You are on page 1of 17

An analysis of three-dimensional ground movements: the Thunder Bay tunnel

K. M. LEEAND R. K. ROWE
Geotechnical Research Centre, Faculty of Engineering Science, University of Western Ontario,
London, Ont., Canada N6A 5B9
Received May 28, 1990
Accepted October 16, 1990

A three-dimensional (3D) elastoplastic finite-element analysis, which is capable of simulating the advance of a tunnelling
shield and the associated ground losses resulting from the tunnelling process, is used to calculate the deformations
caused by the excavation of the Thunder Bay sewer tunnel. The soil parameters adopted in the analysis were based
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

on the results determined from stress-dependent triaxial tests. The results of this analysis are compared with the measured
soil displacements. Reasonable agreement between the calculated and observed 3D settlement distribution and hori-
zontal displacements at different distances from the tunnel face is reported. This overall agreement for displacements
under 3D conditions suggests that the method of analysis may be applicable to design problems involving tunnelling
in soft clays similar to that at the Thunder Bay sewer tunnel provided that the soil parameters are reliably determined.
Key words: tunnelling, three-dimensional analysis, finite element, elastoplastic displacements, ground deformations,
Thunder Bay tunnel.

Une analyse tri-dimensionnelle (3D) Clasto-plastique en Climents finis, qui peut simuler I'avance du blindage d'un
tunnel et les pertes de sol qui risultent de l'opiration de percement d'un tunnel, est utilisie pour calculer les dCforma-
tions causCes par l'excavation du tunnel sanitaire de Thunder Bay. Les paramktres de sol adoptCs dans l'analyse ont
CtC bases sur les risultats obtenus dans des essais triaxiaux fonction des contraintes. Les rbultats de cette analyse sont
comparCs avec les diplacements mesurCs du sol. L'on fait Ctat d'une concordance raisonnable entre les valeurs calculCes
et observCes de la distribution 3D du tassement, de mCme que des diplacements horizontaux a diffkrentes distances
de la face du tunnel. La concordance gCnCrale pour les diplacements dans des conditions 3D donne a croire que la
mCthode d'analyse peut Ctre applicable aux problkmes de conception du tunnel dans les argiles molles similaires a celle
For personal use only.

du tunnel sanitaire de Thunder Bay a la condition que l'on puisse se fier aux paramktres de sol determines.
Mots cl&s : percement de tunnel, analyse tri-dimensionnelle, ClCments finis, deplacements Clasto-plastiques, dCfor-
mation de sol, tunnel de Thunder Bay.
[Traduit par la rCdaction]

Can. Geotech. J. 28, 25-41 (1991)

Introduction
To develop a simple, reliable, theoretically based method tion technique had been used in North America, an exten-
for predicting surface settlements resulting from shield tun- sive field-instrumentation program was undertaken. Two
nelling in soft soils, it is necessary to understand the three- arrays of instrumentation, referred to as array 1 and array 2,
dimensional (3D) nature of the stress changes and defor- were installed. The comprehensive instrumentation and
mations caused by the process of tunnel1 excavation. Field monitoring program has been described in detail by Belshaw
studies (Peck 1969; Cording and Hansmire 1975) have and Palmer (1978) and Palmer and Belshaw (1980). A com-
highlighted the 3D nature of the problem and the effect of prehensive test program, including special tests for the deter-
3D face loss on the subsequent deformations. Clough and mination of deformation and strength parameters for appro-
Leca (1989) have recently reviewed research into soft ground priate stress paths, was also subsequently undertaken (Lo
tunnelling, and it is clear that, to date, very little theoretical and Rowe 1982; Ng and Lo 1985). The availability of both
research has been conducted into the 3D behaviour of appropriate soil parameters and detailed field monitoring
ground movements ahead of the tunnel face, particularly data makes this an excellent case for examining the strengths
for soft soil, where nonlinear elastoplastic soil conditions and limitations of the proposed 3D analysis.
prevail. Consequently, there is currently very little informa- Rowe et al. (1983) and Ng et al. (1986) have described the
tion to aid the design engineer in assessing these 3D effects. use of a two-dimensional (2D) elastoplastic cross-sectional
The authors have developed a 3D elastoplastic finite- analysis for estimating the short- and long-term ground
element analysis (Lee 1989; Lee and Rowe 1990a) that is displacements near the tunnel and have reported reasonable
capable of simulating the advance of a tunnelling shield and agreement between the predicted and observed ground
the associated ground losses resulting from the tunnelling displacements under plane-strain conditions. Thus this pres-
process. The objective of this present paper is to illustrate ent paper will focus on the 3D deformations by comparing
the applicability of the 3D model by comparing observed observed and calculated movements when the tunnel heading
and calculated behaviour for a well-documented case was close to the instrumented sections of the Thunder Bay
history. tunnel. Subsequent papers (Lee et al. 1991; Rowe and Lee
In 1976, a 3.3 km long sanitary trunk sewer was con- 1991a, 1991b) will discuss approximate methods of incor-
structed through soft clay in the city of Thunder Bay, porating these 3D effects into simpler plane-strain analyses
Ontario. This tunnel was constructed using a tunnel-boring as well as the development of simplified design guidelines
machine together with a segmented, precast concrete tun- for use in predicting surface subsidence above shallow
nel lining. Since this was the first time that this construc- tunnels in soft clays.
Prlnted In Canada / lrnprlrne au Canada
CAN GEOTECH. J VOL. 28, 1991 i
I

SHEAR STRENGTH, kPa k

PENETRATION RESISTANCE, BLOWS W Y St


70 t l m
3
PV\T DARK BROWN
SOME SAND A N D SlLT

SAND - SILTY SAND


LOOSE TO DENSE
GREY
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

OCCASIONAL CLAY
SEAMS

SILTY CLAY - CLAY


9
GREY
8.5
SOFT TO FIRM
53.5 11
.it5
OCCASIONAL
SlLT OR SAND
SEAMS
' 55.0
54.8 :
58.3
9.5
10
11
6.5
------------
For personal use only.

5.5
6
FIRM TO STIFF
7
GREY A N D RED
BROWN LAYERS

OCCASIONAL
LAYERS OF SlLT
OR FINE SAND

0 STANDARD PENETRATION TEST


A UNCONFINED COMPRESSION TEST
GEONOR VANE TEIT

SHALE, BLACK

FIG. 1. General soil properties at array 2 (after Palmer and Belshaw 1980).

