and no-drainage at the liner– the liner. For larger values of F, free-field deformations
ground boundary when a steady-state is reached (i.e. result in an unsafe design by a factor given by Eq. (15).
long-term analysis). For the earthquake, it is assumed As an example, for a perfectly flexible and incompress-
that the time for loading is very small compared with ible liner (Peck’s liner; Peck, 1969; i.e. Cs0, Fs`),
the time for dissipation of excess pore pressures, and the ratio is 4(1yn)y(3y2n) which is always greater
thus the soil is sheared undrained (i.e. no change in than one, and hence the far-field deformation would
volume, short term analysis). In the appendices Eqs. always give an unsafe estimate of the tunnel deforma-
(A24) and (A22) give the loading in the liner for the tion. The liner loading is given by Eqs. (A22) and
dry ground case, and Eqs. (A27) and (A26) for the (A24). Because of elasticity, the solution is proportional
saturated case. Tangential stresses and strains in the to the input loading; hence if the liner stresses are
liner can be computed from Eq. (4). computed based on the free-field ground movements,
It has been established in the literature that the free- the actual stresses differ from the computed stresses by
field deformations of the ground caused by a seismic the factor given by Eq. (15).
event may not be fully transmitted to the structure. In
fact the deformations of a tunnel depend on the relative 4.2. Dry ground and S-wave
stiffness between the liner and the ground (Hendron and
Fernandez, 1983; Merritt et al., 1985; Penzien and Wu, The free-field conditions are defined by:
1998; Penzien, 2000; Hashash et al., 2001). This paper
Dro Z 1 Vmax (1qn)
reinforces this concept and expands it to tunnels in Z

Z sy sy sv (16)
saturated ground where the ground deformations occur ro Zfree
Z
field 2 VS E
during an undrained loading (with excess pore pres-
sures), rather than drained (with no excess pore pres- From Eqs. (13), (A21) and (A22), taking Cs0,
sures) as assumed in the literature.
The following shows comparisons between the radial Dro ZZ

Z
deformations of the liner considering ‘free-field’, ‘dry ro Zdry
Z
ground
8(1yn)2F
ground’ and ‘saturated ground’ conditions for P-waves s (17)
Dro Z Z (1yn)(3y2n)Fq12(3y4n)
and S-waves impinging the tunnel. The comparisons are Z
ro Zfree
Z
field
made for the case of a tunnel with an incompressible
liner (i.e. Cs0, where the liner is much less compress- This expression is smaller than one (i.e. the assump-
ible than the ground). This assumption implies that the tion that the structure moves with the surrounding
deformations of the liner are determined by the response ground is conservative) when F-(12(3y4n))y((1y
of the liner to shear rather than to compression; this is n)(5y6n)). For typical values of the Poisson’s ratio,
an approximation since compression and shear are cou- this requires F-10, which may not always apply. Hence,
pled as the equations in the appendices indicate. How- the practice of taking the free-field ground deformations
ever the approximation makes it easier to quantify the as the deformations of the liner is unsafe if F)10, and
differences between the assumptions of free-field, dry, it underestimates displacements and stresses by as much
or saturated ground; the equations obtained can be easily as Eq. (17). For a perfectly flexible and incompressible
rewritten in terms of compressibility, C, and flexibility, liner (Peck’s liner) the design is unsafe by a factor
F, from the appendices. larger than two.
4.1. Dry ground and P-wave Because of the particular interaction that exists
between ground and structure in tunnels, the flexibility
The free-field conditions are defined by: ratio has a significant effect on the liner. Eqs. (17) and
(A24) show that as the flexibility ratio, F, increases,
Dro Z
Z Vmax (1qn)(1y2n) the deformations of the liner increase and the load
Z s s sv (14)
ro Zfree
Z
field VP (1yn)E demand on the liner decreases. For circular tunnels
moderate cracking of the liner increases the flexibility
From Eqs. (12), (A21) and (A22), taking Cs0, of the liner, and as a consequence the load demand in
the liner decreases.
Dro ZZ

Z
ro Zdry
Z
ground
4(1yn)2F 4.3. Saturated ground and P-wave
s (15)
Dro Z (1yn)(3y2n)Fq12(3y4n)
Z
Z
The net effect of a pressure wave in a saturated
ro Zfree
Z
field ground is an increase of the pore pressures in the ground.
386 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

Fig. 6. Comparison between free-field, dry ground, and saturated ground deformations for an impinging S-wave.

