You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265741030

Ground movement and tunnel stability when tunneling in sandy ground

Article  in  Journal of the Chinese Institute of Engineers · November 2004


DOI: 10.1080/02533839.2004.9670957

CITATIONS READS

34 1,355

3 authors, including:

Chung-Jung Lee
National Central University
67 PUBLICATIONS   1,038 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Earthquake-induced permanent settlement of shallow foundation of bridges rested on liquefiable sand: centrifuge and numerical modeling View project

All content following this page was uploaded by Chung-Jung Lee on 13 March 2015.

The user has requested enhancement of the downloaded file.


Journal of the Chinese Institute of Engineers, Vol. 27, No. 7, pp. 1021-1032 (2004) 1021

GROUND MOVEMENT AND TUNNEL STABILITY WHEN


TUNNELING IN SANDY GROUND

Chung-Jung Lee*, Kuo-Hui Chiang, and Chia-Ming Kuo

ABSTRACT
Tunneling may cause ground movements and damage to adjacent buildings and
overlying facilities. In this study, the failure mechanism and ground movement
behavior around tunnels embedded in sandy soil below the water table were investi-
gated in a series of model tunnel tests in a centrifuge. The magnitudes and extents of
the surface settlement troughs for the cases of various ground loss for tunnels buried
at various depths are provided. A new failure mechanism is proposed and validated
by comparison with the test results. The proposed mechanism enables accurate pre-
diction of two of the key quantities in the design of linings for tunnels embedded in
sandy soils, namely the minimum supporting pressure needed to retain tunnel stabil-
ity and the vertical soil pressure acting on the tunnel crown.

Key Words: failure mechanism, surface settlement trough, tunnel.

I. INTRODUCTION Various research groups have performed 1 g


model tests and/or centrifuge model tests to evaluate
Increasing traffic congestion in urban areas in tunnel stability and soil movement in soft ground
Taiwan has prompted transportation authorities to (e.g., Atkinson and Potts, 1977; Zhou et al., 1998).
propose the construction of a mass rapid transit sys- The collapse mechanisms derived from the upper-
tem to slow the growth in traffic volume. When such bound theory of plasticity for a single tunnel and par-
subway systems are constructed in soft ground, shield allel tunnels buried in clayey soils were correlated
tunneling has become a popular method on account with the observed ground movements in the model
of its low impact on traffic congestion in the streets tests (Lee et al., 1999; Wu and Lee, 2003).
above the tunnels. The construction of subways has Nomoto et al. (1998) developed a miniature cen-
also been driven through restricted public areas be- trifugal shield tunneling machine to simulate the ac-
neath existing streets or buildings due to the high tual shield tunneling process as closely as possible in
density of development and the difficulty of land ex- the centrifuge. The sophisticated and bulky tunneling
propriation in crowded cities. machine occupied too much room in the container;
When a tunnel is excavated, the ground around therefore, the model tunnel was limited to a small size
the tunnel inevitably moves towards the tunnel open- (2.5 m diameter in prototype scale) and was only
ing. The engineer must estimate the margin of safety tested at an acceleration of 25 g. In their shield tun-
against tunnel collapse and evaluate the magnitude nel tests, the shield tube was pulled back to expose
of the soil deformation and its effects on nearby struc- the inner tube and the clearance between the shield
tures and public lifeline facilities. The stability of a tube and the inner tube created a tail void. Even so
tunnel with temporary support (e.g., fluid pressure sophisticated a machine used in tests was still not
from air or bentonite slurry) and the deformations that capable of exactly modeling the actual tunnel con-
occur during excavation are the primary concern of struction process. They reported that the experimen-
the engineer. tal equation of the settlement trough to express the
surface settlement profiles was a function of the tail
void ratio (the difference between the radius of the
*Corresponding author. (Tel: 886-3-4227151 ext 34135; Fax:
886-3-4252960; Email: cjleeciv@cc.ncu.edu.tw)
shield tube and that of the inner tube divided by the
The authors are with the Department of Civil Engineering, Na- radius of the shield tube) and cover-to-diameter ratio
tional Central University, Chungli, Taiwan 320, R.O.C. (C/D).
1022 Journal of the Chiness Institute of Engineers, Vol. 27, No. 7 (2004)

100g

D
480
30
10
Tunnel

Rubber bag Tunnel deformation gauge


820
LVDT Colored sand Tunnel deformation
layer gauge
Spaghetti Potentiometer
Unit: mm
Fig. 1 Rubber bag and tunnel deformation gauge
Fig. 2 Test setup and instrumentation configuration used in the
model tests

