Sie sind auf Seite 1von 10

Tunnelling and Underground Space Technology 27 (2012) 101110

Contents lists available at SciVerse ScienceDirect

Tunnelling and Underground Space Technology


journal homepage: www.elsevier.com/locate/tust

Evaluation of tunnel face stability by transparent soil models


Mahmoud Ahmed a, Magued Iskander b,
a
New York State Department of Transportation, Route 9A Project/Lower Manhattan Redevelopment, 115 Broadway, RM 1701, New York, NY 10006, United States
b
Polytechnic Institute of New York University, Six Metrotech Center, Brooklyn, NY 11201, United States

a r t i c l e i n f o a b s t r a c t

Article history: Accurate estimation of tunnel face support pressure is necessary for economical and safe shield tunneling
Received 6 July 2010 in cohesionless soils. This paper presents measurements of tunnel face support pressure and associated
Received in revised form 12 July 2011 soil movements obtained using a transparent soil model that simulates shield tunneling in medium dense
Accepted 1 August 2011
saturated sand. The use of a transparent soil surrogate permits measuring the internal soil deformations
Available online 29 September 2011
within the model soil. Soil deformations associated with various face support pressures are presented for
4 cover-to-diameter (C/D) ratios. Failure is found to be sudden with sand owing into the tunnel leading
Keywords:
to a prismatic wedge in front of the tunnel face and a vertical chimney of soil above. A minimum support
Tunnel
Ground movements
pressure was achieved with support pressures as low as 10 1% of the effective vertical stress at the tun-
Face stability nel axis. The stability of the tunnel face was related to the coefcient of active earth pressure with C/D
Failure mechanisms ratio having a small effect on the magnitude of required pressure at collapse.
Transparent soil 2011 Elsevier Ltd. All rights reserved.
Digital Image Correlation (DIC) cross-
correlation
Particle Image Velocimetry (PIV)

1. Introduction Accurate prediction of ground movements associated with dif-


fering face pressures in soft ground is essential for efcient con-
Tunneling in soft ground is most likely to be carried by either struction and protection of nearby structures and utilities from
slurry/hydro shield (SS/HS) or earth pressure balance (EPB) tunnel damage. Kirsch (2010) compiled a large number of available theo-
boring machines (TBM). Both TBMs work on the principle that retical and experimentally derived approaches to estimate the re-
ground deformations can be greatly reduced if the tunnel face is quired face support pressure. These solution ranged by one order
excavated while being supported. Thus, if the applied pressure of magnitude. The face support pressure must stabilize the soil at
can maintain support to the tunnel face, ground displacements the front. A relatively low face pressures run the risk of excessive
can be minimized. ground deformations (Peck, 1969); therefore it is necessary to
The SS/HS tunneling method balances existing earth and hydro- know the minimum face pressures that result in acceptable ground
static pressures by applying a pressurized slurry consisting of movement. In addition to face collapse (active failure) the face sup-
water, bentonite, and/or polymer conditioners at the tunnel face port pressure must also prevent blow-out (passive failure) of the
(Anagnostou and Kovri, 1994). The slurry also prevents the inl- soil mass near the tunnel face.
tration of groundwater into the tunnel. The EPB method uses the Analytical methods and experimental data to estimate the tun-
excavated soil to balance the soil and water loads to prevent nel face support pressure are in short supply, particularly for tun-
ground failure at the tunnel face (Clough and Leca, 1993). In both nels in saturated sand. Meguid et al. (2008) presented a review of
mechanisms an internal pressure is generated to support the cut- tunnel physical modeling and approaches used to record soil defor-
ting face against the existing overburden and hydrostatic pres- mation that have been developed and used in soft ground tunnel-
sures. The popularity of pressurized shields has been increased ing research. Among the techniques discussed in this study is rigid
by the effective use of additives and other soil conditioning meth- tube with exible face method where a tunnel cavity was sup-
ods such as foams and polymers which increase the range of soils ported by either a Latex membrane (e.g. Chambon et al., 1991;
where SS/HS/EPB methods can be employed (Quebaud et al., 1998; Sterpi et al., 1996; Yoo and Shin, 2003) or a rubber membrane
Vinaia et al., 2008). (e.g. Hagiwara et al., 1999; Wu and Lee, 2003; Lee et al., 2006). A
similar membrane technique was adopted in this study with trans-
parent soil surrogate. The approach applied the recently developed
Corresponding author. Tel.: +1 718 260 3016; fax: +1 718 260 3433. transparent soil technology and associated image processing tech-
E-mail addresses: mmahmoud@dot.state.ny.us (M. Ahmed), iskander@poly.edu niques (Iskander, 2010) to investigate the relationship between
(M. Iskander).

