MUIR WOOD, A. M. (1975). G6otechnique 25, No. 1, 115-127.

The circular tunnel in elastic ground

A. M. MUIR WOOD, MA, FICE”

Neither the ground around a tunnel nor, usually, the Ni le terrain autour du tunnel, ni g&&alement le
tunnel lining, behaves in an elastic manner. Never- tunnel proprement dit, ne se cornportent d’une
theless, the linear characteristic of an elastic or visco- manike Blastique. Neanmoins, les caractbistiques
elastic analysis has great merit, particularly when, as IinBaires d’une analyse Clastique ou visco-Clastique
ont beaucoup de m&ite, particulierement quand,
frequently occurs, limitations of knowledge of the
comme cela arrive frkquemment, les limitations des
behaviour of the ground do not justify greater connaissances du comportement du sol ne justifient
elaboration. pas une plus grande pr&zision.
This Paper sets out to establish the loading on a Cette communication tente d’Ctablir le chargement
circular tunnel in the ‘elliptical’ mode of deforma- sur un tunnel circulaire se dCformant d’une manierc
tion, making certain simplifying assumptions. The elliptique en faisant certaines simplifications dans
effects of shear forces between ground and lining are les hypothkses.
first ignored and subsequently introduced in an Les forces de cisaillement entre le sol et le tunnel
explicit manner. Bending moments in a lining are sont en premiere phase ignortes et ensuite intro-
duites d’une mani&re explicite. Les moments
related to a stiffness factor, R,.
flkchissants dans la vofite sont reliCs ?I un coefficient
While a higher mode of deformation of a tunnel de raideur R,.
lining is only applicable to problems of stability for Alors qu’un mode de d&formation plus important
thin linings, nevertheless, consideration of this con- d’une voate de tunnel est seulement applicable aux
dition allows the coefficient of ground reaction A to problbmes de stabilit6 des tunnels minces, on con-
be evaluated for conditions of asymmetrical deform- sidere nkanmoins que cette condition conduit ?t un
ation. coefficient de rCaction du sol X Bvalu6 pour les con-
Direct radial loading of a tunnel lining is related to ditions de deformation asymktrique.
Le chargement direct radial d’un tunnel est reli6 &
its compressibility factor R,. The component of
son facteir de compressibilitC R,. La composantc
radial loading caused by migration of water towards
de la charge radiale provoqut% par la migration de
the tunnel is also established. l’eau vers le tunnel est aussi Btablie.
Some aspects of application of the method are dis- Quelques aspects de l’application de la m&hode
cussed including the effects of joints in a lining and sont discut& y compris les effets des joints dans UII
the benefits established on account of the deliberate tunnel et le b&fice rCsultant de la prise en comptc
variation of R,. The method has serious limitations des variations d6lib6r6es de R,. La mCthode a dc
but frequently serves as a first, and very useful, indi- sQieuses limitations, mais sert frequemment comme
premiere et trbs utile indication pour mettre en
cation of the relative importance of the principal tvidence l’importance relative des principaux fac-
factors, prior to a more refined analysis. An teurs avant une analyse plus raffinke.
important point, frequently overlooked, concerns Un point important, frbquemment nbgligt, con-
the effect of variation of Poisson’s ratio of the ground cerne les effets de variation du coefficient de Poisson
du sol lors des modifications de charges, ce qui peut
as loads change, which may greatly affect the stress
gravement affecter la distribution des contraintes
distribution in the ground and, in consequence, the dans le sol et en consCquence les charges sur lc
loads on the lining. tunnel.

In a day of high manipulative capability for numerical solutions to problems of increasing

complexity there yet remains an important place for the simple approach. This is especially
true for tunnels for which the degree of confidence in the properties of the ground, including
provisions for heterogeneity and variability, rarely justifies the more prodigious exercises in
number-crunching. A special virtue of the simple method is that it can serve quickly to indi-
cate sensitivity of the solution across the range of possible ground parameters.

* Partner, Sir William Halcrow and Partners.

