ETH Library

Stability analysis for tunnelling with

slurry and EPB shields

Conference Paper

Author(s):
Anagnostou, Georg; Kovári, Kalman

Publication date:
1994

Permanent link:
https://doi.org/10.3929/ethz-a-010819320

Rights / license:
In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection.
For more information, please consult the Terms of use.
, I!

15-1

STABILITY ANALYSIS FOR TUNNELLING WITH SLURRY AND EPB SHIELDS

G. ANAGNOSTQU(l), K. KOV ARI< 2)

(I) Engineer, Inst. of Geotechnics, Swiss Federal Institute of Technology Zurich
2
( ) Professor for Underground Construction, lnst. of Geotechnics, Swiss Federal Institute of Technology Zurich

ABSTRACT

During the excavation of a tunnel through soft water-bearing ground, a temporary support is often required to maintain the stability of the working
face. In a slurry shield this support is provided by a pressurized mixture of bentonite and water. For earth pressure shields, the supporting medium is
the excavated soil. Slurry and EPB-shield tunnelling has been successfully applied worldwide in recent years. Under extremely unfavourable
geological conditions, however, face instabilities may occur. The present paper aims at a better understanding of the mechanics of face failure. It is,
therefore, necessary to consider not only ideal operational conditions but also undesirable, yet sometimes unavoidable effects such as bentonite
infiltration (in the case of slurry-shield tunnelling) or seepage-flow towards the tunnel face (in the case of EPB-shield tunnelling). For the case of
slurry shield, the paper in hand will attempt to cover the complex interrelations between shear strength and ground permeability, suspension
parameters, slurry pressure and the geometric data of the tunnel. Attention will be paid to the time-dependent effects associated with the infiltration of
slurry into the ground ahead of the face. Related topics, such as the stand-up time and the effect of the excavation advance rate, will be quantitatively
discussed. In the case of an EPB-shield, a distinction will be drawn between fluid-pressure and effective pressure in the chamber. The stability of the
tunnel face is controlled through the joint effects of these two parameters. The de-stabilizing effect of seepage-forces will be studied in detail. As
neither the effective support pressure nor the piezometric head in the work chamber are directly adjustable, the properties of the muck are of great
importance. The evaluation of tunnel face stability in actual cases and the choice of the best machine for given ground conditions require
consideration both of the geotechnical and the operational aspects.

1. INTRODUCTION

Tunnel construction in saturated soils is being carried out ever more

frequently with closed shields, a procedure which renders unnecessary
measures such as groundwater lowering, injections or ground freezing.
This modern tunnelling method permits the control of surface -=-1]
settlement and limits the risk of face failure through the continuous
support of the tunnel face during excavation. Lack of sufficient face
support can lead to tunnel-face instability (Fig. la) and, in extreme
cases, global failure through slip lines reaching the surface (Fig. lb).
-=-r (a)

In such a case, the machine excavates more ground than that

corresponding to the theoretical excavation volume. The collapse
propagates towards the surface creating a chimney and sometimes a
crater on the ground surface. The heading failure then results in
excessive subsidence and damage to overlying structures. When
tunnelling under a river, lake or the sea-bed, a direct channel will be
opened between the machine chamber and the water. Seepage flow
towards the face can have other undesirable consequences besides
(b)
tunnel face instability, for the piezometric head in the surrounding
ground decreases and the effective stresses increase. The ground then
experiences consolidation, leading to surface settlement.
Fig. 1. Typical patterns offailure
15-2

(a)

~ BENTONITE
(b)

(c)

Fig. 2. Closed-shield types: (a) Compressed-air shield (b) Slurry Fig. 3. Sliding mechanism (after Hom 1961)
shield (c) EPB-shield

Tunnelling using closed shields has, therefore, the following twofold Earth pressure shields (Fig. 2c) afford continuous support of the tunnel
aim: the support of the excavation face and the prevention of water face using the freshly-excavated soil, which completely fills-up the
seepage flow. Three (Fig. 2) completely different mechanical concepts work chamber (Fujita, 1981; Nishitake, 1990). The supporting
are available to attain these goals (Babenererde, 1988; Distelmeier, pressure is achieved through control of the incoming and outgoing
1987; Jacob, 1986; Stack, 1982). materials in the chamber, i.e. through regulation of the screw-
Through the introduction of compressed air into the work conveyor rotation speed and of the excavation advance rate. Extensive
chamber, the ground is supported either directly by the air pressure on experience with earth pressure shields has been gained in Japan, where
the tunnel face or indirectly through body forces due to air-flow. The this construction method was developed (Stack, 1982). In 1980 earth-
use of compressed air, however, introduces the risk of a blow-out, i.e. pressure shields were used for 27.8% of the total length of tunnels
of a sudden reduction of support pressure on account of rapid loss of constructed in Japan, and by 1985 this figure had risen to 68%
air. the air pressure must be at least equal to the highest prevailing (Nishitake, 1990).
groundwater pressure at the face, i.e., in a homogeneous ground, to the The three types of shield described above have one common
pore water pressure at the level of the tunnel floor. Above the tunnel aspect. This is the demand for stability of the tunnel face during
floor, an excess pressure thus prevails which may cause an escape of maintenance works on the cutter head or other activities at the face
air to the surface - either by leakage through soil pores or by an up- (e.g. removal of large blocks, tree trunks etc.). Usually compressed-air
heaval of the ground mass above the shield - particularly when the is applied in the chamber generally accompanied by additional
tunnel is shallow and its diameter large. measures such as grouting, freezing, relieving of pore water pressure
The so-called slurry shields use a bentonite suspension as or a combination of these measures. Certain types of machine also
supporting fluid (Fig. 2b). By means of an air cushion, the fluid have hydraulically controlled breasting plates available, although
pressure may be accurately applied and maintained without experience has shown these to be useful only to a limited degree. In
fluctuations (Babenererde, 1991; Becker and Sawinski, 1982). As the this contribution, only stability-problems during machine operation in
bentonite density is slightly greater than that of water, the EPB or slurry-mode will be studied. Neither compressed-air
compensation of the groundwater pressure will be uniform along the application, nor deformation-problems (e.g., due to drainage) will be
face, i.e. the excess fluid pressure in the crown will be small. The risk discused further.
of ground upheaval is, therefore, eliminated. Due to the high viscosity
of the slurry, the risk of an uncontrolled escape of fluid is reduced as
well - except for accidental losses through, e.g., unsealed boreholes or
2. THE COMPUTATIONAL MODEL
unexpectedly encountered foreign objects such as old pipes etc ..
Under optimum operational conditions, a filter-cake forms on the
The problem of explanating and predicting face instability using
tunnel face, acting like a membrane and inhibiting the infiltration of
statical calculations has already been addressed by several authors.
the suspension into the ground. For soils with exceptionally high
For a synopsis of the various analysis methods, the reader is referred
permeability or when the shear resistance of the slurry is low,
to Krause (1987), Balthaus (1988), Leca and Panet (1988) and Leca
however, the bentonite will penetrate into the ground to a certain
and Dormieux (1991).
degree.
Here, a simple model which idealises the actual three-dimensional
behaviour of the tunnel face will be used to carry out the numerical
stability investigations. As is common in stability analyses, the
system, the material properties and the loads will be modelled.
 15-3

~ < < < < < < < < < < <<<< < < < <
(d) The resultant effective vertical force of the prism at the
interface DEFC. This force is computed based upon silo-theory,
whereby, in the case of an EPB-shield, the seepage- forces acting
within the prismatic body must be taken into account. As for all
commonly-used calculation procedures (Krause, 1987), no shear force
will be applied to the horizontal limiting surface DEFC of the wedge.
(e) The force resulting from the supporting medium. The nature of
(a) this force depends on whether a slurry or an EPB-shield is used. This
point deserves attention and will be examined in detail in Sections 3
and 4.
The computation proceeds as follows: The known forces acting
upon the wedge are calculated for a specific collapse mechanism (i.e.
for a specific inclination w of the slip surface ABFE) and introduced
into the equilibrium and failure equations. The solution of the system
of equations yields the normal and shear forces acting on the inclined
slip surface, as well as the necessary effective support force. The