0886-7798/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2011.08.001
102 M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110

Atkinson and Potts (1977) also developed upper bound solution


by considering kinematically admissible failure mechanisms. The
unsafe tunnel support pressure or upper bound was found to be
independent of C/D and given by;
 
cD 1 p
rT u 3
4 cos u tan u 2
Leca and Dormieux (1990) expanded Davis et al. (1980) formu-
lation to dry MohrCoulomb (c, u) materials, as follows:

rT cDNd rS NS 4

where Nd and NS are nondimensional parameters that represent the


effect of soil weight and surcharge loading, respectively, similar to
Fig. 1. Important parameters in tunneling. bearing capacity factors. Leca and Dormieux presented two sets of
graphs to estimate Nd and NS corresponding to blow out and col-
lapse. Using these graphs the required support pressures can be
calculated.
tunnel geometry, face support pressure, and ground deformation. A
Anagnostou and Kovari (1996) indicated that at limit equilib-
tunnel is pre-placed in a transparent soil model, which represents
rium state, the required support pressure depends on several fac-
saturated sand. Tunnel face support is simulated using an internal
tors including the tunnel diameter D, the overburden C, the
pressure (rT) applied inside the tunnel (Fig. 1). Tests are conducted
piezomteric head in the chamber, water table level, shear strength
by reducing the tunnel pressure rT in stages until collapse of the
parameters (u and c), the dry unit weight for soil above water table
soil occurs. Because the model is transparent, it can be sliced using
and submerged unit weight for submerged soil. They further devel-
a laser light sheet, along the tunnel axis. Images of the soil along
oped a solution based on dimensional analysis as follows:
the tunnel axis, illuminated by a laser light sheet, were captured
after each decrement of rT reduction and used to obtain corre- rT F 0 c0 D  F 1 c F 2 c0 Dh  F 3 cDh=D 5
sponding 2D deformation elds. Soil deformations perpendicular
to the tunnel axis are available in Ahmed and Iskander (2011). where F1 to F3 are dimensionless coefcients, c0 is the submerged
These model tests are not intended to precisely reproduce a unit weight and Dh is the difference between the original piezomet-
scaled real tunnel during construction with details of the methods ric head at the tunnel invert level and the chamber.
of excavation support. Nevertheless, this simplied approach is Further models based on limit equilibrium method have been
capable of revealing patterns of face collapse and its relation to proposed by Broere (1998) and based on kinematical method of
support pressure and tunnel geometry. In particular the methodol- limit analysis by Mollon et al. (2009).
ogy illustrates the way in which soil around a circular cavity,
resembling a tunnel in saturated sand, deforms when the support
3. Laboratory investigations to analyze stability of tunnel face
pressure within the cavity is reduced. A similar procedure was ini-
tially adopted by Ward and Pender (1981) and has been success-
Tunnel face stability has been investigated by several authors
fully used by Sterpi et al. (1996) and Yoo and Shin (2003) to
using the centrifuge. During centrifuge tests, face support is usu-
investigate tunnel face stability.
ally achieved by either a piston or impermeable membrane sup-
ported by air or uid pressure. For the piston method, the
2. Analytical methods to analyze stability of tunnel face mechanical support is slowly removed to simulate a face collapse.
This method does not conveniently provide information about the
Numerical modeling has been used extensively to investigate required support pressure. Nevertheless, it gives valuable informa-
tunnel face stability (Leca et al., 2000). These methods allow for tion about failure shape and collapse mechanism. The membrane
simulating a wide range of scenarios. However, in this study, an at- method usually provides information about both: failure shape
tempt is made to correlate the experimental results against closed and the required support pressure, which can be found by slowly
form analytical solutions due to the simplicity of the comparison. reducing the support pressure behind the membrane and observ-
A number of authors have described external failure mecha- ing deformations. However, its not possible to examine the effect
nisms of the tunnel face (i.e. failure not related to linings or tunnel of mixing additives (slurry) on the excavated material because the
structural members) and derived formulae to calculate the mini- membrane simulating the tunnel face must be impermeable.
mum required support pressure by analytical or empirical means. Chambon and Cort (1994) conducted a series of centrifuge
In sands, Atkinson and Potts (1977) derived lower bound solu- experiments for tunnels embedded in homogeneous dry sand.
tions for the minimal support pressure of an unlined cavity in a They used model tunnel sealed by membrane and gradually re-
dry cohesionless material for two different cases. The rst case duced the support pressure starting from active earth pressure
(Eq. (1)) for tunnels in a weightless soil (c  0) and surface load down to collapse. Hence their initial support pressure at the tunnel
rS is given by: centerline computed according to Rankines formula is:
 1kp  
2C D p u
rT rS 1 1 rT c C tan2  6
D 2 4 2
where kp is Rankines coefcient of passive earth pressure. The sec- Their investigation focused on the effect of C/D ratio and soil
ond case (Eq. (2)) is for tunnels in soils having unit weight, c > 0 and density on the minimum support pressure and failure mechanisms.
no surface loading is applied, rS = 0. They showed that for the values of C/D (0.54) and unit weight
cDkp (15.3 and 16.1 kN/m3) investigated both, soil density and C/D had
rT 2
2 little effect on the support pressure at failure, perhaps due to the
kp  1
small range of density investigated.
M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110 103

Atkinson and Potts (1977) reported a number of model tests


performed on dense sand in a centrifuge. Tests were carried with
and without surcharge load (rS), which was applied by uid pres-
sure contained within exible rubber membrane applied at the
sand surface. Tunnels were constructed with the tunnel pressure
approximately equal to the vertical stress in the soil at the level
of the tunnel axis; thus their initial support pressure was:
 
D
rT c C rS 7
2
During the tests the tunnel pressure was reduced in decrements
until the tunnel collapsed. Their results indicated that for the case
where rS > 0, the minimum support pressure was found to be well
represented by Eq. (3). For rS = 0 case the observed minimum sup-
port pressure exceeded the lower bound solution given by Eq. (2).
Similar experiments have also been conducted on cohesionless
material by Eisenstein and Ezzeldine (1994), Hisatake et al. (1995)
and Takano et al. (2006).