The construction operation for shield tunneling


consists of excavation of a tunnel face using miners
and machinery protected within a shield. A perma- II. CENTRIFUGE MODELING
nent tunnel lining is erected with the tail skin as the
shield advances. An examination of field records of Experiments were carried out in the geotechnical
subsidence near soft ground tunneling operations in- centrifuge at the National Central University (NCU),
dicates that the majority of soil deformations occur Taiwan. This geotechnical centrifuge has a nominal
due to the ground loss during construction. The radius of 3 m and is capable of accelerating a 1-tonne
ground loss caused by overcutting, due to (1) the dif- model package to 100 g, and 0.55 tonne to 200 g. All
ference between the diameter of tunnel machines and tests reported here were conducted at an acceleration
that of the lining; (2) alignment problems encountered of 100 g. The layout and dimensions of the model
when steering the shield; (3) the effectiveness of tail systems are described in model units; however, the
void grouting; (4) three dimensional soil movements test results are presented in terms of prototype units.
ahead of the tunnel face. The net effect of these fac-
tors may be approximately incorporated into a 2-D 1. Model Tunnel
plane strain case in terms of annual voids. In the study,
the annual void is defined as the tail void. Because The model tunnel was a 1.5-mm-thick cylindri-
the construction process is very complicated, it is ob- cal rubber bag on which a sheet of filament tape of
viously impossible to duplicate all the details of the thickness 0.1 mm was pasted to enhance tunnel stiff-
tunnel process within a small-scale centrifuge model. ness in an attempt to ensure that the tunnel maintained
Approximations need to be made in the model so that a diameter of 60 mm during increase of supporting
key features in engineering practice can be easily in- pressure. It represented a 6-m diameter tunnel at pro-
vestigated. If the model is constructed properly, it totype scale. Four deformation gauges, made of four
will be useful for examining the mechanisms under- thinly sliced strain-gauge cantilevers, were placed
lying tunnel stability and ground deformation. Hence, inside the rubber bag (see Fig. 1). The free end of
it should be possible to evaluate the surface settle- each cantilever was bonded to the appropriate posi-
ment trough caused by tunneling by constructing a tion (see Fig. 2) on the inner surface of the bag, mak-
relation between the subsidence above a shallow tun- ing it possible to measure deformations of the crown,
nel and the magnitude of the ground loss. In the invert, and two sidewalls of the tunnel during the tests.
present study, a series of centrifuge model tunnel tests
was conducted at an acceleration of 100 g on systems 2. Preparation of Saturated Sand Bed
with tunnels embedded in sandy soil at various cover-
to-diameter ratios. Under these conditions, the The model tunnel and the sand around it were
behavior of a 60-mm diameter model tunnel should contained in a rigid strong box constructed from
mimic the behavior of a 6-m diameter tunnel in the aluminum alloy (see Fig. 3). The test setup is illus-
Earth’s gravitational field. This paper examines the trated in Fig. 2. The internal dimensions of the strong-
stability of, and surface subsidence above, shallow box were 820 mm (length) × 225 mm (width) × 480
tunnels in sandy ground below the water table. mm (height). To reduce wall friction, the surfaces of
C. J. Lee et al.: Ground Movement and Tunnel Stability When Tunneling in Sandy Ground 1023

Table 1 Characteristics of the quartz sand used


in the model tests
D 50, in γ max1 γ min1
Sand Gs Cu
(mm) (kN/m ) (kN/m 3)
3

Quartz sand 2.65 0.18 1.58 16.6 14.1


1
The maximum and minimum sand densities were
measured in the dry state according to the method
(JSF T 161-1990) specified by the Japanese
Geotechnical Society.

rose to 2 mm above the sand surface. This saturation


process required 7 hours. The acrylic plate was then
removed and six LVDTs (linear variable displacement
transformers) and one potentiometer (PM) were at-
tached to a mounting unit to measure surface settle-
ments (Fig. 2).
Fig. 3 Sand bed preparation using traveling pluviation apparatus
3. Test Procedures

The model tunnel was connected to an air pres-


the strongbox walls were thoroughly greased and thin sure line and the centrifuge was accelerated to 100 g
latex membranes were attached. The front and back in steps of 5 g. The air pressure was carefully regu-
faces of the strongbox imposed a condition of plane lated to balance the overburden pressure at the tun-
strain on the soil, and the bottom and side faces were nel center to ensure that no crown settlement occurred
sufficiently far removed from the model tunnel that during the acceleration of the centrifuge. Two pore
they did not influence the behavior of the soil around water pressure transducers were used in the pilot tests,
the tunnel. After the model tunnel was set at the de- which were located at the middle of the sand bed near
sired elevation, crushed quartz sand was pluviated the model tunnel at different elevations, to monitor
into the strongbox with a regular path from a hopper the elapsed time needed to reach full self-weight con-
from a falling height of 700 mm and at a constant solidation and pore water pressure changes during
flow rate (Fig. 3). Details of the apparatus are given the tunnel collapse tests later on in the sand bed at an
elsewhere (Chen et al., 1998). The characteristics of acceleration of 100 g. The test results show that it
the crushed quartz sand are summarized in Table 1. took about 2 minutes to reach full consolidation at an
A fairly uniform sand bed (D r=65±2%) was achieved acceleration of 100 g and no excess pore water pres-
and its friction angle was around 38°. The pluviating sure was monitored while reducing the air pressure
process was interrupted as needed to rain a layer of at a rate of 20 kPa/min. in the rubber bag. In the later
colored sand at specified elevations, as shown in Fig. tests, no pore water transducer was used because the
2. Finally a row of marked spaghetti with gradua- embedded transducers would considerably interfere
tions of 10 mm was vertically implanted into the sand with the ground deformation during tunnel collapse.
bed along the central line of the model at regular, pre- But all the test packages in the later tests stayed 5
determined positions. Spaghetti was used because, minutes at an acceleration of 100 g to ensure the sand
after absorbing water, it softens and deforms together bed reaching full consolidation before reducing the
with the surrounding sand. The colored sand layers air pressure in the model tunnel. The tunnel diam-
(2 mm in thickness) and the implanted spaghetti were eter was then reduced by gradually lowering the air
good indicators of the ground deformations caused pressure to zero at a rate of 20 kPa/min. Tunnel de-
by the imposed test conditions. formation and surface settlement were continuously
After the model was prepared, the package was measured during the test.
mounted on the centrifuge platform and the top of The tunnel stability and surface settlements dur-
the strongbox was tightly covered with an acrylic ing closure of the tail voids were assessed for tunnels
plate. A vacuum pump then continuously pumped embedded at various cover-to-diameter ratios (C/D=1,
air out of the strongbox and the model tunnel simul- 2, 3, and 4, where C is the distance between the ground
taneously. At the same time, de-aired water was care- surface and the tunnel crown, and D is the tunnel di-
fully dripped into the strongbox until the water level ameter, and z is the depth of the tunnel axis), as shown
1024 Journal of the Chiness Institute of Engineers, Vol. 27, No. 7 (2004)