The Thunder Bay tunnel A precast, segmented concrete lining was assembled in the
A detailed description of the tunnelling procedure adopted tailpiece, and the machine was advanced by thrusting on the
for the Thunder Bay tunnel has been given by Morton et al. completed lining. The mined diameter was 2.47 m and the
(1977); a brief summary is given below. A full face tunnell- outside diameter of the completed lining was 2.38 m, giv-
ing boring machine (TBM), equipped with a rotary excavator ing a physical gap clearance of 90 mm between the concrete
that supported the face, was used to excavate the tunnel. rings and the excavated surface. During the advance, a clay
The rate of intrusion of spoil through the machine head grout was injected into the tail void; however, the grouting
could be controlled by the size of the opening at the face. was reported to be ineffective. This is to be expected, since
The opening was generally kept small and was adjusted so the rate of plastic deformation in soft clay is usually very
that the rate of advance of the machine was consistent with rapid. Without any internal support pressure, a soft clay
the volume of soil excavated. The maximum rate of prog- soil will fill the gap as soon as the lining leaves the shield
ress was 29 m/8-h shift. tail (Lo and Rowe 1982).
LEE AND ROWE

1
t 9

DIRECTION OF
TUNNEL ADVANCE t8
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

t 7

Is11
A A S12 AS13
AS8 VT6 VT7 As9 Aslo
I
+6 AG5 A 57
v T 4 VT5 gG4 AS6
1
VT1
AG3
VT2 VT3 AG2
AS1 AS2
A s3
mi LD mc A G1
mB A.
BM +30 +29 +28 +27 23nE OD OC O B O A
+ 2 6 ~ + o t ~ + ~ o t Oo ++ o +13 +12 +11 +lo
25 24 22 21 20 19 18 17 16 15 14
For personal use only.

$ TUNNEL ---- + INSTRUMENTED R I N G OF


TUNNEL LINING
L
-

k
+3
TUNNEL
LINING
+ SURFACE SETTLEMENT AUGER
0 DEEP S E T T L E M E N T A U G E R

+ .DEEP BENCH MARK


[7 M A G N E T I C E X T E N S O M E T E R
SLOPE I N D I C A T O R
V THOR P I E Z O M E T E R
A G E O N O R PIEZOMETER
0

0
444" 10 20 ft
t
A S O L I N S T PIEZOMETER

I
FIG. 2. Plan of instrumentation at array 2 (after Palmer and Belshaw 1980).

Results of field measurement


Arrays 1 and 2 are about 1.25 km apart, but the The progress of the tunnel face and the induced set-
stratigraphy is similar. The depth, H, to the tunnel axis was tlements are shown in Fig. 3. The tunnelling procedure at
10.7 and 10.5 m at arrays 1 and 2, respectively, giving rise arrays 1 and 2 was technically identical, but array 1 was only
to similar cover to diameter ratios, H/D, of 4.33 and 4.25, 60 m away from the first shaft. Since the workmen were not
respectively. The stratigraphy and soil properties for array 2 yet familiar with the construction technique, advance at
are shown in Fig. 1. The tunnel is located in a soft grey silty array 1 was slow and intermittent, as may be seen in Fig. 3.
clay deposit that underlies postglacial fluvial deposits of Palmer and Belshaw (1979) have reported that at array 1,
loose sand and silty sand. Beneath the invert of the tunnel alignment difficulties and grouting problems were noted by
is a layer of varved clay that extends to shale bedrock. the site engineer. In contrast, good technique was reported
A plan view of the instrumentation for array 2 is shown in at array 2, and 21 rn of tunnel was completed in an 8-h shift
Fig. 2. at this array.
28 CAN. GEOTECH. J. VOL. 28, 1991
,
I

TIPIE IN D A Y S
I I
A U G U S T 1976 ARRAY 2
19 30 2 2 3
2 24 25 2 4 C

I I
I ,
I I
1
I
1
I I
I
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

PIN 5 ( A R R A Y 2) A T ARRAY 2
For personal use only.

FIG. 3. Time, settlement, and tunnel-progress relationships (after Ng 1984).

The development of vertical soil displacements with depth of soil parameters for the plane-strain analyses has been
is shown in Figs. 4 and 5 for arrays 1 and 2, respectively. given by Ng et al. (1986); only the important points con-
Three stages of movement may be identified during tunnell- cerning the selection of undrained parameters appropriate
ing before the plane-strain condition was reached. In the to the direction and magnitude of stress changes near the
first stage, small downward movement to slight heaving tunnel heading will be included in the following discussions.
(especially for array 1) of the settlement points was observed Since the stratigraphy at arrays 1 and 2 are similar, only the
before the face reached the line of measurement. In the results pertaining to array 2 are reported in detail. The
second stage, a small amount of settlement occurred as the undrained results for array 1 are similar.
shield passed the instrumentation section (i.e., between 0 A number of elastoplastic 3D analyses were performed by
and 5.6 m). However, after the tailpiece cleared the section Lee (1989) and Lee and Rowe (1990a). These results suggest
(i.e., at a distance greater than 5.6 m behind the tunnel face), that the stress path experienced by the soil element ahead
the third stage of settlement occurred rapidly as the soil of the tunnel face is quite different from that at the plane-
invaded the space between the circular tunnel boundary and strain section. In general, the trend of the stress path ahead
the lining. The rapid rate of vertical soil displacement began of the tunnel face is similar to that of triaxial compression.
to level off when the tunnel face was at 10-15 m from the Once the face advances beyond the point of interest, the
instrumented section. At this stage, the lined tunnel section stress path above the tunnel quickly changes to one closer
is considered to approach plane-strain conditions, with the to triaxial extension. The implication of these differences
crown of the tunnel coming into contact with the lining. The of stress path should be considered in the selection of soil
time required to reach this condition was less than 4 days parameters.
at array 1 and only 8 h at array 2. This corresponding set- The results of CIU tests on vertical and horizontal samples
tlement may be considered to be the "undrained" settlement indicate that horizontal to vertical undrained strength,
for this soil (a similar definition is also given by Ng et al. cUh/cu,,is 0.84. As noted by Lo (1965), the vane strength
1986). shown in Fig. 1 approximately corresponds to the strength
cuhof the undisturbed soil. The vertical undrained strength
in unconfined compression tests is also shown in Fig. 1. The
Selection of parameters for undrained analysis implication of the strength anisotropy on surface subsidence
A comprehensive test program including special tests for induced by the construction of shallow tunnels has already
the determination of deformation and strength parameters been discussed by Lee and Rowe (1989), where it is shown
appropriate to the direction and magnitude of stress changes that for a lined tunnel with a moderate value of "gap," the
in tunnelling has been reported by Lo and Rowe (1982) and detailed consideration of strength anisotropy may not be
Ng and Lo (1985), and a detailed discussion of the selection necessary.
LEE AND ROWE
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14
For personal use only.

