Tunnelling and Underground Space Technology 50 (2015) 152–170

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

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

Long-term performance of cast-iron tunnel cross passage in London clay

Zili Li a,⇑, Kenichi Soga b, Peter Wright c,1
a
Civil & Environmental Engineering Department, Colorado School of Mines, Golden, CO 80401, United States
b
Department of Engineering, Trumpington Street, Cambridge CB2 1PZ, UK
c
CH2M Hill UK Ltd, Elms House, 43 Brook Green, London W6 7EF, UK

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

Article history: A series of 3D soil–fluid coupled finite element analyses was conducted to examine the long-term tunnel
Received 18 January 2015 behaviour of an old cast-iron cross passage in stiff London clay. In the proposed geotechnical FE model, an
Received in revised form 27 May 2015 advanced critical state constitutive model was employed to simulate complex soil behaviour, whereas
Accepted 10 July 2015
the details of the tunnel linings was simplified using shell elements. The computed time-dependent soil
Available online xxxx
load derived from the geotechnical model was then applied to a structural finite element model where
the details of the cross passage structure such as bolted-joints and tunnel segments were explicitly mod-
Keywords:
elled. This proposed semi-coupled soil-structure model shows agreement against field observations and
Cast-iron cross passage
Tunnel opening
it was found in this particular case that the structural condition evaluated from the FE model is more crit-
Long-term ground response ical than that of the conventional bedded ring method. Results show that the surface ground settlement is
3D soil arching governed mainly by the twin tunnel construction and the presence of the cross passage does not increase
Semi-coupled soil-structure model the settlement. On the other hand, the construction of a cross passage affects the structural performance
of the adjacent running tunnel in both short term (undrained) and long term (soil consolidation). In par-
ticular, the critical location is identified to be the tunnel segment adjacent to the opening at the axis level
and its stresses and deformation increase with consolidation time.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction tunnel structures. In this approach, the soil loading applied to tun-
nel lining is estimated using a set of reaction springs that considers
London has one of the most extensive networks and oldest sec- the interaction between the soil and the tunnel (i.e. the conven-
tions of railway tunnels in the world (Wolmar, 2004). The tional bedded ring method) (Klappers et al., 2006; Li et al., 2014,
long-term performance of old tunnels, which is associated with submitted for publication). However, such springs are not capable
the change in earth pressure during soil consolidation due to seep- of simulating the complexities of soil behaviour such as soil stress
age into the tunnel, is of great concern (e.g. Mair, 2008). In partic- history, anisotropy and the changes in soil stiffness and earth pres-
ular, the engineering conditions of old cross passages between sures due to seepage induced consolidation around the tunnel in
adjacent cast-iron tunnels are found to be critical by recent assess- the long-term.
ment (Wright, 2010). The long-term behaviour of a tunnel in London clay was inves-
To assess the conditions of a tunnel, analytical equations tigated by Wongsaroj et al. (2007) using soil–fluid coupled finite
derived from 2D plane-strain models are widely adopted element analysis, which allowed the influence of soil strata and rel-
(Murakami and Koizumi, 1980; Duddeck and Erdmann, 1985; ative soil-lining permeability to be considered (Wongsaroj et al.,
RTRI, 1997). Such conventional 2D methods allow engineers to 2013). Later Laver (2010) conducted a further study to account
examine the general behaviour of a full tunnel ring, but they fail for the effect of twin tunnels interaction on long-term behaviour
to investigate the performance of some special structural features and identified that the interaction effect depends on the spacing
such as cross passage openings. To overcome this problem, finite between the two tunnels and cover depth. Many of the past
element analysis is often used to account for the details of such researches (e.g. Addenbrooke, 1996; ITA, 2000; Laver, 2010) pri-
marily focused on ground response induced by tunnel excavation
rather than investigating the tunnel lining behaviour subjected to
⇑ Corresponding author.
long-term changes in earth pressures. In particular, less attention
E-mail addresses: tongjilzl@gmail.com (Z. Li), ks207@cam.ac.uk (K. Soga), Peter.
Wright@ch2m.com (P. Wright).
has been paid on the long-term tunnel behaviour of cross passage
1
Formerly Tube Lines Ltd, 15 Westferry Circus, Canary Wharf, London E14 4HD, section, which is often more structurally critical than the other
UK. non-cross passage running tunnel sections.

http://dx.doi.org/10.1016/j.tust.2015.07.005
0886-7798/Ó 2015 Elsevier Ltd. All rights reserved.
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 153

Nomenclature

RP relative soil-lining permeability rw bulk unit weight of water

DT tunnel diameter DS dimensionless surface settlement
kl lining permeability Sc max(ss) consolidation component of the maximum surface
tl Lining thickness settlement at steady-state
kS soil permeability Sc max(ssi) = Sc max(ss) for an impermeable lining case
Cclay clay cover Sc max(ssp) = Sc max(ss) for a fully permeable lining case
KT lining seepage coefficient

In this study, a series of soil–fluid coupled finite element anal-

yses was conducted to investigate the effect of long-term consoli-
dation on the performance of a cast-iron cross passage section. The
influence of cross passage and relative soil-lining permeability on
ground response as well as tunnel lining behaviour was examined.
Results show that surface ground settlement was similar along the
tunnel rail direction regardless of a cross passage. On the other
hand, the soil load profile applied to the tunnel linings was very
much influenced by the presence of cross passage. To simulate
the tunnel structural performance more realistically without
compromising computational power, a 3D structural model, which
considered the critical tunnel structural features explicitly
(e.g. bolted-joints, tunnel segment), was developed and the
time-dependent soil loading computed from the geotechnical (a) Photo of the cast-iron cross passage
model was applied to this structural model. By doing so, the
long-term structural performance of a cast-iron cross passage sec-
tion in London Underground tunnel was evaluated against field
observations.

