Beruflich Dokumente
Kultur Dokumente
4, 397416
A basic engineering decision to be made in designing a veut realiser un tunnel dans de l’argile molle; il s’agit, en
tunnel in soft clay is whether or not the tunnel can be effet, de decider si un support interne sera ou non
excavated without internal support. The safety of con- ntcessaire pour l’excavation du tunnel. La securitb de la
structing a shallow heading in soft clay can be assessed in construction dune galerie peu profonde dans de l’argile
terms of the fluid support pressure which may be required molle peut itre evalute en fonction de la pression du fluide
to maintain stability. This can be estimated by means of support qui pourrait btre ntcessaire au maintien de la
the lower and upper bound theorems of plasticity. The stabilite. Cette estimation peut se faire a l’aide des
Paper considers three different shapes of shallow under- thtoremes a limites infirieure et superieure. L’article
ground opening relevant to tunnelling and upper and envisage trois formes differentes d’ouverture souterraine
lower bound stability solutions are derived for collapse peu profonde interessant le creusement de tunnels, et des
under undrained conditions. Solutions are also derived solutions de stabilite a limites inferieure et superieure sont
for assessment of the risk of blow-out failure caused by trouvees pour l’affaissement dans des conditions non
excessively high fluid pressures. Conditions are con- drain&es. Des solutions sont egalement trouvees en ce qui
sidered under which local collapse can occur, independent conceme l’evaluation du risque de rupture par eruption
of the cover above the tunnel. due aux pressions excessives du fluide. Les conditions
dans lesquelles un affaissement local peut se produire sont
envisagees, independamment de la couverture sit&e au-
Une decision fondamentale doit Ctre prise lorsque l’on dessus du tunnel.
INTRODUCTION
Figure 1 shows schematically the normal method of constructing a shallow tunnel in soft
ground using a tunnel shield. Soil is removed from the tunnel face either by hand or by cutters on
a machine that forms a complete unit with the shield. The shield is jacked forward using
hydraulic rams which react against the tunnel lining. As the tunnel progresses, new rings of the
lining are installed at the rear of the shield. When the tunnel face is excavated by hand, the shield
protects the tunnellers from a roof collapse and the most obvious threat to the tunnellers arises
from possible instability of the face. When compressed air is used in the tunnel the stability of the
face is increased since the air pressure replaces to some extent the pre-existing in situ ground
stresses.
Compressed air is also often used in conjunction with tunnelling machines but here an
increasingly popular alternative is the use of bentonite or clay slurry under pressure. There is
usually a bead on the tunnelling machine which leads to overcutting, so that the initially
excavated diameter of the tunnel is greater than the diameter of the shield. If the fluid or air
pressure is great enough then the soil will not close up around the machine and an annular gap
will remain.
Under these circumstances it is possible to idealize the process of tunnelling as shown in Fig.
2, where a circular tunnel of diameter D is shown being constructed with a depth of cover C. The
tunnel lining is regarded as rigid and in front of it the tunnel heading is represented by a
Discussion on this Paper closes 1 March, 1981. For further details see inside hack cover.
*University of Sydney.
t Cambridge University.
$ Scott Wilson Kirkpatrick & Partners.
0 Ove Arup & Partners.
398 E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEVIRATNE
Ground surface
I
~~~~_-~~~_~___ -----------
Bead
Tunnel shield
Section X-X
cylindrical cavity of length P in which there acts a uniform fluid pressure cr.. The ground has a
unit weight y and a uniform pressure a, acts on the soil surface: this may be due to a large flexible
footing or an overburden of water (or very weak material). This Paper investigates what tunnel
pressures err are necessary to maintain the stability of the heading for different values of the
parameters that have been defined (D, C, P,y,a,) and the strength of the ground. The collapse of
the tunnel heading will usually be a sudden event (caused, for example, by a sudden loss of tunnel
pressure) and hence it is appropriate to characterize the strength of the ground by its undrained
shear strength cU.In the following analysis it is assumed that c, is constant with depth, although
in practice c, will vary with depth depending on the history of the site. There are, however, many
situations where this assumption will be adequate and the methods of analysis that are used here
can be extended to cases where there is an arbitrary distribution of c, with depth.