For an incompressible liner this leads to zero radial TZdry MZdry


deformations of the liner and to an axial force equal to s
TZsaturated MZsaturated
the pore pressure generated times the tunnel radius. The
(1yn)Fq6
ratio of the axial loads generated by the same pressure s4(1yn) (20)
wave on a tunnel in a dry ground and in a saturated (1yn)(3y2n)Fq12(3y4n)
ground is:
The ratio is smaller than 1 if F-12y(1yn). Hence,
for very flexible liners, the design of the liner with the
TZdry assumption of dry-ground is conservative. This assump-
s2(1yn) (18)
TZsaturated tion is unsafe however for more rigid liners.
Fig. 6 shows a plot of Eq. (17), which gives the ratio
which, given that the Poisson’s ratio is always smaller of the radial deformation of the support and the free-
than 0.5, is always larger than one, and thus the field ground motion as a function of the flexibility ratio,
assumption of dry ground is conservative. and corresponds to an impinging shear wave on a tunnel
with an incompressible liner (i.e. Cs0) and dry ground.
4.4. Saturated ground and S-wave The figure shows that for flexibility ratios larger than
10, the actual liner deformations, and consequently the
Eqs. (A25) and (A26) give the radial displacements stresses in the liner, are much larger than those predicted
of the liner for a tunnel in saturated ground. The ratio by the free-field deformations; as F™` the ratio goes
of radial deformations for dry and saturated ground to 8(1yn)y(3y2n). Fig. 6 also shows a plot of Eq.
conditions is: (19), which gives the ratio of liner deformations for dry
and saturated ground conditions. For flexibility ratios
larger than approximately 15 the assumption of dry
Dro Z
Z
Z
ground is conservative, with an upper limit, for F™`,
ro ZdryZ

4(1yn)w(1yn)Fq6x of 4(1yn)y(3y2n).
s (19)
Dro Z Z (1yn)(3y2n)Fq12(3y4n)
Z 5. Boundary conditions
ro Zsaturated
Z

For the analyses, two assumptions have been made


and correspondingly, relative to the boundary conditions: (1) the external
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 387

Table 2 Eq. (10) are used to investigate the effect of R on


Ryro UryUr±R™`
the drainage towards a circular opening. Again, the
radial displacements are used to study the influence of
ns0.15 ns0.25 ns0.35 the distance to the boundary, R, and can be expressed
100 1.0002 1.0002 1.0001 as:
50 1.0007 1.0006 1.0005
20 1.004 1.004 1.003
(1qn)(1y2n) w
1 2 z
10
5
1.02
1.07
1.02
1.06
1.01
1.05
UrZrsros
2E
ufro

y
xln
R
y B E2
R
C F y1
ro D ro G ~
| (24)

effective stresses and pore pressures are applied far from


the tunnel (i.e. R4ro) and (2) the ground is weightless Eq. (24) has been used to determine that Ur±rsros0
and the external stresses are applied uniformly to the when R4ro. If this is not the case, the deformations at
boundary. For the static analysis, the magnitude of the the tunnel perimeter will introduce additional stresses to
far-field stresses is obtained from the unit weight (buoy- the liner. It is appropriate then to compare this equation
ant unit weight of the ground for effective stresses or with Eq. (6) which gives the displacements of the
the water unit weight for pore pressures) times the ground due to an internal pressure s9ro. This stress s9ro
distance from the ground surface or from the water table can be thought of as the equivalent stress at the tunnel
to the center of the tunnel. perimeter that will give the same deformations as Eq.
In this section, the influence of these two assumptions (24). Then,
on the stresses of the liner is investigated. This is
s9ro (1y2n) w B ro E2z
relevant to determine the depth at which far-field con- s x1yC F |
ditions can be assumed; or in other words, when a uf 2 y DRG ~
tunnel can be treated as a deep tunnel. w
1 2 z
1
5.1. The R4ro assumption
=
x
y
ln
R
y B E2
R | B r E2
C F y1 1q(1y2n)C o F
ro D ro G ~ DRG
(25)