4. Experimental methodology

4.1. Model test setup

Transparent silicas suitable for simulating the geotechnical


properties of a wide range of soils have been developed (Sadek
et al., 2002). One of the developed materials called SG1 is a partic- Fig. 3. Transparent soil tunnel model.
ulate material resembling angular white sand, 0.51.5 mm in size
was utilized. Complete saturation of the particles is required in or- (0.3 mm thick) of negligible strength was attached to the end of the
der to eliminate the refractive index mismatch between silica gel tube to represent the tunnel face. The membrane was made of a
and air. In order to achieve transparency SG1 is saturated with a sheet of Latex that was cut and glued to the PVC tube to fasten
matched refractive index uid. The refractive index, viscosity, it. Additionally, an O-ring was used as a backup to prevent leakage.
and density of the pore uid at room temperature (24 C) were The membrane front was left slack to prevent mechanical inuence
1.447, 5.0 cP, and 800 kg/m3, respectively. SG1 has similar stress- on the displacement of the face. The membrane method might be
strain characteristics to sand (Fig. 2). Detailed properties of SG1 problematic with respect to the failure shape at low stress values,
and the pore uid are available in Iskander et al. (2002). For the because of the uncontrolled shape of the deformed membrane.
experiments described in this paper, the average dry and saturated However, in these tests, the membrane was supported by uid
unit weights were 7 and 8.53 kN/m3, respectively. The angle of fric- pressure that is believed to provide a uniform pressure distribution
tion u of the MohrCoulomb failure envelopes was 36 obtained at the face and consequently controlled the shape of deformation.
by direct shear tests performed on SG1 samples at 60% relative The mineral oil used as pore uid is incompatible with Latex,
density according to ASTM test method D3080. The normal stress and quickly deteriorates thin membranes. A 0.3 mm thick mem-
in theses tests ranged between 40 and 70 kPa. brane (Em = 1.3 MPa) was used to allow enough time for testing.
A Plexiglas model container (Fig. 3) 30.48  25.4  20.32 The effect of membrane stiffness (rm) was determined by measur-
(width  depth  height in cm) was used to contain the transpar- ing the pressure that must be applied inside the membrane to
ent soil. The dimensions of the model have been chosen in such cause it to expand laterally, when no soil is present. A value of
a way, that the inuence of the boundaries was minimized. The rm = 4.8 kPa was obtained for the membrane used in this study
tunnel is modeled by a PVC tube, 2.54 cm in diameter, preinstalled when it exes out a distance of 2.5 mm, with no soil surrounding
inside the model at a depth of 12.7 cm. This allows testing tunnels it. The initial pressure employed is 69 kPa, and reduced to
with cover to diameter ratio (C/D) from 1 to 4.5. A Latex membrane 0.7 kPa, which corresponds to the membrane exing in by
2.5 mm. This indicates that the membrane may have an effect on
the post failure support pressure, but its effect is less than 10%
prior to failure.
Transparent soil was then placed at approximately 20 mm lifts.
The placement technique adopted in this study provided an in-
place uniform relative density of approximately 60% with a unit
weight 8.53 kN/m3. The tube (tunnel) was then lled with water
under pressure to simulate the tunnel support pressure, rT. In real-
ity, such a support can be achieved by use of compressed air, ben-
tonite slurry or earth pressure balance (EPB) TBMs. In this study, rT
is believed to be uniform over the tunnel face. Membrane support
pressure was measured by a tube connected to a pressure port and
a digital monitor with precision in the order of 0.01 Pa.
For application of surcharge or surface pressure, rS, the Plexi-
glas model container was placed between two identical metal
plates connected by four threaded rods (Fig. 4). A rubber tire with
Fig. 2. Stressstrain characteristics of SG1. internal pressure rS was placed on top of the transparent soil.
104 M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110

dows the image before and after each decrement of tunnel pres-
sure are correlated to produce an average particle displacement
vector. Doing this for all interrogation regions produce a vector
map of average particle displacements. The position of the peak
in the correlation plane corresponds directly to the average particle
displacement within the interrogation area investigated. The vec-
tor from the center of the region to the peaks of cross-correlation
function gives the magnitude and direction of the relative move-
ment between two images. An advanced form of DIC which em-
ploys window shifting and window sizing, called Adaptive Cross
Correlation (ACC), has been used (Liu and Iskander, 2004). ACC is
implemented in Flow Manager Software (Dantec, 2001), which is
the software used in this research.