Broms & Bennermark (1967) conducted experiments in which they extruded clay under
pressure through vertical circular openings and they considered field observations both where
failure had occurred and where stability had been maintained. They defined a stability ratio N,
equal to the difference between the total overburden stress in the ground at the axis of the tunnel
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 399
(before the tunnel is constructed) and the tunnel pressure divided by the undrained shear
strength c,.
N = [a,-a,+y(C +D/~)]/c, (1)
They concluded that if N is less than 6 then the opening will be stable. Their results are relevant
to the problem defined in this Paper for the particular case when P = 0.
In practice, the air pressure in a tunnel is often given a value to ensure that there is no flow of
water into the tunnel. This is achieved by applying a tunnel pressure greater than the pore-water
pressure at the tunnel invert. One possible approach to maintaining stability would be to set the
tunnel pressure equal to the overburden stress (i.e. N = 0). The problem with this approach is its
expense and also the health risks for tunnellers working at high air pressures. It is therefore
important to establish the minimum pressure necessary for stability.
It is appropriate to remark here that the Paper does not investigate whether the stability of a
tunnel heading improves or deteriorates in time. For a tunnel in heavily overconsolidated clay, a
reduction in tunnel pressure will be accompanied by the generation of negative excess pore
pressures which will dissipate as time passes. The clay around the tunnel will soften and after a
certain ‘stand-up’ time collapse is possible. Whether or not the heading collapses (and if it does,
the stand-up time) will depend on the geometry of the heading, the magnitude of the negative
pore pressures generated and the consolidation characteristics of the clay. In a tunnel in lightly
overconsolidated clay positive excess pore pressures will be generated and stability of the
heading will not deteriorate with time. Although an assessment of stand-up is of major concern
to tunnelling engineers (and is presently the subject of continued research at Cambridge
University) the Authors believe that the immediate undrained stability problem considered here
is a useful starting point when considering this phenomenon.
Stability solutions will be obtained using the limit theorems of plasticity. The soil is idealized
as an elastic, perfectly plastic material with a cohesion equal to c,. There is considerable
experimental evidence that this is a reasonable assumption for many clays. According to the
theory of plasticity, the collapse load for a particular configuration of loading on a perfectly
plastic body is unique, i.e. the load carrying capacity of the body cannot be changed by applying
the various loads in a different order. The lower bound theorem states that if any stress field can
be found which supports the loads, and is everywhere in equilibrium without yield being
exceeded, then the loads are lower than (or equal to) those for collapse. The upper bound
theorem states that if a work calculation is performed for a kinematically admissible collapse
mechanism then the loads thus deduced will be higher than (or equal to) those for collapse.
Since the tunnel pressure resists the collapse of soil into the tunnel it is a negative load in the
sense discussed above. The lower bound theorem will furnish a safe estimate of the tunnel
pressure required to maintain stability (i.e. higher or equal to that actually required) whereas the
upper bound theorem will provide an unsafe estimate. It is convenient to approach the solution
of the problem via a number of dimensionless parameters or groups which can be formally
derived by dimensional analysis, such as CID,P/D,cssjcu, gT/c,,
and yDfc,. For reasons which will
become clear later GJC, and or/c, can be replaced by the single parameter (a, - crr)/c,. Thus the
problem can be regarded as finding the value of (a, - r~r)/c, for limiting stability once the values
of the independent parameters CID,PJD and yDJc, have been fixed.
Since the solution of the complete problem defined in Fig. 2 is not straightforward (in
particular, it is difficult to find a good lower bound) three simpler cases are considered in turn;
400 E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEVIRATNE
ff.
1 4 1 4 4
!
Unit weighty
C
Undrained shear
strength c, Undrained shear strength cu
II 4. ,
1,
I--‘. . ” ‘7
D
k CT D
1 tf-. . . . . J.
Fig. 3. The plane strain unlined circular tunnel Fig. 4. The plane strain tunnel heading
each of which is relevant to the stability of tunnels or underground openings and from which
some conclusions can be made as to the more general situation. The first and second cases are
shown in Figs 3 and 4: both are problems of plane strain. The stability of the long cylindrical
cavity of Fig. 3 (Case 1) will determine the radial pressure a cylindrical tunnel shield must resist.