The assumption that R4ro has been used in two


types of problems: a circular opening with internal Table 3 shows the values of Eq. (25) as a function
pressure (Fig. 4b); and drainage towards a circular of Ryro. The table shows that Eq. (25) slowly decreases
opening (Fig. 4a). with R. This is due to the term ln(Ryro). For drained
Eq. (6) can be used as an example for the first case conditions, the errors associated with the pore pressures
(opening with internal pressure). The radial displace- are smaller than 10% for boundary conditions in which
ments are used to investigate the influence of the Ryro)20.
distance to the boundary, R. From Eq. (6),
5.2. The weightless ground assumption
(1qn) r2o B R2 E
Urs rC1y2nq Fuf (21) In actual conditions the vertical and horizontal effec-
E R2yr2o D r2 G tive stresses as well as the pore pressures increase with
depth. The differences between these conditions and the
As R™`, Eq. (21) can be written as: assumption that the far-field stresses are constant and
equal in magnitude to the stresses computed at the
(1qn) r2o center of the tunnel is investigated in this section. It is
Urs uf (22) clear from the data shown in Table 3, that the ground
E r
and liner properties may play a role in this matter.
Taking the ratio of Eqs. (21) and (22), and for rs However, to simplify the problem, only case 7 (Table
ro, one gets: 1), has been chosen for this analysis.

Table 3
r2o
(1y2n) q1
Ur R2 Ryro s9yu f
s (23)
UrZR™` ro2 ns0.15 ns0.25 ns0.35
1y 2
R 100 0.07 0.05 0.03
50 0.09 0.06 0.04
Table 2 shows the values of Eq. (23) as a function 20 0.11 0.08 0.05
of Ryro. For values of Ryro)10, the errors produced 10 0.14 0.10 0.06
5 0.19 0.15 0.09
assuming Rs`, are between 1 and 2%.
388 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

Fig. 7. Linear increase of stresses with depth. Comparison between FEM and analytical solutions for drained and undrained conditions.

It is assumed that the water table is at the ground this error increases with decreasing tunnel depth. This
surface, and that the ground extends indefinitely laterally is an expected result. One can conclude that under the
and under the tunnel. That is, the only limiting boundary conditions of the model (i.e. the depth is the only
condition is the distance from the tunnel to the surface; boundary condition), the analytical solution gives errors
the distance from the tunnel to the lateral and bottom below 10% for depths greater than approximately 10
boundaries of the area is modeled with Ryros100. The times the tunnel radius. Note also that the errors are
ground has weight, and hence total, effective and pore smaller than those shown in Table 3; this is so because
pressures increase linearly with depth. The FEM ABA- in the current model only one boundary is being limited,
QUS has been used for the numerical modeling. The while in the discussion concerning the data in Table 3,
same type of elements and steps described for the static all boundaries are limited. Again, the errors shown in
analysis are used here. The most important differences Fig. 7 as a function of depth may change with material
between the previous analyses and this analysis are that properties, although these changes are expected to be
half of the tunnel is discretized instead of a quarter small.
because the vertical axis is now the only axis of Given the results from this discussion and the consid-
symmetry. In addition, the top boundary coincides with erations given for the value of R (i.e. distance to the
the surface of the ground and is free of stress; the boundary), a tunnel can be considered a deep tunnel
stresses along the lateral boundary increase linearly with when the depth of the tunnel centerline is larger than
depth. 10 times the tunnel radius in dry ground. For tunnels
A comparison between the results obtained from below the water table, the depth should be at least 20
ABAQUS and the analytical solution is presented in times the tunnel radius below the water table.
Fig. 7, which is a plot of the maximum tangential
stresses in the liner. The analytical solution, as it should, 6. Discussion and conclusions
gives the same results whether there is drainage or not
at the ground–liner interface. Small differences are The effects of drainage conditions at the ground–
observed in the numerical solution. The error between liner interface are investigated in this paper. It is
the analytical solution and the largest of the drained and assumed that the water table is far from the tunnel
undrained stresses from the numerical method is also opening and that the excavation of the tunnel does not
shown in Fig. 7. It can be observed in the figure that lower the water table. This may not be the case for
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 389