4.3. Modeling technique

Fig. 4. Application of surface loading. Active collapse of the tunnel is triggered by application of sur-
charge, rS, and self-weight, with the tunnel face pressure rT pro-
viding resistance against failure. Under passive conditions, these
In addition to the tunnel container, the set up also included a roles are reversed, and blow-out of the soil mass in front of the tun-
Cohu 2622 black & white CCD camera, 35 mW Melles Griot laser nel face is caused by the tunnel pressure with resistance being pro-
light source, a line generator lens, a test table, and a PC for image vided by the surcharge and self-weight. In this study, only the
processing (Fig. 5). The camera has a resolution of 640  480 pixels active collapse of the tunnel face is considered in the analysis; be-
and controlled by the PC through a Matrox Meteor 2/4 frame grab- cause the blow-out of the soil in front of the tunnel face is less
ber. A macro-zoom lens with a variable focus length from 18 to likely, thus of less practical interest.
108 mm was mounted on the CCD camera. At the beginning of the experiment, the vertical effective stress
acting on the tunnel was negligible. This assumption is found to be
4.2. Displacement eld measurement using Digital Image Correlation reasonable because of the unit weight of the transparent soil (Sa-
dek et al., 2002) and the size of the model which yields negligible
The interaction between laser light and transparent soils pro- magnitude for the quantity c(C + D/2) = 0.50.9 kN/m2. When the
duces a distinctive speckle pattern. This speckle pattern manifests surcharge, rS = 69 kPa, is applied the vertical effective stress in-
the interaction between the transparent soil matrix, impurities, en- creases and is assumed to be equal to rS. The lateral earth pressure
trapped air, and the laser. Small particle movement will result in follows the increase in vertical effective stress, which would cause
change in the speckle distribution in the plane of measurement. the tunnel face to deform into the tunnel, unless the internal tun-
If the deformation is small, the contrast distribution resulting from nel pressure, rT, was increased to balance the increased stresses
the speckle effect will follow the particle movement. Images cap- from surface pressure. A conservative initial internal tunnel pres-
tured before and after deformation are analyzed using the cross- sure, rT = rS was employed in all experiments, such that rT and
correlation function (DIC) which is a classic pattern recognition rS were increased simultaneously.
technique based on using correlation function to locate the best To illustrate that the methods employed in this study were suc-
matching position of two images and thus predicting movements. cessful in obtaining an initial zero face displacement condition,
Sadek et al. (2003) studied the accuracy of DIC application in trans- vectors of resultant ground movement due to application of sur-
parent soil models. The results indicate that the calculated defor- face pressure, rS, and tunnel pressure, rT, are shown in Fig. 7 for
mation using the developed system has an error on the order of one test. The displacement vectors are generated by DIC analysis
0.1 pixel, which is equivalent to a 0.01 mm for optical settings used of images captured before and after application of tunnel and sur-
in this study. face pressures. As seen by the vectors in Fig. 7, only vertical ground
For displacements measurement using DIC, the camera images movements were encountered due to settlement of the soil prole
are divided into rectangular regions called interrogation areas or as a result to application of a surcharge pressure. No horizontal
interrogation windows (Fig. 6). For each of these interrogation win- movement occurred near the tunnel face.
Tests were conducted by reducing the tunnel pressure rT in
stages until collapse occurred. After each decrement of tunnel
pressure (0.7 kPa), the model was sliced optically using laser light
sheet to illuminate the plane of measurements inside the model
and an image was taken by the CDD camera. Later, these images
were analyzed using Digital Image Correlation to obtain corre-
sponding deformation and strain elds.

5. Analysis of test results

5.1. Required support pressure

Plots of face support pressure against face displacements for


four tests with C/D ratios ranging from 1.5 to 4.5 are presented
in Fig. 8, where the tunnel support pressure, rT, normalized by
the vertical effective stress, rV, at the tunnel axis is plotted against
Fig. 5. Test setup. tunnel face displacement, dh, measured at the center of the face,
M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110 105

Fig. 6. Flow chart for Digital Image Correlation algorithm.

the start of collapse and corresponds to rT/rV = 17 1%. Chambon


and Cort (1994) referred to pressure triggering collapse in centri-
fuge experiments on ve tunnels in dry sand as the characteristic
failure pressure, pc. They stated that the characteristic pressure at
the start of collapse, pc, are on the order of 20 kPa for a Cavity 5 m
in Diameter and sufciently deep (C/D > 1). Based on Chambon
and Cort test conditions (c = 15.316.1 kN/m3, pc = 20.1
20.2 kPa, C/D = 12) rT/rV ranges between 11% for C/D = 2 and
18% for C/D = 1. Our tests and Chambon and Cort tests are in sur-
prisingly close range for tunnel support pressure triggering the col-
lapse stage. Further reduction of support pressure beyond Point A
down to Point B, results in small face displacements, which are
approximately proportional to the reduction in rT. At Point B, the
cumulative normalized displacements of the tunnel face range
from 1% to 1.1% (Fig. 8). Again, this range compares favorably with
Fig. 7. Vectors of resultant ground movement due to simultaneous application rS the 0.7% reported by Chambon and Cort (1994) for dry sand.
and rT.
Displacement vectors at Point B are shown in Fig. 9 for C/D ra-
tios 1.54.5. The gure visualizes the extension of ground move-
normalized by tunnel diameter, D. The shape of the curves is sim- ment ahead of the face and above the crown. It appears that
ilar throughout the range of C/D values examined. At the beginning ground movement extends approximately 1D in front of the tunnel
(Point O) rT/rV = 1, where rV (rS in this study) is the theoretical face. Movements are largest near the face but propagate upward to
vertical pressure at the tunnel center line where the measurement form a narrow chimney. The effect of these movements at the sur-
was taken. As rT is reduced, no movement is observed at the face face is varied based on C/D ratio, with shallow tunnels (C/D < 3)
down to pressures that are very low compared to initial support generating more surface disruption, as expected. This behavior is
pressure. For our tests, a distinctive pressure (at Point A) represent recognized again in the contour plots of resultant ground move-
106 M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110

