DISCUSSIONS

The circular tunnel in elastic ground

MUIR WOOD, A. M. (1975). Gt%technique 25, No. 1, 115127.

D. J. Curtis, Mott, Hay and Anderson, Croydon

Muir Wood presents a simple approach to the problem of proportioning tunnel linings,
which can be useful in many situations. A great amount of mathematical analysis is contained
in the Paper, and this contribution questions some of the assumptions and the accuracy of
some of the formulae. The extension of the method to include the parameter of time in
the context of visco-elastic behaviour of ground and lining is indicated, in a manner which
retains the simplicity desired by the Author.

NOTATION
The Author’s notation is followed wherever possible, but some variations and additional
symbols are employed. In general, a symbol with subscript c refers to ground behaviour,
whilst a symbol without this subscript refers to the lining. All stresses and displacement
described occur at the interface between ground and lining, at r=ro.

g gravitational acceleration Ul deformation of lining

H3 water head at radius r3 uo deformation of ground in absence of
N circumferential (hoop) thrust per unit a lining
length of lining ii deformation of ground inhibited by
PW pressure on lining due to water presence of a lining
Q Author’s compressibility factor R, Y rate of creep
Q2 analogous to the inverse of Author’s E strain
stiffness factor R,, expressed thus as P density of water
for a solid lining Qz= Qro2/t2 %, =h pre-existing vertical and horizontal
Qlr seepage factor stress in undisturbed ground
sn,St normal and shear stresses (%--%=Po)
TO time interval between excavation and ratio of uniaxial creep strain to
installation of lining immediate (elastic) strain

BENDING IN THE ELLIPTICAL MODE

Using Airy’s stress function, for a circular opening in an infinite ground, deformations at the
periphery of the opening can be obtained:

uo = w [(S - 6v)S,,+ (4-6v)S,,] cos 28 . . . . . (Al)

0

uoe = -r’$~+V)[(4-6~)S..+(5-6~)S,1sin 20 . . . . (A2)

c
where stresses S,, cos 20 and - S,, sin 28 act normally and tangentially around the opening.
When the opening is excavated, the distortional unloading of the ground is
PO
T cos 28

so that S,, = S,, = PO

T

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.

 232 DISCUSSIONS

The elastic, or immediate deformation of the ground is

uo. = ~(3-4v)+cos28 . . . . . . . (A3)

C

UOBO = -r”~+“)(3-4~)$sin28 . . . . . . . (A4)

C
A thin inextensible lining will deform
Ul = r,4(2S,+S,)cos 28/18EZ . . . . . . - (A5)
uel = -ro4(2sn+ S,> sin 28/18EZ . . . . . . . (A6)
producing forces in the lining
Z&$ (S,+2S,) c0s 28 . . . . . . . . (A7)

M = $(2Sn+S,)c0s 28 . . . . . . . . * WV

Making the Author’s assumption of no shear between lining and ground, (i.e. S,=O) we
proceed as follows.
(a) The ground will deform radially under the action of % cos 28 according to eqn (A3).
(b) The lining will deform radially under a load S, cos 28, from eqn (A5)

ul = gz2S., cos 28.

(c) The ground will be restrained from deforming radially under the action of a load
S,,, cos 28 = -S,, cos 28 from eqn (Al)

no = <~C(l+u)(5-6~)S. cos 28.

(d) Since u. +a0 = ul, we find

s = 3(3 - 4v)P,,/2 ...... (A9)
n 5_6v+4e2 .
where

...... (A101

From eqns (A7) and (A8)

...... W 1)

. . . . . . .
6412)
The foregoing analysis may be compared with the Author’s.
(a) Eqn (12) may be directly deduced from section (a) above.
(b) Eqns (16) and (A12) are not identical because the Author has made two implicit
assumptions.
Muir Wood does not define the cause of the change in deformational loading PO, but the
relationship stated after eqn (14) is equivalent to assuming, as we have, that PO is the stress
which would act in the ground at the periphery if the tunnel had not been excavated. In such

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.

 DISCUSSIONS 233

Fig. 1

an unperforated mass of ground the stress would be the same no matter at what radius the
load P, were applied. However, the further away the load is applied, the greater the deforma-
tion of the periphery of a hole in the ground. The bending moment in the lining is, therefore,
a function of the location of the cause of the distortional stress changes around the tunnel as
well as of their magnitude.
The stress regime in the vicinity of a deep tunnel, before excavation, is usually considered to
be uniform, i.e. uVand uh are each constant at all points. At any point on the periphery of
the tunnel, the normal and shear stresses are obtained by transformation of axes. The
Author assumes that if only normal pressures can be transmitted across the interface between
lining and ground, the shear component of PO will not affect the load on the lining. Since the
shear component makes a contribution to the radial deformation of the ground, uoC(eqn (A3))
this assumption is incorrect.
The values of bending moment produced by the two equations are compared in Fig. 1.

