Académique Documents
Professionnel Documents
Culture Documents
1, 75–90
Following a tunnel excavation in low-permeability soil, it Suite à une excavation de tunnel dans un sol de faible
is commonly observed that the ground surface continues perméabilité, il est fréquent d’observer que la surface du
to settle and ground loading on the tunnel lining changes, terrain continue de se tasser et que la charge du sol sur le
as the pore pressures in the ground approach a new revêtement du tunnel change à mesure que les pressions
equilibrium condition. The monitored ground response interstitielles dans le sol atteignent un nouvel équilibre. Le
following the tunnelling under St James’s Park, London, suivi de la réponse du sol après le creusement du tunnel
shows that the mechanism of subsurface deformation is sous le parc St James, à Londres, montre que le méca-
composed of three different zones: swelling, consolidation nisme de déformation sous la surface comporte trois zones
and rigid body movement. The swelling took place in a différentes de gonflement, consolidation et mouvement en
confined zone above the tunnel crown, extending verti- bloc. Le gonflement survient dans une zone confinée au
cally to approximately 5 m above it. On the sides of the dessus de la voûte du tunnel, s’étendant verticalement
tunnel, the consolidation of the soil occurred in the zone jusqu’à près de 5 m au-dessus. Sur les côtés du tunnel, la
primarily within the tunnel horizon, from the shoulder to consolidation du sol intervient dans la zone comprise
just beneath the invert, and extending laterally to a large essentiellement dans l’élévation du tunnel, de l’épaulement
offset from the tunnel centreline. Above these swelling jusqu’à juste en dessous du radier, et s’étendant latérale-
and consolidation zones the soil moved downward ment sur une large partie à partir de l’axe central du
as a rigid body. In this study, soil–fluid coupled three- tunnel. Au-dessus de ces zones de gonflement et de con-
dimensional finite element analyses were performed to solidation le sol descend vers le bas en bloc rigide. Dans
simulate the mechanism of long-term ground response cette étude, des analyses à éléments finis solide-fluide
monitored at St James’s Park. An advanced critical state couplées en trois dimensions ont été réalisées pour simuler
soil model, which can simulate the behaviour of London le mécanisme de réponse du sol à long terme contrôlé au
Clay in both drained and undrained conditions, was parc St James à Londres Un modèle avancé de comporte-
adopted for the analyses. The analysis results are dis- ment à l’état critique du sol, qui peut simuler le comporte-
cussed and compared with the field monitoring data. It is ment de l’argile de Londres dans des conditions drainées
found that the observed mechanism of long-term sub- et non drainées, a été adopté pour ces essais. Les résultats
surface ground and tunnel lining response at St James’s de ces analyses sont ici discutés et comparés avec les
Park can be simulated accurately only when stiffness données de suivi in situ. On a pu constater que le méca-
anisotropy, the variation of permeability between differ- nisme observé de réponse du revêtement du tunnel et du
ent units within the London Clay and non-uniform drain- sol sous la surface à long terme, au parc de St James, ne
age conditions for the tunnel lining are considered. This peut être simulé de façon exacte que lorsque l’anisotropie
has important implications for future prediction of the de la rigidité, la variation de la perméabilité entre les
long-term behaviour of tunnels in clays. différentes unités dans l’argile de Londres et les condi-
tions de drainage non uniforme du revêtement du tunnel
sont prises en compte. Ces observations ont des consé-
KEYWORDS: ground movements; numerical modelling; per- quences importantes en terme de prédiction du comporte-
meability; pore pressures; settlement; tunnels ment à long terme des tunnels creusés dans les argiles.
INTRODUCTION pore pressures will generally be lower than those prior to tunnel
Tunnelling in low-permeability soil often results in ground construction; settlement will therefore occur as pore pressures
surface settlement that continuously increases over a long reduce to their long-term steady-state values, increasing effec-
period of time. The mechanism for these longer term move- tive stresses and thereby inducing consolidation in the clay.
ments at the ground surface is that the tunnel inevitably The evidence that tunnels in low-permeability soil act as
introduces a new drainage boundary condition. This is because, new drainage boundaries was demonstrated from the field
on the inside face of the tunnel lining, the pressure is usually measurements of pore pressure around tunnels made earlier
atmospheric. The pore pressures immediately after construc- by Ward & Thomas (1965) and Palmer & Belshaw (1980).