 116 A. M. MUIR WOOD

An elastic analysis, entailing linear stress-strain relations throughout, has the great merit
that several factors may be examined individually and the sum of the effects obtained by super-
imposition.
There may be, however, considerable economic benefits in the design of a tunnel lining by
taking advantage of a degree of plastic deformation of the ground, often of a time-dependent
nature; the concept of a competence factor (Muir Wood, 1972), i.e. the ratio of ground strength
in simple compression to the vertical overburden pressure, is at least a first indicator in assessing
whether the assumption of linearity is likely to be acceptable within the limitations of any two-
dimensional tunnel analysis.
Two-dimensional analysis usually assumes that the tunnel has materialized instantaneously
and that the lining has developed its full strength at the instant of excavation. As will be
shown, an elastic analysis may be undertaken in two or more phases, with different values for
the elastic moduli for each, and thus achieve a somewhat greater simulation of actuality.
Several commentators have observed that the paper by Morgan (1961) on stresses around a
circular tunnel contains a basic error, apart from the intentional simplifying assumptions.
This Paper sets out to correct the error, to extend the technique, as a result of twelve years’
further experience in the general approach to tunnel design based on the method, and to clarify
its application. There are several empirical approaches to the problems of tunnel design in
elastic ground but many of these make arbitrary and improbable assumptions which lead to
excessive loadings and in consequence generally to uneconomic designs (Schulze and Duddeck.
1964). The present situation is not aided by several records of strain measurements in tunnel
linings which, on account of instrumental defects, have been incorrectly interpreted in terms of
stress. Records of direct stress measurements known to the Author support the view that
there is a great merit in the light, flexible lining.

NOTATION
a, b, c, d, A, B, constants 111 mode of distortion of lining
c‘ cohesion (in effective stress terms) II number of segments in a ring of lining
E Young’s modulus for lining (replaced I’ normal pressure between ground and
by E/(1 --I+“) where lining continuous lining
along tunnel) /s mean value of p
E,, Young’s modulus for ground 1; maximum value of 5 (p-p)
Z stress function PO excess of p on vertical axis over ZI on
FC competence factor horizontal axis
GC modulus of rigidity of the ground l’v value of p on vertical axis of tunnel
Z second moment of area of lining per ct discharge of water per unit area of
unit length of tunnel ground in unit time
ZP effective value of Z for ajointed lining radius (and as sut%x in radial direction)
ZJ effective value of Z at a joint in a lining
r. radius to extrados of tunnel lining
K constant
RC compressibility factor
k coefficient of permeability for water
RS stiffness factor
KC3 coefficient of earth pressure at rest
T shear stress between ground and lining
I (as suffix) longitudinal direction
M bending moment in lining per unit effective thickness of lining
length of tunnel U radial movement of ground
IV ratio of horizontal to vertical pressures UO value of 21at r = r.
in the undisturbed ground UO maximum value of + u.
THE CIRCULAR TUNNEL IN ELASTIC GROUND 117

/cEnvelope of inkial load

,
on undeformed tunnel

Fig. 1. Reference diagram for initial loading on tunnel prior to deformation

circumferential movement of ground at 8 angle (and as suffix in circumferential

Y=Y, direction)
piezometric pressure at steady state of coefficient of ground reaction
flow of ground water Poisson’s ratio for ground
uniform radial deflexion of lining Poisson’s ratio for lining
ground pressure
uniform variation in p
shear stress in ground in r, 0 plane
strain in ground Airy stress function
ratio of radius of lining centroid to that angle of friction (in effective stress
of extrados terms)

BENDING IN THE ‘ELLIPTICAL’ MODE

Morgan obtained a solution by considering a circular lining deformed into the ‘elliptical’
mode in elastic ground (Fig. l), neglecting shear stresses between extrados and ground; this
omission is conservative and greatly simplifies the solution (Schmid, 1926; Engelbreth, 1961).
The solution then uses the Airy stress function in polar co-ordinates
(I = (ar2+-hr4+cr-2-i_d) cos 20 . . . . . . . (1)
where
*,1 = I?$+%!
rar r2802
= -(2a+6cr-4+4dr-2) cos 20 . . . . . . . . (2)

cD = ‘&i= (2a+12br2+6cr-4)cos20 . . . . . . * (3)

-(6cre4+2dre2)sin 20 . . .