shallow tunnels close to the surface (Ohtsu et al., 1999). used in tunnels, but may be restricted to some ground
The most important conclusion is that for tunnels in conditions where deformations are small, such as stiff
which ground stresses and pore pressures are applied far clays and rocks. If extensive yielding in the ground
from the tunnel center (i.e. R4ro), the drainage con- occurs, the solutions presented here may not be
ditions at the ground–liner interface do not affect the appropriate.
stresses in the liner. It is important to note that although
this is true for the liner, it is not so for the ground. The Appendix A: Notation
stresses and displacements in the ground may change
significantly from drainage to no-drainage conditions. The following is a list of the variables used in the
However, the ground displacements and total stresses at paper:
the ground–liner interface do not depend on the drainage
As, Is area and moment of inertia of the cross-section of
conditions. The stresses in the liner for either drainage the liner
or no-drainage can be obtained with the relative stiffness C, F compressibility and flexibility ratios
method using the far-field total stresses, sv and sh in E, n Young’s modulus and Poisson’s ratio of the ground
Eq. (1), instead of the effective stresses s9v and s9h. Es, ns Young’s modulus and Poisson’s ratio of the liner
f frequency of peak particle velocity
The new analytical solution obtained shows that the R radial distance to the pore pressures boundary
current practice of applying the full water pressure to ro radius of the tunnel
an impermeable liner is conservative. r, u polar coordinates
Another interesting observation is the role that the T, M axial force and moment acting on the cross-section of
the liner
parameter R (the distance from the boundaries to the t thickness of the liner
center of the tunnel) plays on the liner stresses. It has uf far-field pore pressures
been found that for a tunnel in dry ground, the model Ur, Uu displacements of the ground in polar coordinates
has to include a volume of the ground that is at least Urs, Uus displacements of the liner in polar coordinates
u excess pore pressures
10 times the tunnel radius. If a tunnel is modeled in a Vmax peak ground velocity
saturated ground, the model must include a ground VP pressure wave velocity in the ground
volume which should be larger than 20 times the tunnel VS shear wave velocity in the ground
´su tangential strain in the liner
radius. Note that since the solution of the pore pressures
l shear wave length, estimated as VS yf
includes the logarithm of the ratio of the distance from sv, sh far-field vertical and horizontal total stresses
the boundary to the center of the tunnel and the tunnel s9v, s9h far-field vertical and horizontal effective stresses
radius (Ryro), solutions with ratios larger than 20 will sr, su, tru total stresses of the ground in polar coordinates
s9r, s9u effective stresses of the ground in polar coordinates
give similar stresses for the liner, but longer times for ssr, ssu, tsru stresses of the liner in polar coordinates
pore pressure dissipation since they involve larger vol-
umes of ground.
A number of analytical expressions have been derived
to obtain the stresses in the ground and in the liner Appendix B: Relative stiffness solution
during a seismic event. The most important assumptions
are that the ground and the liner remain elastic and that (1) Static analysis with drained conditions (Fig. 4a).
the earthquake can be approximated by a pseudo-static The problem shown in Fig. 4a must satisfy the
loading. Expressions for a deep tunnel in dry and in equilibrium equations, the strain compatibility equations,
saturated ground are presented. The analytical solutions and the boundary conditions. It will be assumed that
obtained reinforce previous observations that showed there is no liner. The solution requires finding the Airy
that the practice of imposing to the liner the far-field stress function, f, that satisfies Eq. (A1) and boundary
ground motions may not be correct; liner deformations conditions:
depend on the relative compressibility and flexibility
1y2n 2
between the support and the ground. For rigid tunnels =2Ž=2f.s =u (A1)
(i.e. the flexibility ratio is smaller than 10), the assump- 1yn
tion that the tunnel displacements are equal to those of where =2 is the Laplacian operator. Stresses in polar
the far-field is conservative; for flexible tunnels, how- coordinates are defined as:
ever, this assumption may be unsafe by a factor larger
than two. Saturated ground conditions have a favorable 1 ≠f 1 ≠2f
effect on the liner compared to dry ground conditions srss9rqus q
r ≠r r2 ≠u2
for flexible tunnels, and an unfavorable effect for rigid ≠2f
tunnels, with differences of approximately 20%. suss9uqus 2
≠r
These conclusions are valid for tunnels in which the
≠ 1 ≠f E
B
liner and the ground have an elastic response. This trusy C F (A2)
assumption may be appropriate for most of the liners ≠r D r ≠u G
390 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