2. Geotechnical finite element model

2.1. Ground condition and geometry

There are many different types of cross passage tunnel openings

in London Underground and most of them do not show structural
distress even after more than 100-year service (Tube Lines, 2008;
Wright, 2010). Of particular interest in this study is a critical
cast-iron cross passage 19 m below ground surface in Northern
Line, which is considered as a case study here. At this site, twin (b) Illustration of the cast-iron cross passage
tunnels with a tunnel diameter D = 3.8 m with a spacing d0 = 6 m
Fig. 1. A critical cast-iron cross passage in Northern Line.
between the two are connected by a small cross passage as shown
in Fig. 1. The length of each ring is 0.508 m. The cross passage is
made by removing parts of five rings and a lintel is installed at Lambeth Group Clay, LG (i.e. Woolwich and Reading Bed Clay,
the top of the opening. The shape of the cross passage opening is WRBC) (20.9–34.9 m). The earth pressure coefficient at rest K0, crit-
rectangular about 1.90 m in height and 2.54 m in width. The twin ical state angle of internal shearing resistance and soil permeability
tunnel sections were constructed in 1904 (Tube Lines, 2005a), kS of each layer are listed in Table 1, while the initial pore water
whereas the cross passage between the twin tunnels was pressure was assumed hydrostatic with the water table at 5.5 m
constructed afterwards as part of the original scheme but clear below the ground surface. The ground condition parameters were
construction records were not documented (Tube Lines, 2006). determined based upon previous investigations by Tube Lines
The tunnel deformation at this site has been monitored for some (2006) and Wongsaroj (2005).
10 years using a real-time sensor system with electro-levels and To save computational time without compromising accuracy, a
potentiometers, and some similar visible evidence of deterioration, coarse mesh was employed at the far boundaries, while a fine mesh
such as segment movement and lintel bending, has been observed was adopted for the key sections (i.e. cross passage). The soil was
at both ends of the cross passage over the recent years. modelled using two types of solid elements: (a) 8-node trilinear
Considering symmetry, a 1/4 finite element model of the displacement and pore pressure elements, and (b) 20-node tri-
cast-iron cross passage as shown in Fig. 2 was developed using quadratic displacement, trilinear pore pressure, reduced integra-
ABAQUS 6.12 (ABAQUS Inc., 2012). The top model boundary was tion elements. The 20-node elements were employed in the zone
set to be free, whereas the vertical movements at the bottom around the tunnel since they gave smoother pore pressure
boundary were fixed. At the sides, the horizontal movements and response than the 8-node elements, which were used for the
longitudinal movements were set to be zero in order to account remaining zone.
for the symmetry plane and model boundary, respectively. The To simulate the ground response more realistically, an
vertical soil profile at the site is as follows: (a) Made Ground and advanced non-linear elasto-plastic critical state soil model
Alluvium, MG (0–3 m), (b) London Clay, LC (3–20.9 m), and (c) (Wongsaroj, 2005; Laver, 2010) was employed for all soil units.
154 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(a) Soil strata

(b) Tunnel structures (Overview)

(c) cross passage section (Local view)

Fig. 2. Finite element model for the cross passage.

The proposed model takes account of many critical mechanical fea- manner similar to the soil elements, each ring near the cross pas-
tures, such as elastic and plastic anisotropy, small strain stiffness sage was modelled individually, whereas a number of rings were
as well as stiffness degradation. Due to lack of site investigation, modelled together using fewer elements in the remaining area.
the soil properties in each strata were assumed to be typical values To take account of segmental joints, Li et al. (2014) conducted 3D
according to Laver (2010) and Tube Lines (2007) as shown in FE structural analysis, which considered some critical structural
Table 2. Further details of the constitutive model can be found in features explicitly (e.g. tunnel segment, bolted joint) and validated
Wongsaroj (2005) and Laver (2010). against past experimental data (Thomas, 1977), and the results
In the geotechnical model, the tunnel linings were modelled were compared to those computed from the 2D plane strain
using 8-node, double curved thick shell elements. The elements method (LUL, 2003). It was found that the bending stiffness of
were numbered along the tunnel rail direction consecutively. In a bolted cast iron lining can be generally considered as a continuous
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 155

Table 1
Assumed soil properties (Wongsaroj, 2005; Laver, 2010).

Stratum Bulk density, c (kN/m3) Critical angle of shearing resistance, /cv0 (°) Coefficient of earth pressure at rest, K0 Permeability, k (m/s)
Made Ground 20 25.0 0.6 1  107
London Clay 20 21.0 1.5 2  1011
Lambeth Group 20 27.0 1.5 2  1011

Table 2
Summary of material properties for all soil units (Wongsaroj, 2005; Laver, 2010).

Strata M e0 u1 m Cb xs qc D r m0 vh m0 hv m0 hh Ghh/Gvh bvv bhh bvh

Made Ground 0.984 0.65 100 0.1 100 15 0.2476 0 0 0.2 0.2 0.2 1 0 0 0
London Clay 0.814 0.7 300 0.05 300 10 0.3 0.05 2 0.015 0.04 0.12 1.5 0.1 0.05 0
Lambeth Group 1.07 0.65 100 0.1 900 50 0.37 0.05 2 0.2 0.2 0.2 1 0 0 0

ring with a reduction factor based upon Muir Wood’s formula Table 4
(Muir Wood, 1975), which equals to 0.44 for the case considered Construction stages assumed in FE simulation.
in this study. Construction Description
The lintel and trackbed at the opening section are also struc- stage
turally critical since they transfer the hoop load from the incom- 1 Geostatic
plete rings to the adjacent full rings. In this study, the lintel was 2 Remove the soil inside the running tunnel and reduce 50% of
modelled using 3-node quadratic beam elements, while 20-node the initial stress around the tunnel boundary (undrained)
quadratic solid elements were employed for the trackbed along 3 Place the lining and further reduce the initial stress to be 0%
(undrained)
the tunnel rail direction. The tunnel linings, lintel and trackbed
4 Remove the soil inside the cross passage tunnel and reduce
were modelled as linear elastic-perfectly plastic material and their 50% of the initial stress around the cross passage boundary
properties are summarised in Table 3. The tunnel shell elements (undrained)
and the adjacent soil elements shared the same nodes at the tunnel 5 Place the lining on the cross passage and further reduce the
initial stress to be 0% (undrained)
boundary; the interface was not modelled. This was considered to
6 Soil consolidation (drained)
be acceptable since the tunnel is unlikely to detach or slip away
from the surrounding ground during soil consolidation
(Wongsaroj et al., 2007). The 3D soil–tunnel–fluid coupled model
closely-space twin tunnels. The ground surface settlement builds
consists of 4648 elements and 9368 nodes. The model boundaries
up with the unloading of initial soil stress. When the soil load
were set to minimise the effects on ground response and tunnel
was decreased to 50% of the original overburden pressure, a
behaviour (Wongsaroj, 2005; Laver, 2010).
volume loss was 1.09%. This is in line with the typically measured
volume loss in London clay (Addenbrooke, 1996; Mair, 2008). In
2.2. Tunnel construction and long-term consolidation this study, an empirical reduction ratio of 50% ground pressure
was adopted for deep tunnels in stiff ground as suggested by
The construction of old cast-iron cross passages in London Muir Wood (1975) and Duddeck and Erdmann (1985).
Underground tunnel were carried out many decades ago and the After the twin tunnel construction, a cross passage was con-
detailed knowledge of tunnelling construction is not available structed in the model. All the construction stages in this study
owing to the absence of historical records. For brevity, a simplified were assumed to be in undrained conditions. Compared to the
construction stage was assumed in the FE modelling and sum- twin-tunnel construction stage, the effect of cross passage con-
marised in Table 4. It was thought that the twin tunnels were likely struction on ground deformation is very small as shown in Fig. 3b.
to be constructed first before the excavation of the cross passage. Tunnel drainage was then allowed in the post-construction
To consider the soil load redistribution induced by excavation, stage. Recent studies (e.g. Wongsaroj et al., 2013) identified the
the elements of excavated soil were removed and the initial nodal importance of soil–tunnel drainage condition on the long-term
loads around the tunnel boundary was reduced to a certain per- ground response due to tunnelling. That is, the magnitude of tun-
centage (e.g. 50%). Then the lining ‘shell’ elements were attached nelling induced consolidation depends on tunnel lining permeabil-
to the tunnel boundary. ity in relation to soil permeability and drainage distance. To
Fig. 3a shows the ground surface settlement immediately after quantify the consolidation effect, Laver (2010) proposed a radial
tunnel construction at the symmetrical plane of the cross-passage. flow model as shown in Fig. 4 and defined the following relative
The maximum ground settlement appears at the centre of the cross soil-lining permeability RP.
passage, where significant soil strain occurs between the two  
DT kl 2C clay
RP ¼ ln þ1 ð1Þ
2t l ks DT
Table 3 kl
Summary of material properties for tunnel structures (after Thomas (1977), Gilbert KT ¼ ð2Þ
(1977) and Tube Lines (2005b)). t l cw

