Underground works in soils and soft rock tunnnelling doc

UNDERGROUND WORKS IN SOILS AND SOFT ROCK
TUNNELING


Eric Leca
1
, Yann Leblais
2
and Karl Kuhnhenn
3




ABSTRACT

Growing needs for modern transportation and utility networks have increased the demand for a more
extensive and elaborate use of underground space. As a result, more underground projects have to be completed
in a variety of ground conditions, including weak water bearing soils and soft rocks. Significant technological
advances have rendered these projects possible, but have also given rise to new challenges as many of these
projects have to be completed in difficult conditions, with very strict environmental constraints, particularly in
urban areas where the potential impact of tunneling on existing structures is a major concern. This report
addresses the main aspects of tunneling and underground works performed in soils and soft rocks. A summary is
presented of the main features related to construction techniques, ground investigations, design methods, and
instrumentation and monitoring practices, as well as of some of the more recent advances in these fields.
Significant progress has been made in the area of soft ground tunneling over the past thirty years, partly because
of advances in computer technologies. The scope of increasing difficult project conditions to be addressed
requires that the best use be made of these technologies, as well as of lessons gained from past experience and
current observational records.

1.0
INTRODUCTION


Growing needs for modern transportation and utility networks have given rise to an increased demand for a
more extensive and elaborate use of underground space. Some of these projects are related to urban
development, which requires the construction of more metro systems, underground water mains, gas pipes,
telecommunication and electric power networks, as well as underground parking facilities. Other applications of
underground construction include the crossing of natural barriers such as rivers and mountains that are found
across the alignment of major road, motorway or railway link projects.
Many of these structures have to be constructed in difficult ground conditions, including soft clays and water-
bearing sands, as well as soft rocks with particular behavioral features such as creep, weathering and swelling.
Additional difficulties may arise because of the occurrence of a variety of heterogeneous ground conditions, with
strong contrasts in the characteristics of materials encountered within the same run, which may require frequent
adjustments to be made in the course of tunneling.
These projects have brought new challenges to the tunneling engineer, and have triggered many technological
and scientific advances over the past thirty years. Reviews of the geotechnical aspects of soft ground tunneling
have been provided by Peck (1969), Cording and Hansmire (1975), Clough and Schmidt (1981), Ward and
Pender (1981), O’Reilly and New (1982), Schlosser et al. (1985), Attewell et al. (1986), Konda (1987), Rankin
(1988), Uriel and Sagaseta (1989), Clough and Leca (1989), Fujita (1989), Cording (1991), Fujita (1994), Mair
and Taylor (1997) and Mair (1998). Some recent developments are also discussed in Leca and Guilloux (1999),
and will be reviewed herewith.
The present report addresses the main aspects of underground works in soil and soft rock, and reviews some
of the more recent advances in this field. The following features of soft ground tunneling are considered:
construction techniques, ground investigations, design methods, instrumentation and monitoring practices. Some
comments on major advances accomplished in recent years, as well as trends for future developments are
provided in the conclusion.


1

Eric Leca, SCETAUROUTE/DTTS, Groupe EGIS, Les Pléiades n° 35, Park Nord Annecy, 74373 Pringy Cedex, France
2
Yann Leblais, EEG SIMECSOL, Consulting Engineers, 18 rue Troyon, 92316 Sèvres Cedex, France
3
Karl Kuhnhenn, BUNG GmbH, Englerstra8e 4, Postfach 101420, 69004 Heidelberg, Germany

2.0 CONSTRUCTION ASPECTS

Soft ground tunneling is often challenging because of the occurrence of soft water-bearing soils and
environmental constraints that require strict ground motion control. Tunneling in such conditions has been made
possible due to significant technological advances that were achieved over the past twenty to thirty years. These
include the development of shield tunneling, as well as major improvements in the more conventional methods
of tunneling or in ground conditioning schemes employed in underground construction.