where r and u are the polar coordinates. where the parameters ao, bo, co, etc. are determined from
Pore pressures are governed by the following the boundary conditions.
equation: The following boundary conditions must be satisfied:

K 2 ≠´vol UuZus0sUuZus p sUuZus2ps0


2
= usy (A3)
gw ≠t s9rZrsrosuf
s9rZrsRs0 (A10)
where K is the permeability of the ground, and gw the
unit weight of the pore fluid. ≠´vol y≠t is the change of The solution of Eqs. (A9), (A10) and (A2) results in
volumetric strain with time. For plane strain, Eq. (A3) the following expressions for the stresses and displace-
becomes: ments in the ground:

≠Žs9xqs9y. R
EK ln
=2usy (A4) 1y2n r2o wB E2
R z
uf r
(1qn)(1y2n)gw ≠t s9rs 2 2
u
2(1yn) R yro y r
fxC
D
F
G
y1 |q
2(1yn) R
~
ln
ro
For steady-state conditions, ≠y≠ts0, and thus =2us0.
wB E2 z
Because of the problem symmetry, ≠y≠us0, and Eq. 1y2n ro2 R
2 fx
s9usy u C F q1|
(A4) can be written in polar coordinates as: 2
2(1yn) R yro yD r G ~

R
B ln q1y2n
d2 1 d E uf r
C 2
q Fus0 (A5) q
D dr r dr G 2(1yn) R
ln
ro
The solution of Eq. (A5) is: (1qn)(1y2n)
Urs
2(1yn)E
uf r w r2o R2q(1y2n)r2 ufr B R Ez
us ln (A6)
ln
R ro
ro
x
y
y 2
R yro
u
2 f
r
q
ln
Cln q1ynF Uus0
RD r
ro
G
~
|
(A11)
which satisfies the boundary conditions: (2) Static analysis with drained conditions using
seepage forces.
uZrsros0 The stresses produced on the liner due only to the
drainage occurring towards the interface (Fig. 3c) can
uZrsRsuf (A7)
be analyzed based on the concept of seepage forces
(Lambe and Whitman, 1969). The pore pressures gen-
Because =2us0, Eq. (A1) recovers the conventional
erated within the ground due to the seepage flow are
form:
given by Eq. (A6). The seepage force, Fr, is radial due
to the symmetry of the flow, and is given per unit
=2Ž=2f.s0 (A8) volume by Frsigw, where gw is the unit weight of
water, and i is the hydraulic gradient, is(1ygw)≠uy≠r,
which has the general solution provided by Timoshenko which can be found by taking the derivate of Eq. (A6).
and Goodier (1970): The equations of equilibrium with body forces are:
fsaoln rqbor2qcor2ln rqdor2uqa9ou ≠sr srysu
Frq q s0 (A12)
1 ≠r r
q a1rusinuqŽb1r3qa91ry1qb91rln r.cosu
2
1 Note that this is a one-dimensional problem due to
y c1rucosuqŽd1r3qc91ry1qd91rln r.sinu the radial symmetry (i.e. ≠y≠us0, ts0, Uus0). In
2
` terms of radial displacements, Eq. (A12) can be written
q 8 Žanrnqbnrnq2qa9nryn as:
ns2
` d2Ur 1 dUr 1 (1qn)(1y2n) uf 1
qb9nrynq2.cosnuq 8 Žcnrnqdnrnq2qc9nryn 2
q y 2 Ursy
ns2
dr r dr r (1yn) E R
rln
qd9nr ynq2
.sinnu (A9) ro
(A13)
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 391