Specification Cast iron Steel Concrete where DT is the tunnel diameter, kl is the lining permeability, tl is
lining lintel trackbed the lining thickness, kS is the soil permeability, Cclay is the clay cover,
Young’s modulus (GPa) 100 200 24 KT is the lining seepage coefficient, and cw is the bulk unit weight of
Poisson’s ratio 0.26 0.28 0.2 water.
Tensile strength (MPa) 180 235 2.5
By increasing the lining seepage coefficient KT, the tunnel
Compressive strength (MPa) 600 235 25
becomes more permeable in comparison to the surrounding
156 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(a) Ground surface settlement due to the unloading of initial soil stress

(b) Ground surface settlement due to cross passage excavation

Fig. 3. Short-term ground surface settlement induced by tunnel excavation.

in the soil flow into the tunnel. But when the pore pressure is neg-
ative, it cuts off the flow since no water is supplied from inside the
tunnel to the soil. Such modelling allows the soil-lining drainage
condition to be modelled more realistically. The parameter RP will
be used in the later section to show the effect of soil consolidation
on the long-term behaviour of a cross passage in stiff London clay.

3. Ground response

Previous efforts (e.g. Wongsaroj, 2005; Laver, 2010) have inves-

tigated the long-term ground response of a single tunnel or twin
tunnels. This study extends this work by examining the
long-term ground response by cross passage construction. Fig. 5
shows how the long term surface settlement (i.e. excluding the
short term settlement due to tunnel construction) of the cross pas-
sage section changes with time for two idealised scenarios: (a)
Fig. 4. Illustration for deriving relative soil-lining permeability (Laver, 2010). impermeable tunnel lining (i.e. relative soil-lining permeability
RP ? 0) and (b) fully permeable tunnel lining (i.e. RP ? 1). In
the impermeable case, the ground surface heaves because the soil
ground (i.e. bigger RP). This allows pore water in the soil ingress swells due to the buoyancy effect of the watertight tunnel and also
into the lining at a faster rate and hence more tunnelling induced the dissipation of the negative excess pore water pressures, which
ground settlement in the long-term. Furthermore, the are developed by soil shearing during tunnel excavation in heavily
drainage-only boundary model in ABAQUS allows only the water overconsolidated clay. When the lining is fully permeable, the pore
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 157

(a) Impermeable case

(b) Fully permeable case

Fig. 5. Long term surface settlement over the consolidation time (short term ground movements are not included).

pressure becomes zero at the edge of the tunnel lining in the long accompanied by greater volume loss. The volume loss can be as
term. The effective stresses of the soil near the tunnel increase and much as 2.64% for the fully permeable case (i.e. 2.4 times the short
thus the soil consolidates, which leads to long-term ground settle- term volume loss), indicating the significance of the long-term
ment until a steady-state flow condition is reached. ground settlement. Wongsaroj et al. (2013) normalised the com-
To compare the ground response of twin tunnels with and with- puted maximum surface settlement against the impermeable and
out a cross passage, Fig. 6 plots the ground surface settlements at fully permeable cases using the following equation:
the long-term steady state for the two cases. There is little differ-
ence in the surface settlement profiles of the two cases. Also, the Sc maxðssÞ  Sc maxðssiÞ
DS ¼ ð3Þ
differential settlement along the longitudinal direction is small as Sc maxðsspÞ  Sc maxðssiÞ
shown in Fig. 6b. This indicates, for closely-spaced twin tunnels,
the effect of cross passage on surface settlement may not be signif- where DS is dimensionless settlement, Sc max(ss) is the consolidation
icant from the engineering assessment point of view. Hence, for component of the maximum surface settlement at steady-state,
brevity, only the surface settlement directly above the cross pas- Sc max(ssi) = Sc max(ss) for an impermeable lining case, whilst
sage is examined in the follow sections. Sc max(ssp) = Sc max(ss) for a fully permeable lining case.
In this study, the permeability of the lining was varied so that Fig. 7b indicates the tunnel is impermeable (i.e. DS = 0, maxi-
the relative soil-lining permeability RP ranged between 105 and mum ground heave) if RP is smaller than 101. As the lining
106. Fig. 7a shows how the ground surface consolidation becomes more permeable, the DS increases and finally levels off
settlement changes with different lining permeability. The more when RP is bigger than 103 (i.e. fully permeable). Based on the
permeable the lining is, the more settlement builds up, which is results of a series of numerical analysis of the long-term ground
158 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

the cross passage itself is very much influenced by the variation

in effective stress and pore pressure around the tunnel. The lining
behaviour of the cross passage section will be examined in detail in
the next section.

4. Tunnel lining behaviour

In this section, both the deformation and internal stresses of the

tunnel lining at the cross passage section are examined. Since Mair
(2008) reports that the measured pore pressure immediately adja-
cent to old tunnels in low permeability London clay usually is close
to zero (i.e. the tunnels act as drains), the tunnel lining is assumed
as fully permeable for this detailed investigation.