DISTANCE TO TUNNEL FACE (m)


FIG. 4. Vertical soil displacement over centreline of tunnel at array 1 (after Belshaw and Palmer 1978).

Results of triaxial compression and extension tests for strata were given elsewhere (Lo and Rowe 1982; Ng et al.
drained and undrained conditions were reported by Ng and 1986). The parameters adopted for these granular materials
Lo (1985). The undrained modulus profile ahead of the tun- in the present analysis are summarized in Fig. 6. The prop-
nel face adopted in the 3D analysis is based on the CKoU erties of the 1 m thick layer of peat were only estimated from
compression tests obtained by Ng and Lo (1985). Behind empirical correlations and are relatively uncertain. Fortu-
the tunnel face, the modulus profile for the clay layers is nately, the surface settlement points were located beneath
represented by data obtained from anisotropically con- the bottom of the peat layer, and the properties of the peat
solidated extension (CKoE) triaxial tests. The stress paths have practically no effect on the results.
of CKoE tests approximately correspond to the field stress
path experienced by th:: soil behind the tunnel face at and
near the crown. Near the springline of the tunnel, the stress Total gap parameter
path is intermediate between compression and extension, and The "gap" is a measure of the volume of material that
the modulus lies between the compression and extension has been excavated in excess of the theoretical volume within
values. A smooth transition in modulus profile (similar to the outer diameter of the tunnel lining. The equation for
that of Ng et al. 1986) was adopted between the modulus the "total gap parameter" that was originally proposed by
at the springline region, and a summary of soil parameters Rowe et al. (1983) and refined by Lo et al. (1984) has been
adopted in this analysis is indicated in Fig. 6 (see Fig. 11 further extended by Lee et al. (1991); this will be briefly
of Ng et al. 1986 for the undrained strength profile adopted outlined below. The total gap parameter (G) may be
in the present analysis). Results from Ng and Lo (1985) expressed as (see Fig. 7)
indicate that under undrained conditions, the Thunder Bay
clay has a ratio Eh/Evapproximately equal to unity. The
values of the independent shear modulus Gvhadopted in As the tunnel face advances, the soil mass in front will move
this analysis were determined using the NGI simple shear towards the face, and some of the soil, which was originally
test and anisotropic parameter tests as reported by Lo and outside the zone to be excavated, moves into the excava-
Rowe (1982) and Ng and Lo (1985). tion zone and is subsequently removed. This gives rise to
a component of the gap GDthat represents the amount of
Parameters for overlying sands "over excavation" owing to the 3D movements ahead of
The silt and silty sand strata for array 2 were divided into the tunnel face. This component of the "gap" arises
three sublayers as shown in Fig. 1. The layers were assigned automatically from the 3D analysis and can be estimated
soil parameters on the basis of soil densities, laboratory for use in 2D analyses as outlined by Lee et al. (1991).
testing, and empirical correlations with SPT data. Details The "over excavation" owing to 3D movements ahead
regarding the selection of parameters for these granular of the face, denoted by U;'D, is determined by assuming the
CAN. GEOTECH. J. VOL. 28, 1991
,
J
\

DLSTANCE TO TUNNEL FACE (m)


\

-15 -10 -5 0 5 10 15 20 25 30 35
-30 I I I I I I I I
INITIAI, DEPTH ( m )
-20 - 2
0 4.5
-10 - A 6.0
+ 7.6
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14
For personal use only.

FIG. 5. Vertical soil displacement over centreline of tunnel at array 2 (after Palmer and Belshaw 1980).

machine is advanced in perfect alignment. If the machine with the sharp corners of the Mohr-Coulomb surface were
is pitched upward or downward, additional material will be overcome by representing it in a slightly modified continuous
excavated. This quantity is related to workmanship and is and differentiable form as proposed by Gudehus (1973).
denoted by w in [I]. The term w, which takes into account Details regarding this form of the soil model, the choice of
the quality of workmanship, is a factor for engineering finite element, the construction simulation, and solution
judgement and may be related to local experience or to technique have been described by Lee and Rowe (1990a) and
empirical corrections derived from case histories such as will not be repeated herein.
those discussed by Lo et al. (1984) and Rowe and Kack One way to simulate the interaction between the soil and
(1983). For the Thunder Bay tunnel there was significant the tunnel shield or lining support is the use of the soil-
loss of ground owing to workmanship effects (i.e., w) at structure interaction theory proposed by Rowe et al. (1983).
array 1 but negligible loss owing to workmanship at array 2 However, because of the high cost associated with imple-
(i.e., w=O). menting this incremental technique in 3D analysis, an
The physical gap denoted by (2A + 6) represents the approximation as illustrated in Fig. 8 is adopted in this
geometric clearance between the outer skin of the shield and study. The soil around the tunnel boundary is initially
the lining, and it is composed of the thickness of the tail allowed to deform freely into the excavated area, once the
piece (A) and the clearance required for erection of the lin- radial convergence of the soil reaches a predefined value
ing (6). It may be seen that the physical gap is readily deter- (which corresponds to the so-called "total gap parameter"),
mined once a machine and lining system are chosen. For a ring of lining element is activated, and full contact between
the Thunder Bay tunnel the physical gap was 90 mm. the soil and the lining support is assumed. A similar tech-
nique is also used to simulate the closure of the soil around
the tunnel shield. The weight of this shield-lining system
Method of analysis is then simulated by activating the body force of the struc-
The analyses reported in the present paper were performed tural elements. (The resulting nodal forces owing to body
using the authors' 3D elastoplastic finite-element program weight of the lining-shield elements are only allowed to
(FEM3D). apply to the soil elements below the springline of the tun-
The soil is assumed to have an elastic - perfectly plastic nel.) The validity of this simplified approach under plane-
constitutive relationship that (for undrained conditions) was strain condition has been discussed by Lee (1989), and the
defined by three independent elastic variables Ev, Eh, and approach is found to be adequate for tunnelling techniques
Gvh (vvh = 0.5 and vhh = 1.0 - 0.5 Eh/Evas noted by similar to the Thunder Bay tunnel case.
Gibson 1974), a Mohr-Coulomb type failure criterion The continuous advance of tunnel face is simulated by
(Tresca type for the undrained case) and a flow rule of the a two-stage analysis. In the first stage, the axial pressure
form proposed by Davis (1968). The problems associated ahead of the tunnel face is released, and the radial pressure
LEE AND ROWE