2.1 Shield Tunneling
Developments related to the shield tunneling technology have been reviewed in Clough (1981), Béjui and
Guilloux (1989), Fujita (1989). A technique using a concept close to shield tunneling was used for the first time
in the mid-1880s in UK, to build a pedestrian gallery underneath the Thames River. Major advances have been
accomplished since then, particularly with the introduction of shields of the pressurized type, which allow
tunnels to be constructed in all types of soils including sands under high water head. These include the slurry
shield and the Earth Pressure Balance (EPB) shield. The main features of the slurry shield technique are
presented in Figure 1 (after Fujita, 1989).

Cutter driving
motor
Agitator
Erector
motor
Tail seal
Shield jacks

Segments
Erector
Cutter face
Slurry
supply
Slurry
return


Figure 1: Principle of the slurry shield machine (after Fujita, 1989)

This technique, which makes use of a bentonite slurry to stabilize the working face of the tunnel, was
introduced in the early 1960s in UK, and in then Japan. The EPB shield was developed a decade later. The
principle used with this latter technique is described in Figure 2 (after Fujita, 1989): in this case, face support is
obtained by retaining the spoils in the working chamber so that sufficient confining pressure is reached.
Compressed air has also been used successfully in some projects to support the working face of the shield, but
this technique is essentially limited to the less pervious categories of soils.
Additional improvements have been made to the shield tunneling technique over the most recent years,
particularly in terms of machine size and ground motion control. Large Tunnel Boring Machines (TBM) are now
common, and shields with diameters up to 14 m and over have been manufactured for projects such as the Trans-
Tokyo Bay Highway (14.14 m diameter) in Japan (JSCE, 1996; Asakura and Matsuoka, 1997) and the 4
th

crossing of the Elbe River (14.20 m diameter) in Hamburg, Germany (Bielecki and Zell, 1999). A 14.87 m
diameter TBM is currently being manufactured for the construction of the Groene Hart High Speed Rail tunnel
in the Netherlands. These advances have allowed the shield technique to be extended to a larger scope of project
conditions, including motorway tunnels that currently require openings in the order of 12 m to be excavated.
Such advances have been accompanied with significant technological improvements that allow a more
appropriate management of adverse conditions to be obtained, when tunneling in difficult grounds
(Herrenknecht, 1998 & 2000). The introduction of foams in EPB shields allows a more appropriate control to be

achieved of ground deformations at the working face and, in turn, of tunneling induced settlements. Similarly,

large diameter TBMs can be equipped with a secondary internal cutting wheel to help excavate through sticky
clays. The introduction of advanced back-filling processes at the shield tailpiece has also strongly contributed to
significantly reduce the potential for tunneling induced settlements in providing a means for limiting the amount
of ground movement into the tail gap.
From a more general standpoint, several developments have been devoted to the design of “mixed-shields”
(Herrencknecht, 2000) that would be capable of handling a variety of heterogeneous materials, which are often
found in urban areas and usually result in major tunneling difficulties. Examples of such difficulties were
reported during the construction, in the mid-1980s, of one section of the Washington Metro, where large
settlements were recorded at several locations with an EPB shield. These excessive ground movements were
mainly attributed to the strong contrast in ground properties found at the face, with a mixture of soft water-
bearing sands and gravel in the crown overlying stiff to hard clays (Clough and Leca, 1993).

Cutter driving
motor
Screw
conveyor
Tail seal
Screw conveyor
driving motor
Belt conveyor
Gate jack
Erector
Shield
jacks
Bulkhead
Cutter
frame
Cutter face

Figure 2 : Principle of the Earth Pressure Balance (EPB) shield (after Fujita, 1989)