tion of a tunnel are not in equilibrium with the modified Because of the large time-scale required for field-based
drainage boundary conditions. If the tunnel is not totally research, long-term data of ground movements are limited to
impermeable, a flow of pore water into the tunnel occurs, and a the ground surface, and long-term comprehensive data on
new steady-state flow condition is eventually reached. The final pore pressure changes around a tunnel lined with expanded
concrete segments are especially scarce. The evidence that
the ground surface continues to settle after tunnel excavation
Manuscript received 5 May 2006; revised manuscript accepted 16 has been illustrated by Peck (1969), O’Reilly et al. (1991),
October 2006.
Discussion on this paper closes on 1 July 2007, for further details
Lake et al. (1992), Bowers et al. (1996), Nyren (1998) and
see p. ii. Harris (2002). As most of the post-construction settlement
data available are only for the ground surface, the mechan-
Geotechnical Consulting Group, London, UK.
y
Department of Engineering, University of Cambridge, Cambridge, ism of the long-term ground response, and in particular at
UK. subsurface level, is not yet well understood.
75
Delivered by ICEVirtualLibrary.com to:
IP: 62.189.175.212
On: Thu, 04 Aug 2011 16:10:26
76 WONGSAROJ, SOGA AND MAIR
This paper discusses the ground movements and pore to follow that same trend, based on field measurements made
pressure changes at St James’s Park greenfield instrumented at Westminster (Burland & Hancock, 1977). However, there
site, which was monitored during construction of twin is no pore pressure measurement made at depths greater
tunnels for the Jubilee Line Extension (JLE) route in Lon- than 27 m below the ground surface to confirm that it is
don (Burland et al., 2001). From the comprehensive mon- under-drained.
itoring data obtained from the site (Nyren, 1998), a
mechanism of long-term subsurface ground movements at
this site is proposed. The monitoring data of subsurface OBSERVED LONG-TERM GROUND RESPONSE
ground deformations and measurements of pore pressure in MECHANISM
the vicinity of the tunnel have provided invaluable informa- The pore pressures around the westbound tunnel, meas-
tion to improve the understanding of long-term ground ured at the end of the excavation and then approximately
response mechanism induced by open-face tunnelling in 240 days later (i.e. start and end of period 3), are shown in
London Clay, particularly for those tunnels supported by Fig. 4(a). Those around the eastbound tunnel, measured at
expanded tunnel linings. the end of the excavation and then approximately one year
Because of the complexity of the problem, which involves later (i.e. start and end of period 5), are shown in Figs 4(b)
soil–tunnel–pore fluid flow interactions, three-dimensional and 4(c). Figs 4(a) and 4(b) show that, after the tunnel
(3D) finite element analysis (FEA) of the first excavation of excavations, the pore pressures above the crown of both
the westbound tunnel and subsequent consolidation prior to tunnels rose initially but subsequently stabilised. Neverthe-
the second excavation of the eastbound tunnel under St less, they were still slightly smaller than their pre-construc-
James’s Park was performed to assess the validity of the tion values. This indicates that the tunnels introduced new
proposed mechanism. An advanced anisotropic soil model, drainage boundaries into the soil, which was more evident
for which parameters were calibrated for both undrained and by the reduction in pore pressures at the springline level of
drained behaviour of London Clay, was used for the analysis. the eastbound tunnel one year after the tunnel excavation
The inclusion of stiffness anisotropy in the soil model was (see Fig. 4(c)).
needed to obtain the pattern of ground movements, the Although the tunnels are acting as drains, it seems that
excess pore pressure development and the tunnel lining the inward flow of the pore fluid towards the tunnel is larger
response similar to those observed in the field. The effects at the side of the tunnel than at the crown. This is confirmed
of two parameters on the computed subsurface long-term by the in-tunnel observations by Nyren (1998) that there are
ground response prior to the eastbound excavation were damp and wet patches on both sides of the tunnel. Some
examined. These parameters were permeability variation be- patches are large, and extend from track bed/key level
tween different London Clay units and lining–soil relative upward around the ring to the springline level (see Fig. 2).
permeability. These damp and wet patches below the tunnel springline
were reported for both the west and eastbound tunnels,
particularly nearby the instrumented section. The observation
ST JAMES’S PARK SITE of these patches below the springline and none above
Detailed descriptions of the St James’s Park instrumented correlates well with the measured reduction of pore pres-
site and the tunnel excavations beneath are given in Nyren sures around the springline (Fig. 4(c)) and the rise to almost
(1998); a brief summary is given in this section. St James’s pre-construction values above the crown (Figs 4(a) and
Park is one of the greenfield instrumented sites along the 4(b)). This suggests that the tunnel lining is likely to be
JLE route and is situated between Westminster and Green more permeable below the springline than above.