--
E
Descrlptlon of S o l 1 Properties
i

\
u EU (PlPa)
sol1 proflle
aJ Y ? C I
61
n
l,O 2P 3q 4Q 5,O KX/6 KO L ~ / EG ~ ~ E : vhh vVh a
0 Peat
G.M.T& 14 .5 1 .2 .33 .30 0 30"
- 2 L o o s e S l l t y Sand
18.7 .57 -8 .2 .33 .35 0 25"
-4
Medlum Sand \ 20.8 .45 .8 .25 .33 .4 0 34'
-6
19.8 .50 .8 .25 .33 .4 0 32'
-8
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

- 10
16.2 .88 1.0 .3 .49 .49 - -
- 12 Transition 18.7 .87 1.0 .25 .49 .49 - -

- 14
Varved C l a y behlnd 18.7 .8 1.0 .25 .49 .49 - -
- 16

- 18
i
- 20
- 22 tunnel face

- 24
For personal use only.

Bedrock

FIG. 6. Soil profile adopted in the analysis of array 2 (modified from Ng et al. 1986). The ratio Gvh/Evrepresents the condition
behind the tunnel face and was adjusted according to the modulus profile adopted ahead of the tunnel face.

INITIAL POSITION OF POINTS ON WHAT


W I L L BECOME THE CROWN AFTER A = GEOMETRIC LOSS OF
EXCAVATION

\
GROUND OVER SHIELD
6 = LOSS OF GROUND
-A'7- - -- -- - \ DUE TO CLEARANCE

GAP = 2 ~ + 6 +u
u = u ; ~+ W

FIG. 7. Definition of gap (modified from Lo et al. 1984).

around the periphery is released until the total radial con- soil-lining interaction is allowed to develop. The interact-
vergence of the soil at tunnel crown and invert corresponds ing effect between the soil and the tunnelling shield is also
to the physical gap (i.e., 90 mm for the case of the Thunder simulated along the portion of soil directly behind the tun-
Bay tunnel). The 3D movements ahead of the face into the nel face and around the shield (see Fig. 8).
zone to be excavated can then be determined, and hence The actual amount of 3D movement may be reduced if
GDcan be evaluated and used to calculate the total gap. pressure is maintained at the tunnel face during excavation;
Once the total-gap parameter is determined, a second 3D this will lead to a corresponding decrease in total gap com-
computation is then performed to approximately simulate pared with that calculated with full stress release. The radial
the complete tunnel-excavation process where the radial con- deformation of the soil found over the tunnel shield will
vergence behind the shield corresponds (see Fig. 8) to the depend on the physical thickness of the cutting bead or hood
total gap G (which now includes the 3D ground-loss effect, attached to the tunnelling machine, alignment difficulty
GD,as well as the physical gap). Once the tail void closure (pitching or diving), and the type of ground; typically this
reaches the value of G, the lining elements are activated and is up to 40% (see Table 1) of the total gap parameter G.
32 CAN. GEOTECH. J. VOL. 28, 1991 1
\

Losses Over
Shield Losses lnto Tall Vold i
t=f

Three-Dimensional
"..-
Losses Ahead -*
of the Face

I TUNNEL StlEILD
I LINING SUPPORT SYSTEM I
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

Full Release of
Axial Stress to
-
----

-
Simulate Face
Loss
&
----
I I I

FIG. 8. Approximate technique to simulate 3D soil-structure interaction.


For personal use only.

TABLE1. Typical partition of the total settlement to the ground


loss at a shield-driven tunnel (after Craig 1975)

Ahead of Over the Tail of


shield shield shield
Ground ("70) ("70) ("70)
Sand above
water table 30-50 10-20 60-80
Sand below
water table 0-25 0-20 50-75
Stiff clay 30-60 0-20 50-75
Silts and
soft clay 0-25 10-40 30-50

A typical finite-element mesh used in the analysis con-


sisted of 4824 11-noded nonconforming brick elements as
shown in Fig. 9 (see Lee and Rowe 1990a for the descrip-
tion of this nonconforming element). Because of symmetry
only half the tunnel is considered. In this study, the struc-
tural behaviour of the tunnel shield and lining were
simulated by continuous rings of 11-noded nonconforming FIG. 9. Finite-element mesh used in the analysis of the Thunder
elements. Theoretically, 3D thin shell elements should be Bay tunnel. Tunnel diameter (D) = 2.47 m.
used if accurate determination of the structural responses
(e.g., bending moment, thrust force) of the lining are
required. However, since the present study is directed and invert, the moment of inertia of the continuous ring was
towards the understanding of the 3D stress and deforma- reduced by a factor of 10 to provide the appropriate flex-
tion behaviour of the soil around the tunnel heading, the ibility for the lining. A similar technique was used by Kasali
use of 11-noded nonconforming elements to simulate the and Clough (1982) and Lo and Rowe (1982) to simulate
structural behaviour of the supporting shield or lining is con- segmented tunnel linings.
sidered to be acceptable.
The tunnel lining was modelled with parameters appro-
priate t o the concrete lining (E = 20 GPa; A = Results of analysis
0.11 m2/m; I = 1.11 x m4/m). Since there were For the Thunder Bay tunnel, the physical gap (24 + 6)
segment joints and crack inducers at the crown, springline, or the minimum gap parameter is 90 mm. Overcutting
LEE AND ROWE

-
i

/ t

APPROACHING MOVING AWAY DISTANCE OF T U N N E L FACE (m) h

-5 0 5 10 15 20 30.5
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

FIG. 10. Predicted and observed surface settlement over centreline of tunnel at array 2.

resulted from 3D face-take, and other construction activi- (2) Full release of axial stress at the tunnel face is assumed,
ties would result in a larger gap parameter. For the case of which would result in the full development of 3D ground
array 2, since good tunnelling technique was reported, the loss ahead and over the tunnel shield. For this case, a value
term w, which takes into account the quality of workman- of annular void over the tunnel shie!d of 32 mm is assumed,
ship, has been assumed as zero. (However, it should be noted which is similar to the value of the equivalent 3D radial
that under normal operating conditions, the tunnel operator displacement, uTD, and represents 30% of the total-gap
is frequently adjusting the pitch of the shield. Therefore in parameters. The total-gap parameter, which now includes
For personal use only.