==teepage Fmce critical inclination Wcris determined by iteratively maximizing the

necessary support force or - for a given-support force - by iteratively
minimizing the safety factor.
(b)

Suspension 3. SLURRY-SHIELD-MODE OF OPERATION

Fig. 4. Stabilizing effect of the suspension: (a) without (b) with 3.1. The Stabilizing Effect of a Bentonite Slurry
penetration into the ground
The supporting effect of slurry has been extensively studied by a
number of authors over the past 40 years - in conjuction with the
As only limit equilibrium states are considered, sliding bodies will be design and the construction of diaphragm walls. For a comprehensive
defined by slip surfaces beginning at the face and reaching the surface. review refer to Xanthakos (1979). Recent research works- related to
In order not to limit the general nature of this study, the ground is the use of slurries in shield tunnelling - include those of Kntipfer and
considered to be homogeneous and isotropic. For both slurry and Meseck (1984) and Krause (1987).
EPB-shields, face stability in homogeneous ground will be assessed To prevent a seepage flow towards the excavation face, the
by considering a mechanism consisting of a wedge and a prismatic pressure Pb in the slurry must be in excess of the pore water pressure
body (Fig. 3). This three-dimensional statical system is based upon Pw in the soil (Fig. 4). The unit weight /1J of the slurry is higher than
the silo-theory (Janssen, 1895) and was first applied by Hom (1961) to that of water. The excess pressure Llp is, therefore, not constant over
the investigation of tunnel face stability. The given rupture mode the face of tunnel, but varies from a minimum at the crown to a
corresponds to the formation of slip surfaces, which may frequently be maximum at the invert. Due to the soil retained in the slurry during
observed during the failure of tunnel faces in shallow tunnels. excavation, a value of Yb =12 kN/m3 will be assumed in the numerical
As the deformation of the ground is disregarded, the soil is examples. It should be noted, however, that here, in contrast to the
idealised as a rigid-plastic material obeying the Mohr-Coulomb failure slurry technique applied to the construction of diaphragm walls, the
condition with cohesion c and angle of internal friction ¢. Then at density of the suspension is not so important from the point of view of
each point on the slip surfaces the mobilised shearing resistance 1' is face stabilization because the slurry is pressurized inside the working
given by area.
It is well known that the stabilizing force of the slurry depends
(1) essentially on the degree of penetration of the slurry into the ground
(cf. e.g. Xanthakos 1979): The less the slurry penetrates, the greater
where a and F denote the normal stress and the safety factor, the support force. In the borderline case of a practically negligible
respectively. Darcy's law is assumed to model the slurry-infiltration penetration distance (Fig. 4a), the hydrostatic pressure of the slurry
(in the slurry-shield-case) or the seepage-flow towards the face (in the acts as though the face were sealed by an impervious membrane (so-
EPB-shield-case). All computations are carried out in terms of called "membrane-model"). The support force results in this case from
effective stresses, i.e. drained conditions will be assumed. In actual the difference in hydrostatic pressure between the slurry and the
case-studies, however, both drained and undrained stability analyses ground water. The penetration distance is small immediately after the
should be carried out (Kovan and Anagnostou, 1994). filling of the working area with slurry, or when the soil is fine-grained
The wedge is acted upon by the following forces: or the slurry has the ability to form a seal, the so-called filter-cake. A
(a) The submerged weight of the wedge. filter-cake is formed when the suspension contains aggregated solid
(b) Seepage forces in the case of an EPB-shield (see Section 4.2). matter which is filtered out at the beginning of the slurry penetration.
(c) The resultant normal and shear forces along the failure A fine-grained low-permeable layer thus builds-up and increases the
surfaces ADE, BCF and ABFE. t1ltration effect. The difference between pw and Pb is lost inside the
filter-cake and a penetration of the bentonite into the soil stops.
 •
15-4

Since the excess pressure Jp is not constant over the tunnel face, the
distance of penetration will vary with depth along the tunnel face. It
should be noted that (3) gives the final distance of penetration, i.e.
after sufficient time has elapsed; time-dependent effects will be dealt
with later (Sections 3.4 and 3.5).
D=10m
Face stability will be studied firstly taking the simple case of the
«<
a.. 100
membrane-model (Section 3.2) and afterwards the more complex case
c H =10m
Hw =5m
involving infiltration of the slurry into the ground (Section 3.3).
a. 80
<:]
Q) 3.2. Results Based upon the Membrane-Model
:; 60
~
~ We carried out a parametric study for the example of a tunnel with a
a.. 40
diameter of D =10m; the overburden amounts to H =10m, the water
~
Q) 20 level is Hw =5 m above the crown. Fig. 5a shows the excess pressure
0
X
(a) Jp at the crown, which is necessary in order to retain equilibrium
w (safety factor F=1), as a function of the soil strength parameters c and
¢. At a given angle of internal friction, a low cohesion can be
Friction Angle <1>
compensated by a higher excess fluid pressure Jp - a main feature of
«1 100.-----------~------------~ slurry shield tunnelling. With a friction angle of approximately 40°,
a.. a: H = 5 m ; Hw = 5 m
the vertical tunnel face would be stable even when the ground is
ca. 80 b : H = 20 m ; Hw = 5 m
cohesionless and the excess pressure at the crown amounts to zero.
<:l .. ~,, g: H =10m; Hw = 5 m
This is due to the assumed unit weight of the slurry - 20% higher than
~ 60 .. ~g ~, c: H = 10 m ; Hw = 0 m
that of water.
~ 40
····~~ ...... ~:H=10m; Hw=10m Fig. 5b illustrates the interrelation between overburden H,
.... ••••..•~.~....... ..~_
a.. ..••. ::.-:llillll-.~ elevation of water table Hw. angle of internal friction ¢and necessary
~ 20 ·····":',,.,~....__
Q) ~ excess pressure Jp at the limiting equilibrium (safety factor F=l).
(b)
~ These results hold for the special case of a granular soil (c=O). It is
UJ 0~--~-----P----~--~--~~
40 readily apparent that neither a doubling of the overburden from 10 m
15 25 30 20
35
Friction Angle <1>
to 20 m (see curves b and g) nor a variation in the elevation of the
water table (see curves g, c and d) will influence considerably the
required excess pressure at a given angle of internal friction. It is,
Fig. 5. Limit equilibrium conditions for membrane model
however, interesting that the necessary fluid pressure increases with
the elevation of the water table but the excess pressure Jp decreases
(curve d is deeper than curve g): Due to the buoyancy inside the
Usually, suspended material is present in the slurry; aggregates (e.g.
prismatic block above the tunnel, the higher the elevation of the water
sawdust; see Fuchsberger 1975, Balmer 1992) may be added to the
table, the smaller will be the effective vertical stress at the interface
suspension when required - for example, in a uniform, coarse and
DEFC (Fig. 3).
poorly-graded ground.
Figure 5b reveals, furthermore, that a low excess pressure of 20
The stability of trenches with a more or less deep penetration of
kPa is enough to ensure the face stability of tunnels in gravels or sands
the slurry into the ground has been studied by Miiller-K.irchenbauer
with ¢-values typically greater than 30°. At an excess pressure of 70
(1972). The stabilizing effect of the slurry is attributed to the mass
kPa, the tunnel face would be stable even in a lake marl with c=O and
forces associated with the pressure gradient inside the suspension-
¢=15°. Such excess fluid pressures can easily be realized in practice,
saturated ground. The resulting support force of the slurry is obtained
for example through adjustment of the air-cushion pressure in a hydro-
by integrating the mass forces over the penetrated zone of the wedge.
shield. The results obtained by the membrane-model imply, therefore,
Due to its yield strength, the suspension comes to a standstill after the
that face instabilities should not occur when using a slurry shield -
penetration reaches a distance emax. The corresponding pressure
even when the ground has a extremely low shear strength. The
gradient !so is called here "stagnation gradient":
question arises, however, to what extent the membrane-model retains
Jp its validity when the slurry penetrates into the ground.
(2) !so = -e-.
max
3.3. Stability Assessment with Slurry Penetration into the Ground
where Jp and emax denote the excess fluid pressure and the final
distance of penetration, respectively. For a specific slurry composition
(bentonite type and concentration, additives etc.) and a specific subsoil 3.3.1. The support force
type, the stagnation gradient is an experimentally measurable constant.
According to (2), the penetration distance will increase linearly with The well-known reduction in the stabilizing effect of slurry once it has
excess pressure: penetrated into the ground can be explained in the following way.
When the suspension penetrates into the ground beyond the wedge-
block, it exerts its thrust on a soil zone which is not involved in the
(3) emax
sliding mechanism. Stabilizing body forces are thus withdrawn from
the wedge, and the resulting support force S (Fig. 6) will decrease
 15-5