Fig. 8. Face support pressure and corresponding face displacements (rV = rS).

Fig. 9. Displacement vectors at Point B.

ment normalized by tunnel diameter (Fig. 10) for four different C/D ment did not reach the ground surface. In the case of tunnels with
ratios, at Point B. For deeper tunnels (C/D > 4), peak ground defor- C/D < 3, ground movements associated with initial instability of the
mation was localized near the tunnel face, and peak ground move- tunnel face extended to the ground surface. Manifestations of
M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110 107

Fig. 10. Contour plots of resultant ground movement at Point B.

ground movement at the surface were clearly observed in shallow-


est tunnels (C/D = 1.5). Additionally, the extension of ground move-
ment ahead of the face increases with C/D ratio as seen in Fig. 10.
Knowledge of when Point B or local tunnel face failure occurs is
not only important to ensure face stability but also to control and
limit ground deformations associated with the tunneling. It ap-
pears that a small face support pressure approximately 911% of
the vertical effective stress, applied to the tunnel face is sufcient
to provide stable conditions against the action of earth pressure at
the face (Fig. 8), which falls between the average of the data pub-
lished by Oblozinsky and Kuwano (2006) and Chambon and Cort
(1994) (Fig. 11). For our test conditions, the required rT/rV at Point
B represents 50% of lateral earth pressure coefcient (ka = 0.26 cal-
culated by Rankine method).
Fig. 11. Comparison between critical support pressure (rTC) measured in transpar-
ent soil and centrifugal models.
5.2. Comparison with theoretical predictions

Atkinson and Potts (1977) solution for tunnels in a weightless


Table 1
soil (c  0) and surface load rS (Eq. (1)) is not widely used. Also Theoretical support pressures from Atkinson and Potts (1977) solution for weightless
the term weightless soil does not describe SG1 fairly knowing that soil.
its unit weight is smaller than natural sands by approximately 40%.
C/D Minimum support pressure, rTmin (kPa)
Nevertheless, a comparison with that solution is presented in Ta-
ble 1 for a range of friction angles. Our results compare best to / = 36 / = 30 / = 28 / = 26

the case of C/D = 1.5 and / = 28, which is lower than the measured 1.5 1.3 4.3 6.9 7.9
angle of friction of SG1. This could have resulted from (1) errors in 2.5 0.4 2 3.5 4.2
3.5 0.2 1.1 2.2 2.7
the theoretical model, (2) experimental error, (3) presence of arch-
4.5 0.1 0.7 1.5 1.9
ing at deeper tunnels, or (4) failure to obey all scaling laws.
108 M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110

5.3. Failure mechanisms proposed by Horn (1961), Chambon and Cort (1994), Leca and
Dormieux (1990) and Mollon et al. (2009) (Fig. 13).
After Point B was reached (Fig. 8), further decrease of the sup-
porting pressure led to sudden collapse (Point F) as transparent soil 6. Relationship between tunnel support pressure and ground
tended to slide into the tunnel and the failure envelope propagated deformation
progressively upward.
Contour plots for resultant ground movements associated with The observed behavior can be divided into three stages in rela-
Point F (failure) are shown in Fig. 12 for four different C/D ratios. tion to lateral earth pressure coefcient, ka, as follows:
The movements observed at failure are substantially larger than
those observed at Point B (Fig. 10). Ground movements at the tun- 1. Face Deformation (1 > rT/rV > 0.8ka): Face deformation is dened
nel level corresponding to failure stage transmitted to the surface by displacement, dh, measured at the center of the face, normal-
with shallower tunnels transmitting more movements than deeper ized by tunnel diameter, D. Although support pressure is
ones. The observed movements compare well with the zone of reduced from its original value no movement was observed at
ground movement reported by Chambon and Cort (1994) using the face till rT/rV = 0.17 .01 was reached. During the face
a centrifuge in dry sand, where the zone of post-instability ground deformation stage, very small face deformations were observed
movements formed a chimney in front of the tunnel face that (less than 0.1% of tunnel diameter). The pressure associated
reached the ground surface. The dashed lines in Fig. 12 indicate with triggering the face movement at the end of this stage rep-
the failed zones at the moment of collapse, corresponding to the resents the start of collapse.
contour of ground displacement equal to the tunnel face displace- 2. Face Slip Stage: (0.8ka > rT/rV > 0.5ka): This stage represents
ment at the center line, at Point F. The failed zone is relatively nar- small deformations leading to local face failure. Face displace-
row and reaches approximately 1D above the tunnel crown, when ment at this stage ranged between 0.1% and 1% of tunnel diam-
arching of the soil prevented further deformation. The failed soil eter. In our experiments local failure occurred when rT/
mass outcropped at the ground surface for the C/D = 1.5 case. How- rS = 10 1%, which we average conservatively as 0.1. Our tests
ever, the chimney shape failure did not outcrop at the ground sur- results compare favorably to centrifuge model tests by Cham-
face for tests with C/D > 2. The observed failure zone is consistent bon and Cort (1994) and Oblozinsky and Kuwano (2006)
with the chimney and wedge combination failure mechanisms (Fig. 11) and falls between the upper bound and lower bound