From the assumed boundary condition, 7 rO=O where r=r,, for small displacements u (u<<rO).
Hence d= -3crOw2 from equation (4) a=6=0 since uB -+ 0 as r --f co. Thus, from equations
(2) and (3)
ul. = -6c(r-4-2ro-2r-2)cos20 . . . . . . * (3
o. = 6cre4 cos 28 . . . . . . . . . (6)
118 A. M. MUIR WOOD

The error in Morgan’s paper concerns the assumption, as for the thick pipe problem, that
plane strain entails plane stress, i.e. that ug + Go= constant (which would follow from the simul-
taneous requirement of c,=O and a,=constant). It can readily be shown that u,+ a,.=con-
stant requires
&S,,
2u,+rar = 0 i.e. 8, cc l/r2

and invariable with respect to 0.

Plane strain requires

E, = $ [uI-V(u.8+ur)] = 0
C
SO

u, = v(ue+ur) . . . . . . . . . (7)
and as

cI. = + [u,-Y(u~+cQ)] . . . . . . . (8)

C
from equations (7) and (8)

EI. = y [ur(l -LJ)-vuO] . . . . . . . . (9)

c
with the convention of positive compression. By substituting in equation (9) for equations (5)
and (6), radial displacement u is given by
au
-zz -y [-6c(l -~)(r-4-2r0-2r-2)-6cr-4~] cos 20 . .
i-‘r _c
i.e.
u= - F [6c(l-~)(+r-~-2r~- 2r-1)+2cr-3v] cos 29+constant

Since u --f 0 as r -+ co, lonstant =O, so

u = -v [r-3-6(1-v)rO-2r-1] cos 28
C
hence

a,, =~(l+v)(5-6v)cos2R or Ug = ti, cos 28 . . (11)

c
where + z& represent the maximum displacements.
The coefficient of ground reaction is defined as
h = 4r = f-3 (121
uo = (1+~)(?6v)r,
. . ’ ’ ’ . .
compared with A= 3&/(1+ v)ro deducible from Morgan. For v =0*25, h = 0.685 EC/r,, com-
pared with A= 24 EC/r0 by Morgan and for v =0*5, A= Eo/ro compared with h = 2EJro by
Morgan.
Taking into account the stiffness of the lining and the loading transmitted to the ground
around the extrados, starting from an applied normal loading to the lining

p = y,-%(l-cos28) . . . . . . . . (13)

it can readily be shown (Morgan, 1961) that, from consideration of change in curvature around
the tunnel, the induced maximum bending moment
EI
A4m*x = +380 2 2 . . . . . . . . . (14)
77 r0
THE CIRCULAR TUNNEL IN ELASTIC GROUND 119
But the corresponding maximum moment applied by the ground loading (equation (13)) is
+ p0T2r0’
-
6
Note that E should be replaced by E/(1 - IQ”)for a continuous lining.
The reduction of zi,, resulting from the stiffness of the lining leads to the following relation-
ship between AI,,,,, and pO
h 3r 4 pOv2r02
M ,,,1+*=*
( !
6 . . . . . . . (15)
i.e. from equation (12)
por,2~2EZ(1 + v)(5 - 6~)
M (16)
msx = ’ 6EZ(l+v)(5-6~)+27~r,,~E, ’ . ’ ’
where equation (14) permits evaluation of Q,, corresponding to the value of Mm,, reduced on
account of passive resistance of the ground.
The stiffness ratio, R, (Muir Wood, 1970) represents the ratio of the stiffness of the tunnel
lining (to deformation in the ‘elliptical’ mode) to that of the surrounding ground. Thus,
R = 3EV + “)(5 - 6~) _ 9EZ
S E crl3r3 (17)
0 hq3r04 . . ’ .
and the reduction in bending moment to be carried by the lining is immediately apparent in
relation to its flexibility, since equation (16) may be written
M,,,,, If:&~oro2r12[&l(1 +&)I . . . . (18)
Lyons and Reid (1974) provide some typical values for I and EC.