4.1. Tunnel deformation

(a) Ground settlement along the transverse settlement trough
Fig. 8 shows the change in tunnel diameter with time along the
full ring section. As the soil consolidates, the vertical diameter of
tunnel lining decreases by 1.20 mm, which is accompanied by
the increase in the horizontal diameter of 1.36 mm at tunnel axis
level (i.e. tunnel squats). Most of the tunnel squatting builds up
within 2000 days (i.e. about 5.5 years) after tunnel construction
as the soil around the tunnel gradually reaches a new
steady-state flow condition. The consolidation time of long-term
tunnel deformation is much shorter than that of ground settle-
ment; the latter usually continues to build up after several decades
due to the longer drainage path from ground surface to tunnel
boundary.
Fig. 9 shows the change in tunnel diameter with time along the
tunnel rail direction near the cross passage opening section.
Compared to the full ring section, the cross passage opening, which
is pushed horizontally by soil loading, tends to deform back to the
original circular shape due to its stiffness reduction in the horizon-
tal direction (i.e. less tunnel squatting). The change in the vertical
diameter at the centre of the cross passage opening (i.e. Ring 0) is
0.94 mm, which is 78.3% of the tunnel diameter change at the full
(b) Ground settlement along the longitudinal direction section (i.e. 1.20 mm). The effect of the opening gradually fades out
along the rail direction and finally disappears at Ring 12 (i.e. 6.35
Fig. 6. Surface settlement at the long-term steady state (short term ground metres away from the opening). Along the railway track, the differ-
movements are not included).
ential tunnel displacement remains localised and the magnitude is
very small (i.e. within a few millimetres).
settlement due to a single tunnel construction in London clay,
Laver (2010) developed an empirical equation of DS = 1/(1 + RP4). 4.2. Tunnel stress
Compared to the hydraulic mechanism of a single tunnel, the
mechanism of a cross passage with twin tunnels may behave either In the geotechnical tunnel model, the tunnel lining is simplified
similar to a small single tunnel (i.e. lower bound, D = 3.8 m) or a big as a continuous cylinder with a stiffness reduction factor to
one circumscribing the twin tunnels (i.e. upper bound, D = 9.8 m); account for the effect of segmental joint. Li et al. (2014,
the geometries are illustrated in Fig. 7c. In this manner, the equiv- submitted for publication) argue that a continuous ring model only
alent RP of a cross passage at both lower and upper cases can be considers compressible stiffness of segmental lining, but it fails to
derived based on the mathematical model of a single tunnel by express the detailed bending behaviour (e.g. the local bending stiff-
Laver (2010) (see Fig. 4 and Equations (1) and (2)). Results show ness at flexible joints). For accuracy, only hoop thrust but not bend-
that the hydraulic performance of the cross passage behaves more ing moment along the tunnel lining is examined from the results of
like a bigger tunnel rather than a smaller one in the middle (see the geotechnical tunnel model. Later in this paper, the bending
Fig. 7b and c). The data forms an S-shape curve, while the assumed behaviour is examined separately by applying the computed soil
upper bound case (i.e. the big tunnel with a diameter of 15.8 m) loading into a detailed structural model.
agrees well with the equation proposed by Laver (2010). The find-
ing is confirmed by pore water pressure profile as shown in Fig. 7d, 4.2.1. Long term behaviour of the twin tunnels section
where the pressure contour of the cross passage resembles more To understand the soil arching effect induced by tunnel excava-
like a bigger tunnel than a smaller one, particularly above the tun- tion before ground consolidation, Terzaghi (1943) proposed a sim-
nel crown. plified soil strip model to illustrate the load transfer above the
In summary, the effect of cross passage on the long-term sur- tunnel as shown in Fig. 10a. In this figure, the sliding of the over-
face ground settlement is relatively negligible compared to the burden soil strip III over the tunnel is resisted by the friction along
magnitude of the surface settlement generated by twin tunnel con- the vertical sections ad & bc and consequently transferred to the
struction only. That is, the seepage into twin tunnel dominates the adjacent soil strips I & II. In a similar manner, Dimmock (2003) pro-
consolidation induced settlement. However, the performance of posed a load column mechanism to explain the effect of
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 159

(b) Dimensionless settlement over the relative permeability

(a) Ground consolidation settlement at long-term steady state

(short term ground movements are not included)

(c) Illustration of the equivalent tunnel diameter

(i) Cross passage (ii) Big tunnel (D=9.8m) (iii) Small tunnel (D=3.8m)
(d) Pore water pressure profile due to seepage into the tunnel
Fig. 7. Ground surface settlement at different relative soil-lining permeability.

on the two soil columns at the sides of the tunnel. As the soil col-
umns shorten with time during consolidation, the soil columns
retract support, which results in more load transferred through
the lining.
For the closely-spaced twin tunnels in Northern Line, there are
three soil columns along the cross-section; one between the two
tunnels and two at the other two sides. The middle soil column
(see Column I in Fig. 10c) consolidates more rapidly between the
twin tunnels than the soil columns at the outer sides (Column II)
(i.e. an equivalent shorter drainage path). This differential consoli-
dation condition may cause soil loading along the lining to be
asymmetric.
Fig. 11a gives an illustration of the segmental lining in detail.
Fig. 8. Change in tunnel diameter at the full ring section during soil consolidation Along the lining, six tunnel elements were numbered
(Day 0 is the end of the cross passage tunnel construction). anti-clockwise starting from crown Segment A, invert Segment C
& D, and then finally back to crown Segment F. In the geotechnical
consolidation on tunnel lining as shown in Fig. 10b. In his mecha- model, each tunnel segment was modelled by two shell elements
nism, the overburden above the lining is sustained by the interac- named using subscript r and l, respectively; for example, shell ele-
tion between three load columns; one on either side of the tunnel ments A.r and A.l represent Segment A (see Fig. 11b). At tunnel axis
(i.e. soil column I & II) and one through the tunnel lining itself. level, Segment B is adjacent to the middle soil column I, whilst
During tunnel excavation in clayed soil, the overburden first relies Segment E is next to the outside soil column II.
160 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(a) Illustration of tunnel deformation

(a) The soil strip model for illustrating the
soil arching effect (Terzaghi, 1943)

(b) Tunnel vertical deformation (b) Single tunnel load-analogy of three

load columns (Dimmock, 2003)

(c) Twin tunnel load-analogy

(c) Tunnel horizontal deformation Fig. 10. The effect of consolidation on tunnel lining.