Table 2 Test arrangements and the estimated i/D


Test No. C/D Depth, z (m) Estimated i/D
STEST1 1 9 0.94
PTEST2 1 9 0.97
STEST4 2 15 1.56
PTEST3 2 15 1.51
STEST3 3 21 1.73
STEST5 4 27 1.92

respectively. The determination of p c will be ex-


Fig. 4 Surface and subsurface settlement troughs plained in detail in the next section. As the tunnel
stability varies from stable to critical, the value of
LF varies from 0 to 1.
As the supporting pressure was reduced during
in Fig. 4. Details of the test arrangements and condi-
the model tunnel test, the crown settled, leading to
tions are given in Table 2.
an increase in the magnitude of the surface settlement
The sand bed was cautiously excavated to ex-
measured by each of the LVDTs. Fig. 5 shows the
pose the implanted spaghetti and the colored sand
plots of the measured surface settlements along a tun-
layers after the model test. The graduations on the
nel transverse section versus the supporting pressure,
spaghetti and the positions of the colored sand layers
p, and the plots of the deformations at the crown, in-
were then traced onto a transparency, and changes in
vert, and side walls versus p, for the tunnel with C/D
the positions of these marks before and after testing
= 2. The LVDT array and the positions of the tunnel
revealed ground displacements around the tunnels.
deformation gauges (gauges 1, 2, 3, and 4) are dis-
These measurements allowed investigation of the col-
played on the right-hand side of Fig. 5. The crown
lapse mechanism.
deformation, δ c, and the surface settlements at vari-
ous lateral distances from the centerline, S(x), increase
III. TEST RESULTS
slowly with decreasing supporting pressure up to the
onset of tunnel collapse, after which they increase
1. Surface Settlement Profile
rapidly. The following procedure was used to define
the supporting pressure at collapse, p c. Extending the
The transverse surface settlement trough is com-
straight-line portions of the first and the second parts
monly approximated by a Gaussian distribution (Peck,
of the S(x) ~ p curve measured at the tunnel centerline
1969):
to intersect at the point as shown in Fig. 5, the hori-
2 zontal ordinate of this critical point is defined as the
S(x) = S maxexp( – x2 ) (1) supporting pressure at collapse, p c.
2i
Figure 6 shows the surface settlement troughs
where Smax is the maximum surface settlement at the measured in STEST4 (represented by symbols) to-
tunnel center-line, x is the lateral distance from the gether with the settlement troughs estimated from Eq.
tunnel center-line, and the width parameter i is the (1) (represented by solid lines) at various supporting
distance from the center-line to the inflection point pressures along with the corresponding load factors.
of the curve and may be employed as a measure of Comparison of the measured and predicted data re-
the shape and extent of the settlement trough (see Fig. veals that Eq. (1) quite accurately predicts the settle-
4). Two and half times i is commonly used to repre- ment trough. The estimated value of i did not appear
sent the half width of the settlement trough. In prac- to change during a particular model test before the
tice, a key concern of construction engineers is ground tunnel collapsed. However, as shown in Table 2, the
movement prior to tunnel collapse. Hence, a load magnitude of i derived from the model tests increased
factor (LF), defined as below, is regarded as the re- with increasing tunnel burial depth. Numerous em-
ciprocal of the safety factor. pirical methods have been developed to evaluate the
width of the settlement trough. For instance, Mair et
σ –p
LF = σ vo – p (2) al. (1981) proposed the formula for clay soil
vo c

where σ vo is the overburden pressure at the depth of 2i = z (3)


D D
the tunnel axis, and p and p c are the current support-
ing pressure and the supporting pressure at collapse, where z is the depth of the tunnel axis, and Clough
C. J. Lee et al.: Ground Movement and Tunnel Stability When Tunneling in Sandy Ground 1025

Supporting pressure, (kPa)


pc
300 250 200 150 100 50 0
0.0
LVDT48 LVDT44 LVDT43
Surface settlement, (m)

0.5 STest4 (C/D=2) LVDT41 LVDT47 LVDT42 PM5

LVDT41
1.0 LVDT42
LVDT43 6.5 4.5 4 4 4.5 6.5
LVDT44
1.5
LVDT47
LVDT48 unit: m 12
2.0 PM5

2.0 6
Tunnel deformation, (m)