This case is equivalent to the case illustrated in Fig. 2 when the ratio P/D is large. The case
shown in Fig. 4 (Case 2) is a ‘plane strain heading’; the excavated volume is not cylindrical but
instead is similar to a long wall mining excavation. The third case to be considered is the Broms
& Bennermark problem which has the configuration of Fig. 2 when P/D = 0. Of course the
results for this case are directly relevant to the stability of the hand excavated tunnel referred to
at the start of the introduction. The three problems defined above will be referred to as Cases 1,2
and 3 in the remainder of this Paper.
c I =G T +2c ” In 2
0D
rJ@= fJ,+2c, (2)
?I3 = 0
using the normal notation. Outside the annular region there is an isotropic stress field a,. Thus
the lower bound solution is
Fig. 5. A lower bound stress field for the plane strain circular tunnel (yD/c, = 0)
another stress field which satisfies the stress boundary condition at the surface and extends to
infinity without violating yield. This construction is illustrated in Fig. 7.
Lower bound solutions for values of yD/c, from 1 to 4 are shown in Fig. 8. For low values of
C/D when yD/c, = 3 or 4 it was not possible to complete the solution as described above
without violating yield and these solutions are not illustrated in Fig. 8. For values of yD/c,
greater than 4 the stress characteristics overlapped above the tunnel and no solution is
Fig. I. Extension of the lower bound stress field for a plane strain circular tunnel (yD/c, > 0)
Fig. 8. Lower bounds for the plane strain circular tunnel under gravity loading
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 403
presented. It can be shown (see the section entitled Local collapse) that as yD/c, is increased a
point is reached when it becomes impossible to maintain stability regardless of the magnitude of
the applied uniform tunnel pressure. The overlapping stress characteristics or the failure to
complete the extended stress field for low values of C/D when yD/c, = 3 or4 does not necessarily
mean that these cases are inherently unstable. No attempt has been made to find alternative
lower bounds since the generated solutions cover most of the range of parameters which is of
practical significance.
Upper bound
Four upper bound mechanisms are shown in Figs 9 to 12. Mechanisms A and B are simple
‘roof and ‘roof and sides’ mechanisms each containing one variable dimension (or angle) and
were deduced from model tunnel tests at Cambridge University (Cairncross, 1973; Mair, 1979).
The procedure for determining the critical collapse load is to derive an expression for (6, - (or)
(involving the variable dimension or angle) and then to minimize the value of (a, - oT) with
respect to the variation of the dimension or angle. This can be done either analytically or
numerically (e.g. by a digital computer program).
The reason that a, and rrr appear only in the form (oS- rrT)in the upper bound calculations is
that since a kinematically permissible mechanism for cohesive material involves no volume
change then the decrease in area of the tunnel must equal the area of ground loss at the surface.
Hence the work done by the pressures in the work calculation will be (0, -Q=) multiplied by that
area.
Mechanism C has four variable angles in its specification and includes mechanisms A and B
as special cases. Mechanism D is a ‘roof, sides and bottom’ mechanism with three variable
angles. Figures 13 and 14 show the results of a numerical optimization to discover the critical
mechanisms for yD/c, = 0 and 3. In both cases mechanism C is more critical for low values of
C/D and is superseded by mechanism D for high values of C/D. It can be seen that the value of
C/D at which this changeover takes place is lower for the greater value of yD/c,. Figures 13 and
14 also show the lower bounds which lie close to the best upper bounds indicating that the exact
collapse loads have been closely bracketed. In the neighbourhood of the optimum upper bound,
changes in the variable angles lead to small changes in the collapse load. There is not much
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 405
Lower bound
1 1 8
1 2 3 4 5
CID
Fig. 13. Stability solutions for the plane strain circular tunnel (yD/c, = 0)
2
t CID
Lower bound
-12
t
Fig. 14. Stability solutions for tbe plane strain circular tunnel (yD/c, = 3)
406 E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEVIRATNE
- - -- - Lower bound
Upper bound
Fig. 15. Upper and lower bound stability ratios for plane strain circular tunnels
Fig. 16. A lower bound stress field for the plane heading (C/D = 4)
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 407
Fig. 17. A lower bound stress field for the plane strain heading (C/D = 2.875)
difference between mechanisms B, C and D for practical purposes (i.e. in estimated collapse
load) although the mechanics of deformation are very different. In general it seems that a
mechanism with one variable will yield an adequate upper bound (i.e. close to a good lower
bound) providing an appropriate pattern of collapse is chosen.