Park Stations. The alignment of the tunnels and locations of The measurements of subsurface vertical movements after
the instruments in plan and elevation views, together with the westbound tunnel excavation during the rest period
the soil profile interpreted from boreholes near the instru- (Table 1) are shown in Fig. 5(a). The data obtained from
mented section, are shown in Fig. 1. The twin tunnels are extensometer Bx, which is located directly above the west-
referred to as the westbound and the eastbound, with their bound tunnel centreline, indicates that there is a larger
axes located at approximately 31 m and 20.5 m below the increase in displacement at 27 m below the ground surface
ground surface respectively. The tunnels were excavated than that at 22.5 m. This increase in the relative displace-
using a 4.2 m long open-face tunnel boring machine (TBM) ment between these two points suggests that the soil between
with a backhoe excavator that has a 1.9 m reach when fully these two points is swelling. This swelling corresponds well
extended. They are supported by a 200 mm thick expanded to the rise in pore pressure above the tunnel centreline
precast concrete lining, which was erected near the end of during this period (see Fig. 4(a)).
trailing fingers, with a standard assembly to achieve an Although there is no pore pressure measurement at the
external diameter of 4.85 m (see Fig. 2). Measurement of springline level, the wet patches on both sides of the tunnel
ground response was divided into five distinct periods below this level indicate that there must be a reduction in
(Nyren, 1998), as summarised in Table 1. pore pressure, leading to consolidation of the soil on the
Two aquifers exist in the London Basin: (a) a deep aquifer sides of the tunnel. The amount of consolidation decreases
comprising Thanet Sand, Chalk and Basal Sand, overlain by with increasing offset from the tunnel centreline, as indi-
the Lambeth Group (formerly known as the Woolwich and cated by the data from extensometers Ax and Cx to Hx in
Reading Beds), which in many areas is overlain by London Fig. 5(a). These extensometer measurements also show that
Clay; and (b) a perched water table in Terrace Gravel on top the consolidation occurs only within the tunnel horizon in
of either London Clay or the clay of the Lambeth Group. the region close to the side of the tunnel.
This upper aquifer is recharged from surface precipitation Above this consolidation zone (i.e. at a depth less than
and locally from the River Thames. The observations from 20–25 m below the ground surface in this case), the soil
the piezometers installed in London Clay at different depths was settling as a rigid body. This was also noted by
in the Westminster area indicate that the pore pressures at Dimmock (2003). Based on this interpretation, a mechanism
the lower part of London Clay are slightly below hydrostatic, for the long-term ground movement at the westbound tunnel
as shown in Fig. 3 (Nyren, 1998). Hence there is usually can be split into three zones—a swelling zone, a consolida-
under-drainage at depths greater than 30 m below the ground tion zone and a rigid body movement zone—as schemati-
surface. The data for the St James’s Park site are also likely cally illustrated in Fig. 5(b).
Legend
Tree canopy
To Green Park
Eastbound outline Shallow surface
running tunnel monitoring point (SMP)
(4·85 m diameter)
Electrolevel inclinometer
Rod extensometer
Pneumatic piezometer
2m
2m 2m 4m 4m
2m I J K
2m
8m 6m
2m
8m
2m
4m 6m 6m 5·5 m 4·5 m 6m
4m 1 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
3
ST1
ST1
~2·5 m ~21·5 m
N
Westbound 5m 5m To Westminster
running tunnel
(4·85 m diameter)
(a)
0 A B C D E F G H I J K
Made Ground/alluvium
5
Terrace Gravels
10
FP1
Legend
London Clay
FP2 Rod extensometer
15
FP3 Electrolevel inclinometer
20 SP1/
SP4/
Piezometer
SP3/ SP2/ SC1
SC4 SC3 SC2
BP1
25 Combined pneumatic
BP2 spade cell/piezometers
30 Eastbound
running tunnel 5m 5m
(4·85 m OD)
35
Composite borehole log
from JLE BH109 and instrument 40
installation records
Westbound
Woolwich and running tunnel
Reading Beds Clay (4·85 m OD)
(b)
Fig. 1. (a) Plan and (b) elevation view of instrumented section at St James’s Park (after Nyren, 1998)
The extensometer measurements recorded after the east- the soil above the westbound tunnel appears to be consoli-
bound tunnel excavation are presented in Fig. 6(a). A ground dating (see extensometer Bx in Fig. 6(a)), which is probably
movement mechanism similar to that in the rest period was caused by the flow of pore fluid into the eastbound tunnel.