1.5 Gage1
Gage2
Gage3
1.0 Gage4
Gage1

0.5
Gage4 Gage2

0.0 Gage3
300 250 200 150 100 50 0
Supporting pressure (kPa)

Fig. 5 Surface settlements and tunnel deformations at various supporting pressures

Normalized distance from tunnel center-line, X/D 2i/D


0 1 2 3 4 5
-5 -4 -3 -2 -1 0 1 2 3 4 5 0
0.000 Eq. (4)
Normailzed settlement, S/D

Eq. (3)
STEST4 Peck (1969)
0.005 (C/D=2) 1 peck (1969)
Eq. (5)
0.010 Eq. (7) in the study
Pi LF Eq. (6)
308.03 kPa 0.5 2 Chang et al. (1997)
0.015 288.41 kPa 0.6 measured in the study
z/D

268.79 kPa 0.7


0.020 249.17 kPa 0.8 3 Rock, hard clays,
229.55 kPa 0.9 Estimated from Eq. (1) sands above
210.10 kPa 1.0 ground water table
0.025 Sands below
4 ground water table
Fig. 6 Evolution of surface settlement trough with changing sup-
porting pressure (STEST4 C/D=2)
5
Soft clay to
stiff clays

and Schmidt (1981) suggested the empirical relation Fig. 7 Relations of the width of the settlement trough and the
in clay tunnel depth

2i = ( z ) 0.8 (4)
D D embedded in soft clayey soils:
Atkinson and Potts (1977) derived the following em-
2i = 1 + 0.58( z ) (6)
pirical equation from centrifuge model tests on a tun- D D
nel buried in dry sand
Figure 7 shows the above empirical relations be-
i=0.25(C+D) (5) tween the tunnel depth and the width of the settle-
ment trough [i.e., Eqs. (3), (4), (5), and (6)] along
and Lee et al. (1999) derived yet another empirical with the test results obtained in the present study
relation from centrifuge model tests on a tunnel (solid circles). Also shown in this figure are the
1026 Journal of the Chiness Institute of Engineers, Vol. 27, No. 7 (2004)

Load factor, LF 1.0


Measured in this study
0.0 0.2 0.4 0.6 0.8 1.0 Smax/ δc=0.29+0.97exp(-0.50*C/D)
0.000 R2=0.98
0.8 dry sand (Atkinson & Potts, 1979)
Soft clay (Lee, et al., 1999)
0.005
0.6

Smax/δ c
STEST4 (C/D=2)
Smax/D, δ c/D

0.010
0.4

0.015
Smax/D 0.2
δ c/D
0.020
0.0
0 1 2 3 4
0.025 Cover-to-diameter ratio, C/D

Fig. 8 Load factor versus S max /D and δ c/D Fig. 9 Relation of S max / δ c and cover-to-diameter ratios

curves derived by Peck (1969), who obtained the factor of 0.6, which corresponds to a safety factor of
value of i from field observations of settlement caused 1.7. However, on further increase of the load factor,
by tunneling in a variety of soils, and related these the values of both S max/D and δ c /D measured in sandy
values empirically to the burial depth and the soil soil increase faster than has been previously observed
type. The i values measured in the present study lie in experiments on model tunnels embedded in clayey
above the upper bound for clayey soils suggested by soils (Lee et al., 1999). Similar findings were ob-
Peck, indicating that the settlement trough of a tun- tained in model tests on tunnels with different cover-
nel embedded in sandy ground below the water table to-diameter ratios. Fig. 9 shows the variation of the
is wider than that in clayey soils. On the basis of the ratio of S max to δ c in a tunnel with changing cover-to-
measurements reported here, we developed the fol- diameter ratio obtained from the model tests. The
lowing empirical relation for the settlement trough measured data are well fitted by the expression:
width of tunnels embedded in sandy soils below the
water table: S max
= 0.29 + 0.97 × exp(– 0.50 × C ) (8)
δc D
( 2i ) = 1.602( z ) 0.6113 (7) Comparison of the present results with those obtained
D D
by other researchers (Lee, et al., 1999; Atkinson and
This empirical relation shows good agreement with Potts, 1977), which are also shown in Fig. 9, reveals
field data (represented by hollow squares in Fig. 7) that the sandy soil considered here gives larger values
measured at Contracts CH218, CT210, and CH221 of Smax/ δc. This indicates that the closure of tail voids
of the Taipei MRT project (Chang et al., 1997; Ou et in a tunnel embedded in sandy ground below the wa-
al., 1998), where the tunnels were mainly driven ter table produces a greater degree of surface settle-
through sandy ground. This agreement confirms that ment than in a tunnel embedded in other soil types.
the proposed relationship can be used to predict the The magnitude of δ c depends on many factors,
width of surface settlement troughs in sandy ground including the volume of the tail voids, construction
below the water table in engineering practice. method, quality of workmanship, the presence or lack
of back grouting, and the time delay of grouting op-
2. Maximum Surface Settlement and Crown De- erations. However, the crown settlement may be more
formation difficult to evaluate in engineering practice than in
the model tests. Instead, contractors can estimate the
The maximum surface settlement, S max, and the ground loss in advance by taking into account past
crown deformation, δ c , were continuously measured experience and the tunneling method used. In fact,
during the model tests. The measured crown settle- designers often use the ground loss, V t, in place of
ment can be taken as a measure of the annulus clear- δ c. The ground loss is defined as the ratio of the soil
ance between the cutting surface and the lining dur- volume flowing into the tunnel, ∆ V, during tunneling
ing tunneling. The closure of the clearance (tail voids) divided by the volume of the tunnel per unit tunnel
was the primary cause of ground movement. Fig. 8 length, V, expressed as a percentage. Prior to tunnel
shows the dependence of S max/D and δ c/D on the load collapse, the magnitude of δ c is relatively small com-
factor for STEST4 (C/D=2). Both S max/D and δ c/D pared with the tunnel diameter; therefore, V t can be
vary linearly with increasing load factor up to a load written as:
C. J. Lee et al.: Ground Movement and Tunnel Stability When Tunneling in Sandy Ground 1027