EFFECT OF SHEAR FORCES BETWEEN GROUND AND LINING

For simplicity, shear forces between the ground and the lining have so far been neglected,
and it should be noted that this always leads to conservative solutions. However, it is known
that appreciable shear may be developed (Muir Wood, 1969) and the theory is not greatly
complicated if an explicit value for ground/lining shear stress is inserted and a means provided
for establishing its compatibility with these criteria
(a) its maximum value must not exceed permissible shear stress between tunnel and
ground
(b) the related circumferential movement of the ground at r = ro, assumed to be caused by
drag as the ring deforms, must not exceed that of the corresponding point on the deformed
ring. The introduction of shear stress between ground and lining implies that for
r=ro
Q-,.~
= T sin 20
where T is the value of shear stress at 0=rr/4. Thus, from equation (4)
d = -(3crom2++Tro2)
and hence,
ol. = -2(3crm4- 6cr-2ro-2-Tr-2r02) cos 20 . . . . (W
co = 6crm4 cos 20 . . . . . (201
and
7Vu = -(6cre4- 6cr-2ro-2-Tr-Zro2)sin28 . . . . . (21)
The equation corresponding to equation (10) may then be written
au
-- = -i [-(6cre-2- 12cr-2rO-2-Tr-2r02)(1 -v2)-66cr-4v(1 +v)] cos 28
i,
120 A. M. MUIR WOOD

i.e.

u=- Qp [(2rr-3--12cr-+o-2 -Trw1r02( I -~)+2cvr-~] cos 2O+constant

r
so

u. =y(i+v)[(5-6v)+~~r04(h)] ~0~20 . . . (221

and from equations (19) and (22)
(3 + Tro4/2c)E,
A = r,(l+v)[(5-6v)+T(l-v)r,4/2c] . . . . . (23)
Circumferential movement of a point on the ground/lining contact on radius 0=x/4, 3n/4 is
given by
- Tra
roFdr = 2F
uoo = ,I -(6cr-4-6cr-2ro-2-Tr-2r02) dr
s c c s

= 2~[2~~-3-6cr_‘ro-2_T~-lr,Zl,
c
i.e.

llO@= -~(4crom3+Tro) . . . . . . . (24)

Considering the lining as inextensible, then
0=14
where zloO= (dsO) - 3 where d, = [~$~z+r,2]1”d0
s o=o
r1 = r,+Oocos20
SO
*,4

UOQ = (4zio2 sin2 28+ro2+2rotio cos 20+tio2 cos 20)1’2 de-2

I 0
or since
u. << r.
.ff/l
uoe N o (ro2 + 2roziocos 20)lj2 de - nr,,/4 N (r. + ti, cos 28) d 0 - mo/4
J
i.e.
Trrcj 22, =ro UO
lJOON-----= -- . . . . . . (25)
4 2 4 2
Criterion (b) then gives, from equations (24) and (25)

v(4crom3+Tro) > Go/2 . . . (26)

(7
criterion (a) implies that
T + c’+$tan+’ . . . . . . . (27)
where j is the value of the normal pressure between ground and lining at 0=n/4, 3~14, i.e. the
mean value of this pressure between its extreme values at 6 =0, rr and at t?=rr/2, 3nj2. The
initial pressure between the ground and the lining, given by equation (13) comprises a uniform
component (pr-p0/2)( =p), which is unvaried by the deformation of the lining in bending, and
a variable component p0/2 cos 28. It may readily be shown by the use of equation (18) that,
in equilibrium, the variable component

g = “?”
R,/( 1 + R,)
 THE CIRCULAR TUNNEL IN ELASTIC GROUND 121
Thus the normal pressure on the extrados of the deformed tunnel may be represented as
1
p = ~+~cos20c . . . . . . . . . (28)