Figure 15 shows the bounds for different values of yD/c, all plotted as stability ratio N against
CID. For values of CID greater than 3 the upper and lower bounds of N do not change
significantly with yD/c,. Below C/D = 3 there is a larger spread but adopting the lower bound
for yD/c, = 0 as a criterion for deciding the tunnel pressure should be a safe procedure. This is
because it always corresponds to a lower value of N, or a higher value of oT, than those
prescribed by the lower bounds for higher values of yD/c,. It is important, however, to consider
also the possibility either of local failure at high values of yD/c, (see the section entitled Local
collapse) or of failure caused by ‘blow-out’ for very shallow tunnels (see the section entitled
Blow-out).
yD/c, = 0
Lower bound. Lower bound solutions for the plane strain heading can be constructed from three-
sided and four-sided areas of constant stress at or below yield. Two such typical fields of stress
are shown in Figs 16 and 17. These are systematically extended from two radial zones emanating
from the top and bottom of the heading (Gunn, 1980). These fields transmit a shear load from the
soil to the tunnel lining. Figure 16 shows 87% of the undrained shear strength being mobilized
on the soil-lining interface. This degree of mobilization would be reasonable for rough linings in
soft clay but solutions based on smooth linings would be more appropriate in other cases.
Solutions for a smooth lining can be obtained by adapting the analyses of Booker & Davis
(1973) or Ewing & Hill (1967) for the problems of bearing capacity near a vertical face and a V-
notched tension bar respectively. In particular the addition of an isotropic stress field to Ewing
& Hill’s slip line solution gives as a lower bound
(a, - a,)/~, = 2 + 2 In (C/D + 1) (4)
408 E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEVIRATNE
Fig. 18. An upper bound mechanism for the plane strain heading
tar
a-
6- _&_---
>
01 I
0 1 2
CID
3 4 5
Upper bound. When the upper bound mechanism illustrated in Fig. 18 is optimized with respect
to the three variable angles, the critical collapse load is found to be
If the work done by the self weight of the soil is accounted for in the work calculation for the
mechanism shown in Fig. 18 then the same optimum is obtained and
yD/c, = 0
Lower bound. The first lower bound stress field is shown in Fig. 20(a). Within the cylindrical
volume of soil which is the continuation of the already excavated tunnel the axial stress is equal
to cr. In planes perpendicular to the tunnel axis, the two principal stresses are equal to a,+ 2c,.
Outside this cylinder there is a radially symmetric stress field similar to that adopted for the
plane strain circular tunnel with the axial stress being some intermediate value between the
radial and circumferential stresses. Outside the larger cylinder of diameter (C +(D/2)) there is an
isotropic stress field a,. The lower bound is
(Go- rrr)/c, = 2 + 2 In (2C/D + 1) (7)
Figure 20(b) illustrates an alternative lower bound stress field. Within a hemispherical cap on
the end of the tunnel there is an isotropic stress field cr. Outside this hemisphere there is a
spherically symmetric stress field given by
0, = uT + 4c, In (2r/D)
00 = a, +2c, (8)
1
Outside the sphere (to which the surface is a tangent plane) there is an isotropic stress field a,.
The lower bound is
(a, - G&C, = 4 In (2C/D + 1) (9)
It can be seen from Fig. 21 that lower bound (a) allows higher loads to be supported for values
of CIDcO.86, whereas for values of C/D> 0.86 lower bound (b) is better.
Upper bound. The mechanism of Fig. 18 is adopted but the plane strain sliding blocks are
replaced by blocks with elliptical cross-sections. The length of the semi-axes perpendicular to
the plane of the diagram is equal to D/2. Optimization of the upper bound with respect to the
three variable angles leads to the line shown in Fig. 21.
E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEWRATNE
Section X-X
Section X-X
Fig. 20. Lower bound stress fields for the circular tunnel heading
The effect of the self weight of the soil can be added in exactly the same way as for case 2
leading to the interpretation of Fig. 21 in terms of stability ratio as well as (a, - c~)/c,, for the
weightless case. The relatively large gap between the upper and lower bounds for this case is a
reflection of the increased complexity of the three-dimensional problem.