The boundary conditions are: (3) Dynamic analysis: dry ground.


The problem is shown in Fig. 1, where sv and sh
UrZrsRs0 are the far-field stresses produced by the earthquake.
srZrsros0 (A14) The solution must satisfy Eq. (A8) or Eq. (A9) and
boundary conditions.
with the assumption that there is no liner (the effect of The stress–displacement relations for the liner are
the liner is added later). The solution of Eqs. (A13) (Flugge,
¨ 1966):
and (A14) is:
d2Usu dUsr C(1yn2)
q sy rotsru
C2 du 2
du E
UrsC1rq qC3rln r
r dUsu C B d4Usr d2Usr E
w z
qUrsq C 4 q2 2 qUrsF
R du F du
D du G

x |
(1yn)yln C(1yn )2
1y2n ro s rosrs (A19)
C1s 2
y(1yn)yln ro C3 E
ro 1y2n 1
q
y ro2 R2 ~
with the assumption that t<ro (the liner thickness is
R
(1yn)yln much smaller than the tunnel radius). C and F are the
ro compressibility and flexibility ratios given by Eq. (3):
C2s C3
1y2n 1 Eqs. (A9) and (A19) are combined with the following
q
r2o R2 boundary conditions:
1 (1qn)(1y2n) uf
C3sy (A15) tZus0stZus" p stZusps0
2 (1yn)E R 2
ln
ro UuZus0sUuZus" p sUuZusps0
2

The radial displacements at the interface, and when Uground


r ZrsrosUliner
r Zrs ro
R™`, are: groundZ linerZ
sr rsrossr rsro
ground liner
uf t Zrsrost Zrsro
UrZrsros(1qn) ro (A16) suZr™`ssv
E
srZr™`ssh (A20)
These displacements are reduced by the presence of
the liner, which as a reaction to the soil movements will The stresses and displacements in the ground are:
apply a constant radial compressive stress, ssr to the
ground. The liner deforms inwards because of the radial svqsh B r2 E s ysh
stress ssr; the ground deforms outwards because of ssr, srs C1qC1 o2 Fy v
2 D r G 2
and thus the radial deformations given by Eq. (A16) B 2 4E
ro ro
need to be reduced accordingly. The liner displacements, =C1q2C2 2 q3C3 4 Fcos 2u
Usr, are given by Eq. (7), and the ground displacements, D r r G
Ur, by Eq. (6). Compatibility of displacements between svqsh B r 2E
s ysh
sus C1yC1 o2 Fq v
the ground and the liner requires: 2 D r G 2
B r4o E
uf =C1q3C3 4 Fcos 2u
(1qn) royUrsUrs (A17) D r G
E
svysh B r2 r4 E
Combining Eqs. (6), (7) and (A17) results in a trus C1yC2 o2 y3C3 o4 Fsin 2u
2 D r r G
compressive radial stress for the liner equal to:
1qn T svqsh
S B r2 E s ysh
Urs rUT C1y2nyC1 o2 Fq v
Es(1qn)t E V 2 D r G 2
srss uf (A18) w 2 4z
Es(1qn)tqEŽ1yns2.ro ro ro W
=xy1q2(1yn)C2 2 qC3 4 |cos 2uTX
T