Even though the “universal machine” is yet to be invented, concepts such as the “mixed-shield” can help in
adjusting to the variety of grounds encountered along a tunnel alignment, particularly at shallow depth. An
extension of this concept was been used recently in the design of the TBM that is being manufactured for the
SOCATOP project (underground section of the second ring of Paris Beltway, in the city’s western suburb). This
11.57 m diameter machine will have the capability of being operated in an open mode or as a slurry or an EPB
shield, depending on the grounds found at the face (Carmes and Athenoux, 1998).
Construction would, however, need to be halted several days to allow modifications to be made on the
machine, which means that alternating tunneling modes would only be possible if they occurred a limited
number of times along the project. Moreover, sufficient knowledge of ground conditions should be available so
that the location of changeovers could be identified and planned ahead of time.
Advances in the shield technology have allowed significant improvements to be made in terms of ground
motion control, and tunneling induced settlements can now be kept under relatively low values (Mair and Taylor,
1997) in comparison with previous records (New and O’Reilly, 1991). Significant experience has also been
gained over the past years in shield operation know-how. As a result, and provided an appropriate machine is
selected and skillful workmanship is available, high performances should be expected for most shield tunneling
jobs, with limited impact on the environment (Richards et al., 1997).
Ground collapse may, however, be experienced - even when using the most elaborate machines - in situations
where unexpected conditions are encountered, or when face pressures fail to be maintained at the design level.
Some attempts have been made to anticipate and prevent localized face collapse through a more systematic real
time use of shield parameters recorded during tunneling (Aristaghes and Blanchet, 1998). The system, termed
CATSBY, was installed and tested on a 10.8 m diameter slurry shield used for the construction of the Sydney
metro in Australia, and lead to promising results, in extremely heterogeneous grounds - ranging from hard rock
to soft marine clays and dune sands - below the water table.

This system allows all recorded data to be either stored, or used to estimate some pre-established key
parameters that could, in turn, provide some indication of the ground-structure interactions associated with the

tunneling process. These include pressures in the muck chamber, as well as characteristics of the thrust resultant
acting on the tunnel face. Measurements are taken at regular intervals (typically every 3 minutes), and each
parameter is characterized in terms of mean value and standard deviation. These data can be used by the shield
operator to check that mean values remain within acceptable levels and that no sharp changes occur in the time
response of pre-established key parameters. The concept can be applied to a variety of project conditions, and is
designed with sufficient flexibility to allow adjustments to be made as required in the course of the project.
One particular difficulty to be emphasized with shield tunneling is the operation of the machine through the
entrance and exit shafts, as these junctions usually result in reduced confining pressures in the surrounding
ground, which could lead to critical conditions in grounds such as water-bearing soils with low cohesion. Break-
in/Break-out transition zones need to be introduced in these areas; these should serve five main purposes:
(1) Ground support in the direction perpendicular to the opening;
(2) Face reinforcement to ensure ground stability ahead of the shield, using a confining pressure limited to
the thrust reaction capacity;
(3) Ground support in the vault to limit decompression effects, so that settlements can be controlled despite
reduced confining pressures;
(4) Control of water pressures and water ingress, so that blow-in and flood can be prevented;
(5) Guidance to the TBM along the first meters of drive, to prevent sinking of the machine to occur.
Several treatment solutions have been made available to cope with these difficulties (Richards et al., 1996); in
particular, techniques based on partial ground substitution have proven to be fully efficient in soft water-bearing
grounds.

2.2 Conventional Methods
Considerable progress has also been achieved in the more conventional methods of tunneling, mostly in
relation with an extensive use of ground reinforcement and improvement techniques. Recent advances in this
area include: the development of pre-lining techniques; improvements in tunnel support systems, the
introduction of advanced ground conditioning methods, and new developments in compensation grouting.

2.2.1 Pre-lining techniques
Using a similar approach to fore-poling, several techniques have been developed over the most recent years,
that consist in placing some reinforcement over the tunnel face so as to obtain some kind of structural support at

the front prior to proceeding with ground excavation. This support can be made of an “umbrella” of peripheral
bolts or jet grouted columns, or a concreted vault. The latter is usually referred to as the “precutting” method.
This technique was first introduced during the construction of the Paris metro (Bougard et al., 1977; Péra et
Bougard, 1978) in weak rock and then extended to softer materials.
When used in soft ground, the method is completed in three stages (Figure 3): (1) excavation of a curved
shaped cut over the tunnel face, using a large excavator; (2) concreting inside the cut so as to obtain a vault,
termed “pre-vault”, ahead of the tunnel face; (3) ground excavation underneath the “pre-vault”.