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 411
24-
20 -
16-
0 1 2 3 4 5
CID
LOCAL COLLAPSE
It can easily be demonstrated that if the ratio yD/c, is sufficiently large then collapse will
take place for any value of uniform tunnel pressure. For Cases 1, 2 and 3 which were defined
earlier optimization of the mechanisms shown in Figs 22(a), (b) and (c) results in upper
bounds on yD/c, for collapse of 8.71,8-28 and 1096 respectively. Even though these mechanisms
involve no immediate subsidence of the ground surface, it is likely that this would be the first
step of a progressive failure which would eventually propagate to the surface. Since for a given
site the value of y/c, would be predetermined, the limiting value of yD/c, can be viewed as
specifying the maximum height of tunnel heading which can be constructed under uniform
tunnel pressure. To calculate safe values of yD/c, for the various cases lower bound solutions are
required. For the plane strain circular tunnel the evidence of the stress characteristics lower
bounds is that, for any value of C/D, stability is maintained provided that yD/c, < 2. Stability can
still be maintained for values of yD/c, between 2 and 4 when C/D exceeds certain minimal values
(see Fig. 8).
412 E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEVIRATNE
Fig. 23. A lower bound stress field for local collapse (Cases 2 and 3)
3,62c,/y(Pastot’s solution)
Fig. 24. A lower bound stress field for local collapse (Case 2) using Pastor’s solution
For both the plane strain tunnel heading and the circular tunnel heading, stress fields based
on Fig. 23 show that there will be no possibility of local collapse for tunnelling when yD/c, < 4
and uniform tunnel pressure equals y(C[D/2) (i.e. stability ratio N = 0). It is possible to improve
the solution for the plane strain heading by using Pastor’s (1978) lower bound for the critical
height of a vertical cut in cohesive material. Using a linear programming approach, an allowable
stress field with 208 zones of linearly varying stresses and a safe height of 3.63 c,/y was obtained.
Adopting this field as shown in Fig. 24 then the heading is stable for yD/c, = 5.63
when eT = y(C+O.3550). In contrast to the plane strain circular tunnel these solutions are
valid for any value of C/D.
BLOW-OUT
The problem of causing a failure by having a tunnel pressure which is too large has already
been mentioned. For Cases 2 and 3 the solutions already obtained can be directly adapted as
follows.
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 413
Upper bound
For both the plane strain and circular heading the result can be written in the form
Lower bound
If the directions of major and minor principal stress are reversed in the stress fields considered
for collapse-in and the gravity field added as before then equivalent results to the upper bounds
are obtained
B Ii
Force
m73-y+7(C+qa
+
Moment
703/i 2
of the tunnel. Indeed, adoption of the Broms & Bennermark criterion (N < 6 for stability) would
indicate that some shallow headings would be stable with no support pressure whereas an upper
bound calculation demonstrates that collapse would be inevitable without a supporting
pressure.
A possible objection to the lower bound or safe tunnel pressures suggested for the headings is
that they are based on stress fields where the tunnel pressures vary linearly with depth. Although
the adoption of the average pressure as safe seems reasonable, it is not satisfactory from the
point of view of the theory of plasticity. At first sight the adoption of the pressure at the bottom
of the heading as safe might seem to deal with this objection because now there is a greater
overall pressure supporting the heading than in the actual stress field. This process is equivalent
to a reduction in the safe value of N by &D/c,) but is again not satisfactory from the point of
view of plasticity theory because the greater constant support pressure is not proportionally
greater than the linearly varying support pressure. Indeed both these approaches could be used
to predict safe tunnel pressures when the results of the section entitled Local collapse indicate
that collapse would definitely occur. There is, however, a simple way in which the results already
presented can be adapted to predict safe tunnel pressures which are theoretically justified.
Figure 25 shows the loads acting when the effect of gravity and the surface pressure is
subtracted (according to plasticity theory the collapse load is not affected by this subtraction). A
linearly varying tunnel pressure is now the only load acting and it can be regarded as consisting
ofa tensile load equal to D[o, - a,+ y(C + D/2)] and a moment equal to yD3/12. The tensile load
is clearly responsible for failures involving the surface (as in Fig. 18) whereas the moment is
responsible for local collapse. These two loads can be regarded as independent and according to
plasticity theory the corresponding failure locus on the interaction diagram must be convex.
