Professional Documents
Culture Documents
ABSTnACT: Few theoretical methods and experimental data exist for the analysis
of tunnel-face stability in cohesionless soils. The present paper addresses a series
of practical questions by using centrifugal-model tests. The values of limit internal-
support pressures are given for various conditions (density of the sand, position of
the tunnel with respect to the ground surface). These values are shown to be low
as predicted by the latest limit-calculation models and collapse is shown to be
sudden. The geometry of the failure zone is depicted for different embedment
depths. The initial mechanism appears to be of a bulk shape with a limited extent
in front of the face. The presence of a short unlined length of tunnel at the face
is also investigated. Data are presented about its effect on the failure mechanism,
on the limit pressure, and on stress transfer onto the tunnel lining at collapse. The
results obtained in the model tests are in general agreement with present knowledge
of full-scalesituations.
INTRODUCTION
1148
J. Geotech. Engrg., 1994, 120(7): 1148-1165
depth and type of soil or to make sure of the repeatability of tests. On the
other hand, there are various experimental techniques by which this problem
can be approached using small-scale models (Cort6 1989).
For underground cavities, past experimental investigations, in most cases,
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
covered plane cross sections [e.g., Atkinson et al. (1977)]. But analysis of
tunnel-face behavior requires a three-dimensional modeling approach. There
have been model studies of cavities in normal gravity, with or without
equivalent material [e.g., Egger (1985)], to observe long-term behavior.
The first set of tests on small-scale models that satisfied similarity conditions
and dealt with face behavior was conducted with the Cambridge centrifuge
on cavities in clay (Mair 1979). Studies at the Ruhr University at Bochum
(RUB), Germany, focused on the special case of boring by the "new Aus-
trian tunneling method." These studies mostly examine stress transfer in
the lining (K6nig et al. 1991).
The purpose of the small-scale models conducted with the Laboratoire
Central des Ponts et Chaussdes (LCPC) centrifuge (Cort6 1984) (Fig. 1)
was to address tunnel-face stability when excavated by a tunnel-boring ma-
chine in a cohesionless medium. The study concentrated on depths shallow
enough for the stress gradient in the soil to play a decisive role. Previous
results are described by Chambon and Cort6 (1989) and by Chambon et al.
(1991).
The present paper summarizes a stability study of a tunnel face carried
out since 1987. The following questions are addressed:
Two types of pressure on the face were used, and different depths and
densities of material were tested. The size of the models was kept constant
to eliminate edge effects from the container walls.
Tunnel-boring machines move forward with a stop-and-go rhythm to allow
1149
J. Geotech. Engrg., 1994, 120(7): 1148-1165
placement of segments. During boring, there are many periods of immo-
bility. It was accordingly decided to isolate the main factors of influence in
a quasi-static approximation of the problem.
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
DESCRIPTION OF MODELS
Two tunnel models were built (Fig. 2), one to investigate the collapse of
the tunnel face itself, the other to examine unlined lengths between the face
and the shield.
The first model, the LCPC model, consists of a rigid metallic tube 100
mm in diameter. The end representing the face is covered with a thin (0.2-
mm) latex membrane left slack to prevent mechanical influence on the
displacement of the face. The membrane provides tightness between the
inside and outside of the cavity while leaving the soil free to move. The far
end of the tube has openings for the pressure piping and wiring to the inner
displacement sensors that track face movement.
The second model, the RUB model, was designed to allow strain mea-
surements of the tunnel for different unlined lengths near the face. It consists
of two nested tubes; the larger one is fitted with strain gages and the smaller
holds the membrane separating the interior of the tunnel from the ground.
To represent a certain unlined length, the membrane was made of a latex
thicker (1 mm) than that of the first model, so that the membrane would
retain its shape during test preparation. The displacements at the face were
measured by sensors embedded in the soil mass.
Fontainebleau sand (ds0 = 0.17 mm, Cu = d 6 o / d l o = 1.47), a fine ho-
mogeneous sand often used in France for rheological investigations, was
strain gages
i4-~4 I,I
t20 mm~
lm,~, --
40 mm~~'L ~ ...........
11~5mm
100 m m
Latex
1 't~em?ran"~ Fluid inlet
. ..........................
t ........................................................
..H I
"RUB model"
L
300
~ m m
1150
J. Geotech. Engrg., 1994, 120(7): 1148-1165
nt
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
I
I
36d mm
I
I
I
I
T
1200 m m
FIG. 3. Centrifuge Model in Container (1,200 x 800 ram)
used. It is the same material found in the Villejust tunnel on the western
high-speed rail line, a structure for which a large number of measurements
are available (Bochon 1990).