It will be noted that equation (13) allows constant c to be evaluated in equations (22) or (26)
in terms of the variable loading $ by way of equation (19), since for r = r. and 6’= 0
$12 = a, = (6crom4+2T)
allowing the substitution
c = (‘J/2 - 2T)/6ro - 4 ........ (2%
with a direct evaluation of
c = j/12rom4 . . ........ (30)
for equation (11) when T= 0 and since u. = Gocos 28, fi, may be evaluated from equation (22)
whence

T < ;(6v-1)/(5-9v) ........ (3 1)

From equation (23) the general expression for h is thus given by
3(8+ 57%
(32)
where R, may be evaluated by equation (17) R, = 9EI/hq3r04.

DIRECT COMPRESSION OF THE LINING

The assumption of linear stress-strain relationships of elasticity permits the effect of bending
and compression of the lining to be evaluated separately and the overall effect to be determined
by superimposition. The initial normal load between ground and tunnel is provided by
equation (13), and will be seen to vary between py at crown and (pV-po) at axis. The ratio
(pV--po)/pV (i.e. (j--po/2)/(@+po/2)) may be referred to as K. but the Author prefers to desig-
nate this ratio as N, a term more commonly used in rock mechanics, leaving K. for the axi-
symmetric case which is by no means the normal condition for tunnels in competent ground,
which has been subjected to tectonic activity. The constant component of loading is p
(equation (28)).
The general expression of stress in elastic ground around a circular tunnel causing a change
in uniform normal ground/lining pressure of magnitude dp is
CS=
= -a, = Apro2/r2
where initially oI.= uB=j.
The radial deflexion at r. is thus

Ar,= -
dl E+ v)ro = (p-Ap)(l+v)r,/E, . . . . . (33)
c
The change in pressure Ap will also give rise to compression of the tunnel lining yielding
Are = Apro2(1 - q2)/$E . . . . . . . . (34)
where f is effective thickness of lining. Thus assuming no initial clearance between ground
and tunnel equations (33) and (34) must give
(B - Ap)(l + v)/EC = Apr,(l - y2)/r]tE
or
122 A. M. MUIR WOOD

The compressibility factor R, is defined as the compressibility of the tunnel in relation to that
of the surrounding ground (this is not the same as the compressibility ratio of Peck et al. (1972)
which considers the cylinder of ground displaced by the tunnel).
Thus
R = r,E,(l -v1”)
0 . (36)
r)tE(l+v) ’ * ’ ’ * * * ’
and equation (35) may be rewritten as
dp = j/7/(1+ R,) . . . . . . . . . (37)
whence is apparent the capability of reducing loading on the tunnel lining by providing a high
compressibility factor.

GROUNDWATER
The effect of a tunnel at depth, providing drainage for groundwater, is to impose radial body
forces represented as the pressure gradient
du, 1
dr - Kr
where K has to be evaluated in relation to rate of flow and permeability. For the special case
where permeability k is isotropic and constant (i.e. unaffected by changes in ground loading)
and 2vq is the flow into unit length of tunnel,
duw
K = k]q and -&J . . . . . . (38)
If the permeability of the tunnel lining is effectively less than the ground, it must directly
support the full hydrostatic pressure, assuming there are no other ‘sinks’. Where the permea-
bility of the tunnel lining is greater than that of the ground, it is possible to determine the
loads transmitted to the tunnel on account of the groundwater. For the special case of
k = constant, equation (38) allows the condition of equilibrium to be stated thus
du, d
rz+-&r dr)-a, = 0 . . . . . . . . (39)

For the axisymmetrical case,

du
and Q = -- u
%= -Ti; r
thus
Eo-cr+r-
dce= 0
dr
or, since

i.e.
due
oO-a,+(l -v)rT-w% = 0

From equation (39) a stress function F may be selected such that ro, = F and
dF du,
0, =%+rdr
THE CIRCULAR TUNNEL IN ELASTIC GROUND 123

whence
,.2 F+ -=qr
daF+,.d_F-.
dr2 dr (l-v)k O
The solution is:
F = /f,.+E_qr loge ’
r 2k(l-V)
and

and

Since
a, = a0 = 0 where r-+00, A=0
For free drainage through the lining, q can be determined for a given tunnel radius r,,.
Without support from the lining a, = 0 where r = r. and
ro29log, r.
B = 2k(l-v)
whence
q(ro2 log, r. - r2 log, r)
up = . . . . . . . (40)
2k(l -v)r”
and