practical design situations, it is suggested that as a conser- the 3D ground-loss effect,-@& as well as the physical gap,
vative measure a nominal value of the workmanship factor is assumed to be 122 mm (referred to as "gap of 122 mm").
w of say 10 mm should be allowed for good workmanship.) These two cases will represent the bounds of the possible
In this case it only remains to determine the equivalent 3D construction conditions where deformation can occur owing
radial displacement, GD.A preliminary 3D analysis was to various degrees of stress release at the tunnel face and
performed, as previously described, from which the radial movement of the soil to fill the gaps between the
equivalent 3D radial displacement, G D , was determined to excavated surface and tunnel shield-lining support system.
be 32 mm for full stress relief at the face of the tunnel. For The findings from these two cases would be expected to
this case the total-gap parameter G becomes 122 mm. bracket the possible response of a full construction condi-
(A similar value of the total-gap parameter was adopted by tion (provided that the tunnelling machine is not pushed into
Ng et al. (1986) to predict the plane-strain behaviour of the the soft soil).
Thunder Bay tunnel; however, Ng et al. had to estimate
GDusing engineering judgement, whereas in this paper it Vertical soil displacement along centreline of tunnel
was determined directly from a 3D analysis.) Observed surface-settlement profiles along the
To model the interaction between the soil and the tunnel longitudinal axis from plan of instrumentation numbers
shield, a value of annular void over the shield is required. (PIN) 1-5 (see Fig. 2 for the location of each surface settle-
However, information on this parameter was not available. ment PIN) are shown in Fig. 10 together with the predicted
Observations from previous case histories (see Table 1) profiles with gap parameters of 90 and 122 mm. The
indicate that the radial displacement of soil developed over predicted settlement profiles are generally in good agreement
the tunnel shield is usually in the range of 10-40% of the with the observed settlements, and the measured range of
final settlement. settlement profiles are generally within the bounds of
In this study, because of the uncertainty regarding the calculated settlements. The data for centreline surface set-
amount of overcutting over the shield, two cases were tlement reported by Palmer and Belshaw (1980) are shown
examined. in Fig. 11. It may be seen that substantial variations occurred
(1) The tunnel is assumed to be excavated under perfect along the centreline of the instrumented array. The range
alignment, and the tunnelling machine is pressed hard of centreline surface settlements predicted by the 3D anal-
against the face so that 3D movements will be minimized. yses with the total-gap parameters of 90-120 mm is also
The only source of ground loss is assumed to be the physi- plotted in Fig. 11. Within the instrumented array the
cal clearance between the diameter of the excavated surface measured range of centreline settlements lies within the
and the lining. Under this condition, the equivalent 3D radial bounds of calculated settlements. Since the minimum settle-
displacement, G D , is assumed to be zero. The radial ment measured corresponds very well to that predicted by
displacement of soil developed over the tunnel shield has the physical gap (i.e., minimum gap) of 90 mm, it indicates
also been assumed as zero. The total-gap parameter, that the adjustable size of opening at the machine face
therefore, is equivalent to the value of physical gap of (through which the spoil intrudes into the bulkhead) was
90 mm. (In the remainder of the paper, this construction effectively controlled so that the rate of advance of the
simulation will simply be referred to as "gap of 90 mm.") machine was consistent with the volume of soil excavated.
CAN. GEOTECH. J. VOL. 28, 1991 \
i

PIN LOCATION (m)


10
I
20
I
30
I
40
I
50
I
'
ARRAY 2
Q1 #2 113 t4 85 #6 #7 #8 #9

PREDICTED FOR GAP = 90mm


Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14
For personal use only.

I
FIG. 11. Predicted and observed ranges of surface centreline settlement at array 2.

I The variation of settlements between each PIN may therefore crown coming into contact with the lining. The calculated
I be attributed to the variation of 3D face-take owing to dif- maximum surface settlement behind this point is in very
ferent degrees of face control at various sections or good agreement with the observed settlement.
variability of the shield alignment during excavation. In this For the case of array 1, substantial variations of surface
paper, this can be simulated by varying the value of the settlement were observed along the instrumented area, and
equivalent 3D displacement, GD, in the range of 0 (full the measured settlements were consistently larger than those
face support) and 32 mm (full stress release at the face). It of array 2. Palmer and Belshaw (1979) suggested that the
may be noted that the detailed instrumented section at array 2 difference in behaviour can be attributed to a difference in
is located in the area where maximum settlement was workmanship at the two arrays. Since at array 1 the crew
recorded. This leads to the use of a gap parameter of 122 mm was not yet familiar with the tunnelling technique, steering
for evaluating the performance at the detailed instrumented and alignment problems were experienced. This "workman-
section for the rest of the analysis. ship" factor can be incorporated into the total-gap param-
The calculated vertical-settlement profiles with a gap eter in terms of o as indicated in [I]. An approximate tech-
parameter of 122 mm that occurred at the ground surface nique to estimate the maximum value of this
and along the tunnel crown axis are shown together with "workmanship" factor under plane-strain conditions has
the observed settlements in Fig. 12. The predicted settlement been developed by Lee et al. (1991) and Rowe and Lee
profiles are in encouraging agreement with the observed (1991a). 3D simulation of this "w~rkmanship'~ factor, how-
settlements. Near the tunnel the finite-element result ever, is a complex and expensive task and is beyond the scope
overestimates the soil settlement as compared with the of this paper. It is likely that the steering and alignment
observed behaviour. This discrepancy may be partially problems encountered in array 1 would result in deviation
attributed to the approximate technique adopted in the from its original design alignment. Furthermore, the
simulation of soil-shield interaction and the uncertainty adjustable size of opening at the machine face may not have
regarding the maximum allowable value of annular displace- been controlled in a proper manner. For instance, if the
ment over the tunnel shield. However, as the soil clears the opening size is too small so that the rate of advance of the
tailpiece of the tunnel shield (at a distance of about 5.6 m machine is inconsistent with the volume of soil excavated,
behind the working face), the general trend of the settlement high outward pressure would be developed ahead of the face.
profile predicted by the 3D finite-element analysis is very This would result in the outward displacement of soil in front
close to the observed profile. The rapid rate of vertical soil of a progressing advanced shield. This is evident by the heav-
displacement begins to level off at a distance of about 15 m ing of soil up to 10 mm ahead of the face, as shown in Fig. 4
behind the tunnel face (i.e., about six tunnel diameters for array 1; however, a similar phenomenon was not
behind the tunnel face). At this stage, the lined tunnel sec- observed for array 2 (see Fig. 5). The heaving process at the
tion is considered to attain plane-strain conditions with the shield face would induce positive excess pore pressure and
LEE AND ROWE ,\
,, ' I \

LONGITUDINAL DISTANCE TO TUNNEL FACE (m)


"?
-1 5 -10 -5 FACE 5 10 15 LO 25 30 35
-30 I I I I I I I I
-20 -
-10 -
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14
For personal use only.