Fig. 9. Tunnel deformation at different sections along the longitudinal direction

(Day 0 is the end of the cross passage tunnel construction). The soil pressure applied to the tunnel lining is evaluated from
the computed nodal forces. Fig. 12b shows the normal soil stress
profile acting on the lining in both short term and long term.
Fig. 12a shows how the computed hoop thrust forces of With time, the vertical soil stress applied to the lining builds up
Segment B and Segment E change over time. As expected, the as more overburden is transferred to the lining and consequently
thrust of Segment B increases at a more rapid rate than that of increases hoop forces in the axis level tunnel segments. The inte-
Segment E due to a shorter drainage path during soil consolidation. gral of the increased vertical pressure along the top half of the tun-
The hoop thrusts of both segments develop from an identical mag- nel equals to the increment of hoop force as illustrated in Fig. 12b.
nitude (i.e. 377.1 kN/m; 55.0% full overburden) and consequently In comparison to the normal soil stress, the tangential stress is very
reach the same level (i.e. 450.7 kN/m; 65.7% full overburden) in small.
the long term. The computed magnitudes of the hoop thrusts are
in line with the typically measured value (i.e. about 60% full over- 4.2.2. Long term behaviour of the cross passage section
burden) in London Underground (Addenbrooke, 1996; Tube Lines, To illustrate the soil arching mechanism above a tunnel open-
2007). Apart from the 65.7% full overburden applied on tunnel lin- ing, a ‘‘table’’ load column model, which extends Dimmock’s
ing, the rest (34.3% full overburden) is then sustained by the soil model, is proposed as shown in Fig. 13a. Similar to the twin tunnel
arching within the cross section perpendicular to the tunnel rail mechanism, the soil overburden is sustained by the tunnel struc-
direction. On the other hand, the hoop thrusts at the crown also ture together with four ‘‘table legs’’ (i.e. soil load columns) around
build up during soil consolidation, but the increment is less than it considering the symmetry of the problem. To focus on the struc-
the axis level. Such difference is a result of the change in earth tural response along the lining, Fig. 13b shows the detailed struc-
pressure during soil consolidation and this leads to tunnel ture of the tunnel cross passage at X–X cross section, which is a
squatting with time. side view of segmental linings in Fig. 13a. Furthermore, the side
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 161

development of 3D soil arching in both transverse cross section

and longitudinal section. As consequence, the overburden is trans-
ferred to the adjacent soil columns (e.g. II, IV) in the short term.
Most of the arching is developed in the longitudinal direction to
Soil column IV. This results in large increase in normal stress on
the left side of the non-opening rings adjacent to the opened sec-
tion (Ring 3, 5 and 7 in Fig. 14b1, c1 and d1, respectively). A small
increase in normal stress due to soil arching into Soil column II can
be observed on the right side of Ring 1 and 3.
The increased loads in Soil column II and IV during cross pas-
sage construction generate large excess positive pore water pres-
sures. As the pore water pressures dissipate, the soil columns
(i) The tunnel segments along the lining consolidate and the overburden load redistributes. The normal
stresses in the non-opening rings (Ring 3, 5 and 7 in Fig. 14b2, c2
and d2) reduce with time, whereas those in the opening rings
(Ring 1 in Fig. 14a2) increases. Ring 7 now has soil stresses similar
to the far end Ring 12. Hence, in the short term, the soil arching
effect influences up to five times the width of the opening size
(i.e. 12 rings). However, in the long term, the extent of influence
becomes smaller (up to 6 rings).
To quantify the 3D soil arching effect, Table 5 lists the soil pres-
sures applied on different tunnel rings in both short-term and
long-term. When immediately after cross passage construction,
the earth pressure applied on the lining at the far end (Ring 12)
is 208.7 kPa, which equals to 55.0% full overburden. At the tunnel
opening section (Ring 0–2), the total load decreases from
521.8 kN (=208.7 kPa  1.01 m2 (Tunnel surface area above the
(ii) The tunnel ringconsists of shell elements in the geotechnical model opening)/ring  2.5 rings) to 438.6 kN (=164 kPa
(a) Numbering of the segmental ring (Ring0)  1.01 m2  0.5 rings + 169 kPa (Ring1)  1.01 m2  1
rings + 183.3 kPa (Ring2)  1.01 m2  1 rings). After the
long-term soil consolidation, the soil pressure at the opening sec-
tion reloads back to a magnitude similar to the far-end full ring
section. This reloaded overburden applied to the opening is trans-
ferred through the lintel and consequently to the segments
underneath.
Fig. 15a and b shows the hoop thrust at different rings both in
the short term and long term, respectively. Of particular interest
is the critical Segment B in Ring 3 adjacent to the opening at the
(i) Shell elements for a tunnel segment (ii) A tunnel segment axis level. The short term hoop thrust is 128.8% of the overburden,
which is significantly higher than the thrust in the non-cross pas-
(b) Numbering of a tunnel segment
sage section (i.e. 55.0% overburden). After the long term, the thrust
Fig. 11. Illustration of a tunnel ring in FE models. continuously builds up to 191.1% of the overburden due to the
overburden redistribution as discussed before. This segment is
obviously considered to be one of the most critical sections of
view of cast-iron tunnel segments and lintel is illustrated at Y–Y the cross passage section and its behaviour will be discussed in
cross section in Fig. 13b. At the right end of the figure, the detail in the next section.
half-ring above the tunnel opening is numbered as zero, while
the other rings are numbered consecutively along the longitudinal
railway direction. In accordance with the ‘‘table’’ analogy model, 5. A semi-coupled soil-structure model
the left-hand side of Ring 0, 1, 2 adjoins to soil column I, while
its right-hand side stands for the soil column II. Away from the 5.1. Methodology
opening, the left-hand side and right-hand side of the rest of the
rings are at the position of soil column IV and III, respectively. Previous investigations (e.g. Li et al., submitted for publication)
Fig. 14 shows the change of soil load on the linings at (a) the suggest that the whole life-cycle of a tunnel cross passage should
short term stage, which is immediately after cross passage excava- consider the following four stages: (1) self-weight loading stage
tion (undrained), and (b) the long-term stage (consolidation steady (i.e. lining construction stage), (2) soil loading stage, (3) tunnel
state). For brevity, the data from the odd-numbered rings in the opening in short term (cross passage construction stage) and (4)
longitudinal direction are given here. Both normal and tangential soil pressure reloading stage (i.e. long-term soil consolidation
soil stresses acting along the circumference of the lining are plot- stage). Ideally any model should aim to model all these stages.
ted. Also for comparison, the normal soil stresses acting on the In the geotechnical model described in the previous sections,
far end ring (i.e. Ring 12) is plotted, which is not affected by the the soil behaviour, such as nonlinear plasticity and small strain
cross passage construction. stiffness, were simulated as accurate as possible. It modelled the
When a cross passage is excavated, the lining stiffness of the tunnel linings as a continuous ring using shell elements with
opened section (Ring 0, 1 and 2) is lost and the normal soil stress reduced stiffness to consider joints while beam elements were
reduces to zero as shown in Fig. 14a1 (see the left-hand side of used for the lintel above the tunnel opening. Such simplification
Ring 1). Soil column I then settles accompanied by the was acceptable to understand the general structural behaviour,
162 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(a) Tunnel load-analogy of a table with four load columns

(a) Long-term increase of hoop thrust (Day 0 is the end of the cross
passage tunnel construction)

(b) Side view of tunnel opening subjected to soil

overburden at cross section X-X
Fig. 13. 3D cross passage tunnel opening.