Fig. 12. Contour plots of resultant ground movements at Point F.


M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110 109

Cover C 1 Chimney
Chimney
C/D
0
Wedge
D
Prismatic
wedge
-1 0 1 2 3
Diameter D
Normalized distance from tunnel face (y/D)

Failure mechanism by Horn (1961) Failure observed in transparent soil models

s s

2
C
C

T
D D T

Multi-block failure mechanism by Mollon et.al., (2009) Two-block failure mechanism by Leca and Dormieux (1990)

Fig. 13. Comparison between theoretical failure mechanisms and failure observed in transparent soil.

solutions proposed by Leca and Dormieux (1990). This nding  Transparent soil used in these model tests has a smaller unit
also conrms Chambon and Cort (1994) and Leca and Dormi- weight than natural soils, but has a similar stress-strain behavior.
eux (1990) conclusion that the upper bound solutions are closer The discrepancy in unit weight may affect the scaling of stress
to the actual pressures at failure than the lower bound values, based problems such as scaling of liner deformations or soil rein-
and can provide a reasonable estimate of critical tunnel face forcement, to natural soils. For deformational problems such as
pressures (rTC). Theoretically, the upper-bound theorem (kine- the ones presented here, it is believed that the role of geometry
matical approach) provides a critical (unsafe) estimate of the is more signicant than that of self-weight, but small differences
tunnel support pressure required to maintain stability. in failure geometry may occur due to the discrepancy in unit
Collapse was triggered, by surface loading unlike Chambon weight. At any rate a surcharge has been used in lieu of the self
and Cort (1994) experiments and the assumptions employed weight to minimize the effect of the small unit weight.
in theoretical solutions proposed by Anagnostou and Kovari  Model tests performed in this study did not investigate the
(1996) and Leca and Dormieux (1990). This fact does not change effect of gravity and assumed it to be minimal. Therefore, a sur-
signicantly the analysis of the results due to several reasons: face loading (69 kPa) was introduced, by means of air pressure
First, because the comparison was based on the state of stress contained inside a rubber tire, in order to trigger instability
represented by the ratio of the support pressure to the vertical and to create a situation where a measurable pressure is needed
effective stress at the tunnel axis (rT/rV). Second, SG1 has a sim- for stabilizing the face. Such approach is necessary for two rea-
ilar stress-strain behavior to sand. Finally, face collapse resulted sons: First, because the unit weight of silica gel is smaller than
from passive failure due to insufcient support pressure at the natural sands by approximately 40%. Second, the small dimen-
face. Therefore, estimating the minimum required support pres- sion of the model yields negligible magnitude for the quantity
sure based on effective stress, is reasonable. of the gravitational force (i.e., cH = 0.91 kPa).
The value of rTC could have been inuenced by the boundary  Although the centrifuge is better at scaling the parameters
conditions of the small scale experiment, but the consistency of involved in tunneling the method presented in this paper is less
the results with previous studies is re-assuring. In any case, the costly, and permits visualizing internal deformations without
results point to the importance of holding or increasing the face the use of sensors that may affect the measured deformations.
pressure above the value of rT/rV > 0.1 in order to stabilize the The technique can be employed with a centrifuge but with added
face and prevent progression of ground failure. difculty and cost. For the time being surface loading is employed
3. Failure Stage (rT/rV < 0.5ka): This stage represents total failure. as a means of increasing the effective stress. Surface loading has
Face collapsed when transparent soil slide into the tunnel and been successfully employed in the past with natural soils to over-
large displacements occurred without decreasing the support come model size effects in a variety of 1 g model tests.
pressure. Collapse occurs suddenly, with the soil owing into  The study was performed at a low conning stress, so the effect
the tunnel, which leads to a collapse chimney developing in of dilation may have been over emphasized.
front of the tunnel face. It would be difcult to control failure  The effect of the pore pressure was small due to the small size of
once collapse begins. the model which yields negligible magnitude for the quantity
(cuid  H) where H is the depth to tunnel axis and cuid is the
7. Limits of the study unit weight of pore uid.
 SG1 particle size (0.150.5 mm) is somewhat large for the tun-
Although it has been shown that tunneling process can be well nel diameter (25 mm) used in these test. The size effect of the
modeled by transparent soils, the use of the above-described pro- particles is probably relevant and may have been over
cedure has some limitations as follows: emphasized.
110 M. Ahmed, M. Iskander / Tunnelling and Underground Space Technology 27 (2012) 101110