gradually during penetration. By neglecting the variation of kPa. The authors carried out a parametric study concerning the effect
penetration-distance e over the tunnel face, the following fairly simple of fso on the safety factor F. In these computations, the variation of
expression can be derived: penetration distance over the face was taken into account. At any /so-
value, the critical inclination Wcr was iteratively determined. The
(4) ~= 1---e- if e <.D tan w; computational results are shown in Fig. 7. On the abscissa, the
So 2D tanw
stagnation gradient as well as the corresponding distance of
S D tanw penetration at the level of the tunnel axis is given. According to the
(5) So =--re- if e >D tanw;
membrane model, the safety factor amounts to 1.50. With a decreasing
where So denotes the support force of the membrane-model (i.e. at e stagnation gradient (i.e. an increasing penetration distance of the
=0). The support force accordingly decreases linearly with e during suspension into the ground), the safety factor gradually decreases. At a
the saturation of the wedge (i.e. at e<.D tanw) and sub-linearly critical stagnation gradient fer of 12 kN/m3 (i.e. a mean penetration
thereafter. In contrast to the membrane model, the resulting support distance of 3 m), the safety factor equals I; at lower stagnation
force depends on the value of w, i.e. on the specific wedge. The gradients the tunnel face cannot remain stable.
critical sliding mehanism (i.e. the value of Wcr) will, therefore, be
influenced by the degree of slurry penetration.

< < < <

10m Suspension
2,0--------------------------~

He STABLE :uNSTABLE

0
0 MEMBRANE-MODEL
«S
0 1,0 u. 1 5 -----S2"---------------:----------
(/)
.........
~·
Q)
I

(/)
0,8 (ij
Q) (f)
~
0 0,6 ;fa
u. 1 '0 -+-----.---....-------.......-=--r-------1
-e
0 0,4 500 150 50 15 5 1.5
a.
a.
::J
(/) 0,2 Stagnation gradient f50 (kN/m3)

0,0
0 0.2 0.4 0.6 0.8 1.0 0.06 0.20 0.60 2 6 20
Penetration Distance e/0 Penetration Distance e (m)

Fig. 6. Loss of effective support force due to suspension infiltration Fig. 7. Safety factor as a function of stagnation gradient
for a wedge with w=20° (f/J =37.5°)

Fig. 6 shows the graphical representation of (4) and (5) for a These results can be better elucidated with the help of the following
mechanism with w =20°. In this example, approximatelly 10% of the empirical formulae suggested by the German Standards DIN 4126:
force of the membrane model is lost per meter of penetration. After
the complete saturation of the wedge under consideration (ate= 4 m), (6) - 3__!j__
!so- dJO
the force will decrease more slowly. This is, however, unimportant
because the slurry will have already lost approximately 50% of its Eq. (6) links the stagnation gradient fso to the slurry type,
stabilizing effect and, furthermore, because the respective fluid losses characterized by its yield strength 'rf, as well as to the subsoil type,
are in practice intorelable. characterized by its effective size d 1o, i.e. the grain diameter at which
10% of. the soil weight is finer. The extent of slurry penetration does
3.3.2. The safety assessment not, accordingly, depend on the complete particle size distribution but
is governed by the finer particle fraction. For the theoretical and
Consider again a tunnel (D=H=10 m; Hw=5 m) in a granular soil empirical background of (6) refer, e.g., to Mtiller-Kirchenbauer (1972)
(gravel) with an angle of internal friction of f/J=37.SO - the highest and Xanthakos (1979). As far as the yield strength 'fJOf the suspension
value for preliminary designs suggested by the German Standards is concerned, the bentonite concentration is essential (see Krause
(DIN 1055). At a given excess pressure, the final distance of 1987, DIN 4127). Table I shows the concentrations as well as the
penetration will depend solely on the stagnation gradient/so (3). The assigned y-values which will be used in the examples offered; for the
fluid excess pressure at the crown of the tunnel is chosen to be L1p=20
15-6

design of a tunnel, of course, the actual relation between yield strength

and concentration has to be experimentally determined.

TABLE 1. Parameter Values

Soil:
Dry unit weight 'Yd (kN/m3) 20.0
Submerged unit weight 1 (kN/m3) 12.0
Porosity n 0.20
Strength parameters c, ¢ variable
Permeability k variable
1 ,0-r-----,---r-------r--L.-.....---l......l
Unit weight of slurry:
0,06 0,2 0,6 2,0 6,0 20
Including aggreg. solid matter 'i1J (kN/m3) 12.0
Without aggregates rs (kN/m3) 10.3 Grain Size d1 o (mm)
Yield strength 'l'f(Pa) of slurry
(cf. Krause 1987, DIN 4127):
"SAND" "GRAVEL"
With 4% Bentonite 15
With 7% Bentonite 80
Dynamic viscosity Jlb (cP) of slurry Fig. 8. Safety factor as a function of characteristic grain size d 10 ( cp
(after Xanthakos 1979): =37.5 °,' Curve A: Lip = 2a kPa, 4% bentonite; Curve B: Lip = 4a kPa,
With 4% Bentonite 3
4% bentonite; Curve C: Lip= 2a kPa, 7% bentonite)
With 7% Bentonite 7

3.3.4. The effect of excess pressure ~p

On account of (6), it is possible to represent the safety factor as a
function of the characteristic grain size d 1a (Fig. 8). On the abscissa,
the grain size range for gravels and sands is given as well; a d 1 a Curve B in Fig. 8 holds for a bentonite concentration of 4% and an
-value of 0.60 mm, for example, could belong to a poorly-graded excess pressure of 40 kPa - twice as high as curve A. Raising the
medium sand with a small silt or clay fraction. The curves A, B and C excess pressure, accordingly, causes an increase in safety, but only in
in Fig. 8 correspond to different values of bentonite concentration and fine porous soils - i.e. in soils in which the lower pressure would be
of excess pressure Lip at the crown of the tunnel, i.e. to different enough as well. When the characteristic grain size is bigger than
values of those parameters which can be directly controlled by the approximately 2 mm (i.e. poorly graded gravel without a fine-grained
engineer in practice. fraction), increasing the fluid pressure will only cause further
infiltration and fluid loss.
The considerable influence of excess pressure on face stability in
3.3.3. The effect of the characteristic grain size d 10
the case of a small d1a -value can easily be understood: in fine-
grained soils, the membrane model represents a reasonable
Let us focus our attention on curve A in Fig. 8 (excess pressure Lip=20
approximation (Fig. 7), i.e. the effective support force S is only
kPa, bentonite concentration 4% ). At low d 1a-values, the deviation
slightly lower than the support force S0 of the membrane model which
from the membrane model is negligible; the safety factor in this range
increases linearly with excess pressure. In a coarse porous medium,
remains approximately constant. In a subsoil with a d1o-value which
however, increasing the excess pressures causes a proportional
is smaller than the grain size of medium sand, the slurry acts as though
increase of the force S 0 , but at the same time a decrease in the ratio
the face were sealed, actually regardless of whether or not a filtercake
S/So due to the deeper penetration (see Fig. 6). In the final analysis,
has built up. As a matter of interest, in a subsoil in which the
the support force S remains constant and, consequently, the safely
conditions for the formation of a filtercake are fulfilled, the filtercake
factor as well.
will be statically unimportant.
To sum up, the deeper penetration of slurry into the ground
Soils with a d 1 a-value higher than 0.60 mm occur rather
represents a serious safety risk that cannot be compensated by raising
infrequently. We will study, however, the safety in the range
the fluid pressure. In extremely coarse and poorly-graded soils, an
d J0>0.60mm as well because we wish to exlore the limits of slurry-
important technique for stabilizing the face - the control of the fluid
shield tunnelling. As shown by curve A, a steep decrease in the safety
pressure- becomes ineffective.
factor occurs at d 1 a-values in the range of coarse sand (0.60-2.00
mm). At a characteristic grain size in the magnitude of fine gravel, the
safety factor becomes equal to 1; in a coarser subsoil, heading failure 3.3.5. The effect of bentonite concentration
will occur.
In the example discussed above, the concentration of bentonite is The difference of curve A and C lies in the differing bentonite content
4% and the excess pressure 20 kPa. The question then arises as to (7% and 4%, respectively). Accordingly, by selecting a higher
what extent the stability of the tunnel face can be increased by raising bentonite concentration, the grain-size range of soils which can be
the excess pressure or the bentonite concentration. supported by a bentonite slurry becomes- in this example - larger by
an order of magnitude. The bentonite concentration influences
 15-7