The models were prepared by raining the sand in place in order to control
both the density and the homogeneity of the sand mass. The tests described
were all performed with dry sand. The sand unit-weight values for the
various tests range from 15.3 to 16.1 kN/m 3, corresponding to relative den-
sities of 0.65-0.92.
The tests were performed with and without a uniformly distributed load
on the ground surface (Fig. 3). Two types of support for the tunnel face
were reproduced: (1) A hydrostatic pressure, to simulate the pressure dia-
gram in the case of a slurry shield; and (2) a uniform pressure for the
compressed-air shield.
The height of soil cover to the outside tunnel diameter ratio C/D varied
between 0.5 and 4.
The dimensions of the prototype was dictated by the level of acceleration
used. Prototype diameters of 5 m, 10 m, and 13 m were chosen, corre-
sponding to accelerations of 50 g, 100 g, and 130 g. The LCPC model allowed
precise monitoring of face fall out in the case of a fully lined tunnel. The
configurations in which part of the tunnel is unlined were simulated by using
different membranes in the RUB model.
To prepare the test, sand is spread evenly in a rigid container. The tunnel
model is placed on a first layer of sand, and sand placement is resumed,
with layers of colored sand inserted at known intervals. The pressure in the
tunnel model during preparation and during acceleration on the centrifuge
is kept equal to the geostatic stress at the centerline of the tunnel.
The pressure on the face is gradually reduced, starting from the active
earth pressure, until failure occurs. Failure is detected by measurements of
horizontal movement at the face. Additionally, observations of surface set-
tlements, deformations of the lining, and pressure in the tunnel were made
1151
J. Geotech. Engrg., 1994, 120(7): 1148-1165
during reduction of internal pressure before collapse. The geometry of the
failure mechanism is recorded after the test by wetting the sand and cutting
the soil mass along different vertical planes. For each of these planes, the
failure pattern is identified from a careful tracing onto paper of the position
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
The results are presented for the prototype structure. For a test at ng the
displacements measured on the model are multiplied by n for the prototype,
and the stresses, in particular the internal pressure, are the same for the
model and the prototype.
0
"E 10
u~ 20- .................. ..................................................................
Z
30-
40-
SO-
7O i
o 5 10 15 20 25 30
INTERNAL PRESSURE (kPa)
FIG. 4. Evolution of Tunnel Face: Horizontal Displacementswlth Decreasing Con-
fining Pressure (C/D = 2)
1152
J. Geotech. Engrg., 1994, 120(7): 1148-1165
If p, is the active earth pressure at the tunnel's centerline computed ac-
cording to Rankine's formula
p. = ~l(C + D/2)tanZ('rr/4 - qb/2) (1)
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
Influence of Geometry
The limiting pressures in the tests performed at the LCPC and at the
RUB, with a dry sand having a unit weight between 16.0 and 16.2 kN/m 3,
are summarized in Table 2. Some tests were repeated two or three times
and yielded a very good repeatability for all features. Judging from these
repetitions, the experimental uncertainty on the limiting pressure is about
1 kPa.
Some of these tests were carried out with the RUB model and a small
unlined length (L/D = 0.2). Given the slight influence of this length on the
limiting pressure, the results are included.
Collapse
Model Diameter D Depth/diameter pressure Pl Unit weight
centrifuge (m) (C/D) (kPa) (kN/ma)
(1) (2) (3) (4) (5)
N/N 5 0.5 3.6 16.1
N/N 5 0.5 3.3 16.1
N/N 5 1 3.5 16.1
N/N 5 1 3.0 16.1
N/B 5 1 3.3 16.1
N/B 10 1 7.4 16.0
N/N 5 2 4.0 16.1
N/B 10 2 8.0 16.0
N/B 10 4 8.2 16.0
B/B 13 4 13.4 16.2
1153
J. Geotech. Engrg., 1994, 120(7): 1148-1165
For the relative depths investigated (C/D = 0.5-4), depth does not have
a large influence on the limiting pressure for a given diameter.
The most significant parameter is tunnel diameter. The minimum support
pressure increases directly with the diameter (Fig. 5). This finding prompts
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
14
12
I,
I o C/D=0.5
[] C/D=1
10 o C/D=2
==
x C/D=4
8
[]
==
6 ............................ i ......................: i i
4 ...........o ............................................
i .............................i...................................
~...............................
O 2 i i
4 6 8 10 12 4
T U N N E L D I A M E [ ' E R (m)
FIG. 5. Influence of Tunnel Diameter on Collapse Pressure p:
0.1
0.08
i O
0.06 -
........................................................ i .................................................. i ....................................................
o 9
0.04
0.02 ........................................................ i.