 Soil-tunnel interaction was studied with one tunnel size only, Atkinson, J.H., Potts, D.M., 1977. Stability of a shallow circular tunnel in
cohesionless soil. Gotechnique 27 (2), 203215.
like most studies. Use of different tunnel sizes would conrm
Broere, W., 1998. Face stability calculation for a slurry shield in heterogeneous soft
the lack of a size effect. soils. In: Proc., World Tunnel Congress 98 on Tunnels and Metropolises, vol. 1.
Balkema, Rotterdam, The Netherlands, pp. 215218.
8. Conclusions Chambon, P., Cort, J.F., 1994. Shallow tunnels in cohesionless soil: stability of
tunnel face. Journal of Geotechnical Engineering 120 (7), 11481165.
Chambon, P., Corte, J.F., Garnier, J., 1991. Face stability of shallow tunnels in
The evaluation of the stabilizing face pressure in mechanized granular soils. In: Proceedings of an International Conference on Centrifuge. A.A.
tunneling is a critical aspect for both, design and construction Balkema, Rotterdam, pp. 99105.
Clough, W., Leca, E., 1993. EPB shield tunneling in mixed face conditions. ASCE
phases. It has direct impact on ground movements and consequently Journal of Geotechnical & Geoenvironmental Engineering 119 (10), 16401656.
the safety of nearby structures and utilities. Potential problems Dantec Dynamics, 2001. Flow Manager Software User Guide. Tonsbakken, DK.
caused by instability and associated ground movement could lead Davis, E.H., Gunn, M.J., Mair, R.J., Seneviratne, H.N., 1980. The stability of shallow
tunnels and underground openings in cohesive material. Geotechnique 30 (4),
to public protests and eventually may impact the course of the pro- 397416.
ject. In this paper an experimental technique and procedures were Eisenstein, A.R., Ezzeldine, O., 1994. The role of face pressure for shields with
developed to analyze the progression of ground movements with positive ground control. Tunneling and ground conditions. Balkema, Rotterdam,
pp. 557571.
decreasing tunnel face pressure. A tunnel is pre-placed in transpar- Hagiwara, T., Grant, R.J., Calvello, M., Taylor, R.N., 1999. The effect of overlying strata
ent soil resembling medium dense saturated sands. Tunnel face sup- on the distribution of ground movements induced by tunneling in clay. Soils
port is simulated using an internal pressure applied inside the and Foundations 39 (3), 6373.
Hisatake, M., Eto, T., Murakami, T., 1995. Stability and failure mechanisms of a
tunnel. Tests are conducted by reducing the tunnel pressure in
tunnel face with a shallow depth. In: Fujii, T., (Ed.), Proceedings of the 8th
stages until collapse of the soil occurs. Images of the soil, illuminated Congress of the International Society for Rock Mechanics, pp. 587591.
by a laser light sheet, were captured after each decrement of tunnel Horn, M., 1961. Horizontal earth pressure on perpendicular tunnel face. In:
pressure and used to obtain corresponding 2D deformation elds. Proceedings of the Hungarian National Conference of the Foundation Engineer
Industry, Budapest, Hungary, pp. 716.
Four tests were performed at the ordinary gravitational acceleration Iskander, M., 2010. Modeling with Transparent Soils, Visualizing Soil Structure
(1 g) with various cover to depth ratios (C/D = 1.54.5). The use of Interaction and Multi Phase Flow, Non-Intrusively. Springer Verlag, Heidelberg,
transparent soils permitted visualization of actual internal ground Germany.
Iskander, M., Sadek, S., Liu, J., 2002. Optical measurement of deformation using
deformations under the test conditions. The tests results were then transparent silica gel to model sand. International Journal of Physical Modeling
compared with centrifuge tests performed at various gravitational in Geotechnics 2 (4), 1326.
accelerations (P1 g) reported by Chambon and Cort (1994) and Kirsch, A., 2010. Experimental investigation of the face stability of shallow tunnels
in sand. Acta Geotechnica 5, 4362. doi:10.1007/s11440-010-0110-7.
Oblozinsky and Kuwano (2006) and the analytical models proposed Leca, E., Dormieux, L., 1990. Upper and lower bound solutions for the face
by Leca and Dormieux (1990) and Mollon et al. (2009) and found to stability of shallow circular tunnels in frictional material. Geotechnique 40
be in a good agreement. (4), 581606.
Leca, E., Leblais, Y., Kuhnhenn, K., 2000. Underground works in soil and soft rock
A minimum support pressure, rT, was achieved with pressures as tunneling. In: Proc., Int. Conf. on Geotech. and Geolog. Eng., vol. 1. Australian
low as 10 1% of the effective vertical stress, rV, at the tunnel axis, Geomechanics Society, Institution of Engineers, Melbourne, Australia, pp. 220
with C/D having a small effect on the magnitude of the pressure at 268.