significantly the heading stability in a poorly-graded, coarse subsoil -

actually, an empirical fact (see Babendererde 1991).
As a comparison of curves A and C reveals, increasing the
bentonite content has little effect when the soil is fine-grained. From
the point of view of face stabization, therefore, there can be no
X
objection to the use of a slurry with a low concentration of bentonite;
such a suspension is favourable due to the easier separation and better e(z,t)
handling of the excavated material. Conflicts may occur, however, 4
E
when the ground consists of different layers with extreme variations in
(J)
the finer fraction. (J) 3
0
c
«S
3.4. The Stand-up Time of the Tunnel Face (ij
0 2
In Sections 3.2 and 3.3, the stability of the tunnel face was studied for c
0
two borderline cases concerning penetration distance (e=O or ·~
(i)
e=emax). Penetration distance, however, increases gradually over the c(J)
course of time. Face stability is, therefore, time-dependent as well, i.e., Q..
0
the safety factor decreases gradually from its maximum value 0 5 10 15 20
(corresponding to e=O) to its minimum value (corresponding to
e=emax). Subsequently, time effects will be studied for the simple Time (hours)
case where excavation is interrupted, e.g., when installing the lining
segments. Fig. 9. Penetration distance as a function of time (4% bentonite
according Table 1, d10 = 6 mm, k =10-4m/sec)

3.4.1. The Penetration Distance

From (7), (8) and ( 10), one obtains the following expression for the
In the first step, we compute the time-development of the penetration
infiltration velocity de/dt :
distance based upon the theory of multiphase flow of immiscible
fluids (Bear 1977). Concerning the stability of the tunnel face, only de = k f.lw ( !1p(z) _!so ) .
the area close to the face is important. It is, therefore, reasonable to ( 11)
dt 11 J.lb 1\-v e(z,t) '
model suspension infiltration as a one-dimensional process.
In Fig. 9, e(z,t) denotes the penetration distance at time t and or, due to (2):
elevation z. The hydrostatic pressure at the tunnel face (x=O) and at the
suspension-water interface [x=e(z,t)] is equal to Pb(Z) and Pw(Z), de = k J.lw !so ( enuu{ z) -e(z, t) )
(12)
dt 11 J.lb 1\-v e(z, t) ·
respectively. The displacement of the interface is given by the mean
microscopic velocity de/dt of the suspension: Accordingly, infiltration velocity deldt decreases with increasing
de_ q(e,t) penetration distance e and it becomes equal to zero when e=emax·
(7) dt- n This is due to the gradually decreasing pressure gradient. Eqs. (11)
and (12) illustrate the importance of ground permeability and
where n and q denote the porosity and the (macroscopic) filtration
suspension viscosity.
velocity, respectively. The filtration velocity q is equal to zero when
Integrating ( 11) and taking into account the initial condition
the pressure gradient {)p!()x is lower than the stagnation gradient/so; at
e(z,O)=O leads to the following expression:
higher pressure gradients, the validity of Darcy's law is assumed:
11
(13) t= J.lb 1\-v { - e - f1p In ( 1- e !so ) } ;
(8) q = k Jlw .!_ ( I * I -!so), if I* I >!so ;

*
k J.lw !so !so !1p
Jlb,.....,
(9) q = 0, if I I ~!so ; or, on account of (2):

where k is the permeability of the ground with respect to water; f.lw (14) t = n J.lb 1\-v !1p { - _e_- In ( 1- _e_)}.
und J.lb denote the dynamic viscosity of water(= 1 cP) and bentonite k J.lw fso2 emax emax
(see Table 1), respectively; and 1\-v is the unit weight of water. For the These equations give the time t which must elapse in order that the
sake of simplicity, both ground and suspension are assumed to be suspension infiltrates up to a distance e. Eqs. (13) and (14) cannot be
incompressible. In this case, the filtration velocity is spatially constant solved in a closed form with respect to e. One can easily verify,
((}qlttr=O) due to the one-dimensional mass conservation equation. however, that the penetration distance reaches asymptotically emax
Consequently (8), the pressure decreases linearly within the when t~oo. Fig. 9 shows the graphical representation of (13). It is
suspension-saturated area [0 ~ x ~ e(z,t)], i.e. the pressure gradient is apparent that, after an initial phase with rapid infiltration, the
constant: penetration distance increases more and more slowly up to its final
~ Pw(z) - Pb(Z) f1p(z) value (here 4 m). In this example, the suspension will infiltrate up to
(10) 1-2 meters within a few hours- provided that, despite of the high fluid
dx e(z,t) e(z.t)
loses, fluid pressure in the working area can be maintained.
15-8

face may, therefore, be more critical than the ones studied in this
paper (Fig. 3).
By taking into consideration the time-dependency of the
0 penetration distance and the pressure gradient, we compute the safety
0 factor as a function of time. Fig. 10 shows the computed relation for a
tf 1,5
tunnel with the geometric data D=H=lO m and Hw=5m and for two
~ different excess fluid pressures. As in the examples of Section 3.3, the
1ti ground is assumed to be cohesionless with a friction angle of 37.5°.
en
1 ,o+-.....,._..,..._ _...jh-_,...........,...,.,._...,. The safety factor decreases gradualy from its maximum value under
the membrane model (approximately 2 for ~p=40 kPa). The initial
0 2 3 4 5 6 7 8 safety margin vanishes after a critical time-petiod lcr. i.e. at lcr the
Time (hours) limit equilibrium is achieved. Accordingly, tcr represents the stand-up
time of the slurry supported tunnel face. Of course, such a stand-up
Fig. 10. Safety factor as a function of time (4% bentonite according time exists only when the face is instable at the maximum penetration
Table 1, d 10 =6 mm, k =1o-4 mlsec, granular soil with cp =37.5°)
distance emax, i.e. only when the stagnation gradient !so is lower than
the critical gradientfcr (see Fig. 7). Fig. (10) reveals, furthermore, that
the higher the excess pressure, the longer the stand-up time. This can
be explained on the basis of eq. (15). Neglecting the variation of the
5 days
excess fluid pressure L!p over the tunnel section and replacingfs by the
Q) 1 day critical gradient/cr. eq. (15) gives the stand-up time lcr; one can
E readily verify that lcr increases linearly with the excess pressure L!p.
i= 8 hr
a. Obviously, permeability has a decisive influence on the stand-up
:::> 3hr
time. The lower the permeability, the slower the infiltration of the
-oc 1 hr suspension and consequent loss of effective support force and thus the
tU 30min
en 10 min
longer the stand-up time will be. Based on a simple dimensional
analysis, one can readily verify that the stand-up time is proportional
to the reciprocal value of permeability: A one order of magnitude
2min
higher permeability results in a correspondingly lower stand-up time.
10-5 10 4 10 -3 10-2
This relationship is represented by a straight line in a double
Permeability (m/sec) logatithmic plot (Fig. 11 ): In this example (cohesionless ground and
insufficient fines), the tunnel face will remain stable during an
Fig. 11. Stand-Up Time as a function ofpermeability (4% bentonite
interruption in excavation of up to several hours, provided that the
permeability is lower than 1o-4 rnlsec. In highly-permeable ground
according Table 1, d 10 =6 mm, granular soil with cp =37.5°)
(e.g. k=to-2 rnlsec), face instability will occur after a few minutes.