observed: slight swelling above the tunnel crown, which This leads to a mechanism of swelling, consolidation and
corresponds well with the rise in pore pressures above the rigid body movement zones, as depicted in Fig. 6(b).
tunnel, and consolidation at the sides of the tunnel due to In summary, the mechanism of long-term ground move-
the decrease in pore pressures, as shown in Figs 4(b) and ments at St James’s Park, as interpreted from the field data,
4(c) respectively. After the eastbound tunnel construction, is that the soil swells in a concentrated zone above the
D
33·6°
O
mm
4850
200 mm
A A⬘
T1 T2
Observed zone
of damp and 12·0° Observed zone
wet patches of damp and
Key Key wet patches
segment segment
T2 T1
Section A–A⬘
Leading edge
1000 mm
T1 T2 T1 T2
Key details
800 mm
400 mm
B B⬘
Section B–B⬘
Key type Dimension L
L Approximate scale
A 507 mm
B 487 mm 800 mm
Fig. 2. General arrangement of expanded tunnel lining used under St James’s Park (after
Nyren, 1998)
Note that the period is divided according to the survey number of precision levelling surveys (Nyren, 1998, table F1): period 1, survey nos
1–16; period 2, survey nos 16–29; period 3, survey nos 29–62; period 4, survey nos 62–85; period 5, survey nos 85–101.
tunnel crown and simultaneously consolidates on either sides Table 1) to examine the time-dependent behaviour of a
of the tunnel in the zone close to tunnel axis level. Above single tunnel construction. This was necessary to avoid any
these zones, the soil was settling as a rigid body, which complexity due to interactions of the two tunnels, so that a
resulted in continued settlement at the ground surface. time-dependent ground deformation mechanism can be pro-
posed based on a simple boundary condition. As future
research, the numerical study can be extended to include the
FINITE ELEMENT ANALYSIS effects of tunnel–tunnel interaction on the long-term ground
The construction of the westbound tunnel was modelled, deformation behaviour.
and the subsequent ground response prior to the eastbound
excavation was examined. All analyses presented in this
paper were performed using the soil-fluid coupled 3D finite Soil conditions and geometry
element (FE) method with the commercial finite element The FE model for the westbound tunnel excavation with
package ABAQUSTM . For brevity, this numerical study dimensions shown in Fig. 7 was assumed to have the
utilises the data mainly obtained during Period 1 to 3 (see following soil profile:
⫺5 London Clay/gravel
interface
Top of London Clay ⫺10
⫺10
Depth: m
St James’s Park ⫺15
⫺25
z: m
⫺30
⫺30 WB
Initial pore pressures
⫺35 240 days after excavation
Hydrostatic One year after excavation
Bottom of London Clay profile
⫺40 (a)
(estimated) Under-drained
Lambeth Group profile Pore pressure (kPa)
0 50 100 150 200 250
0
⫺50
⫺5 London Clay/Gravel
Fig. 3. Measured pore pressures at instrumented section in St interface
James’s Park and in Westminster area, London (Nyren, 1998)
⫺10
Depth: m
Ax Bx Cx Dx Ex Fx Gx Hx
5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15 5 0 ⫺5 ⫺10 ⫺15
⫺5 ⫺5 ⫺5 ⫺5 ⫺5 ⫺5 ⫺5 ⫺5
Depth: m
Rigid body movement ⫺15
consolidation ⫺25
occurs ⫺30
⫺35
⫺40
(b)
Fig. 5. (a) Field measurement of long-term ground movement after westbound tunnel construction; (b) schematic diagram
indicating zones of different response during consolidation
Ax Bx Cx Dx Ex Fx Gx Hx
0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30 0 ⫺10 ⫺20 ⫺30
⫺5 ⫺5 ⫺5 ⫺5 ⫺5 ⫺5 ⫺5 ⫺5
(b)
Fig. 6. (a) Field measurement of long-term ground movement after eastbound tunnel construction; (b) schematic diagram
indicating zones of different response during consolidation
WRBC p⬘
Subloading
surface
140 m
120 m
Region of 20-node
elements Normal yield surface
Fig. 7. Finite element model employed for analysis Fig. 9. Concept of subloading surface within normal yield
surface and ratio R
Unsupported length
during excavation
The tunnel lining thickness of 200 mm, which is the Soil constitutive model