Concreted Pre-vaults Cut

a. Longitudinal Section b. Cross-Section


Figure 3 : Mechanical precutting with “pre-vault”


Some face bolting may be required to help stabilize the face when large tunnel sections have to be excavated
in soft materials. This can be accomplished by means of fiberglass bolts that have the capability of offering
sufficient tensile resistance without impeding the excavation process (Lunardi, 1993). This process, which has
been used in Italy since the late 1980s, allows the excavation of large tunnel sections (over 100 square meters) to
be performed, without recurring to any partition of the face. This approach tends to be preferred to the more
conventional top-and-heading method, as it is perceived to be more efficient, both in terms of construction time
and ground motion control.
Full-face tunneling, with combined fiberglass bolting and pre-vault support, has been extensively used in
France for the excavation of large tunnel openings, since the construction of the La Galaure High Speed Train
(TGV) tunnel, which was successfully completed in molasses, with a 150 square meter profile (EMCC, 1993).
Recent experience in this field includes the construction of the Pech Brunet motorway tunnel in southwestern
France, which required the excavation of a 155 square meter profile in marls (Gaudin et al., 1999).
Combined use of pre-lining and face bolting has also been developed, during the same period of time, in
conjunction with the “umbrella” vault technique, with primary application to large transportation tunnel projects

in Italy. A typical layout for the “umbrella” technique is shown in Figure 4. It refers to the San Vitale tunnel,
which was constructed as part of the Caserta-Foggia railway line project in Italy (Lunardi, 1998). This 4.2 km
long tunnel was excavated under 150 m of ground cover, in soils consisting of sands, silty clays, clay-marls and
limestone.
The reinforcing system for this project included: 18 m long fiberglass bolts installed in the face; peripheral
bolts sealed under high pressure, to form an “umbrella” arch over the tunnel face; and ground drainage from the
face. Another important feature of this project was the introduction of a reinforced concrete invert right behind
the tunnel face, which contributed to achieve adequate ground motion control. This concept was proposed after
several unsuccessful attempts had been made to excavate the tunnel with staged excavation at the face.

a. Longitudinal Section b. Cross-Section

Figure 4 : Peripheral and Face Bolting (after Lunardi, 1998)

Design of the reinforcement system was achieved by means of a combination of experimental work
(extrusion laboratory tests and pullout tests) and numerical studies (three-dimensional Finite Element analyses).
Construction with the “umbrella” arch technique is usually accompanied with extensive tunnel instrumentation,
to check for the adequacy of bolt design. Instrumentation includes settlement markers, convergence pins, and
pressure cells to monitor the overall ground response to tunneling. In addition, borehole extensometers are
installed at the front, so that ground deformations ahead of the tunnel face can be better anticipated, and the
amount of bolting adjusted accordingly.

2.2.2 Tunnel support systems
Some progress has also been made in conventional methods, with the development of more flexible tunnel
support systems. This includes the increased use of shotcrete in “hand-mined” tunnels. A comprehensive review
on sprayed concrete liners for tunnels has been produced recently by the Institution of Civil Engineers (ICE) in
UK (ICE, 1996). The use of sprayed concrete as primary liner, particularly when reinforced with steel fibers,
allows early support to be applied to the tunnel walls (and/or face) after excavation, which contributes to
achieving reduced construction time and improved ground motion control. The amount of support can also be
Pre-vaults