(b) Actual soil load on the lining

Fig. 12. Long-term increase of hoop thrust.

In the conventional bedded ring model, on the other hand, the
ground surrounding the tunnel is simplified as a set of elastic or
elasto-plastic reaction springs (see Fig. 16a). For modelling cross
passage (Stage 3 and 4), however, there are some critical issues
but it failed to examine the effect of segmental joints and complex in this soil spring-structure approach. For example, since the stiff-
tunnel geometry on the structural behaviour in a precise manner. ness of each soil spring is independent, it cannot consider the 3D
As an alternative, a conventional bedded ring model can be per- soil arching developed at the cross passage section. Moreover,
formed to account for the soil–tunnel interaction (Duddeck and when modelling the construction process, it is also unreasonable
Erdmann, 1985). In this approach, the complex 3D tunnel structure to neglect the change of earth pressure caused by the effect of
including flanges, panels and joints is modelled explicitly. Fig. 16a unloading – reloading on the lining during consolidation.
shows that the 3D structural model consists of twelve tunnel rings, In this study, a semi-coupled soil-structure model as shown in
trackbed and a lintel across the opening, which were modelled Fig. 16b was developed to investigate the complex soil-structure
explicitly using solid elements. On the other hand, the bolts in interaction. In correspondence to the geotechnical model proposed
the segmental joints were modelled using a new bolt-spring model in this study, a 1/2 structural model identical to the conventional
proposed by Li et al. (2014). In this approach, a bolt is simplified as bedded ring model was used for the simulation except the sur-
a set of normal and tangential springs (Fig. 17) to simulate rota- rounding soil springs. When conducting the soil-structure analysis,
tional and shear behaviour as shown in Fig. 17c and d, respectively; the self-weight of the tunnel lining was first applied on the struc-
their parameters were calibrated from the experimental data as tural model to simulate Stage 1 (lining construction). In the follow-
well as detailed modelling of bolts (Li et al., 2014; Thomas, ing stages (Stage 2–4), the soil load derived from the soil-pore fluid
1977). The adopted bolt-spring model in conjunction with mod- coupled geotechnical model at each stage was sequentially applied
elling of the tunnel segment were well validated against laboratory to the structural model. Theoretically it is more appropriate to
test by Thomas (1977). By using this proposed approach, Li et al. model both structure and soil in one fully-coupled finite element
(submitted for publication) conducted further numerical analysis model. Unfortunately this was too expensive computationally to
of a cast-iron tunnel ring and the computed results showed good be implemented in this study.
agreement with classical analytical solution (Duddeck and
Erdmann, 1985) and typical field measurement (Wright, 2010).
Based upon the abovementioned verified efforts, the structural 5.2. Critical tunnel segment
model in this study consequently extends to simulating the cross
passage tunnel opening more realistically by considering the criti- Fig. 18a shows the maximum stresses of compression and ten-
cal structural features (e.g. segmental joints). Due to limited space sion computed inside the most critical Segment B in Ring 3 from
for this article, details of the structural model and the associated Stage 1 to 4. The results from two cases are given; one is the
discussion were described in Li et al. (submitted for publication). semi-coupled soil-structure model (Fig. 16b), whereas the other
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 163

(a.1) Ring 1 short term (c.1) Ring 5 short term

(c.2) Ring 5 long term

(a.2) Ring 1 long term

(b.1) Ring 3 short term (d.1) Ring 7 short term

(b.2) Ring 3 long term (d.2) Ring 7 long term

Fig. 14. Soil load on different rings in short term and long term.
164 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

Table 5
Soil load applied on tunnel lining near the cross passage.

Soil load Ring number

12 ... 7 6 5 4 3 2 1 0
Far end Opening section
Short-term
Magnitude (kPa) 208.7 ... 223.7 230.4 231.5 233.4 233.8 183.3 169.0 164.0
Load/overburden (%) 55.0 ... 59.0 60.7 61.0 61.5 61.6 48.3 44.5 43.2
Long-term
Magnitude (kPa) 252.9 ... 258.2 260.3 262.4 262.8 280.7 248.1 246.6 244.6
Load/overburden (%) 65.7 ... 67.1 67.6 68.2 68.3 72.9 64.5 64.1 63.5

is the soil spring-structure model (i.e. bedded ring model) (see

Fig. 16a). In the former model, the soil loads computed from the
geotechnical model were applied. In the latter model, the external
earth pressure was applied and the soil reaction forces applied to
the lining was computed by the soil spring-structure interaction.
In the soil spring model (i.e. the latter model), the earth pressure
applied to the lining is estimated. For tunnel design, some
researchers (e.g. Duddeck and Erdmann, 1985; Lee et al., 2001)
assume it as 100% primary soil stress in the undisturbed ground,
since it is thought that at the final steady-state stage (years after
construction), the ground will eventually return to the same condi-
tion as before the tunnelling. In fact, however, during tunnel exca-
vation, the soil stress around may redistribute due to soil arching
effect (Terzaghi, 1943). Furthermore, Mair (2008) demonstrates
that after years of consolidation, there is no evidence that the earth
pressure will return to the initial stress state due to various relative
soil-lining permeability and ground conditions. Considering these,
other researchers (Muir Wood, 1975; Addenbrooke, 1996) suggest
to apply a reduction factor to the primary soil stress as the earth

(a) Short term hoop thrust

(a) Conventional bedded ring model

(b) Detailed structural model in the soil-structure model

(b) Long term hoop thrust
Fig. 16. The semi-coupled soil-structure model versus the conventional bedded
Fig. 15. Hoop thrust in the tunnel lining for different rings. ring model.
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 165

X-X cross section

(a) (b)

(c) (d)
Fig. 17. Structural model of a cast-iron joint: (a) Bolt in a joint, (b) Illustration of nine springs model at X–X cross section, (c) Spring model in rotational load, and (d) Spring
model in shear load (Li et al., 2014).