Load factor, LF
0.0 0.2 0.4 0.6 0.8 1.0 0.020
0 CH218 (Ou et al., 1998) Vt=0.5%
CH221 (Chu, 1995) Vt=1%
CT201 (Chang et al., 1997) Vt=2%
2
Ground loss, Vt (%)

0.015 CN218 (Chang et al., 1997) Vt=3%


4

Smax/D
STEST5 (C/D=4) 0.010
6 (2.20%)
STEST3 (C/D=3)
STEST1 (C/D=1) (1.73%) (2.38%)
8
0.005 (1.64%) (1.35%)
(1.11%)
10 (0.799%)

Fig. 10 Load factor versus ground loss at various cover-to-diam- 0.000


eter ratios 0 1 2 3 4
Cover-to-diameter ratio, C/D

π D 2 – π (D – 2δ ave) 2 4Dδ ave – δ ave


2
V t = ∆V × 100 =
Fig. 11 S max /D versus cover-to-diameter ratio at various values
= of ground loss
V πD 2 D2
400δ ave 220δ c
× 100 ≅ = (9) sandy ground below the ground water table, a stabil-
D D
ity number, N s, is defined as:
Here δ ave is defined as the average of the tunnel
deformations measured at the crown, invert, and two pc – pw pc′
Ns = = (11-a)
sidewalls ( δ ave ≈0.55 δ c in the model tests). As indi- σ v′ σ v′
cated in Fig. 10, the ground loss increased rapidly at
where pw and σ v′ are the measured pore water pres-
load factors larger than 0.7. The test results also show
sure and the effective overburden pressure at the level
that surface settlement increases rapidly once the
of the crown, respectively, and pc′ is the measured
ground loss exceeds 3%.
effective supporting pressure. For a tunnel buried in
Substituting Eq. (9) into Eq. (8) gives
sandy ground above the ground water table, Eq. (11-
S max V a) can be rewritten as:
( ) = (0.29 + 0.97exp(– 0.50 × C )) × t
D D 220 pc
Ns = (11-b)
γC
where 0≤V t ≤3% (10)
where γ is the unit weight of sand above the tunnel
Fig. 11 shows the solutions of Eq. (10) for the cases axis. A larger value of N s corresponds to a higher
of Vt=0.5%, 1%, 2%, and 3% for various burial depths, ratio of the supporting pressure to the overburden
together with the values of Smax measured at contracts pressure, which promotes tunnel stability. Table 3
CN218, CT201, CH218, and CH221 (Sung, 1995) of summarizes the supporting pressures at tunnel col-
the Taipei MRT project. Those tunnels were driven lapse and the values of Ns measured in the present
mainly through sandy ground. For each of the se- study. As shown in Fig. 12, the measured stability
lected contracts, the value of the ground loss esti- number (solid circles) decreases with increasing C/D
mated directly from the volume of the measured sur- ratio. The variation in Ns with changing C/D is well
face settlement trough is shown in parentheses next fitted by the equation
to the corresponding symbol. The results predicted
using the empirical equations developed in the present N s = 0.095 + 0.7198exp(– 0.6234 C )
D
study are in good agreement with the measurements
from the field projects. This agreement confirms that, C
for 1≤ ≤4 (12)
D
in situations where the ground loss can be determined
in advance, Eq. (10) together with Eq. (4) provide a The test results show no apparent decrease in N s as
viable empirical approach for predicting the magni- the C/D ratio is increased above a value of 3.
tude and extent of the surface settlement trough dur- Direct observation of the test by CCTV through
ing tunneling in sandy ground below the ground wa- the window of the strongbox revealed that, as the sup-
ter table. porting pressure was lowered, failure planes first ap-
peared slightly above the tunnel crown and then pro-
3. Tunnel Stability and Failure Mechanism gressively grew upward before finally reaching the
ground surface. The collapse mechanisms for tunnels
To evaluate the stability of a tunnel buried in buried at various depths (C/D from 1 to 4), as obtained
1028 Journal of the Chiness Institute of Engineers, Vol. 27, No. 7 (2004)

Table 3 Stability numbers and supporting pressures at collapse under the different experimental condi-
tions
Test pc pw pc′ Ns Ns
C/D Ns*
No. (kPa) (kPa) (kPa) (measured) Eq. (18)
STEST1 1 89.1 60.7 28.4 0.48 0.36 0.64
PTEST2 1 89.5 60.7 28.8 0.48 0.36 0.64
STEST4 2 155.1 117.6 37.5 0.32 0.28 0.44
PTEST3 2 154.6 117.6 37 0.31 0.28 0.44
STEST3 3 210.1 176.4 33.7 0.19 0.23 0.32
PTEST6 3 212.8 176.4 36.4 0.20 0.23 0.32
STEST5 4 274.9 235.2 39.7 0.17 0.19 0.25
*
The values of N s were calculated by substituting the effective supporting pressure at collapse into Eq. (11-a).