! [
a,=-
l-
ro2 log, r. + r2 log, r + r2
2(1- v)r” (41)
Clearly, the stresses caused by water pressures may be added to ground load stresses to estab-
1 ......
lish the overall condition of loading to be supported by the tunnel. A relatively impermeable
lining will reduce q in equations (40) and (41) and thus lead to reduced ground loading as
partial compensation for the fraction of the water load which is then carried directly by the
lining.

HIGHER MODE DISTORTIONS OF THE TUNNEL

The first mode of distortion of the tunnel has been represented earlier in the Paper as
r = r,+zi,cos20 . . . . . . . . . (42)
The general symmetrical case is given by
r = r,+ii,cosm0 . . . . . . . . * (43)
where m is an integer and m > 1.
From the foregoing, the basic Airy equation corresponding to equation (43) is
4 = (cre2 +d) cos me
for the boundary conditions that gB= up=0 where r --f co. By a similar process to that
represented by equations (2) and (11)
u0 = F (1 +v)[(4m2- l)-(4m2f2)v] cos MB
C
whence in combination with the equivalent to equation (5) the coefficient of ground reaction
may be determined
3(m2 - l)E,
h(=“r(r=ro~lUO) = (1+v)[(4m2_11)_((4m2+2)v~ * * * * (44)

In Fig. 2 Xis plotted against m as a continuous curve applicable to the special circumstances of
thin pipes and of tunnel deformations in orthotropic ground (Brownrigg and Wood, 1973)
124 A. M. MUIR WOOD

Fig. 2. Relationship between coefficient of ground reaction A and mode of deformation m for different values
of Poisson’s ratio v

which do not precisely correspond to the condition of m being an integer. It will be noted that
the assumption of h being constant for any number of deformation lobes, or for distortion
represented by a Fourier series, is unlikely to introduce error beyond the uncertainty of
the other physical factors.

APPLICATION OF THE METHOD

The design method described in this Paper is applied to changes of Ioading in the ground and
not as most other design methods (Hewett and Johannesson, 1922; Schulze and Duddeck,
1964) directly to the overburden pressure at the level of the tunnel. Clearly if a tunnel were
inserted into the ground without relaxation of the initial state of stress then the values of pv
and (pv-po) to be inserted into equations (13) would represent the initial conditions of vertical
and horizontal ground load; where the degree of relaxation is unknown, it is safe but possibly
very conservative to select these initial values. Consideration of a circular tunnel seen as a
hole being drilled through an elastic solid suggests that at the face the state of stress in the r, 0
plane will be intermediate between the intact and the perforated condition, e.g. the value of
1
Qr=ri$ = p = 2 p”-F [l-cos 281 . . . . . (45)
( >
and the loads on the tunnel lining can be derived by considering such an initial value. Panet
(1973) suggests that, as the face is not maintained effectively plane only about 3 (and not 3 as
suggested above) of inward movement of ground had occurred at the plane of the face. Panet
does not consider how this fraction is affected by the distance between the face and the leading
edge of effective tunnel support. Even when the problem is treated as an elastic one, in
reality, for cohesive ground, it is known that a visco-elastic condition should be considered.
At its simplest, this may imply a reduction of E, with time so that it would be unacceptable to
assume any appreciable reduction in effective load as suggested by equation (44). The three-
dimensional time-dependent state of stress around an advancing tunnel has a complexity be-
yond the scope of this Paper but reasonable upper and lower limits of the fraction of pv and
(pv-p,,) may be postulated for particular circumstances to check whether the range of uncer-
tainty may be so great as to merit its reduction by experiment and analysis in the interest of
economy.
It is generally far easier to control the stiffness ratio R, of the tunnel lining (equation (17))
than its compressibility factor R, (equation (36)) as for a given lining material, E, the ratio
between stress and strain at acceptable working load limit cannot easily be varied; but hinges
 THE CIRCULAR TUNNEL IN ELASTIC GROUND 125