pressure applied to the lining. In this study, the external earth little tensile stress is found at most area of the segment. In partic-
pressure was considered to be 60% full overburden in both vertical ular, significant stress concentration occurs at the top radial flange
and horizontal direction based upon typical tunnel measurements (i.e. the arrowed area) due to the contact pressure transferred from
summarised by Addenbrooke (1996) and Tube Lines (2008). The the lintel. This computed tunnel behaviour is also noted in the field
applied external earth pressure will be sustained by the soil– observation as indicated by the dotted white circle in
tunnel interaction: usually, majority of the pressure (e.g. 90% Fig. 20a and b, where the segment under the lintel develops a local
according to the force factor n = 0.9) is carried by the tunnel lining distortion. In this figure, it is also found that the web of the lintel
(Duddeck and Erdmann, 1985; Li et al., submitted for publication), moves away from the soil side towards the tunnel side (see the
whilst the rest is by the soil springs. In this study, the soil spring red circle), which is illustrated in Fig. 20c. Likewise, the FE model
stiffness is 120 MPa/m for Stage 3 immediately after construction predicted a similar structural deformation mode of the lintel (see
and decreases to 60 MPa/m for Stage 4 during long-term consolida- Fig. 20d). In addition, large bending deformation occurs at the
tion, based upon relative soil–tunnel stiffness (Duddeck and end of the lintel as shown in Figs. 20 and 21, which is caused by
Erdmann, 1985; USACE, 1997; Tube Lines, 2007). the lintel-segment contact pressure; 7.9 mm bending deformation
In the first two stages (i.e. self-weight loading stage and soil occurs at left end (Fig. 21a), while 5.6 mm is at the right end
loading stage), all the linings consisted of full rings and the two (Fig. 21b). The computed tunnel model predicts a bending
approaches resulted in similar stress states. In the last two stages deformation of 6.9 mm generally in agreement against field obser-
(i.e. tunnel opening stage and long-term reloading stage), several vations (see Fig. 21c).
tunnel segments were removed to model cross passage opening Fig. 22 shows the observed dislocation of the segments above
and the soil subsequently consolidated. The soil spring-structure the opening modelled in this study. The inclination of the glass
model predicted a stress state within the permissible range both (see Fig. 22a and b) indicates that the segments above the lintel
in compression and tension. On the other hand, the semi-coupled moves inward to the tunnel about 2 mm over many decades after
soil-structure model gave greater stresses than the soil construction. The observed segment movement is also noted in the
spring-structure model because more soil load was applied to FE model (see as circled in Fig. 22c and d).2 At the full ring section,
Ring 3 due to the longitudinal soil arching effect described in the the rings squat by having positive horizontal displacement (red
previous section. This was also noted in the internal forces. The colour) at the right side of the tunnel and negative horizontal dis-
bending moments and hoop thrusts in Ring 3 for the two cases placement on the opposite side. At the cross passage section, more
are shown as shown in Fig. 18b and c, respectively. Again the horizontal movement builds up (light blue to dark blue colour) at
semi-coupled soil-structure model predicted greater moment and the opening side due to its weaker structural integrity after the
hoop thrust than the soil spring-structure model when the cross removal of tunnel segments. At the opposite of the opening side
passage was constructed. (the right side), the positive horizontal displacement reduces
Fig. 19 shows the compressive and tensile stress contours in the slightly. This shows that the tunnel at the cross passage section
critical Segment B of Ring 3 predicted by the semi-coupled
soil-structure model. Large compressive stress appears at the cir-
cumferential flange (see the black circle in Fig. 19b) and build 2
For interpretation of color in Fig. 22, the reader is referred to the web version of
ups in the long term due to the earth pressure reloading, whereas this article.
166 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(a) Maximum compressive / tensile stresses

(b) Maximum hoop thrust (c) Maximum bending moment

Fig. 18. Maximum compressive/tensile stresses in the tunnel segment B of Ring 3.

(i) Short-term (ii) Long-term (i) Short-term (ii) Long-term

(a) Tensile stress contour (b) Compressive stress contour

Fig. 19. Principal stresses contour in the tunnel segment B of Ring 3.

shifts towards the opening as shown schematically in the plan view The soil load then transfers through the end of the lintel to the seg-
given in Fig. 22e. Further detailed discussion on the cast-iron tunnel ment underneath. The significant lintel-segment contact pressure
behaviour can be found in Li et al. (2014, submitted for publication). results in local deformation and stress concentration at the contact
In summary, the soil pressure at the tunnel opening section area. In the segment, the transferred overburden load causes sig-
overloads the lintel and bends its web away from the soil side. nificant compressive stress at the circumferential flange. When
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 167

(a) The front view of the lintel

(b) The side view of the lintel

(c) Illustration of the lintel distortion

(i) short-term (ii) long-term

(d) The FE model (the distortion is magnified to 3 times)
Fig. 20. Buckling deformation of the lintel.
168 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(a) The left end of the lintel

(b) The right end of the lintel

(c) The numerical model (the distortion is magnified to 3 times)

Fig. 21. Bending deformation of the lintel above the opening.

the soil consolidates (Stage 4), the soil load applied to the lining long term in stiff London clay. The computed ground response is
increases, which leads to further stresses and movement of the in line with the findings of previous efforts (Wongsaroj, 2005;
structure. Laver, 2010; Wongsaroj et al., 2013), while the calculated tunnel
Ideally, more sophisticated quantitative assessment of the com- behaviour shows agreement with field measurement (e.g. Tube
puted tunnel behaviour should be made against the field measure- Lines, 2005b). Results provide new insights into the mechanism
ment. However, most of the cast-iron cross passages in London of the long-term behaviour of a cross passage and the main conclu-
Underground were constructed more than 100 years, whereas the sions can be drawn as follows:
field monitoring data or construction records, if they were docu-
mented in the past, are now not available. The critical cast-iron (1) For closely-spaced twin tunnels, the effect of a cross passage
cross passage in Northern Line investigated in this study provided on the long-term surface ground settlement is limited. That
an opportunity to assess the potential long-term performance of is, the soil consolidation due to drainage into twin tunnels is
cross passage tunnels, whereas similar critical cross passage site much greater than the additional drainage effect by the cross
or relevant case study, to the authors’ best knowledge, is very rare. passage. The hydraulic field of the twin tunnels behaves
more like a large circular tunnel case circumscribing the
6. Conclusions whole cross passage area.
(2) After tunnel construction, the tunnel lining develops a fur-
A series of three dimensional soil-pore fluid coupled FE analyses ther squatting as the surrounding soil consolidates. The
was conducted to investigate the long-term behaviour of a cast magnitude of ovalisation at the cross passage section is less
iron tunnel cross passage in London Underground Tunnels. The than that of the full ring section. When segments are
computed soil load acting on the tunnel lining was applied to a removed, the flexible horizontal stiffness allows soil pres-
structural model that included the structural details such as sure to push it back to the original circular shape. The extent
flanges, pans and bolts. This allowed to examine the complex of influence of cross passage construction is generally
soil-structure interaction of the cross passage in both short and localised within five times of the opening size.
 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170 169

Fig. 22. The inward movement of the opening: (a) the left end of the opening, (b) the right end of the opening, (c) the overview of the inward movement in the FE model, (d)
the local inward movement (the distortion is magnified to 10 times), and (e) Illustration of the tunnel distortion (plan view).
170 Z. Li et al. / Tunnelling and Underground Space Technology 50 (2015) 152–170