0.7

0.6
Stability number, Ns

0.5

0.4

0.3

0.2 Ns (meaured)
Ns (predicted with Eq. (24))
0.1 Ns (predicted with Eq. (18))
Ns=0.095+0.7198exp(-0.6234C/D) R2=0.99
0.0
0 1 2 3 4
Cover-to-diameter ratio, C/D

Fig. 12 Stability number versus cover-to-diameter ratio


Fig. 14 Deformation pattern on the colored sand layers for the tun-
nel embedded at a depth of C/D=4

implanted spaghetti and the colored sand layers for


the model tunnel buried at the depth of
C/D=4. The displacement field taken after the test for
a tunnel buried at the depth of C/D=4 is shown in Fig.
17. The displacement vectors developed in the area
confined by the dashed lines are mainly downward.
Taking this into account, the failure mechanism can
be simplified to the geometry depicted in Fig. 18.
The failure mechanism for a tunnel buried in
sand consists of three sliding wedges (Wedge abcd,
Wedge ebgh, and Wedge edfm), as depicted in Fig.
18. Wedge abcd moves vertically downward along
Fig. 13 Deformation pattern on the colored sand layers for the tun-
nel embedded at a depth of C/D=2
the sliding interfaces ab and cd. On the other hand,
Wedges ebgh and edfm move along the correspond-
ing sliding interfaces df and bg, where these sliding
from the movements of the colored sand layers (Fig. interfaces lie at a tangent to the tunnel. We can rea-
13 for C/D=2 and Fig. 14 for C/D=4) after the tests, sonably assume that reducing the supporting pressure
are depicted in Figs. 15-a to 15-d. The tunnels buried in the tunnel will result in active failures in Wedges
at greater depths show the highest number of failure edfm and ebgh. Therefore both the sliding interfaces
planes; however, regardless of the depth at which the bg and df were in the Rankine active state.
tunnel is buried, the region showing major downward Below we analyze only the mechanism of the right-
movement is confined to about ±5 m from the tunnel hand side of the system due to the symmetry of the
centerline. Fig. 16 shows the profile containing the system about the tunnel centerline. The magnitude of
C. J. Lee et al.: Ground Movement and Tunnel Stability When Tunneling in Sandy Ground 1029

Horizontal distance from tunnel center, (m) Horizontal distance from tunnel center, (m)
-20 -10 0 10 20 -20 -10 0 10 20
0 0

-5 -5

-10 -10
Elevation, (m)

-15 -15

-20 -20

-25 -25

-30 -30

(a) C/D=1 (c) C/D=3


Horizontal distance from tunnel center, (m) Horizontal distance from tunnel center, (m)
-20 -10 0 10 20 -20 -10 0 10 20
0 0

-5 -5
Elevation, (m)

-10 -10

-15 -15

-20 -20

-25 -25

-30 -30
(b) C/D=2 (d) C/D=4

Fig. 15 Tunnel failure mechanisms at various C/D raios

Distance from the the center line (m)


-20 -18 -16-14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20
0
2
4
6
8
10
12
Depth (m)

14
16
18
20 scale: 1m
22
24
26
28
Fig. 16 Profile containing the colored sand layers and the im- 30
planted spaghetti for a model tunnel buried at the depth of
C/D=4 Fig. 17 Velocity field of soil around a tunnel (C/D=4). Dashed
line marks the region within which the velocity vectors
are predominantly directed downwards
∠dfn is equal to 45°+1/2φ and the length of ed can be
expressed as
with an effective friction angle of 38°, the length of
B1 = ed = D (13) ed is 4.80 m according to Eq. (13). This value is rea-
π
2tan( + 1 φ) sonably consistent with the measured value of 5 m.
8 4
The effective vertical stress on a horizontal section
For a 6 m diameter tunnel embedded in sandy soil (Element ijkl in Fig. 18) at any depth z below the
1030 Journal of the Chiness Institute of Engineers, Vol. 27, No. 7 (2004)