FIG. 12. Comparison of predicted and observed settlement profiles along the longitudinal axis.

a mechanically disturbed zone in the clay where subsequent slightly overestimate the maximum centreline settlements.
consolidation would occur. If the positive outward pressure However, at distances farther away from the face, the max-
in front of the face is controlled in a proper manner, as in imum settlements predicted by the finite-element analysis
the case of the slurry shield and the Earth Pressure Balance are very close to the observed values.
shield, reduction of the overall surface settlements would The calculated surface-settlement profiles at a plane strain
result (Kasali and Clough 1982; Finno 1983). This heaving section are shown together with the observed settlement in
process is apparently designed to partially offset the inward Figure 14a. The normalized shape of the curves, obtained
movements due to collapse of the soil into the tail void. by dividing the settlements by the centreline settlement, are
However, very large heaving displacements would likely have shown in Fig. 14b. These normalized plots provide a more
a detrimental effect on the soil adjacent to the tunnel owing direct comparison of the trough shapes.
to its remoulding. Terzaghi (1942) observed such problems It may be seen from Fig. 14a that the maximum surface
where a shield was shoved blind into Chicago clay. As much settlement is underestimated if a gap parameter of 90 mm
as 250 mm of heave was induced, and subsequent set- is adopted. The settlement curve for a gap parameter of
tlements up to 300 mm occurred over a 2-year period. 122 mm is in good agreement with the observed settlements.
(Similar results were also obtained by Ng et al. 1986 by a
plane-strain finite-element method. However, it should be
Surface-settlement troughs along transverse sections noted that the gap parameter of 120 mm adopted by Ng
The development of transverse surface-settlement troughs et al. was based on engineering judgement, whereas in this
(i.e., settlement profile along the ground-surface perpen- paper it was determined directly from a 3D analysis.) Both
dicular to the direction of tunnelling) at various distances curves have the appearance of an error function. The shape
behind the tunnel face are shown in Fig. 13. A relatively of the observed surface-settlement profile is slightly narrower
flat settlement profile is initially developed ahead of the face. than the predicted profiles. The width of the settlement
This gradually changes to the well-recognized bell-shaped trough can be more directly compared by considering the
profile as the tunnel passes the section. The predicted set- trough width i at various distances from the tunnel face. The
tlement troughs are in relatively good agreement with the observed and calculated values at various distances from the
observed settlements for all five distances behind the face. tunnel face are shown in Table 2. The values of the trough
The shapes of the predicted surface-settlement troughs are width gradually reduce as the distance behind the tunnel face
consistently slightly wider than the observed troughs. Near increases. The calculated width of the settlement trough con-
the tunnel face, results predicted by finite-element analysis sistently overestimates the observed values by 10-20%.
CAN. GEOTECH. J. VOL. 28, 1991 3
i
f

DISTANCE I'ROII CI:>II'REI,I'iE 01.' 1'IJNSl:L 1t;i)

Field llensurcmont
( 1 . 5 rn f r o n ~ t t i n n e l f a c e )
F-L l'rediction

(b) SJUTtl +- I --) NORTH


Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

DIS'rAYCE FRO:! CENTRELINE OF TUNNEL (nl)


10 - 8 6 2 0 2 4 6 8 10

F i e l d >leasurc:nent

- ki::
m frotn t u n n e l f a c e )
4 0 precliction

SOUTtl I NORTli

1)ISTANCE FROil CENTRELINE OF TUiJNEL (m)


10 - 8 6 4 2 0 2 4 6

h e l d >leasuremerrt
For personal use only.

(6.7 m from t u n n e l f a c e )
F-E Pred~ction

(d )
VERTICAL SCALE: SETTLEEIENT (mm)

-- I NORTH

DISTANCE FRON CENTRELINE OF TUNNEL (m)

F i e l d Neasurement
(9.8m from t u n n e l f a c e )

(el SOUTH L( -
DISTANCE FROEl CENTRELINE OF TUNNEL
NORTH

(m)

F i e l d Measuremen:
X= 15.2. (15.2m from tunnel f a c e )

VERTICAL SCALE: SETTLE'iENT

"-f-"'
(mm)

-L
60
\F-E Prediction

FIG. 13. (a-e) Predicted and observed surface-settlement troughs at various distances from the tunnel face (array 2). F-E, finite
element.
LEE AND ROWE \

I _ . \

(a) DISTANCE FROM CENTRELINE (m) x


Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

Array 2
"Undrained"

L o c a t i o n = PIN 5
T u n n e l F a c e a t 15.2111 I
(b) DISTANCE FROM CENTRELINE (m) ;-

I n i t i a l Ground S u r f a c e I
For personal use only.

FIG. 14. ( a ) Surface-settlement troughs: ( b ) Normalized surface-settlement troughs at the plane-strain section (predicted with
GAP = 90 and 122 mm).

TABLE
2. Observed and calculated trough widths at array 2

Distance behind
the tunnel face Observed (obs), Calculated (cal), -
*,,I
(m) i/a i/a i~bs
1.5 3.2 3.5 1.1
4.6 3.1 3.3 1.1
6.7 2.9 3.1 1.1
9.8 2.5 3.0 1.2
15.2 2.4 2.9 1.2
NOTE: The trough width i is the distance from the centre of the trough t o a point where the
settlement is 0.606 of the maximum settlement at the centre; a is the radius of the tunnel.

The plots of calculated normalized settlement (Fig. 14b) Variation of lateral displacement with depth
show that their shapes do not vary significantly with the gap The observed and calculated lateral deformations when
parameter. This is consistent with the observation that the the tunnel face was 5 m away from the slope indicators (i.e.,
shape of the surface-settlement trough is similar from sec- - 5 m) and 4 and 20 m past the slope indicators are shown
tion to section (Belshaw and Palmer 1978) and also with the in Figs. 15a-15c. Noticeable deformations begin to occur
numerical results obtained by Rowe et al. (1983) and Ng when the face is at a distance of 5 m (two tunnel diameters)
et al. (1986). away from the slope indicators and gradually increases as
CAN. GEOTECH. J. VOL. 28, 1991
i
I

HORIZONTAL klOVPIENT (mm)


Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14
For personal use only.