EFFECT OF SHEAR FORCES BETWEEN GROUND AND LINING

If we assume full shear interaction between lining and ground, and repeat the procedure
given in an earlier section of this contribution, but equate both radial and tangential deforma-
tions, we find
s = (1- QzPoP . . . . . . . . . . (Al3)
n l+Q,(g
1
s = (1+2QzPoP . . . . . . . . . . (A14)
t l+Q,(g
1
N msx _ -+Poro . . . . . (Al%
2
=+/{l+Q2[~]} . . . . .
M max (A16)

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.

 234 DISCUSSIONS

Author’s equation (31: ,

O.11
Fig. 2

It may be noted that Qz usually possesses a numerical value well in excess of unity, and then
S, is a negative quantity. Provided that the uniform component of the total load on the
lining, which will be a positive number, exceeds S, there is no danger of locally separating the
lining from the ground. Since S, =ji/2, the Author’s equation BPR
z = 2 I+R
is invalid. He has assumed that the normal and shearing components of P,, are shared between
ground and lining in identical proportions, which by virtue of eqns (A13) and (A14) is not the
case.
If the shear stress in the interface is limited to, say
S, = T . . . . . . . . . . (A17)
then

3(3 - 4~) $‘- {2Qz + (4 - 6v))T

s, = . . . . . . 64w
4Qz+5-6~
Equation (24) should be derived as follows. If we write

Eg = y [(l -“)Os--va,]
c
and

lO =
u l&u,
-+--
r r80
we eventually obtain for direct comparison with eqn (24).
uoo = --Cl+4 F~~)$+(L+)T~~] . . . . . (A19)
EC [
The Author’s criterion (a) is expressed by eqn (27), in which the mean pressure jj should
surely be replaced by the mean normal pressure Ap from his eqn (37).
Criterion (b) is expressed by eqns (A14) and (A17), i.e.

Since

l!=S
2 n . . . . . . . . . . (A21)

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.

 DISCUSSIONS 235
we find from eqns (A13) and (A20)

* . . . . . . .

<+fi when Q.$+l . . . . . . . (A23)

Figure 2 shows the difference between eqns (A23) and (31).
The general expression for h is obtained from eqn (Al) as
X = 3S,,E,/r,(l +v)[(5 - 6v)&,+ (4- 6v)S,,] . . . . (A24)
Since there is coupling between radial and tangential deformations, this should be reflected by
using eqns (Al) and (A2) to derive coupled ground reaction coefficients.

EFFECT OF THICKNESS OF THE LINING

In all the expressions in the first two sections of this contribution, the Author’s distinction
between r. and the centroidal radius qr, has been ignored. Comparisons between analogous
formulae derived for a continuous thick ring suggest that for bending moments M, Mccp,~2r02,
but in the expression for Q2, r. should be retained. For thrusts N, NccPOr,,, but in the expres-
sions for Qz or R,, vro should be used for ro.

GROUNDWATER
It is difficult to see why the integration constant A in the Author’s solution should be zero.
If the ground is infinite in extent the flow is also zero and therefore B must be zero. We can
proceed along lines similar to those of the Author, and indeed extend the method to account
explicitly for the effect of the lining. If it is assumed that although the ground is infinite in
extent, it is freely permeable beyond radius r3 from the axis of the tunnel where r3 > r. and that
the water head at r3 is Ha, the pressure on the lining due to seepage forces through the ground
is given by

1
. . . . . . (A25)

where

Q=&&-$yR,
-t
k rl’o
Qk = E Wdro> ro-ri
The pressure on the lining due to direct water loads is
p N pgH,v/(l + Q><l + Qg) . . . . . . . (A26)
(If the lining is impermeable p = 0 and (p = pgH,/(l + Q)).
The hoop thrust in the lining is
N = (p+pW)ro . . . . . . . . (A27)
and the flow into the tunnel

(A281
Over a wide range of ground and lining properties it will be found that the total effect of
water loads upon the lining, (eqn (A27)) will be very close to the full hydrostatic head. The
steady state simple seepage condition, is, however, a conservative design case, because it

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.