Fig. 3. Diagrammatic representation of the benefit of an initial phase of high compressibility of a lining in
relation to its ultimate condition of loading, i;-Ap

can be introduced into the lining. For steel or concrete E is about an order of magnitude
too high for optimum load sharing in a wide range of ground calling for permanent support.
There are several methods which might be adopted to increase the compressibility ratio,
which are as follows.
Temporary peripheral jacks can be used in the lining, for example allowing a controlled de-
gree of closure of the ring by keeping a constant load on each jack. The period of jacking
must depend on the speed of tunnel advance and the time-dependent ground properties. This
method could be used for an expanded lining.
Frangible packing between ground and lining could be used to control maximum radial
loading. Such a method, which could be attempted by the use of aerated grout outside a ring
of lining, would require much care to avoid uneven loading.
Use could be made of a first stage compressible lining of a shotcrete variety with deliberately
reduced E.
Yielding arches and supports could be used, a long standing practice in mines (usually
hydraulic supports) and small diameter rock tunnels (usually depending upon friction joints),
placed before a permanent lining.
Compressible radial packings could be inserted in rings of lining to achieve reduced effective
value of E for the lining until a certain predetermined circumferential strain has been reached,
after which the value of E appropriate to the principal lining material would supervene (Muir
Wood, 1973).
The criterion for stability of the ground, however, must always satisfy the Mohr-Coulomb
condition
_(l +sin $‘)
u~(~~~~)~~(~_s~n~~)+c’cos~’ ’ . ’ ’ . .

Such expedients permit an analysis of loading, using equation (37) up to a certain value of
dr,,/r, provided by the deliberately increased compressibility ratio with subsequent loading for
the ultimate ring compressibility added thereto. Fig. 3 shows simply how this analysis may
be combined with a reduced value for E, to provide a more realistic provision for the effects of
quasi-elastic creep. For a high enough competence factor F, (Muir Wood, 1972) the lining
may alternatively be delayed until a large fraction of inward movement of the ground has been
completed. In all such expedients regard must be paid to the effects of relaxation on other
structures. For a simple approach to this problem see Muir Wood (1970).
126 A. M. MUIR WOOD