3.4.2. Face Stability 3.5. Face Stability during Continuous Excavation

The support force of the suspension is obtained by integrating the In this case, the infiltration of slurry takes place simultaneously with
pressure gradient fs over the suspension-saturated area of the wedge the removal of ground at the face. The penetration distance is,
(see Section 3.3). It should be noted that the extent of this area, as well therefore, governed not only by the filtration velocity deldt but also by
as the pressure gradient, are time-dependent. From (13), one obtains the advance rate v, i.e. the infiltration is partially compensated by the
the following relationship: continuous excavation. Subsequently, only the simplified case with a
constant advance rate v will be considered.
(15) t = L1p(Z) ll f.ib 'Yw { _/so _ In ( 1 _ /so ) j.
k f.iw fso2 fs Is
3.5.1. The Penetration Distance
The pressure gradient at an arbitrary time t can be computed
iteratively from (15). One can readily verify thatfs decreases over the Fig. 12 shows schematically the infiltration velocity deldt as a
course of time and thatfs~!so when t~. Furthermore, apart from function of the penetration distance e. Note that the infiltration
the borderline case of t~oo, the higher the excess fluid pressure Jp(z), velocity deldt decreases with an increase in the penetration distance e
the higher the pressure gradient. Since L1p increases with depth (Section 3.4); at e~max, deldt will be equal to zero. According to
(Section 3.1), the pressure gradient will increase from its minimum (11 ), the infiltration velocity increases to infinity fore~, i.e. it does
value at the tunnel crown to its maximum value at the tunnel floor. not have an upper limit. Fig. 12 shows, furthermore, the constant
The variation of the pressure gradient over the tunnel face is more advance rate v.
pronounced at the beginning of the infiltration process (low !-values), Assume that at time t penetration distance and infiltration velocity
especially when the excess fluid pressure t1p at the crown is low. The are given by point A in Fig. 12. Accordingly, the infiltration velocity
fact that the pressure gradient is lower in the upper part of the tunnel is lower that the advance rate. In this case, the penetration distance -as
section suggests that an instability of the face would first occur there. measured from the instantaneous location of the tunnel face - will be
Sliding mechanisms involving a wedge in the upper part of the tunnel gradually reduced because the excavation advances more rapidly than
the infiltration of the suspension. Consequently, point A will move on
 15-9

the curve towards the left (lower penetration distance) and up (higher pressure gradient, the lower the safety· factor, and that face instability
infiltration velocity). After some time, point 0 will be reached. occurs when the pressure gradient is lower than a critical value fer
Subsequently, the penetration distance will remain constant because (Fig. 7). According to (17), however, an insufficient stagnation
infiltration velocity and advance rate are equal. The opposite will gradient fso can be compensated by a higher excavation advance rate.
happen if the infiltration velocity at some particular time is higher than Consequently, in coarse-grained and poorly-graded soils (i.e. in soils
the advance rate (Point B). In this case, the penetration distance will with fso<fcr ), the advance rate must be higher than the critical rate
increase over the course of time and, consequently, the infiltration vcr (Fig. 12) for the tunnel face to remain stable. The critical advance
velocity will decrease. Point B will move on the curve towards point rate is obtained from ( 17) through replacing fs by fer:
0.
J.lw
(18) Vcr = k (fer -!so ) ---''-'-'--
J.lb n IW
Accordingly, a permeability higher by one order of magnitude means
an correspondingly higher critical advance rate.

c 100
.E
E
_§_
CD
(ij
a:
CD
g 10
~
"0
<{
c
0
Penetration Distance e

Fig. 12. Relation between penetration distance, infiltration velocity

and excavation advance rate
1
w 1+---~--~~~~----~~~~~

1~ 1~ 1~
Permeability (rrv'sec)

One can see that, during continuous excavation, a quasi steady state Fig. 13. Critical excavation advance rate as a function of permeability
occurs in which the infiltration velocity is equal to the advance rate. (7% bentonite according Table 1, d10 =20 mm, granular soil with¢
The corresponding penetration distance is obtained from ( 11) by =37.5°)
setting deldt equal to v:
e 1
(16)
emax Fig. 13 shows the critical advance rate as a function of permeability
for the example of a granular soil with very high d 1o-value of 20 mm
(e.g. poorly-graded coarse gravel) and a high bentonite concentration
Accordingly, the higher the ratio vlk of advance rate to permeability, of 70 kg/m3 (material constants according to Table 1); for reasons
the smaller the penetration distance will be. The term on the right hand already mentioned in Section 3.4, the excess fluid pressure Jp does
side of this equation represents a reduction factor which expresses the not have any influence on fer and, consequently (18), on vcr·
influence of continuous excavation on penetration distance. From the According to Fig. 13, in a soil with a permeability of, e.g., 1o-3 rnlsec,
results of the previous sections, it becomes clear that the face stability the advance rate should amount to at least 17 mm/rnin. At higher
during continuous excavation will be higher than if there is an advance rates a safety margin will be present; at lower advance rates,
interruption in the excavation work. the tunnel face becomes unstable.

3.5.2. Face Stability

To quantify this effect, we need again the respective pressure gradient.

Through reformulating (16), one obtains the following interesting
relation:

(17)
J.lb v
fs =!so + n IW - k ·
J.lw
Accordingly, during continuous excavation, the pressure gradientfs is
higher than the stagnation gradient fso by an amount which increases
linearly with the dynamic viscosity J.lb of the suspension and with the
ratio of advance rate v to permeability k. Remember that the lower the
15-10

4. EPB-MODE OF OPERATION state potential field and it is based upon the following assumptions
concerning the hydraulic boundary conditions. At the tunnel face, a
4.1. The face support pressure in earth pressure shields constant piezometric head h :::; h0 is prescribed. In the extreme case,
the tunnel face represents a seepage face, i.e. the pressure there is
The term "support pressure" needs to be clarified in the case of earth atmospheric and the piezometric head is equal to the elevation of each
pressure shields. As the work chamber is filled with excavated soil point (h=z). At a sufficient distance from the tunnel face (far-field
under pressure, a distinction must be drawn between the total and boundary condition), the piezometric head corresponds to the
effective stress acting upon the face. Only the effective normal stress, groundwater table elevation (h 0 ). Assuming a waterproof tunnel
according to soil mechanics, can be denoted as actual support pressure lining, a no-flow boundary condition applies to the tunnel walls.
on the excavation face. This will be termed "effective support
pressure" and denoted in the following text by s'.
/)7/7///7/7///7/7/7//Y///////JJ.
I I
I
I

"1 E

T
0
C')
II
E .s=

1 UJ (~h=2 m)
(a}

p (kPa)
250 - - - - - T ~o- - -
Fig. 14. Face support when using an EPB-shield 200
150

What effects does the piezometric head h (Fig. 14) in the work ====100
50
chamber have from a stability point of view? If it is lower than the
piezometric head ho in the undisturbed state, then the groundwater 5 10 15 x (m)
will seep through the tunnel face. Thus, seepage forces will act (b)
towards the tunnel and could endanger its face stability.
Consequently, when using an earth pressure shield, both the Fig. 15. Seepage in the vicinity of the tunnel face (a) Contour-lines of
effective support pressure s' and the piezometric head h influence piezometric head; (b) Porewater pressure distribution along the
tunnel face stability, and should, therefore, be controlled and adjusted tunnel axis
according to the encountered hydrological and soil mechanics
conditions. Contrary to the situation with compressed-air and slurry- For all of the numerical examples in this paper, it will be assumed
shields, where only one parameter has to be regulated (air or slurry- that, in spite of drainage through the face, there is no drawdown of the
pressure, respectively), two such parameters exist for earth pressure water table. This assumes a continuous groundwater recharge by, e.g.,
shields which, in addition, may be difficult to control from a process rainfall or an adjacent river, lake or well. The potential field was
engineering point of view. As both parameters depend on the calculated using the three-dimensional finite element code HYDMEC
characteristics of the excavated ground, the way the ground is mixed (Anagnostou, 1991).
in the work chamber, the rotational speed of the screw conveyor and Fig. 15a illustrates the groundwater conditions (contour-lines of
the excavation advance rate, both geotechnical and operational aspects piezometric head) in the vertical plane of symmetry. In this numerical
can be seen to affect tunnel face stability. The interface between the example, the pressure in the chamber is atmosphelic. The increasing
construction equipment and the surrounding ground forms at the density of the potential lines close to the tunnel face shows the
tunnel face with the parameters s' and h prevailing there. The increasing value of the seepage forces. This is also clarified when
influence of the ground ends only on the conveyer belt with the plotting the pore water pressure along the tunnel axis (Fig. 15b).
transfer of the excavated material (Fig. 2c). In the following pages, the As the seepage forces are oriented perpendicularly to the potential
interplay of s' and h on tunnel face stability for given soil parameters, lines, the resultant seepage force acting on the wedge slopes slighlly
groundwater table and tunnel diameter will be numerically downward, while that in the prism above is practically vertical. The
investigated. The question of controlling of these two critical destabilizing effect of the seepage forces acting on the wedge may
parameters to assure stability will not, however, be treated here. thus be clearly recognised: An approximately holizontal load is
exercised on the wedge, while the vertical load from the prism is
4.2. Calculation of the seepage forces simultaneously increased.
In order to provide a better understanding of the face stability
Determination of the seepage forces calls for a numerical seepage- conditions, results of parametlic studies which were carried out using
flow analysis. Such a calculation yields the three-dimensional steady- the numetical model described above will now be considered.
 15-11