Fig. 19 Stress state of an element on the sliding surfaces ab and


cd

Solving Eq. (15-b) we obtain

B1 γ ′
σ v′ = (1 – e – kz/B1tanφ ) (16)
k tanφ

Fig. 18 Failure mechanism for a tunnel buried in sandy ground The effective vertical pressure (loosening earth pres-
sure) on the elevation of the crown for a tunnel with
a cover thickness of C is given by
surface is σ v′ , and the corresponding effective normal B1 γ ′
(σ v′ ) z = C = (1 – e – kC/B1tanφ ) (17)
stress on the vertical sliding surface (interfaces ab k tanφ
and cd) is
This pressure can be regarded as the minimum sup-
σ h′ = kσ v′ (14) porting pressure needed to retain tunnel stability or
as the loosening soil pressure acting on the tunnel
where k is the ratio between the effective horizontal crown. By substituting Eq. (17) into Eq. (11-a), the
and the effective vertical pressures. stability number of the tunnel can be expressed as:
The weight of Element ijkl with a thickness dz
at a depth z is 2B 1 γ ′dz per unit of length perpendicu- (σ v′ ) z = C B1
Ns = = (1 – e – kC/B1tanφ ) (18)
lar to the plane of the drawing. The forces acting on γ ′C Ck tanφ
Element ijkl are indicated in Fig. 18. Invoking the
condition that the sum of the vertical components of Terzaghi (1943) suggested that k has a value of
the forces acting on Element ijkl must equal zero 1. However, the estimates of Ns as a function of tunnel
yields the expression burial depth (square symbols in Fig. 12) obtained from
Eq. (18) using k=1 differ considerably from the values
2B1γ ′dz = 2B1(σ v′ + dσ v′ ) – 2B1σ v′ + 2kσ v′ dz tanφ obtained in the model tests. This discrepancy prompted
us to develop a new approach to derive the stability
(15-a) number for the case of sandy ground based on analysis
or of the failure modes in the system shown in Fig. 18.
First, we examine the importance of the ratio k
dσ v′ tanφ in the evaluation of the stability number. The shear-
= γ ′ – kσ v′ (15-b)
dz B1 ing resistances are fully mobilized on the interfaces
ab and cd; thus, the stress state of a soil element on
If the sliding surfaces (ab and cd) that develop dur- the vertical plane along the interfaces ab and cd would
ing tunnel collapse reach the ground surface, as oc- be on the Mohr-Coulomb failure line, as shown in
curs for shallow tunnels, the effective vertical pres- Fig. 19. Hence the ratio k can be expressed as:
sure on the surface is equal to zero, i.e.,
sinφcosφ
k= (19)
σ v′ =0 for z=0 2tanφ – sinφcosφ
C. J. Lee et al.: Ground Movement and Tunnel Stability When Tunneling in Sandy Ground 1031

The variation of k with φ is quite obvious and it is N


equal to 0.451 for the sand ( φ =38°) used in the tests.
Thus, contrary to the suggestion of Terzaghi (1943)
that k=1, the value of k varies markedly with φ , tak-
e
ing on a value of k=0.451 for the sand used in the d
model tests ( φ =38°). Now, substituting k=0.451 into
45+1/2 φ
Eq. (17) to calculate the vertical pressures (σ v′ ) z = C
T1
and assuming that those pressures act uniformly on m
o
the interfaces ed and eb, the forces acting on Wedge
P
edfm are found to be those indicated in Fig. 20. The
forces per unit length perpendicular to the plane of N1
the drawing are as follows: f

B12γ ′ W
N = (σ v′ ) z = C × B1 = (1 – e – kC/B1tanφ ) (20)
k tanφ D

P = pc′D sin(67.5 – 1/4 φ) (21) Fig. 20 Forces acting on the wedge edfm

π (135 – 1/2φ)
W = 1 γ ′D 2[cot(π /8 + 1/4φ) – ] (22)
4 360
resulting stability numbers are listed in Table 3. As
T 1=N 1tan φ (23) indicated in Fig. 12, the stability numbers obtained
from Eq. (26) (hollow circles in Fig. 12) are in rea-
where N is the vertical force acting on the interface sonable agreement with those measured in the model
cd, W is the submerged weight of soil of Wedge edfm, tests. Thus, the proposed failure mechanism provides
P is the force provided by the supporting pressure at a viable method for estimating the supporting pres-
collapse, N 1 is the normal force acting on the sliding sure at collapse and the stability number when
surface fd, and T 1 is the frictional force mobilized on tunneling in sandy soils.
the sliding surface fd.
The force equilibrium equations for the vertical IV. SUMMARY AND CONCLUSION
and horizontal directions are expressed as
The ground movement behavior around tunnels
W+N=Psin(22.5+1/4 φ )+N 1cos(45+1/2 φ ) embedded in sandy soils below the ground water ta-
+T 1 sin(45+1/2 φ ) (24-a) ble was investigated in a series of model tunnel tests
in a centrifuge. The degree of ground movement was
Pcos(22.5+1/4 φ )+T 1cos(45+1/2 φ ) closely related to the load factor and increased dra-
matically when the load factor exceeded 0.7. The
=N 1sin(45+1/2 φ )] (24-b) relation between i and the ratio C/D was derived by
regression of the centrifuge model test data; this re-
Substituting Eq. (23) into Eqs. (24-a and 24-b), we lation can be used to estimate the width of the sur-
obtain the following two equations with two un- face settlement trough for a tunnel of a particular
knowns (P and N 1): depth. The maximum surface settlement can be evalu-
ated using the proposed relations of S max/D and C/D
Psin(22.5+1/4 φ )+N 1[cos(45+1/2 φ ) ratio at various ground loss. Importantly, the pro-
+tan φ sin(45+1/2 φ )]=W+N (25-a) posed relations are simple and easy to use in engi-
neering practice. A new failure mechanism was also
Pcos(22.5+1/4 φ )+N 1 [tan φ cos(45+1/2 φ ) proposed and validated by comparison with the test
results. The proposed failure mechanism enables ac-
–sin(45+1/2 φ )]=0 (25-b) curate prediction of two of the key quantities in the
design of linings for tunnels embedded in sandy soils,
After solving for the unknown P, the effective sup- namely the minimum supporting pressure needed to
porting pressure at collapse, pc′, is retain tunnel stability and the vertical soil pressure
P
acting on the tunnel crown.
pc′ = (26)
D sin(67.5 – 1/4 φ)
ACKNOWLEDGMENTS
Finally the stability numbers can be found by
substituting this estimate for pc′ into Eq. (11-a); the The authors gratefully acknowledge the financial
1032 Journal of the Chiness Institute of Engineers, Vol. 27, No. 7 (2004)