RELATIVE HORIZONTAL LOCATION (m)

HORIZONTAL PIOVEMENT (mm)

RELATIVE HORIZONTAL LOCATION (m)

FIG. 15. Predicted and observed lateral displacement (a) 5 m ahead of the tunnel face, (b) 4 m behind the tunnel face, and (c) 20 m
behind the tunnel face. F-E finite element.
LEE AND ROWE \

FIORIZONTAL I\IOVE~IENT ( r n ~ n )
,
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

Distance of + 2 0 rn from the Tunnel Face


t o the L i n e of Slope Indicators

- - - - - - - Field Measurement

F-E Prediction
For personal use only.

RELATIVE HORIZONTAL LOCATION (rn)


FIG.15. Continued.

VERTICAL DISPLACEMENT (mm)

the tunnel approaches. Once the tunnel passes the

il'l
instrumented section, rapid development of lateral displace-
ment occurs. The results predicted at 15 m from the tunnel
face (Fig. 15c) are very similar to those obtained by Ng et al. 8 Observed (Undrained)
(1986) using plane-strain finite-element analysis. This,
together with previously presented results, indicates that for F-E Prediction
the case of the Thunder Bay tunnel, plane-strain conditions (GAP= 122 mm)
are reached at a distance of about 15 m (six tunnel 3 Undrained Analysis
diameters) behind the tunnel face. Thus, once the tunnel face
advances beyond this distance, plane-strain analysis can be
used to predict the resulting ground deformations (such as
Rowe et al. 1983; Ng et al. 1986). Furthermore, it is noted
that the observed deformation are not symmetric. This may
reflect nonhomogeneity of this soil and (or) nonuniform
excavation by the TBM owing to "workmanship" problems.
The calculated results were found to closely follow the trend
of the measured displacements. The magnitude of movement
was slightly underestimated before the face reached the
instrumented section and slightly overestimated after the face
passed the instrumented section. Both the observed and
calculated displacements indicated that the maximum lateral
soil deformation occurred close to the springline. The
magnitude of soil displacement decreases very rapidly below FIG.16. Predicted and observed vertical displacement along
this level. Similar behaviour involving maximum horizon- the vertical centre axis with depth. F-E, finite element.
tal movement towards the tunnel at the axis level has been
reported for a number of other tunnelling case histories Variation of centre settlement with depth above the tunnel
(Barrett and Tyler 1976; Attewell et al. 1986) and has been axis
examined theoretically, as discussed by Lee and Rowe The observed and calculated undrained settlement
(1990b). distributions with depth at 15 m from the tunnel face are
40 CAN. GEOTECH. J. VOL. 28, 1991
\

0 4 (m)
L 2 L J
SCALE
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

SETTLEMENT CONTOUR

ALIGNMENT \
For personal use only.

0 2 4 (m)
u
SCALE

FIG. 17. Predicted and observed surface-settlement contour (settlement bowl) above the advancing tunnel.

shown in Fig. 16. The observed and calculated settlements analysis and the comparison of observed and calculated
agree well, except within a distance of one tunnel diameter deformations can be summarized as follows.
from the crown where the observed settlements are some- (1) The maximum 3D radial displacement, GD,was
what higher. As suggested by Ng et al. (1986), the 32 mm, giving rise to a maximum total-gap parameter of
discrepancy may be due to an overestimation of the strength 122 mm. This value if very similar to the gap parameter
of the sand immediately above the clay cover adopted in the estimated by Ng et al. (1986).
analysis. Lowering the friction angle of the sand would lead (2) The observed and calculated displacement profiles
to an increase of calculated settlement in this region. It along the longitudinal and transverse cross sections were in
should be noted that the entire clay layer above the crown reasonable agreement.
is within the plastic zone; therefore, decreasing the shear (3) The calculated shape of the surface-settlement trough
strength of clay above the crown would not result in a sig- is relatively insensitive to moderate changes in the gap
nificant increase in centreline displacements. Darameter. The s h a ~ e of
s the observed and calculated set-
Settlement contour at the ground surface ilements at various distances from the tunnel face are in rea-
A contour plot of the surface settlement predicted by the sonable agreement.
finite-element analysis is shown in Fig. 17. A similar plot, (4) Both the observed and calculated horizontal
which was deduced from field observations by Ng (1984), displacements at various distances from the tunnel face
is also shown for comparison. The general trend of the sur- indicate that the maximum lateral displacement occurs near
face contour is very similar between the predicted and the springline level. The calculated profiles of horizontal
observed behaviour. The finite-element analysis displacements compare reasonably well with the observed
overestimates the extent of 3D displacement ahead of the data.
face and to the sides of the tunnel. Both the calculated and (5) Reasonably good agreement is also observed between
observed contours indicate that the maximum undrained set- the calculated distribution of undrained settlements with
tlement occurred at a distance about 15 m behind the tun- depth and the corresponding observed field settlements.
nel face (i.e., about six tunnel diameters). Reasonable agreement has been obtained between
calculated and observed 3D settlement distribution and hor-
Summary and conclusions izontal displacements at different distances from the tun-
A 3D elastoplastic finite-element undrained analysis of nel face. This overall agreement for displacement under 3D
the displacement caused by excavation of the Neebing- conditions suggests that the method of analysis is applicable
McIntyre Thunder Bay tunnel in Thunder Bay, Ontario, has to design problems involving tunnelling in soft clays similar
been performed. The findings from the 3D finite-element to that at the Thunder Bay sewer tunnel.
LEE ANID ROWE 41