4.3. Results of parametric studies 2

r
Because of its great importance, the influence of seepage-flow was
777/---=- 77777777---===-
numerically investigated for two borderline-cases concerning
hydraulic boundary conditions (Fig. 16). In the first case, the
piezometric head h in the work chamber is equal to the elevation of 20m
the groundwater table (h=h 0 ), i.e. the work chamber and the ground
are in hydraulic equilibrium. In the second case, the pore water
pressure pin the chamber is assumed to be atmospheric (p=O, h=z). In ---r;J f8
case 1, no seepage forces occur. In case 2, the seepage forces obtain
their maximum values. Fig. 16 shows the necessary support pressure s' ~ .....y
at the limit equilibrium for the cohesion-range 0 < c < 70 kPa (other s' s' I
parameters: see Table 2). p=O
For a granular soil (c = 0), it may be seen that the necessary t? 140
c..
support pressure for limit equilibrium without seepage (case 1) is ==- 120
-(/)
about 30 kPa, and in the presence of seepage (case 2) is about 150
~ 100
kPa. The seepage forces necessitate accordingly an additional ::::J
(/)
(/)
effective support pressure of about 120 kPa. Approximately 75% of ~ 80
c.
the required support pressure in case 2 is due solely to the horizontal 1: 60
0
component of the resultant seepage force. Without any effective c.
c.
support pressure (s' = 0), a minimum cohesion is required whose value
::::J
(/)
40
Q)
may be found on the abscissa axis in Fig. 16. Accordingly, the > 20
·u
minimum cohesion should be equal to 12 kPa when the groundwater Q)
:t:
UJ
pressure is completely compensated by the water pressure in the
10 20 30 40 50 60 70
chamber (case 1). For open-mode operation (i.e., an empty work
Cohesion c (kPa)
chamber with s' = 0 and p = 0), the surrounding ground should have a
minimum cohesion value of approximately 70 kPa (Fig. 16, case 2).
The sliding of a wedge in the work chamber is not the only Fig. 16. Effective support pressure s' as a function of cohesion c
tunnel-face failure mechanism observed in the field. Failure can also ( ¢=30°): 1: Without seepage forces; 2: Max. seepage forces
take place in the form of shell-shaped soil bodies caving in at the face.
This is due to a low or completely non-existent soil tensile strength
and the immediate effect of the seepage forces. This failure-mode TABLE 2. Parameter Values
becomes relevant at high cohesions and low tensile strengths. Point Z
in Fig. 16 shows the cohesion value (here approximately 50 kPa) at Tunnel Diameter (m) 10
which the failure-mode shifts from shear failure (with a sliding wedge Overburden (m) 20
according to Fig. 3) to tensile-failure. For high cohesion values, the Dry unit weight Yd (kN/m3) 20
assumption of a low soil tensile strength is, nevertheless, questionable. Submerged unit weight 1 (kN/m3) 12
In the following investigation only the shear failure mechanism shown Cohesion c variable
in Fig. 3 will be considered. Friction Angle ¢ 30°
Fig. 17a and 17b show the necessary effective support pressures'
and the total support pressure s, respectively, as a function of the
water table elevation h0 . The average total support pressure s is higher In conclusion, the fluid pressure (or, equivalently, the head h) and the
than the effective support pressure s' by the amount of the pore water effective support pressure s' affect face stability in a different
pressure prevailing in the work chamber at the level of the tunnel axis. manners. Consequently, both quantities, s' and h, should always be
In case 2 (atmospheric water pressure in the chamber), the effective measured in the work chamber of earth pressure shields. The
and the total pressure are identical. The difference between s and s' in monitoring of s encounters, however, great difficulties.
case 1 is due to the piezometric head ho in the work chamber. Note Further results for border-line case 2 (max. seepage forces) and
that in case 1 (without seepage) the total support pressures is high, but different values for cohesion c are shown in Fig. 18. For example, if
the effective support pressure s' is low and, moreover, it remains the ground has a cohesion of 20 kPa and the groundwater table is only
unaffected by the elevation of the groundwater table h0 . a few meters above the tunnel roof, a very small effective support
Fig. 17 also shows that, for case 2 (maximum seepage forces), the pressure s' will be sufficient to stabilise the tunnel face- in spite of the
necessary support pressure for stability is only half as large as the seepage forces acting towards the face. The lowering of the
undisturbed groundwater pressure at the level of the tunnel axis. This piezomettic head in the ground (by pumping) thus represents a very
is a result of the decrease in the piezometric head due to drainage efficacious measure if the settlements induced by drainage can be
conditions (Fig. 15b). The generally represented conception according tolerated (Biggart et al., 1993).
to which the support pressure should correspond to the "sum of the It may be shown that it is not the absolute values of h and h 0 , but
groundwater and earth pressures" (Krause, 1987) is valid only for the difference (ho -h), which is important for the required effective
border-line case 1, i.e. only for the ideal case in which full hydrostatic support pressure s' (Fig. 19). From the slopes of the lines in Fig. 19, it
pressure is maintained in the work chamber. may be seen that a decrease in piezometric head difference (ho - h) of
15-12

about 10 m leads to a reduction in necessary support pressure s' of ~ 7777777777'

about 40 kPa. The distances between the lines show that the same
reduction in necessary support pressure results when the cohesion 20m
v
T
~~ 1
increases from 0 to 20 kPa (for example through ground
improvement). In conclusion, a decrease in hydraulic head difference
ho
of about 10 m is statically equivalent to an effective support pressure
of about 40 kPa or a cohesion value of 20 kPa. Seeing the great
importance of the piezometric head h, the principle of the so-called
10 ~y'
"water-pressure-type EPB shield" is immediately obvious. For this s' I
type of machine, the piezometric head should always be maintained p=O
through the application of pressurized water in the work chamber
200
(Abe et al., 1978; Fujita, 1981; Stack, 1982). I?
a..
~
-Ill

~
150
1 2 ::I
(/)
(/)
Q)
a 100
7777777 77777777/o~ 't:
8.
~
::I
(/)
Q)
> 50
·u
Q)

ffi

]i]~i
0
10 20 30 40 50
Water Level h 0 (m)
s' s' I
p=O
500 Fig. 18. Effective support pressure as a function of initial head h0 and
I? 450 of cohesion c (f/J=30°). Borderline case 2 witH max. seepage forces
a..
~ 400 (i.e., atmospheric pressure at the face).
-Ill

~ 350
::I
(/)
(/)

~
~
300
250
2 -1-- 77 777777777777

~
t:
8. 200
~
::I
(/)
150
(I)
> 100
-~
50
-~., ______________ _1
=
w
0 I I (a)
10 20 30 40 50
10m +
...
Water Level h 0 (m)
500 1 j__
...
450 s'
I?
a.. 400 180 c (kPa) =0
~ I?
a.. 160
.... ·······
(/)
350 ~
~
::I
(/) 300 -Ill 140
(/)
2 Q)