actual thickness of the tunnel lining, was used for all All soil units in this study were modelled using the non-
simulations (neglecting the effect of the joints). The tunnel linear elasto-plastic critical state soil model with the sub-
lining was modelled using a linear elastic model with a loading surface concept (Hashiguchi & Ueno, 1977;
Young’s modulus of 28 GPa and a Poisson’s ratio of 0.15. Hashiguchi & Chen, 1998). The concept assumes that there
The expansion of the tunnel lining was not modelled. The is a small yield surface called the subloading surface, which
placement of the tunnel lining was simply modelled by always passes through the current stress point within the
activating the shell elements at the deformed tunnel heading. conventional yield surface (also termed the normal yield
As the expansion of the lining was not modelled, the lining surface), and the ratio of their size is denoted by R, as
was not perfectly circular with the initial opening diameter shown in Fig. 9. The soil behaviour is elastic if the subload-
of approximately 4.85 m. ing surface is contracting (i.e. R is reducing), and is elasto-
When the tunnel face advanced 100 m into the model, the plastic if the subloading surface is expanding (i.e. R is
long-term consolidation was set to begin by assigning a increasing). The ratio R and its evolution law control the
drainage-only flow surface to the tunnel boundary along the amount of plastic strain generated within the normal yield
entire excavated tunnel length, and consolidation was al- surface. Generation of plastic strains within the normal yield
lowed until the new steady-state pore pressure regime was surface was necessary to model accurately the development
reached. The drainage-only flow surface controls the drai- of excess pore pressures during undrained shearing. The
nage condition at the tunnel boundary by assuming that the constitutive model also incorporates the following features:
pore fluid velocity in the direction outwards from the soil non-linear elasticity, anisotropic stiffness, the Matsuoka &
element, vn (m/s), is proportional to the pore pressure at the Nakai (1985) failure criterion, and a re-invoked small-strain
tunnel lining extrados, uw, when it is positive. The propor- stiffness at stress reversal points (SRP; i.e. unloading and
tional constant is termed the seepage coefficient, KT , and is reloading). Further details of the model are described in
equivalent to kl /tl ªw , where kl and tl are the permeability Appendix 1.
and thickness of the tunnel lining respectively. Additionally, The model was implemented in the ABAQUSTM user
no flow is allowed across the tunnel boundary when uw is interface. The model was adopted for all soil units in the
negative (i.e. water is not supplied from inside the tunnel). analyses. As the Made Ground is far from the tunnel, and
Hence varying the magnitude of KT (m3 /kN.s) is the same as the strains occurring within this material are small during
varying the permeability of the tunnel lining for a given tunnel excavation, its behaviour is more or less elastic. For
tunnel lining thickness, which results in different drainage the Terrace Gravel and Woolwich and Reading Beds clay,
conditions at the tunnel boundary. Damp patches on the the parameters were selected such that the stiffness degrada-
tunnel lining suggest that the pore pressure in the tunnel tion curves are within the bounds adopted by other studies
lining segment may be negative. However, the suction within such as Addenbrooke et al. (1996). The critical state lines
the tunnel lining is difficult to determine. Therefore atmo- for the Made Ground and Terrace Gravel were calculated
spheric pressure was assumed inside the tunnel lining for all from the angles of friction given in Potts & Zdravkovic
analyses presented in this paper. (2001). The parameters for London Clay were determined
*Only adopted for the simulations to investigate influence of permeability profiles and tunnel drainage conditions.