Bolts

modified as required in view of the observed tunnel response, and ancillary reinforcements, consisting of radial
bolting or steel ribs can be incorporated when necessary, as excavation proceeds.
Shotcrete has been extensively used in the completion of tunnel support systems in conjunction with the New
Austrian Tunneling Method (NATM). This approach (Rabcewicz, 1964), which was originally introduced for
the construction of rock tunnels in the Alps in the 1950s, has been more recently extended to softer materials. It
is based on the principle that much of the tunnel stability comes from the self-supporting capability of the
surrounding ground, and that some optimization of tunnel liners can be achieved by continuously adjusting the
type and amount of support to that required to enhance the ground’s ability to reach equilibrium around the
opening.
The deformations of the tunnel walls are continuously monitored during construction, and adjustments made
on the basis of the observed ground response to tunneling. When used in soils and softer rock, where early
installation of support systems is necessary, this approach results in making extensive use of shotcrete with steel
fiber and/or lattice girder reinforcement, in combination with radial ground bolting when required, in view of the
observed tunnel response.
For instances such as shallow soft ground tunnels in urban areas, real time optimization of the tunnel support
system becomes hardly possible because of major concerns for preventing any potential damage to existing
structures. In such cases, shotcrete liners would be used without formally recurring to the NATM, and this
technique could be preferably be referred to as Sprayed Concrete Lining (SCL) rather than NATM, as proposed
by the ICE (1996).
Another major advantage of sprayed concrete is its ability to adjust to variable tunnel geometries, which can
save from complex form-works and contribute to more cost-effective design, particularly when large size
openings have to be built. A typical example for such application was provided by the Chauderon railway station
in Lausanne, Switzerland, which required the construction of a funnel shaped opening where railway tunnels
merged into the station. Advanced shotcrete specifications had to be used for this project, to allow high short-
term mechanical characteristics with durable strengths to be obtained (Tappy et al., 1994)
Recent improvements in shotcrete characteristics have also allowed a more extensive use to be made of this
material in tunneling, including for the long-term stability of underground structures (Leca et al., 2000). These
developments, which should eventually result in reduced steel reinforcement and improved cost-efficiency for

tunnel projects, tend to be counterbalanced by current trends to systematically recur to fully reinforced concrete
liners.
Whereas these trends are primarily governed by concerns for concrete cracking and liner water-tightness,
they probably also result from most concrete codes being mainly intended for aboveground structures. This
emphasizes the need for more exchange between geotechnical and structural engineers to be organized, so that
our standards could appropriately reflect the experience gained by practicing engineers. An attempt in that
respect has been made by the AFTES (French Tunneling Society), with the preparation of recommendations for
the use of plain concrete in tunnel liner design (Colombet et al., 1998).

2.3 Ground conditioning methods
Major progress has been achieved over the past years in the applications of ground conditioning techniques to
tunneling projects. A general review of recent advances in this field can be found in the works published by the
Soil Improvement and Geo-textiles (SIG) Committee of the American Society of Civil Engineers (ASCE, 1997a
& 1997b). Using the same classification, the criteria summarized in Table 1 can be proposed as for the potential
impact of each technique in terms of improvement in mechanical and hydrological ground properties.

Table 1 : Effects of ground conditioning schemes on ground properties
TREATMENT REINFORCEMENT IMPROVEMENT
Dewatering
➎
Bolts
➊ ➍
Compaction grouting
➋
Fracture grouting
➋ ➌
Jet grouting
➊ ➋ ➍

Freezing

➊ ➌ ➍
Micro-piles
➋ ➍

Jet grouting
➊ ➌ ➍
Pre-vault
➊

Permeation grouting
➊ ➌ ➍
Soil mixing
➊ ➍

Effect on stiffness
➊
Effect on displacement
➋

Effect on permeability
➌

Effect on strength
➍

Effect on water level
➎

Additional insight into the ranges of application of grouting products was provided by the European Standard
Committee (CEN, 1998), as reproduced in Figure 5 and Table 2.
principle
grouting
method
injection
(impregnation)
grouting
displacement
displacement
hydraulic
compaction
penetration
bulk filling
fissure/contact
grouting
bulk filling
subprinciple
fracturing
permeation
with ground
without ground

Figure 5 : Grouting principles and methods (after CEN, 1998)

Other conditioning techniques include drainage and ground freezing. Pumping, with generalized water draw-
down, tends to be avoided or limited because of the potential for consolidation settlements to take place in soft
cohesive soils. Conversely, some localized water draw-down, with drains installed at the front of open-face
advancing tunnels, is often used to help stabilize the ground and limit water inflows during construction.