~ 250 :s 120 20

...······
(/)
(/)
t: 200
. 100
Q)

8. a
.. ··
~
::I
150 't: 80 40
(/)
100
8.
<a ~
::I 60
0 (/)
1- 50 Q)
40 60
0 (b) ·u> 20
=
Q)

10 20 30 40 50 L.U
0
Water Level h 0 (m)
0 5 10 15 20 25 30
Head difference h 0 -h (m)
Fig. 17. Effective (a) and total (b) support pressure as a function of
initial piezometric head ho (cohesionless ground with) f/J=30°): 1: Fig. 19. Effective support pressures' as a function of head difference
VVithoutseepageforces; 2:A1ax. seepageforces (ho-h) and of cohesion c ( f/J=30 °).
 15-13

problem of non-uniform distribution. of support pressure becomes

even worse in the case of a mixed tunnel face due to the widely
differing stiffnesses of the rock and soil layers (Eisenstein, 1993).
b) For a full work chamber, the cutting wheel not only separates
(a) the ground from the tunnel face, but also shears the compacted muck
during the rotation (Fig. 20b ). The larger the internal friction angle of
the muck and the effective stress in the work chamber, the larger the
shear resistance which must be overcome. Thus, larger effective
support pressures s' and larger internal friction angles result in greater
torque and in excessive cutter-wear (Babendererede, 1989).
c) For high support pressures and low clay percentages, arching of
the muck occurs at the entrance to the screw conveyor (Fig. 20c) and
inhibits further discharge (Fujita, 1981 ). If this is not noticed by the
operator immediately, the excavation continues with impeded
(b) discharge, and the muck becomes further compacted. It will be
obvious that a silty, sandy soil is especially susceptible to arching and
that the cutting wheel could then be brought to a complete halt due to
the very high necessary torque.

4.5. The maintainance of a high pore water pressure in the work

chamber

(c) For the reasons presented above, the effective support pressures' must
not exceed a certain upper limit. This is especially important for muck
with a large internal friction angle, e.g. when tunnelling in sandy soils.
Nevertheless, a low effective support pressure may not be sufficient to
Fig. 20. Problems caused by high effective support pressures' and assure tunnel face stability. Depending on the cohesion and the
high friction angle of the muck: (a) Uncontrollable support pressure elevation of the groundwater table, a higher or lower piezometric head
distribution; (b) Excessive cutter-wear and torque; (c) Arching at the must be maintained in the work chamber (cf. Section 4.3).
entrance to the screw conveyor. On what factors, then, does the piezometric head in the work
chamber depend? Consider the system represented by the machine and
the surrounding ground (Fig. 21). Let ko and k denote the permeability
4.4. Operational effects of a high effective support pressure s' of the ground and of the muck, respectively. The piezometric head at a
large distance from the tunnel face corresponds to the undisturbed
The effective support pressure s' cannot be controlled directly, but groundwater table h0 , and at the exit of the screw conveyor is equal to
rather indirectly through the mass balance in the work chamber. hA. The piezometric head difference h0 - hA is dissipated partially
Excavation advance rate and the rotation speed of the screw conveyor within the ground and partially within the machine.
If the muck is much more permeable than the undisturbed soil
determine the incoming and outgoing soil quantities in the work
chamber per time-unit. If the incoming quantity is greater than the (k>>ko), then the head difference will be dissipated mainly within the
outgoing quantity, then the average soil porosity in the work chamber ground ahead of the face. In the extreme case, the hydraulic potential
decreases and the effective stress s' increases. On the contrary, if too within the machine is equal to the head hA at the exit of the screw
much material is extracted from the working chamber (under high conveyor. The more impervious the muck, the larger the portion of the
screw-conveyor rotation speed and low advance rate), then the head difference which is dissipated within the machine, and the higher
porosity will increase and the effective stress decrease. Changes in the the piezometric head in the work chamber. In the border-line case of a
incoming and outgoing quantities cause porosity changes with time practicaly impervious muck (k<<k 0 ), the hydraulic head at the tunnel
and, thus, effective stress fluctuations. The stiffer the excavated face corresponds to the undisturbed piezometric head ho and decreases
material, the larger the fluctuations. Irregularities in machine operation along the screw conveyor to the value hA.
(e.g. varying excavation advance rate, discontinuous flow of the muck Two basic possibilities for maintaining a high water pressure in
within the screw-conveyor due to temporary blockage etc.) can lead to the work chamber result from these considerations: (a) through
large fluctuations in the effective support pressure if the muck has a maintaining a high piezometric head at the exit of the screw conveyor,
high stiffness modulus (e.g. sandy soil). A constant effective support or (b) through decreasing the permeability of the muck.
pressure is difficult to maintain for such materials. The first solution requires either maintaining of a soil plug or the
The muck exhibits a shear resistance which, for a given internal installation of a pump at the discharge port of the screw conveyor. For
friction angle, increases with the effective stress s'. If the shear the formation of a plug, various "water-cut-off'-devices have been
resistance is large, the following problems may occur: developed (Nishitake, 1990). Plug formation, however, presupposes a
a) Since muck with shear resistance does not behave like a fluid predominately clayey, compressible muck. In a sandy or gravelly soil,
the stress field in the work chamber, and thus the distribution of th~ a high pressure can be maintained only through a pump, but the
effective stress s' along the tunnel face, is not under control. This is a excavation advance rate may become dependent on the performance
non optimal situation from a stability point of view (Fig. 20a). The of the pump.
15-14

5. CLOSING REMARKS

1
The computational method presented in this paper is based upon the
wedge model of Horn (1961), together with existing knowledge
concerning the interaction between bentonite suspensions and the soil
and seepage-flow. In this way, one derives quantitative relations
ho between the shear strength parameters of the ground, the properties of
the supporting medium, the pressure within the work chamber, the

1 geometric data of a tunnel, and the safety factor. The results of the
parametric studies presented here accord with experience in the field.
Even when - in the absence of laboratory or field investigations - the
data necessary for a stability analysis are incomplete, the
computations are useful, for they improve understanding of the
complex interrelations involved and, therefore, enable a better
assessment of the residual risks or- in the case of face instabilities - of
(a) HIGH-PERMEABLE
SPOIL (k >> k 0 ) the effectiveness of various countermeasures.
X When using a slurry shield, soil permeability as well as the
dynamic viscosity of the suspension are decisive with respect to time
(b) LOW-PERMEABLE effects; in very coarse-grained soils, the excavation advance rate also
SPOIL (k << ko) exerts an influence. The effectiveness of slurry support depends
essentially on the possible infiltration depth of the suspension into the
ground. With increasing penetration distance, the effective support
force as well as the stability of the tunnel face - as expressed by the
safety factor - will decrease. Since infiltration takes place gradually
over the course of time, the safety factor is time-dependent.
When using an EPB-shield, a distinction must be drawn between
Fig. 21. Loss of head difference within the ground and along the the effective stress s' perpendicular to the tunnel face and the pore
screw conveyor: (a) High-permeable muck; (b) Low-permeable muck water pressure pin the work chamber. The stability of the tunnel face
is guaranteed through the joint effects of s' and p (or, equivalently, s'
and head h). The larger the piezometric head h, the smaller the
The addition of bentonite suspensions or polymers to the excavated necessary effective support pressure s', and vice versa. As neither the
material (conditioning) can reduce both its permeability and shear effective support pressure s' nor the piezometric head h in the work
strength (internal friction angle) (Babendererde, 1989; Nishitake, chamber are directly adjustable, the properties of the muck are of great
1990). Similar muck treatment problems may then be encountered as importance. Both geotechnical and operational aspects should,
for slurry shields. Recently, environmentally acceptable additives have therefore, be considered when assessing tunnel face stability in an
been developed for use in conditioning. actual case or when evaluating the optimal machine for given ground
The importance of thoroughly mixing of excavated material in the conditions .
work chamber is especially great with an inhomogeneous ground. As For several operational reasons, the shear resistance of the muck
seen in Fig. 22a, the overall permeability of the ground may be high should be kept as low as possible. Depending on the soil encountered
due to the presence of sand layers. Nevertheless, the muck may have a during tunnel excavation, conditioning by means of lubricants
smaller permeability than the undisturbed ground when the sand (rheological foams, polymers etc.) may be necessary in order to reduce
grains are dispersed within the clay matrix (Fig. 22b). If complete its tiiction angle. Furthermore, tunnel face stability should be achieved
mixing does not take place and the piezometric head in the work with the smallest possible effective support pressures'. This requires
chamber is too low, then such permeable layers unavoidably lead to that the head difference between the chamber and the ground should
tunnel face instability (a leaking out of the sand and breaking up of the be kept as small as possible. This can be achieved either through
cohesive layer). permeability-reduction of the muck within the working chamber
(additives, thorough mixing), or through additional measures (such as
installation of a pump at the exit of the screw conveyor).