0
0 5 10 15
TUNNEL DIAMETER (m)
FIG. 6. Investigation of Size Effect on Failure Condition
1154
J. Geotech. Engrg., 1994, 120(7): 1148-1165
Ground surface
./ -,,.
~/sr...,~../'2"./.',,2./~/x" y/7"~../'2y',Z'//i~./y/
FRONT VIEW SIDE VIEW
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
/ \ / \
I~I,
! , l I
I I
\ \ / .
/,,-~/-,,v/,~/-~/~,
FIG. 8. Failure Bulbs for Different C/D Ratios (Cut Made along Plane of Symmetry
of Tunnel)
1155
J. Geotech. Engrg., 1994, 120(7): 1148-1165
limiting pressure of failure. The influence of this factor on the limiting
pressure of collapse is found to be small (Table 1). While the differences
are of the order of the experimental uncertainty, the repeatability of the
results seems to show that loose soil leads to a less stable configuration than
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
dense soil. As regards the pressure at which the first face movements are
observed, the scatter of the results made it impossible to demonstrate any
significant influence of density.
Failure Pattern
Cross sections of the model in different vertical planes after the test
provide information about the geometry of the mechanism of failure. In all
cases, the failure envelope is bulb-shaped (Figs. 7 and 8). Strains are lo-
calized along the surfaces bounding this bulb, which is limited by a vertical
plane passing through the tunnel face and by a concave envelope that starts
near the floor, extends a half-diameter in front of the face, and rejoins the
vertical plane at a height of about one diameter above the face. The ground
surface is affected only with relative depths C / D less than 1 (Fig. 8). With
deeper tunnels, the bulb closes and the stability of the soil is ensured by
arching with force transfer to the crown of the tunnel. At a more shallow
depth, the bulb can no longer close. The instability of the face is accom-
panied by the triggering of secondary mechanisms on the surface.
Given the brittle character of the failure, interrupting the collapse as soon
as it starts presents experimental difficulties. However, the cross sections
made on the models show that the interior of the bulb itself undergoes little
deformation. This confirms the stability estimates based on movements of
blocks and investigations of behavior assuming a tendency to strong local-
ization of strains in this type of soil.
The evolution of surface movement above the tunnel face during decon-
finement is similar from one test to another. However, loose soils do ex-
perience slightly larger settlements. Below the characteristic pressure Pc
(Table 1), for which the first movements of the face are observed, the
evolution of settlements is comparable to that of the horizontal displacement
of the face. In the case of a tunnel 5 m in diameter at a depth C / D = 1
,_,~ 0 ~ 2.0
2 .............. ............................................. .............................. ......................... .....................
r~
10
Four tests were conducted to study the case where the face is supported
by a fluid rather than by air. Loading conditions of a slurry shield were
simulated. The fluid used in the tunnel model is water, and the soil is again
dry Fontainebleau sand. The interior of the tunnel is connected to a tank
of water carried aboard the centrifuge. The filling of the tank can be adjusted
to control the pressure on the tunnel face.
The failure occurs suddenly when the free water surface goes below the
tunnel-crown level. For settlements, the same evolutions are observed as
with pneumatic support.
Experiments in which the relative depth C/D and the load applied on the
surface are varied show that, for the ranges of variation of C/D and density
in question, neither the depth nor the surface load significantly affect the
limiting pressure of failure. This essential result is also found by limit cal-
culation.
Investigation of failure after the test shows that the soil is swept into the
zone of low pressures near the arch. There is no overall failure of the face.
There is a chimney-shaped flow that tapers toward the surface and flares
again to form a subsidence (Fig. 10). Although confined to the top of the
tunnel, the mechanism of collapse develops similarly to what is observed
with a uniform support pressure.
/~_~ C/D=2
. __.,,
\\\\\\\\\\\\
FIG. 10. Observed Lines of Failure with Hydrostatic Internal Pressure
1157
J. Geotech. Engrg., 1994, 120(7): 1148-1165
The small-scale model tested in the centrifuge does not represent field
conditions precisely but attempts to simulate the worst configuration, that
is, when deconfinement occurs not behind the skirt but at the tunnel face.