Soil descriptions by Standing & Burland (2006) Permeability profile 1: m/s Permeability profile 2: m/s
kv kh kv kh
13 Profile 2, kh LC Div B shown in Fig. 13. The zone of negative excess pore
18
Profile 2, kv pressures with the isotropic model has contours that spread
Burland & Hancock (1977) in the horizontal direction, whereas those computed with the
23 LC Div A3ii anisotropic soil model give contours that spread more in the
28 vertical direction. The computed effective stress paths at
LC Div A3i
33 the tunnel crown and springline closely followed that of an
38
undrained triaxial compression test on a horizontally and
LC Div A2
vertically cut sample respectively (Yimsiri, 2001; Hight et
43 al., 2003). That is, the soil adjacent to the tunnel crown
48 WRB clay extrados experiences a dilation tendency whereas the soil
adjacent to the tunnel springline extrados experiences a
Fig. 11. Assumed permeability profiles tendency to contraction prior to dilation. This initial differ-
ence is due to the initial elastic stiffness anisotropy (e.g.
for these layers based on field measurement data reported in Graham & Houlsby, 1983) before the plastic dilation devel-
Burland & Hancock (1977) and Hight et al. (2003). The ops at larger strains. The isotropic model gave similar stress
assumed anisotropic permeability ratio (kh /kv ) was 2, except paths above the tunnel crown and near the springline: both
for division A3ii, for which the assumed kh /kv ratio was 5. displayed some contraction tendency followed by dilation,
This higher ratio was assigned because water strikes were defined by the isotropic soil model. Soil elements adjacent
encountered in this division (Standing & Burland, 2006). As to the tunnel springline extrados experienced more shearing
discussed before, the tunnel lining above the springline was and greater negative excess pore pressure than soil elements
assumed to have finite permeability whereas the lining below adjacent to the tunnel crown extrados; this was due to the
the springline was assumed to be fully permeable, which is initial K0 (.1) condition.
considered to be reasonably realistic based on the in-tunnel The computed pore pressure profile above the tunnel
observations. The under-drained initial pore water pressure crown is plotted in Fig. 14. The figure shows that, at a depth
profile derived from permeability profile 2 is shown in of 27 m, the simulation with the anisotropic model gave a
Fig. 12. larger reduction in pore pressure than that computed with
the isotropic soil model. The computed pore pressures above
the tunnel crown remain positive at most depths apart from
the zone very close to the tunnel crown. Based on Fig. 14,
IMPORTANCE OF SOIL STIFFNESS ANISOTROPY FOR which includes field measurements of pore pressure, and the
LONG-TERM GROUND RESPONSE development of excess pore pressures during tunnel excava-
It has been reported widely that stiffness anisotropy has tion (not shown), the trend of change in pore pressures
an important influence on predicted short-term ground move- computed with the anisotropic soil model agrees more
ments caused by tunnel excavation (e.g. Addenbrooke, 1996; closely with the field data than that computed with the
Simpson et al., 1996; Lee & Ng, 2002; Franzius, 2003). isotropic model.
Similar findings were made in this study for both surface When the tunnel lining is acting as a partially permeable
and subsurface ground movements, the use of stiffness boundary, the pattern of excess pore pressures generated
anisotropy providing a better match to the field data on around the tunnel with the anisotropic model led to larger
short-term ground movements (Wongsaroj et al., 2004). The consolidation settlement above the tunnel centreline, causing
influence of stiffness anisotropy on long-term ground re- the tunnel lining to squat in the long term. This is indicated
sponse is presented in this paper, based on results of by the increase in the horizontal diameter, as shown in
analyses assuming K0 ¼ 1.5 for London Clay. Further Fig. 15.
investigation showed that the agreement between predictions The direct comparison between this prediction and field
measurements cannot be made owing to the lack of field
data for tunnel lining deformation during the rest period.