Table 2 : Types of grouts applicable for grouting different types of ground (after CEN, 1998)
HOST

RANGE NON-DISPLACEMENT GROUTING DISPLACEMENT
GROUTING
MEDIUM

PERMEATION FISSURE OR
CONTACT GROUTING
BULK FILLING
Gravel, coarse sand
and sandy gravel
k> 5*10
-3
m/s
Pure cement
suspensions,
Cement based
suspensions

Granular soil Sand
5*10
-5
<k< 5*10
-3
m/s
Microfine suspensions,
Solutions
Cement based
suspensions,

Mortar
Medium to fine sand,
5*10
-6
<k< 1*10
-4
m/s
Microfine suspensions,
Solutions,
Special chemicals

Faults, cracks, karst
c > 100 mm
Cement based mortars,
Cement based
suspensions (clay filler)
Mortars,
Cement based
suspensions with short
setting time,
Expansive polyurethane,
Other water reactive
products

Fissured
rock
Cracks, fissures
0.1 mm < c < 100 mm
Cement based
suspensions,

Microfine suspensions

Microfissures
c < 0.1 mm
Microfine suspensions,
Silicate gels,
Special chemicals

Cavity Large voids Cement based mortars,
Cement based
suspensions with short
setting time,
Expansive polyurethane,
Other water reactive
products

(c = fissure width; k = ground permeability)




Gonze (1989) discussed the applications of ground freezing in underground projects. This technique is rarely
considered in practice, because of the expenses involved in its implementation in the field, but can prove reliable
and cost-effective when used appropriately. One recent application of this technique was related to the
construction of the northern section of the Lyons beltway in France. This section includes a twin tube tunnel,
with cross-passages installed at regular intervals between each tube, for safety purposes. Ground freezing was
used on this project to excavate one of the cross-passages, which had to be hand-mined in an urban environment,
through grounds consisting of mixed molasse and water-bearing pervious soils underlain with granite, with 25 m
of water head. This technique was found appropriate in view of the strong contrast in mechanical and hydraulic
properties between the two ground formations, and allowed the cross-passage to be safely executed.

Recent advances in grouting techniques have been mostly associated with the introduction of extremely fine
grained components (ultra-fine cement or mineral based chemicals with low viscosity) in injection products so
that better groutability could be achieved in finer soils. These products allow significant and durable increases in
the mechanical characteristics of grouted soils to be obtained, which was practically impossible with
conventional products.
An interesting application case of mineral based grout, of the Silacsol
TM
type, was provided by contract
D3M10 of the Paris metro extension project (Joho and Morand, 1995; Gouvenot et al., 1994). These works were
completed in a densely inhabited area, and included the construction of large span openings (15,60 m) in coarse
Seine alluvium, under 5.50 m of ground cover.
Pressuremeter and plate tests, performed on the site during the ground improvement works, allowed to
evidence a sharp increase in the mechanical properties within the grouted soil, with cohesions in excess of
200 kPa and Young’s modulus values in the order of 350 MPa (i.e. seven times higher than before grouting).
Construction could proceed safely, with no noticeable settlement at the ground surface; excavation took place in
a concrete-like material, which allowed open face tunneling to be used with a perfectly stable 5 m high front.
Significant advances have also been achieved in the field of jet grouting, with applications in both ground
improvement and seepage control. An example of extensive use of jet grouting in underground projects was
provided by the construction of two major railway stations, the Magenta (Fauvel, 1997) and Condorcet (Vignat,
1998) stations, as part of the EOLE subsurface rapid rail transit system in Paris. Each station comprised three
vaults, with an overall span of 53 m, and were excavated in the central part of Paris with limited ground cover.
Heterogeneous ground conditions were present on both sites, with fill and alluvium in the crown and sands or
limestone in the invert, underlain with fine water-bearing sands.
The project included the construction of four pillar and side galleries, which were used to stabilize the ground
by means of a network of vertical and inclined jet grouted columns, prior to proceeding with the excavation of
the three main openings. The design and construction procedure associated with the completion of the jet
grouting works was adjusted on the basis of three real scale in situ tests (Fredet and Leblais, 1997). These
confirmed that a significant increase in cohesion and furthermore ground modulus would be achieved through
the jet grouting process, with improved ground characteristics typically 2 to 5 times higher than initially
measured. This ground conditioning work was essential in achieving adequate surface settlement control during