Furthermore, the results of the study indicate that foun- ground deformations due to iunnelling in soft cohesive soils.
dations o r structures are subjected t o a wave of 3D Part I1 - Results. Computers and ~eotecknics,in press.
displacements and strains before the tunnel reaches the LEE, K.M., ROWE,R.K., and LO, K.Y. 1991. Subsidence due to
plane-strain conditions. Thus, information based o n plane- tunnelling: Estimating the "Gap" parameter. ASCE Journal of
Geotechnical Engineering Division, submitted.
strain analysis of a transverse section of a tunnel (e.g., Ng
Lo, K.Y. 1965. Stability of slopes in anisotropic soils. ASCE Jour-
et al. 1986) only represents part of the required informa- nal of the Soil Mechanics and Foundations Division, 91(SM4):
tion. 3D finite-element analysis such as that described in this 85-106.
study can be used t o predict the spatial 3D ground Lo, K.Y., and ROWE,R.K. 1982. Prediction of ground subsidence
displacements within the soil mass. However, because of the due to tunnelling in clays. Research Report GEOT-10-82, Faculty
high cost and processing time associated with this analysis, of Engineering Science, University of Western Ontario, London,
simplified procedures a n d (or) empirical techniques are Ont .
usually adopted. T h e validity o f various simplified Lo, K.Y., NG, M.C., and ROWE,R.K. 1984. Predicting settlement
Can. Geotech. J. Downloaded from www.nrcresearchpress.com by Université de Sherbrooke on 06/20/14

approaches has been evaluated by Rowe a n d Lee (1991b) due to tunnelling in clays. Tunnelling in Soil and Rock, ASCE
using, in part, the 3D results presented in this paper. Geotech I11 Conference, Atlanta, GA, pp. 46-76.
MORTON, J.D., DUNBAR, D.D., and PALMER, J.H.L. 1977. Use
Acknowledgement of a precast segmented concrete lining for a tunnel in soft clay.
Proceedings, International Symposium on Soft Clay, Bangkok,
This paper forms part of a general programme of research Thailand, pp. 587-598.
into soft ground tunnelling being conducted by the NG, R.M.C. 1984. Ground reaction and behaviour of tunnels in
Geotechnical Research Centre a n d made possible by the soft clays. Ph.D. thesis, University of Western Ontario, London,
award of grant A1007 and a Steacie Fellowship to R.K. Rowe Ont .
by the Natural Sciences a n d Engineering Research Council NG, R.M.C., and LO, K.Y. 1985. The measurements of soil
of Canada. parameters relevant to tunnelling in clays. Canadian Geotechnical
Journal, 22: 375-391.
ATTEWELL,P.B., YEATES,J., and SELBY,A.R. 1986. Soil NG, R.M.C., Lo, K.Y., and ROWE,R.K. 1986. Analysis of field
movements induced by tunnelling and their effects on pipelines performance - the Thunder Bay tuinel. Canadian Geotechnical
and structures. Blackie and Son Ltd., Glasgow. Journal, 23: 30-50.
BARRATT, D.A., and TYLER,R.G. 1976. Measurements of ground PALMER,J.H.L., and BELSHAW,D.J. 1980. Deformations and
pore pressures in the vicinity o t a precast, concrete-lined tunnel
For personal use only.

movements and lining behaviour on the London Underground


at Regents Park. TRRL Laboratory Report No. LR684. in clay. Canadian Geotechnical Journal, 17: 174-184.
BELSHAW, D. J., and PALMER, J.H.L. 1978. Results of a program PECK,R.B. 1986. Deep excavations and tunnelling in soft ground.
of instrumentation involving a precast segmented concrete-lined Proceedings, 7th International Conference on Soil Mechanics
tunnel in day. Canadian Geotechnical Journal, 15: 573-583. and Foundation Engineering, Mexico City, Mexico, pp. 225-290.
CLOUGH, G.W., and LECA,E. 1989. Models and design methods ROWE,R.K., and KACK,G.J. 1983. A theoretical examination of
with focus on use of finite element methods for soft ground tun- the settlements induced by tunnelling: four case histories. Cana-
nelling. General report. Proceedings, International Conference dian Geotechnical Journal, 20: 229-314.
on Tunnelling and Microtunnelling in Soft Ground-from Field ROWE,R.K., and LEE, K.M. 1991a. Subsidence due to tunnell-
to Theory, session IV, Paris, pp. 531-573. ing: Evaluation of a prediction technique. ASCE Journal of
CORDING, E.J., and HANSMIRE, W.H. 1975. Tunnels in soils- Geotechnical Engineering Division, submitted.
General Report. Proceedings, 5th Pan American Conference of -1991b. An evaluation of simplified techniques for
Soil Mechanics and Foundation Engineering, session IV, estimating 3-dimensional undrained ground movements due to
Buenos Aires, Argentina. tunnelling in soft soils. Canadian Geotechnical Journal,
CRAIG,R. 1975. Discussions at the meeting of the British Tun- submitted.
nelling Society. Tunnels and Tunnelling, 7: 61-65. ROWE,R.K., LO, K.Y., and KACK,G.J. 1983. A method of
DAVIS,E.H. 1988. Theories of plasticity and the failure of soil estimating surface settlement above shallow tunnels constructed
masses. In Soil mechanics selected topics. Edited by I.K. Lee. in soft ground. Canadian Geotechnical Journal, 20: 11-22.
Butterworths, Sydney, Australia. TERZAGHI, K. 1942. Shield tunnels of the Chicago Subway. Jour-
FINNO,R.J. 1983. Response of cohesive soil to advanced shield nal of the Boston Society of Civil Engineers, 29: 163-210.
tunnelling. Ph.D. thesis, Stanford University, Stanford, CA.
GIBSON,R.E. 1974. The analytical method in soil mechanics. List of Symbols
GCotechnique, 24: 115-140.
horizontal undrained shear strength
GUDEHUS,G. 1973. Elastoplastiche stoffgleichungen fur
trockenen sand. Ingenieur Archiv. 42. vertical undrained shear strength
KASALI,G., and CLOUGH, W. 1982. Three dimensional finite ele- anisotropically consolidated extension triaxial test
ment analysis-Advanced and conventional shield tunnelling. horizontal elastic modulus
Development of a design technology for ground support for tun- vertical elastic modulus
nels in soils, vol. 2, United States Department of Transportation. independent shear modulus
LEE, K.M. 1989. Prediction of ground deformations resulting total-gap parameter
from shield tunnelling in soft clays. Ph.D. thesis, University of physical gap
Western Ontario, London, Ont. equivalent 3D movement ahead of t h e tunnel face
LEE, K.M., and ROWE,R.K. 1989. Effects of undrained strength 0 workmanship parameter
anisotropy on surface subsidences induced by the construction
A thickness of the tail piece
of shallow tunnels. Canadian Geotechnical Journal, 26: 279-291.
-1991a. Finite element modelling of the three-dimensional 6 clearance required for erection of the lining
ground deformations due to tunnelling in soft cohesive soils. W water content
Part I - Method of analysis. Computers and Geotechnics, Y unit weight of soil
in press. st sensitivity of clayey soil
-1991b. Finite element modelling of the three-dimensional SPT standard penetration test

You might also like