y r r ~ Y
Note that Eq. (A18) is identical to Eq. (8), which 1qn svysh
gives the stresses of the liner when there is no-drainage. Uus r
E 2
This proves that the seepage forces in the ground caused
by drainage at the tunnel perimeter induce in the liner
w
r2o r4o z
=x1y(1y2n)C2 2 qC3 4 |sin 2u (A21)
stresses that are identical to the stresses induced by the y r r ~
full water pressure for no-drainage conditions.
392 A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393

with

1y2ny(1yn)Cq(1y2n)CyF
C 1s
1q(1yn)CqCyF
(1yn)2Cq(1yn)yw(1yn)Cq4x3yF
C2sy2
(1yn)2Cq(1yn)(3y2n)qw(1yn)(5y6n)Cq4(3y4n)x3yF
1 (1yn)Cy2yC2w(1yn)Cq4nx
C 3s (A22)
3 (1yn)Cq2

¨
Since, Flugge (1966): with:
dT dM
ro y syro2trsu
du du (1yn)C
d 2M C1sy
roTq 2 sro2srs (A23) 1q(1yn)CqCyF
du (1yn)2Cq(1yn)yw(1yn)Cq4x3yF
where T and M are the axial force and moment acting C2sy
w2q(1yn)Cxw(1yn)q6yFx
on a cross-section of the liner. Thus: 1 (1yn)2Cy12yF
svqsh svysh C 3s (A26)
Ts Ž1qC1.roq 2 w2q(1yn)Cxw(1yn)q6yFx
2 2
=Ž1yC3.rocos 2u
and for the liner:
svysh
Ms Ž1qC2qC3.r2ocos 2u (A24)
4
svqsh svysh
(4) Dynamic analysis: saturated ground. Ts Ž1qC1.roq
2 2
If the ground is saturated, excess pore pressures, u,
=Ž1y2C3.rocos 2uM
may be produced during rapid loading (e.g. during a
seismic event). It is assumed that the duration of the svysh
s Ž1q2C2q2C3.ro2cos 2u (A27)
loading is very small relative to the permeability of the 4
ground and no pore pressure dissipation occurs (i.e.
undrained analysis). This condition implies that the
ground does not change volume, ´vols´rq´us0, which
References
in terms of stresses requires: s9rqs9us0. Equilibrium
and compatibility are satisfied by Eq. (A1), and pore
ABAQUS, 1998. Finite Element Program, from Hibbit, Karlson,
pressure dissipation by Eq. (A4). Because of the Sorensen, Inc. 1080 Main Street, Pawtucket, RI, USA.
assumption of short term analysis, the right-hand side Aristorenas, G., 1992. Time dependent behavior of tunnels excavated
of Eq. (A4) is zero, and Eq. (A8) is recovered. The in shale. Ph.D. Thesis, Massachusetts Institute of Technology.
solution is still given by Eqs. (A9) and (A2), with the Bobet, A., 2001. Analytical solutions for shallow tunnels in saturated
same boundary conditions as Eq. (A20). The stresses ground. ASCE J. Eng. Mech. 12, 1258–1266.
and displacements at any point in the ground are given Bobet, A., Aristorenas, G., Einstein, H.H., 1999. Feasibility analysis
for a radioactive waste repository tunnel. Tunn. Undergr. Space
by: Technol. 13, 409–426.
svqsh r2o svysh B r2 r4 E Einstein, H.H., Bobet, A., 1997. Mechanized tunnelling in squeezing
s9rs C1 2 y C1q2C2 o2 q6C3 o4 Fcos 2u ground-from basic thoughts to continuous tunnelling. Proceedings
2 r 2 D r r G of the World Tunnel Congress’97. Vienna, Austria, pp. 619–632.
s9usys9r Einstein, H., Bobet, A., Aristorenas, G., 1995. Feasibility study
svysh B r2 r4 E opalinuston. NAGRA Interner Bericht 61-95.
trus C1y2C2 o2 y6C3 o4 Fsin 2u Einstein, H.H., Schwartz, C.W., 1979. Simplified analysis for tunnel
2 D r r G supports. J. Geotech. Eng. Division, ASCE 105, 499–518.
svqsh svysh r2o Fotieva, N., Bulychev, N.S., Sammal, A.S., 2001. Design of tunnels
us y2 C2 2 cos 2u located near slopes in seismic areas. Fourth International Confer-
2 2 r
ence on Recent Advances in Geotechnical Earthquake Engineering
1qn ST svqsh r2o svysh and Soil Dynamics, paper No. 8.03, pp. 1–6.
Ursy rUT C1 2 q
E V 2 r 2 ¨
Flugge, W., 1966. Stresses in Shells. Springer-Verlag, Inc, New York,
w 2 4z NY.
ro ro W
=x1y2C2 2 y2C3 4 |cos 2uXT
T
Hashash, Y.M.A., Hook, J.J., Schmidt, B., Yao, J.I., 2001. Seismic
y r r ~ Y design and analysis of underground structures. Tunn. Undergr.
Space Technol. 16, 247–293.
1qn svysh w ro4 z
Uus r x1q2C3 4 |sin 2u (A25) Hendron, A.J., Fernandez, G., 1983. Dynamic and static design
E 2 y r ~ considerations for underground chambers. In: Howard, T.R. (Ed.),
A. Bobet / Tunnelling and Underground Space Technology 18 (2003) 377–393 393