Figure 26 shows (for the particular cases of a plane strain heading with C/D = 2 and
C/D = 2.875) the three upper bound lines and lower bound points (A, B, C) on the appropriate
(nondimensionalized) interaction diagram. From the convexity property AB must represent
safe values of N, and this line can be regarded as a criterion for the reduction of N with
increasing yD/c,. For the case of Fig. 26(a) with yD/c, = 3 the average pressure criterion gives a
safe N of 4, i.e. o*/c,, > 1.5, whereas the above criterion gives a safe N of 2.3, i.e. GJC, > 3.2.
The practical tunnelling engineer will want to know what experimental evidence there is for
the solutions presented here. Tests on unlined circular model tunnels carried out at Cambridge
University (Mair, 1979; Seneviratne, 1979) have shown that the experimental collapse loads lie
between the lower and upper bound solutions that have been presented (see Fig. 27). A series of
STABILITY OF SHALLOW TUNNELS IN COHESIVE MATERIALS 415
6 1
(a) C/D = 413
Upper bound (blowout)
- - - -- -_--- --- i
(b) C/D = 2,875
5-
4-
3-
4 2-
b"
Upper bound
D Experimental
CID
-1 -
Fig. 27. Comparison of the stability solutions presented in this Paper with experimental observations (Case I:
yD/c, = 2.6)
416 E. H. DAVIS, M. J. GUNN, R. J. MAIR AND H. N. SENEVIRATNE
tests on model tunnel headings (Mair, 1979) has resulted in collapse loads which lie fairly close
to the lower bounds suggested for Case 3. For shallow tunnels (Case 1) the experimentally
observed collapse mechanism is usually very close to the optimum upper bound mechanism.
For deeper, i.e. (C/D)> 2, Case 1 tunnels and for most of the Case 3 tests, however, significant
differences are observed between the sliding block mechanisms presented here and the pattern of
deformation observed experimentally. The sliding block mechanisms suggest that a movement
of soil inwards near the tunnel is accompanied by an (approximately) equal settlement of the
surface. In practice, however, large movements near the tunnel are accompanied by much
smaller settlements of the surface, perhaps indicating a zone of plastically yielded soil near the
tunnel which is supported by a surrounding elastic region.
Nevertheless the Authors believe that the experimental evidence currently available indicates
that the results presented in this Paper can be used with confidence for the calculation of the
undrained stability of tunnels when C/D < 3.
ACKNOWLEDGEMENTS
The work presented in this Paper forms part of a wider programme of research into the
behaviour of shallow tunnels in soft ground at Cambridge University supported by the
Transport and Road Research Laboratory. The research has been conducted under the overall
direction of Professor A. N. Schofield and the Authors are grateful to him for several stimulating
discussions.
REFERENCES
Booker, J. R. & Davis, E. H. (1973). Some adaptations of classical plasticity theory for soil stability problems.
Proceedings of the symposium on the role of plasticity in soil mechanics, ed. Andrew C. Palmer, 2441. Cambridge,
1973.
Booker, J. R. & Davis, E. H. (1977). ‘Stability analysis by plasticity theory’, in Numerical methods in geotechnical
engineering, eds C. S. Desai and J. T. Christian, Chapter 21. London: McGraw-Hill.
Broms, B. B. & Bennermark, H. (1967). Stability ofclay in vertical openings. J. Soil Mech. Fdns Div. Am. Sot. Cio. Engrs,
193, SMl, 71-94.
Cairncross, A. M. (1973). Deformations around model tunnels in stiff clay. PhD thesis, Cambridge University.
Ewing, D. J. F. & Hill, R. (1967). The plastic constraint of V-notched tension bars. J. Mech. Phys. Solids, 15, 1155124.
Gunn, M. J. (1980). A note on the centred-fan stress field and its use in plasticity problems relevant to geotechnical
engineering. Cambridge University Engineering Department Internal Report.
Pastor, J. (1978). Analyse limite: determination numirique de solutions statistiques completes. Application au talus
vertical. J. Mtc. appl. 2, No. 2, 167-196.
Mair, R. J. (1979). Centrqial modelling of tunnel construction in soft clay. PhD thesis, Cambridge University.
Seneviratne, H. N. (1979). Deformations and pore pressure uariations around shallow tunnels in sqfi clay. PhD thesis,
Cambridge University.