support of the National Science Council of the Repub- Tunnel Excavation by Shield Machine,” Techni-
lic of China (Grant NSC 89-2211-E-008-101). cal Report, Sinotech Engineering Consultants,
Ltd. (in Chinese).
NOMENCLATURE Chen, H. T., Lee, C. J., and Chen, W. H., 1998, “The
Traveling Pluviation Appartus for Sand Specimen
B1 half width of failure wedge, m Preparation,” Proceedings of the International
C distance between ground surface and tunnel Conference Centrifuge 98, Kimura, Kusakabe and
crown, m Takemura (eds.), Balkema, Tokyo, pp.143-148.
C/D cover-to-diameter ratio Clough, G. W., and Schmidt, B., 1981, “Excavation
Cu uniformity coefficient and Tunneling,” Soft Clay Engineering, Brand &
D tunnel diameter, m Brenner (eds.), Chap. 8, Elsevier, Amsterdam.
Dr relative density, % Lee, C. J., Wu, B. R, and Chiou, S. Y., 1999, “Soil
D 50 diameter corresponding to 50% finer, mm Movements around a Tunnel in Soft Soils,” Pro-
Gs specific gravity of soil solids ceedings of the National Science Council, Part
i width parameter, m A: Physical Science and Engineering. Vol. 23,
k ratio between effective horizontal and effec- No. 2, pp. 235-247.
tive vertical pressure Mair, R. J., Gunn, M. J., and O’Reilly, M. P., 1981,
LF load factor “Ground Movements around Shallow Tunnels in
N vertical force on Interface cd, kN Soft Clay,” Proceedings of the 10th International
N1 normal force on sliding surface fd , kN Conference on Soil Mechanics and Foundation
Ns stability number Engineering, pp. 323-328.
P force provided by supporting pressure at col- Nomoto, T., Imamura, S., Hagiwara, T., Kusakabe,
lapse, kN O., and Fujii, N., 1998, “Shield Tunnel Construc-
p supporting pressure, kPa tion in Centrifuge,” Journal of Geotechnical and
pc supporting pressure at collapse, kPa Geoenvironmental Engineering, ASCE, Vol. 125,
pc′ effective supporting pressure at collapse, kPa No. 4, pp. 289-299.
pw pore water pressure, kPa Ou, C. Y., Hwang, R. N., and Lai, W. J., 1998, “Sur-
S(X) surface settlement at lateral distance x from face Settlement during Shield Tunneling at
tunnel center-line, m CH218 in Taipei,” Candian Geotechnical Jour-
Smax maximum surface settlement, m nal, Vol. 35, No.1, pp. 159-168.
T1 horizontal force on sliding surface fd, kN Peck, R. B., 1969, “Deep Excavations and Tunneling
Vt ground loss, % in Soft Ground,” Proceedings of the 7 th Interna-
V volume of tunnel per unit tunnel length, m3/m tional Conference on Soil Mechanics and Foun-
W submerged weight of soil of wedge edfm, kN dation Engineering, Mexico City, State-of-the-
x lateral distance from tunnel center line, m Art Volume, pp. 225-290.
z depth of tunnel axis, m Sung, C. L., 1995, “Ground Movements due to Shield
∆V soil volume flowing into tunnel, m 3 Tunnelling of Construction CH221 for Taipei
γ unit weight of sand, kN/m 3 MRT System,” Master Thesis, Department of
γ max maximum unit weight of sand, kN/m 3 Civil Engineering, National Chiao Tung Univer-
γ min minimum unit weight of sand, kN/m 3 sity, Taiwan, R.O.C. (In Chinese)
γ′ submerged unit weight of sand, kN/m 3 Terzaghi, K., 1943, Theoretical Soil Mechanics,
δc crown deformation, m Willey, New York, USA.
δ ave average crown deformation, m Wu, B. R., and Lee, C. J. , 2003, “Ground Movements
σ vo overburden pressure, kN/m 2 and Collapse Mechanisms Induced by Tunneling in
σ v′ effective overburden pressure, kN/m 2 Clayey Soil,” International Journal of Physical
σ h′ effective horizontal stress, kN/m 2 Modelling in Geotechnics, Vol. 3, No. 4, pp. 13-27.
φ friction angle Zhou, X., Pu, J., and Yin, K., 1998, “A Study of Sta-
bility and Failure Mechanism of Sand around a
REFERENCES Tunnel,” Proceedings of the International Con-
ference Centrifuge 98, Kimura, Kusakabe and
Atkinson, J. H., and Potts, D. M., 1977, “Subsidence Takemura (eds.), Balkema, Tokyo, pp. 727-731.
above Shallow Tunnels,” Journal of Geotechnical
Engineering, ASCE, Vol. 103, No. 4, pp. 307-325. Manuscript Received: Jun. 12, 2003
Chang, C. T., Wang, J. J., Chen, C. C., and Wu, Revision Received: Jan. 06, 2004
T. C., 1997, “Acquisition and Interpretation of and Accepted: Feb. 09, 2004
Monitored Data of Ground Subsidence due to

View publication stats

You might also like