 236 DISCUSSIONS

ignores the historical variation in ground properties from the time of excavation of the tunnel
to final equilibrium.
The seepage model referred to earlier is not very satisfactory as it will overestimate the water
inflow compared with a more realistic model. The stress calculations are reasonably accept-
able when compared with finite element analyses.

EXTENSION OF THE METHOD

We can now continue the Author’s search for a simple initial approach to tunnel design, and
indicate how easy it is to employ a visco-elastic model. First, the ground movement which is
delayed because it is supported by the face at the time the lining is installed is examined. The
interaction formulae derived above apply directly and it is merely necessary to modify the
modulus values, so that
E’ = E/(1++) . . . . . . . . (A29)
where 4 = E,/E, (Fig. 3)
The ground movement which has occurred before the lining is installed is now considered.
(The derivation is given here for distortional loads, but is similar for the uniform load situation).
If the creep strain of a uniaxially stressed column is expressed as

where F is a function of time T, which has a value of unity when T= 0 and zero when T = co,
at time To when the lining is installed the ground movements are given by replacing EC in
eqns (A3) and (A4) by EC/{1++,[l - F(TJ]}.
If no lining had been installed, the movements for infinite time would have been calculated
from the same equations, writing for the modulus EC’= EC/(1+q&).
The load due to creep gradually builds up on the lining. If the final values are S, 20 and S,cos
28 the deformations of the lining are given by eqns (A5) and (A6), substituting E’= E/(1 + 4) if
the lining is also a visco-elastic material.
The deformations of the ground are restrained by the lining, and eqns (Al) and (A2) are used
for stresses - S, and - St, and substituting ECfor the modulus.
Equating deformations and solving for the two unknowns S, and St we find that the resulting
expressions are those resulting from the purely elastic analysis, e.g. eqns (A15) and (A16), but
with modified modulus values as above and multiplied by F(T&,/( 1 + CO).
We can substitute into eqns (A7) and (A8) to find the forces in the lining. For example,
eqn (A16) becomes in the visco-elastic case

M max (A30)

where
EC 1 l++ ro3
Qz=BiTviq-1210
Both F (To) and &/(l -+I&) cannot exceed unity. In the simple Kelvin visco-elastic model
F(T,J = e - YTowhere y is a ground property defining the rate of creep. The parameter 4
defines the amount of creep.

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.

 DISCUSSIONS 231

Fig. 3. Kelvinvisco-elastic cell

These parameters are all that are required in addition to the linear elastic ground (or lining)
properties. In principle they can be obtained from geotechnical studies, just as are the
elastic properties.

CONCLUSIONS
The simplicity of the visco-elastic relationships provides a means of examining numerically
the sensitivity of the load to be carried by the lining. In tunnelling terms we can assess the
circumstances in which it will be profitable to delay the installation of a lining, as well as the
implications of the amount of creep expected.
As the Author has emphasized in his Paper, however, the simplicity which is the attraction
and usefulness of the analytical relationships presented must not be allowed to distract
attention from the difficulty of modelling the behaviour of real ground.

Author’s reply to Curtis

The difference between my eqn (16) and his eqn (A12) stems entirely from my simplification
in ignoring initial shear stresses between ground and lining. It is of interest to note (Curtis,
Fig. 1) that my solution lies intermediately between the two cases he considers for zero shear
and for fully developed shear respectively between ground and lining, the validity of the latter
case requiring confirmation in respect of criteria (a) and (b) (p. 119 of the Paper) using Curtis’
eqn (A22) for the latter.
Consideration of seepage forces within a finite cylinder provides a practical first approxi-
mation for assessment of inflow and seepage forces, although it is often more valid to consider
the familiar form of potential flow towards the tunnel, considered as a sink, from an imaginary
source, with the upper surface of the aquifer in which the tunnel is situated constituting a plane
of symmetry.
For the extension of the elastic to the visco-elastic case, while the substitution
E’ = E/(1 ++[l --F(T)]}
provides a simple first parametric test, there is only a direct application of this approach where
the time effects are long-term in relation to the period during which the tunnel advances through
the influence zone. Otherwise, the time base for F(T) cannot be measured from a unique
instant and Curtis’ 57, at the time of placing the lining will relate to a lower fraction of total
creep strain as the distance from the tunnel increases. Even for an assumed Kelvin model, a
valid solution will require a numerical integration in recognition that creep will occur as the
stress pattern is varying. Thus the problems of three-dimensional time dependence unfortu-
nately persist. However, Curtis’ simplification will generally lead to a conservative result.

Downloaded by [] on [06/11/21]. Copyright © ICE Publishing, all rights reserved.