The thickness of this gap is set at 100 mm (prototype), a value substantially
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
greater than the thickness of most skirts (50 to 60 mm). On the 1:50 model,
the gap is 2 mm. To produce it, the initial model is completed by a semi-
cylindrical shell that can slide back (Fig. 11). This movement of the external
shell takes place at 50 g, before the pressure inside the tunnel is lowered
to failure. The test is conducted at a relative depth C/D = 1 in a dense
sand. At the initial support pressure, when confinement is reduced, surface
settlement occurs that has an amplitude smaller than the thickness of the
20
30
40-
50-
0 10 20 30 40 50 60
INTERNAL P R E S S U R E (kPa)
FIG. 12. Vertical Ground Surface and Horizontal Tunnel-Face Displacements ver-
sus Internal Confining Pressure
1158
J. Geotech. Engrg., 1994, 120(7): 1148-1165
i Groundsurface
//-,7/x,,,/y/5~,/ ,7/v/y/"9"/Nv/TM'
SIDEVIEW
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
t !
G A P ~
a
//'////,/,/
i
--,k___
9 / j !
t
Y////////
gap theoretically left by the skirt (Fig. 12). When the internal pressure is
lowered (from 36 to 32 kPa), a first movement of the face is observed. This
corresponds to a local return of the soil around the tunnel and does not
affect the ground surface.
With regards to failure, the same characteristics are found as in the pre-
vious tests. These include the following:
9 A stage of settlement and gradual fall out of face; this stage im-
mediately precedes failure
9 Identical limiting pressure of failure (approx. 5 kPa)
9 Similar collapse bulbs (Fig. 13).
In the collapse bulb, zones of incipient failure that did not occur in the
confguration without deconfinement can be observed; these incipient shear
failures are probably caused by the settlement of repositioning observed
when the annular gap is created. The size of the annular gap produced by
the model described is of the order of 10-20 grains of sand, a size which
is small and partially explains why the effect on the extent of the zone of
failure is slight.
ExcavationbyNewAustrianTunnelingMethod
Originally used for tunneling in squeezing rock, the new Austrian tun-
neling method involves the following sequences:
J %),
~
--"~rltll/rllr
JrHt,,,,llllJlitlt
L=0.1D L=0.2D L ~- 0.4 D Unlined length
FIG. 14. Influence of Tunnel Unlined Length on Extent of Failure Mechanism (L/D
= 0.1, 0.2, and 0.4; C/D = 4)
1160
J. Geotech. Engrg., 1994, 120(7): 1148-1165
0 I I Nl~ m --
_ -2 ~ ' i , i i i
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
2"~ -4 i
-6
! _,o ?mo
..............................I ~
-14 I I I E [
0 100 200 300 400 500 600 700 800
I N T E R N A L P R E S S U R E (kPa)
FIG. 15. Changes in Compressive Stress at Extrados of Lining for Different Dis-
tances to Tunnel Face (C/D - 4)
1161
J. Geotech. Engrg., 1994, 120(7): 1148-1165
1, and 2 m from the tunnel face (prototype dimensions) as the pressure
against the tunnel face decreases.
\
I
Experimental mechanism ~ ,
~ C/D=I
I I
I
2Danalytiealmechanism,~ I / ~ i1
xN\\\\\NX\\\\'~
FIG. 16. Comparison between Experimental and Theoretical Failure Mechanisms
1162
J. Geotech. Engrg., 1994, 120(7): 1148-1165
CONCLUSIONS
Tunnel-face behavior in sand as failure is approached can be investigated
using small-scale models in the centrifuge. The test results have provided
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
ACKNOWLEDGMENTS
The writers wish to thank D. K6nig (Bochum, Germany) for the test
results he obtained and shared kindly. We also wish to express our gratitude
to A. Zelikson (Univ. Haifa, Israel), who contributed to devising the first
tests of this study. Finally, we wish to say that this work could not have
been done without the precious help of J. Gamier and C. Favraud (LCPC,
Nantes, France).
APPENDIX I. REFERENCES
Atkinson, J. H., Potts, D. M., and Schofield, A. N. (1977). "Centrifugal model
tests on shallow tunnels in sand." Tunnel and Tunneling, (Jan.) 59-64.
1163
J. Geotech. Engrg., 1994, 120(7): 1148-1165
Bochon, A. (1990). "Tunnel de Villejust, m6moire de synth~se des mesures," PhD
thesis, Conservatoire National des Arts et M6tiers (CNAM), Paris, France (in
French).
Briglia, P., Pantet, A., and Kastner, A. (1989). "Evolution de l'emploi et domaine
Downloaded from ascelibrary.org by National Inst of Tech - Jamshedpur on 09/23/16. Copyright ASCE. For personal use only; all rights reserved.
A P P E N D I X II. NOTATION
1165
J. Geotech. Engrg., 1994, 120(7): 1148-1165