Pore pressure: kPa
0 50 100 150 200 250 300 350
0 Isotropic Anisotropic
⫺10
⫺5
0
0
Depth below ground surface: m
⫺10
⫺20
⫺15 ⫺40
⫺20
Depth: m
⫺40
⫺30
⫺25
Fig. 12. Pore pressure profile derived from permeability profile Fig. 13. Excess pore pressure contours around tunnel computed
2 (see Fig. 11) with isotropic and anisotropic soil models in 3D
Depth: m
⫺5 0 days
⫺15
100 days
Depth below ground surface: m
Depth: m
⫺15 0 days
100 days
However, the trend of squatting with time has been observed ⫺20 Steady state
in other tunnels in London Clay (e.g. Ward & Thomas, ⫺25
1965; Nyren, 1998 (during period 5); Dimmock, 2003). The ⫺30 Pre-construction profile
analysis with the isotropic model gave the opposite trend: Distance from springline extrados: m
0 10 20 30
that is, decreasing horizontal diameter with time, as shown 300
⫺4
IMPORTANCE OF PERMEABILITY PROFILE AND
TUNNEL DRAINAGE CONDITIONS FOR LONG-TERM
⫺6 GROUND RESPONSE
Days after excavation
Simplified permeability model case
Fig. 15. Influence of stiffness anisotropy on tunnel lining The drainage condition of the tunnel boundary is often
deformation following tunnel excavation (increase of horizontal assumed to be influenced by the relative magnitude of the
diameter shown positive) tunnel lining permeability and the surrounding soil per-
⫺5 ⫺5
⫺10 ⫺10
Consolidation
time
⫺15 ⫺15
0 day
Depth (m)
⫺25 ⫺25 86 days
175 days
⫺30 ⫺30 215 days
274 days
⫺35 ⫺35
⫺40 ⫺40
⫺45 ⫺45
⫺50 ⫺50
(a) (b)
⫺5 ⫺5
⫺10 ⫺10
Consolidation
time
⫺15 ⫺15
0 day
Depth (m)
⫺40 ⫺40
⫺45 ⫺45
⫺50 ⫺50
(c) (d)
Fig. 18. Computed subsurface ground vertical movement with simplified permeability model: (a) Bx along tunnel centreline; (b) Cx
4 m from tunnel centreline; (c) Dx 10 m from tunnel centreline; (d) Ex 16 m from tunnel centreline
level. This suggests that the tunnel is significantly more able boundary at the top half and permeable at the bottom
permeable below the springline level: this may be asso- half. Fig. 19(a) shows the distribution of pore pressure
ciated with the key segments being smaller than the other around the tunnel extrados when KT of the upper part was
types (see Fig. 2), because there were holes at the key 1 3 1012 m3 /kN.s. The partial recovery of pore pressure is
locations of the tunnel. Although these holes were filled computed above the tunnel springline, whereas nominally
with in situ concrete, their permeability would probably be zero pore pressure is computed at the bottom half of the
much higher than that of the precast concrete segments tunnel lining extrados.
manufactured under controlled conditions. For demonstra- Figures 19(b) and 19(c) show the recovery of pore pres-
tion purposes, the lower part of the tunnel below the sures with time above the tunnel crown for a range of KT
springline was modelled as a permeable boundary in this values. At BP1, which is approximately 4.5 m above the
analysis. To ensure the permeable condition at the tunnel tunnel crown extrados, the recovery of pore pressure seems
below the springline, a large value of KT relative to the soil to be less dependent on the relative lining–soil permeability.
permeability of 1 3 105 m3 /kN.s was assigned to this This is because the magnitude of pore pressure depends
region. As a result, the tunnel can still act as an imperme- mainly on the horizontal flow of pore fluid due to the
100 263 days simplified model case. This is clearly shown by the larger
Steady state ⫺90°
pore pressure at BP1 computed with the permeability profile
50
2 (Fig. 19(b)) than that with profile 1 (Fig. 17(a)).