(3) After cross passage construction, a 3D soil arching mecha- Dimmock, P.S., 2003. Tunnelling-Induced Ground and Building Movement on the
Jubilee Line Extension. Ph.D. Thesis, University of Cambridge.
nism develops and the soil adjacent to the cross passage sec-
Duddeck, H., Erdmann, J., 1985. On structural design models for tunnels in soft soil.
tion carries the additional overburden load and positive Underground Space 9, 246–259.
excess pore pressure develops. However, as the soil consoli- Gilbert, G.N.J., 1977. Engineering data on grey cast irons – SI units. BCIRA
dates, the overburden load gradually transfers back to the Alvechurch, Birmingham.
I.T.A. Working Group No. 2. (ITA), 2000. Guidelines for the design of shield tunnel
crown section of the tunnel opening, which is then trans- lining. Tunnelling Underground Space Technology, vol.15, pp. 303–331.
ferred to the ring next to the opening section via the lintel. Klappers, C., Grubl, F., Ostermeier, B., 2006. Structural analyses of segmental lining-
A tunnel load-analogy of a ‘‘table’’ with four load columns coupled beam and spring analyses versus 3D-FEM calculations with shell
elements. Tunn. Undergr. Space Technol. 21, 254–255.
was proposed to account for the 3D soil arching mechanism Laver, R., 2010. Long-Term Behaviour of Twin Tunnels in London Clay. Ph.D. Thesis,
observed in the geotechnical model. The University of Cambridge, Cambridge, UK, 2010.
(4) The tunnel segment adjacent to the opening at the axis level Lee, K.M., Hou, X.Y., Ge, X.W., Tang, Y., 2001. An analytical solution for a jointed
shield driven tunnel lining. Int. J. Numer. Anal. Meth. Geomech. 25 (4), 365–390.
is the most critical structural section. To better simulate the Li, Z., Soga, K., Wang, F., Wright, P., Tsuno, K., 2014. Behaviour of cast-iron tunnel
soil–tunnel interaction, a semi-coupled soil-structure model segmental joint from the 3D FE analyses and development of a new bolt-spring
that applied the soil load computed in the geotechnical model. Tunn. Undergr. Space Technol. 41, 176–192.
Li, Z., Soga, K., Wright, P. 2014b, Behaviour of Cast-Iron Bolted Tunnels and their
model directly to the detailed structural model was used. Modelling. Tunn. Undergr. Space Technol. (submitted for publication).
The computed tunnel behaviour derived from the proposed L.U.L. London Underground Ltd. (LUL), 2003. Review of Engineering Standard E3322
model shows agreement against the field observation and A2 Deep tunnels and shaft – assessment. LUL Report 4058-PMN, London.
Mair, R.J., 2008. Tunnelling and geotechnics: new horizons. Geotechnique 58 (9),
predicts a more critical stress state against the conventional
695–736.
bedded ring method. Results show that the lintel, which is Muir Wood, A.M., 1975. The circular tunnel in elastic ground. Geotechnique 25 (1),
overloaded by the soil pressure at the tunnel opening sec- 115–127.
tion, develops large bending deformation at the web section Murakami, H., Koizumi, A., 1980. On the behaviour of the transverse joints of a
segment. In: The 35th Annual Conference of the JSCE, pp. 73–86. (in Japanese).
as well as the contact area with the segment underneath. Railway Technical Research Institute. (RTRI), 1997. Design Standard for Railway
The lintel-segment contact pressure results in local defor- Structures (Shield-Driven Tunnel). Tokyo. (in Japanese).
mation and stress concentration at the radial flange of the Terzaghi, K., 1943. Theoretical Soil Mechanics. John Wiley and Sons, New York, pp.
66–76.
segment, whilst it transfers the overburden load to the cir- Thomas, H., 1977. Measuring the Structural Performance of Cast Iron Tunnel Linings
cumferential flange, where significant compressive stress in the Laboratory. Geotechnics Division, Building Research Station, Building
occurs. The deformation of the lining increases with time Research Establishment, Department of the Environment. pp. 29–36.
Tube Lines, 2005a. Tunnel Assessment Project Annual Works Plan 2: Historical
as the soil pressure and hoop load increases due to soil con- Study Report, TLL-L001-N416-DTAAWP2-TUN-RPT-00013, London, UK.
solidation. Hence it is important to evaluate the evolution of Tube Lines, 2005b. Deep Tube Tunnel Knowledge & Inspection Programme Annual
the distress in the segments of the ring next to the opening. Works Plan 2, Validation Report Elastic Continuum Analysis, Finite Element
Modelling Ultrasonic Testing, Strain Measurement, BEM Testing, ACSM Testing,
TLL-L001-N416-DTAAWP2-TUN-RPT-00006, London, UK.
In this study, the properties of tunnel structure was assumed Tube Lines, 2006. Projects Directorate Permanent Way Delivery Team AWP4:
not to change with time. By adopting a reasonable structural dete- Conceptual Design Statement, TLL-N098-P100-TUN-CDS-00001, London, UK.
Tube Lines, 2007. Deep tube tunnel knowledge & inspection programme Soil
rioration relationship, further investigation can be performed to
Parameters Report, TLL-L001-N416-DTAAWP2-TUN-RPT-00001, London, UK.
consider the effect of ageing process on the time-dependant lining Tube Lines, 2008. Deep tube tunnel knowledge & inspection programme annual
behaviour. work plan 2: Cast iron Openings report. TLL-L001-N416-DTAAWP2-TUN-RPT-
00003. London.
United States Army Corps of Engineers (USACE), 1997. Engineering and Design
Acknowledgements Tunnels and Shafts in Rock EM 1110-2-2901, Department of the Army,
Washington, DC, USA.
This work was supported by the Major State Basic Research Wolmar, C., 2004. The Subterranean Railway: How the London Underground was
Built and How it Changed the City for Ever. Atlantic.
Development Program of China (973 Program) (No. Wongsaroj, J., 2005. Three-Dimensional Finite Element Analysis of Short and Long-
2011CB013803) and the Cambridge Trust scholarship. Term Ground Response to Open-Face Tunnelling in Stiff Clay, PhD Thesis.
University of Cambridge.
Wongsaroj, J., Soga, K., Mair, R.J., 2007. Modelling of long term ground response to
References tunnelling under St James’ Park London. Geotechnique 57, 75–90.
Wongsaroj, J., Soga, K., Mair, R.J., 2013. Tunnelling-induced consolidation
ABAQUS Inc., 2012. ABAQUS User’s Manual, Version 6.12. SIMULIA. <www. settlements in London Clay. Geotechnique 63, 1103–1115.
simulia.com>. Wright, P., 2010. Assessment of London Underground tube tunnels – investigation,
Addenbrooke, T.I., 1996. Numerical Analysis of Tunnelling in Stiff Clay. PhD Thesis, monitoring and analysis. Smart Struct. Syst. 6, 239–262.
Imperial College of Science Technology and Medicine, University of London.