Lee, C.J., Wu, B.R., Chen, H.T., Chiang, K.H., 2006. Tunneling stability and arching
collapse. The observed ground deformations can be divided into
effects during tunneling in soft clayey soil. Tunnelling and Underground Space
three stages in relation to lateral earth pressure coefcient, ka, rep- Technology 21 (2), 119132.
resenting (1) Face Deformation (1 > rT/rV > 0.8ka), involving little Liu, J., Iskander, M., 2004. Adaptive cross correlation for imaging displacements in
or no deformation; (2) Face Slip Stage (0.8ka > rT/rV > 0.5ka) involv- soil. Journal of Computing in Civil Engineering, ASCE 17 (2), 8896.
Meguid, M.A., Saada, O., Nunes, M.A., Mattar, J., 2008. Physical modeling of tunnels
ing increasing but stable deformation; and (3) Failure Stage (rT/ in soft ground: a review. Tunnelling and Underground Space Technology 23,
rV < 0.5ka) where excessive irrecoverable deformations occurred. 185198.
The observed failure mechanisms resembled a prismatic wedge Mollon, G., Dias, D., Soubra, A.-H., 2009. Probabilistic analysis and design of circular
tunnels against face stability. International Journal of Geomechanics 9 (6), 237
in front of the tunnel face extending upward in the form of a vertical 249.
chimney conrming the mechanisms proposed by Horn (1961), and Oblozinsky, P., Kuwano, J., 2006. Centrifuge experiment on stability of tunnel face in
Chambon and Cort (1994). Ground movements at collapse trans- sandy ground. In: Proceedings of the 5th International Conference on
Geotechnical Aspects of Underground Construction in Soft Ground, TC28 of
mitted to the surface with different extents depending on C/D value. the ISSMGE, The Netherlands, pp. 271275.
The chimney failure outcropped at the ground surface for C/D < 2. Peck, R.B., 1969. Deep Excavations and Tunneling in Soft Ground. In: Proceedings of
These results are believed to be of practical interest for predict- the 7th International Conference on Soil Mechanics and Foundation
Engineering, Mexico City, State-of-the-art Volume, pp. 225290.
ing surface and internal ground deformations induced by tunneling Quebaud, S., Sibai, M., Henry, J.P., 1998. Use of chemical foam for improvements in
under various support pressure scenarios. Internal ground defor- drilling by earth pressure balanced shields in granular soils. Tunnelling and
mations are particularly important for ensuring stability of under- Underground Space Technology 13 (2), 173180.
Sadek, S., Iskander, M., Liu, J., 2002. Geotechnical properties of transparent material
ground utilities.
to model soils. Canadian Geotechnical Journal 39, 111124.
Acknowledgements Sadek, S., Iskander, M., Liu, J., 2003. Accuracy of digital image correlation for
measuring deformations in transparent media. ASCE Journal of Computing in
Civil Engineering 17 (2), 8896.
Transparent soils were originally developed with NSF support Sterpi, D., Cividini, A., Sakurai, S., Nishitake, S., 1996. Laboratory model tests and
under Grant No. CMS 9733064. Continued NSF support under Grants numerical analysis of shallow tunnels. In: Proc. Eurock 1996, Balkema, pp. 689
696.
DGE 0337668 and DGE 0741714 is gratefully acknowledged. Re- Takano, D., Otani, J., Nagatani, H., Mukunoki, T., 2006. Application of X-ray CT
search using transparent soils is presently funded by the Defense boundary value problems in geotechnical engineering research on tunnel face
Threat Reduction Agency (DTRA) Grant No: HDTRA1-10-1-0049. failure. In: Proc., Geocongress 2006, ASCE, Reston, VA.
Vinaia, R., Oggeria, C., Peila, D., 2008. Soil conditioning of sand for EPB applications:
a laboratory research. Tunnelling and Underground Space Technology 23 (3),
References 308317.
Ward, W.H., Pender, M.J., 1981. Tunneling in soft ground. In: Proc. 10th Conf. Soil
Ahmed, M., Iskander, M., 2011. Analysis of tunneling induced ground movements Mech. Found. Eng., Stockholm, vol. 4, pp. 261-275.
using transparent soil models. Journal of Geotechnical and Geoenvironmental Wu, B.R., Lee, C.J., 2003. Ground movement and collapse mechanisms induced by
Engineering, ASCE 137 (5), 525535. tunneling in clayey soil. International Journal of Physical Modeling in
Anagnostou, G., Kovri, K., 1994. The face stability of slurry-shield driven tunnels. Geotechnics 3 (4), 1327.
Tunnelling and Underground Space Technology 9 (2), 165174. Yoo, C., Shin, H., 2003. Deformation behaviour of tunnel face reinforced with
Anagnostou, G., Kovari, K., 1996. Face stability conditions with earth-pressure- longitudinal pipeslaboratory and numerical investigation. Tunnelling and
balanced shields. Tunnelling and Underground Space Technology 11 (2), 165173. Underground Space Technology 18 (4), 303319.

Das könnte Ihnen auch gefallen