Seismic Design of Embankments and Caverns. ASCE, NY, pp. Peck, R.B., 1969. Deep excavations and tunnelling in soft ground.
157–197. Proceedings Seventh International Conference on Soil Mechanics
International Tunnelling Association (ITA), Working Group No. 2, and Foundation Engineering, State-of-the-Art Volume, The Socie-
2000. Guidelines for the Design of Shield Tunnel Lining, Tunnel- dad Mexicana de Mecanica´ de Suelos, Mexico City, Mexico, pp.
ling and Underground Space Technology, 15 (3), pp. 303–331. 225–290.
Lambe, T.W., Whitman, R.V., 1969. Soil Mechanics. Wiley, New Penzien, J., 2000. Seismically induced raking of tunnel linings.
York. Earthquake Eng. Struc. Dynamics 29, 683–691.
Lee, I., Nam, S., 2001. The study of seepage forces acting on the Penzien, J., Wu, C.L., 1998. Stresses in linings of bored tunnels.
tunnel lining and tunnel face in shallow tunnels. Tunn. Undergr. Earthquake Eng. Struc. Dynamics 27, 283–300.
Space Technol. 16, 31–40. St. John, C.M., Zahrah, T.F., 1985. Applicable analytical tools for a
Merritt, J.L., Monsees, J.E., Hendron, A.J., 1985. Seismic design of seismic design of underground excavations. In: Mann, C.D., Kelley,
underground structures. In: Mann, C.D., Kelley, M.N. (Eds.), M.N. (Eds.), Proceedings of the Rapid Excavation and Tunneling
Proceedings of the Rapid Excavation and Tunneling Conference, Conference, Society of Mining Engineers of the American Institute
Society of Mining Engineers of the American Institute of Mining, of Mining, Metallurgical, and Petroleum Engineers, New York,
Metallurgical, and Petroleum Engineers, New York, NY, pp. 104– NY, pp. 84–103.
131. Suwansawat V., 1998. Using the Relative Stiffness Method (Tunnel-
Ohtsu, H., Ohnishi, Y., Taki, H., Kamemura, K., 1999. A study on liner) and the Finite-Element Method (PLAXIS) for Tunnel Design.
problems associated with finite element excavation analysis by the MIT Internal Report.
stress-flow coupled method. Int. J. Numer. Anal. Methods Geo- Timoshenko, S.P., Goodier, J.N., 1970. Theory of Elasticity. third ed.
mech. 23, 1473–1492. McGraw-Hill Book Co., Inc, New York, NY.

Potrebbero piacerti anche