0 0° The findings from the results of different permeability
models emphasise that appropriate tunnel drainage condi-
⫺50 tions should be assigned for a long-term analysis. The
90°
⫺100 drainage conditions should take into account the presence of
⫺100 ⫺50 0 50 100 joints, gaps and the relatively high-permeability region in
Position from springline: degrees the tunnel lining, although how to account for joints and
(a) gaps is far from straightforward. Furthermore, the rate of the
consolidation settlement depends principally on the per-
180 meability of the soil at the tunnel axis level, because the
160 majority of the consolidation takes place on either side of
Pore pressure: kPa
⫺5 ⫺5
⫺10 ⫺10
Consolidation
⫺15 ⫺15 time
0 day
⫺20 ⫺20 5 days
Depth: m
Depth: m
23 days
⫺25 ⫺25
71 days
⫺40 ⫺40
⫺45 ⫺45
⫺50 ⫺50
(a) (b)
⫺5 ⫺5
⫺10 ⫺10
Consolidation
time
⫺15 ⫺15
0 day
Depth: m
23 days
⫺25 ⫺25 71 days
135 days
⫺30 ⫺30 263 days
⫺35 ⫺35
⫺40 ⫺40
⫺45 ⫺45
⫺50 ⫺50
(c) (d)
Fig. 20. Computed subsurface ground vertical movement with refined permeability model: (a) Bx along tunnel centreline;
(b) Cx 4 m from tunnel centreline; (c) Dx 10 m from tunnel centreline; (d) Ex 16 m from tunnel centreline
APPENDIX 1 where u1 and m are the material constants, and då p is the change in
The normal yield surface adopted in this study is the Modified plastic strains tensor, which is obtained from the flow rule. For
Cam Clay yield surface, which can be described in q–p9space as simplicity, the associated flow rule is assumed in the model and,
therefore, the plastic potential is described by equation (2).
F ¼ q 2 M 2 ð p90 p9Þ p9 ¼ 0 (1) The conditions for the elasto-plastic process are given by
9
where M is the gradient of the critical state line and a function of the R ¼ 0: dR ¼ þ1 > >
=
Lode angle using the Matsuoka & Nakai (1985) failure criterion, and 0 , R , 1: dR . 0
for p 6¼ 0 (4)
p90 is a constant defining the size of the normal yield surface (i.e. R ¼ 1: dR ¼ 0 > >
;
isotropic preconsolidation pressure). For a given value of R, the R . 1: dR , 0
subloading surface can then be expressed as
The elastic modulus of the model is assumed based on experimental
F R ¼ q 2 M 2 ð Rp90 p9Þ p9 ¼ 0 (2)
data plotted in log10 e–log10 p9 space, where swelling behaviour is
Two different evolution laws for R were proposed by Hashiguchi described by the gradient of the swelling line in log10 e–log10 p9 as
& Chen (1998). The one adopted in this study is expressed as (after Pestana, 1994)
1 1 þ øs s p9 1=2
dR ¼ u1 1 kdå p k (3) rr ¼ þDð1 r Þ (5)
Rm Cb pa
Ghh /Gvh
dimensional (1D) swelling line respectively; D is the gradient of the
1.5
1
1
1
isotropic swelling line at low mean effective pressure; r controls the
rate at which the isotropic swelling line reaches the gradient D; and
atmospheric pressure pa of 100 kPa is assumed for this study. and
s are used to re-invoke small-strain stiffness at stress reversal points
(i.e. unloading and reloading). They are the dimensionless distances
0.12
0.2
0.2
0.2
9hh
in space as defined by
s ¼ ðç çrev Þ: ðç çrev Þ 1=2 (6)
p9= p9rev for p9 , p9rev
0.16
¼ ;<1 (7)
0.2
0.2
0.2
9hv
stress reversal.
0.2
0.2
0.2
9vh
(8)
r
˜1
where is the accumulated strain relative to the previous stress
reversal state, which can be decomposed into its volumetric and
deviatoric components, ˜1 p (¼ p prev ) and ˜1 ås (¼ ås åsrev ).
0.05
0.05
D
–
–
From equation (5), the tangent bulk modulus, K9, during swelling
is defined as
1 þ e p9
0.2476
K9 ¼ (9)
0.556
e rr
0.37
rc
0.3
9hv
29vh 29hh 9vh
E9v ¼ K9 1 49vh þ (11)
9hv 9hv
15
15
20
50
øs
Under the assumption that the soil is isotropic in the plane of the
deposition, the value of Ghh becomes
E9h
Ghh ¼ (12)
2ð1 þ 9hh Þ
100
400
200
900
Cb
The values of 9vh , 9hv , 9hh and Gvh are evaluated from published
data.
For the present model, it is assumed that the normal consolidation
0.1
0.1
0.2
0.1
therefore becomes
1þe
d p90 ¼ p90 dpv (13)
eðrc rr Þ
100
100
300
100
u1
0.65
0.5
0.7
e
NOTATION
Table 4. Model parameters for all soil units
London Clay