Where a lining is composed of segments, the stiffness at the joints may be appreciably less
than elsewhere and, for abutting joints, clearly the effective stiffness will be increased as the
ratio of hoop stress to bending stress increases. If the second moment of area at the joint is
designated as 4, the corresponding effective value of I for the ring, Z,, to be used in equation
(17) for determining R,, may be approximately calculated. While generally the load around
the tunnel will have two maxima and two minima, for four or fewer segments the existence of
joints may not affect ring stiffness. For an increased number of equal segments, say n, we
may assume as a first approximation a parabolic envelope to bending moment around a seg-
ment such that
Ze=Z,+(4/n)2(Ze,Z,n>4) . . . . . . . (47)
and thus, where Z, << Z for an expanded and articulating lining of, say, eight segments, Z, = Z/4.
For the special case of an elliptical tunnel in orthotropic ground whose elastic moduli are
approximately in the ratio of the axes of the ellipse (the greater ground modulus being parallel
to the major axis) it is possible, for a first approximation, to transform the problem into that of
a circular tunnel with appropriate factors for correction. Where the value of N corresponds
approximately to the ratio of orthotropic ground moduli, there will be no bending moment in
the lining. An approach of this nature has served as a first guide to the behaviour of elliptical
tunnels with horizontal axis approximately 1.5 times the vertical axis in Lias Clay and in
London Clay. For the general case of orthotropic ground Brownrigg and Wood (1973) per-
mit estimation of the value of effective E, to be assumed for best application of the simple
elastic case for the circular tunnel.
CONCLUSIONS
There is real merit in applying a simple method of analysis to a tunnel design problem which
may give far more insight into the relative importance of the relevant factors than immediate
recourse to computer methods, which may as a second stage far better serve in reproducing
a more valid representation of the problem.
An elastic method may serve to indicate upper and lower limits, even where cyclic tests on
the ground indicate considerable departures from a linear relation between stress and strain.
From multi-cyclical load tests, lines of slope of EC may be selected as corresponding to the
lowest and the highest reasonable value for EC.
The most awkward factor in any elastic method relates to the choice of value for Poisson’s
ratio v. A simple illustration may be provided by assuming that elastic ground of low per-
meability has, when unsaturated or for slow loading, values of Young’s modulus and Poisson’s
ratio, EC and v respectively. For rapid loading in plane strain of total loads a, and ue the
strains lp and ce will each be reduced by [(l +v)/&] U, where U, is pore-pressure (assuming a
value of unity for pore-pressure factor A), and thus the apparent behaviour in total stress terms
will imply that Poisson’s ratio is variable with values varying between 0.5 and v. Where
drainage occurs towards the tunnel excess pore-pressures will have a higher value away from
the tunnel and thus the distribution of load in the ground will be more favourable than that
assumed for the simple elastic case. Clearly the variation of v in effective stress terms merits
considerably better understanding; ground testing and sample testing should be designed to-
wards this end. Provided sufficient regard is paid to variability of the ground, jacking tests
with circular and elongated pressure plates of comparable area (approximately representing a
plane strain condition) may help in separately evaluating EC and v.
Generally, tunnel loading problems should be treated in effective stress terms. In the
absence of better information, a part of this Paper may serve to indicate the pattern of piezo-
metric pressures for cohesive ground, on completion of consolidation.
THE CIRCULAR TUNNEL IN ELASTIC GROUND 127
ACKNOWLEDGEMENTS
The Author acknowledges much benefit from discussions over the years with colleagues and
others concerned with the application of these design methods. He also wishes to thank Mr
Frank Cooper, MSc, MICE, for criticism and comment during drafting.

REFERENCES
Brownrigg, D. R. K. & Wood, W. L. (1973). Deformation of tunnel with variable ground reaction. Proc.
Znstn Civ. Engrs 55, 479-486.
Engelbreth, K. (1961). Correspondence on Morgan. Geotechnique 11, No. 3, 246-248.
Hewett, B. H. M. & Johannesson, S. (1922). Shield and compressed air tunnelling. McGraw-Hill.
Lyons, A. C. & Reed, A. J. (1974). Modern cast iron tunnel and shaft linings. 2nd Rapid Ext. and Tunnel.
Con& San Francisco.
Morgan, H. D. (1961). A contribution to the analysis of stress in a circular tunnel. Geotechnique 11,
37-46.
Muir Wood, A. M. (1969). Proc. VZZth Conference of Znt. Sot. of Soil Mechanics and Foundation Engng,
Mexico 3, 363-365.
Muir Wood, A. M. (1970). Soft ground tunnelling. Technology and Potential of Tunnelling, Johannesburg
1, 167-174, and 2, 72-75.
Muir Wood, A. M. (1972). Tunnels for roads and motorways. Q. Jnl Engng Geol. 5, 111-126.
Muir Wood, A. M. (1973). High relaxation tunnel lining. Provisional patent No. 36062173.
Panet, H. (1973). La stabilite des ouvrages souterrains-soutenement et revetement. Rapport de recherche
No. 28. Paris: Laboratoires des Ponts et Chaussees.
Peck, R. S., Hendron, A. J. & Mohraz, B. (1972). State of the art in soft ground tunnelling. 1st Rapid Ext.
and Tunnel Conf., Illinois 1, 259-286.
Schmid, H. (1926). Statische Probleme des Tunnel-und Druckstollenbaues. Berlin: Springer. (Engl.
trans. US Bureau of Reclamn. Tech. Memo. 262, 1931).
Schulze, H. & Duddeck, H. (1964). Spannungen in schildvorgetriebenen Tunneln. Eeton und Stahlbeton,
l-7.