SAND

CLAY

GROUND MUCK

Fig. 22. Reduction of muck-permeability due to mixing within the

work chamber
 •
15-15

REFERENCES (16) Distelmeier, H. (1987). Tunne-lauskleidungen hinter grund-

(1) Abe, T., Sugimoto, Y., Ishihara, K. (1978). Development and wasserverdrangenden Schildvortri8bmaschinen. Bauingenieur,
application of environmentally acceptable new soft ground 62, 487-492 (in German).
tunnelling method. In: Proc. Int. Symp. "Tunnelling Under (17) Eisenstein, Z.D. (1993). Large undersea tunnels and the progress
Difficult Conditions, Tokyo, 315-320. of tunnelling technology. In: Int. Symposium on Technology of
(2) Anagnostou, G. (1991). Untersuchungen zur Statik des Bored Tunnels Under Deep Waterways, 3.-5. Nov. 1993,
Tunnelbaus in quellfahigem Gebirge. Dissertation Nr. 9553, Copenhagen, Denmark.
Eidg. Techn. Hochschule ZUrich (in German). (18) Fuchsberger, M. (1975). Some practical aspects of diaphragm
(3) Anagnostou, G., Kovari, K. (1992). Ein Beitrag zur Statik der wall construction. In: Diaphragm walls and anchorages, 75-79.
Ortsbrust beim Hydroschildvortrieb. In: Tunnelbau-Symposium Inst. Civ. Eng., London.
"Probleme bei maschinellen Tunnelvortrieben ? - Geriite- (19) Fujita, K. (1981). Use of Slurry and Earth-Pressure-Balance
hersteller und Anwender berichten ", TU MUnchen (in German). Shield in Japan. In: Int. Congress on Tunnelling, Tunnel 81, Bd.
(4) Anagnostou, G., Kovan, K. (1994). The Face Stability of 1, 383-406. DUsseldorf. (Ed.: DUsseldorfer Messegesellschaft
Slurry-shield-driven Tunnels. Tunnelling and Underground mbH -NOWEA - in Zusammenarbeit mit der Deutschen
Space Technology, 9, No.2, 165-174. Gesellschaft ftir Erd- und Grundbau e.V.)
(5) Anagnostou, G., Kovan, K. (1994). Die Stabilitiit der Ortsbrust (20) Horn, M. (1961). Alagutak homlokbiztositasara hat6 vizszintes
bei Erddruckschilden. In: Stabilitiitsprobleme in der Geotechnik, foldnyomasvizsgalat nebany eredmenye. Az orszagos melye-
Fruhjahrstagung der Schweiz. Gesellschaft fur Boden- und pit6ipari konferencia el6adasai, Kozlekedesi Dokumentaci6s
Felsmechanik, 29. April 1994, Mitt. der SBGF,129, 27-34 (in Vallalat, Budapest (in Hungarian). See also "Horizontaler Erd-
German). druck auf senkrechte Abschlussflachen von Tunneln", In:
(6) Babendererde, S. (1988). Die Erhaltung der Primarspannungen Landeskonferenz der ungarischen Tiefbauindustrie, Budapest
beim Schildvortrieb im wasserftihrenden Lockerboden. In: (German transalation, STUV A, DUsseldorf).
Sicherungs-, Stutz- und Gebirgsverbesserungsmassnahmen im (21) Jacob, E. (1986). Der Mixschild: Entwicklungstendenzen unci
Untertagbau. ETH-Symposium. 61-67 (in German). erste praktische Erfahrungen. STUVA, Forschung+Praxis, 30,
(7) Babendererde, R. (1989). Stand der Technik und Entwicklungs- 20-31 (in German).
tendenzen beim maschinellen Tunnelvortrieb im Lockerboden. (22) Janssen, H. A. (1895). Versuche tiber Getreicleclruck in
STUV A, Forschung+Praxis, 33, 78-84 (in German). Silozellen. Zeitschrift des Vereins deulscher lngenieure, Band
(8) Babendererde, S. (1991). Kritische Betrachtungen zum Einsatz XXXIX, No. 35, 1045-1049 (in German).
von Hydroschilden. Berichte des Int. Symp. Sicherheit und (23) KnUpfer, J. and Meseck, H. 1984. Schildvortrieb bei fltissig-
Risiken bei Untertagebauwerken (Hrsg. R. Fechtig & K. Kovan), keitsgestUtzter Ortsbrust. Mitt. des lnst. fur Grundbau und
47-51 (in German). Bodenmechanik, TU Braunschweig, 16 (in German).
(9) Balmer, P. (1992). Einsatz des Mixschildes Grauholztunnel in (24) Kovari, K., Anagnostou, G. (1994). Storebrelt Railway Tunnel-
sehr wechselhafter Geologie. In: Symp. "Probleme bei Assessment of face stability. Report prepared for the MT-Group,
Maschinellen Tunnelvortrieben ? - Geriitehersteller und Consortium for Storebrelt Crossing.
Anwender berichten". 22/23 Oktober 1992, MUnchen (in (25) Krause, T. (1987). Schildvortrieb mit flUssigkeits- unci
German). erdgesttitzter Ortsbrust. Dissertation TU Braunschweig. (in
(10) Balthaus, H. (1988). Standsicherheit der flUssigkeitsgestUtzter German)
Ortsbrust bei schildvorgetriebenen Tunneln. In: Festschrift H. (26) Krause, T. (1991). Erfahrungen mit einem Erddruckschild beim
Duddeck. In st. fUr S tatik TU Braunschweig, 4 77-492 (in Bau eines Mischwasserkanals in Bremen. STUVA,
German). Forschung+Praxis, 34, 53-59 (in German).
(11) Bear, J. (1977). Dynamics of Fluids in porous Media, p. 764. (27) Leca, E. and Dormieux, L. (1990). Upper and lower bound
New York: American Elsevier. solutions for the face stability of shallow circular tunnels in
(12) Becker, C.H., Sawinski, J.W. (1982). Operational Experience frictional material. Geotechnique, 40, 581-606.
and development of bentonite shield tunnelling in Europe since (28) Leca, E. and Panet, M. (1988). Application du Calcul a la
1975.1MM, Tunnelling'82, 41-48. Rupture a la stabilite du front de taille d' un tunnel. Revue
(13) Biggart, A.P., Rivier, J.-P., Sternath, R. (1993). Storebrelt Franfaise de Geotechnique, 43, 5-19 (in French).
Railway Tunnel - Construction. In: Proc. Int. Symp. 011 (29) Mtiller-Kirchenbauer, H. (1972). Stability of slurry trenches. In:
Technology of Bored Tunnels under Deep Waterways, Proc. 5th europ. conf SMFE, Madtid, vol. I, 543-553.
Copenhagen, Denmark, 63-93.
(30) Nishi take, S. ( 1990). Advanced Technology realize high-
(14) DIN-4126 (1986). Ortbeton-Schlitzwiinde. Konstruktion und performance earth pressure balanced shield. In: Franchissements
Ausfuhrung (in German). souterrains pour I' Europe, 291-302, (Legrand, eel.), Balkema,
(15) DIN-4127 (1986). Schlitzwandtone fur stutzende Flussigkeiten. · Rotterdam.
Anforderungen, Priifveifahren, Lieferung, Guteuberwachung (in (31) Stack, B. (1982). Handbook of Mining and Tunnelling
German). Machinery. John Wiley & Sons, Chichester u.a ..
(32) Xanthakos, P. (1979). Sluny Walls, p. 622. New York a.o.:
McGraw-Hill.