construction.
Ground conditioning schemes may also be used to assist in the completion of shield driven tunnels in difficult
ground conditions. Campo et al. (1997) reported on various grouting and jet grouting works being completed in
fine water-bearing soils, during the construction of Line 2 of the Cairo metro, with a slurry shield. This case
history emphasizes the requirement for a thorough examination of all specific situations that may arise along the
completion of a tunneling project, even when the most sophisticated fully mechanized techniques are used.
From a more general standpoint, grouting and jet grouting schemes can be appropriately used, as remedial or
preventive techniques, where weaker materials are found along the tunnel alignment. Such applications include
occurrences of weak water-bearing grounds in rock tunneling projects. An example of ground conditioning
works associated with rock tunneling was provided by the Freudenstein tunnel, which was constructed in
Gypsum Keuper, as part of the Mannheim-Stuttgart railway line project in Germany (Kirschke et al., 1991b).
Gypsum Keuper, as found in Baden-Württemberg, Germany, is composed of two layers with distinct rock
attributes, separated by the gypsum upper limit. The rock is leached above the gypsum level, because of its
sulfatic components being dissolved and washed away through progressive decomposition. It tends to be split
into jointed masses and partially disassembled, and usually looses its original competence. Some overstressing
and fracturing is also produced in the overlying grounds, as a result of stress rearrangements associated with this
process.

The “active leaching” zone is subject to high water pressures, and shows no or short term stability when
exposed to excavation works. Conversely, the underlying gypsum rock can be described as compact and nearly
watertight. The main part of the eastern drive of the Freudenstein tunnel had to pass through the leached gray
Estherien layers, with gypsum level only a few meters below the tunnel invert. Along a 450 m stretch of the
tunnel, a 1-3 m thick layer of weak water-bearing ground was present at various levels at the face (from crown to
invert). Extensive grouting works were used to cut through this area.
Grouting was performed using a pilot adit, and targeted so as to form a sealed zone in the tunnel area
(Figure 6). Some 38000 m of grout-holes were drilled and more than 2000 tons of cement were injected. The
procedure allowed the tunnel to be successfully excavated. Water inflows were reduced and durably controlled.
Nevertheless, water-flows in the order of 60 l/s had to be pumped during construction, along the 2 km long
stretch that linked the treated zone to the closest portal.

Figure 6 : Freudenstein tunnel - Grouting works completed in unleached gypsum

Additional difficulties were found on this project with the excavation of an intermediate shaft, which had to
be introduced for ventilation purposes during construction. Because anhydrite was present in this area, with the
potential for swelling to occur in this formation, if exposed to water, the shaft had to perfectly sealed. The sealing
works were successfully completed using jet grouting columns driven from the bottom of the advancing shaft
excavation.

2.3.1 Compensation grouting
Advances have also been made in the field of ground improvement applied to tunneling, and particularly with
the compensation grouting approach (Mair and Hight, 1994). This technique was introduced in the early 1980s,
in the form of compaction grouting, to assist in controlling tunneling induced settlements in dense sands (Baker
at al., 1983). The same principle was used more recently for shallow tunnels excavated underneath sensitive
buildings, as part of the construction of the Vienna metro (Pototschnik, 1992) and the Jubilee Line Extension in
London (Harris et al., 1996). In both cases, tunneling was completed in clays, using fracture grouting with fluid
grout to limit the impact of settlements on existing structures. Extensive monitoring was used to adjust the
amount of grouting to observed ground deformations.
The main construction features are illustrated in Figure 7, which refers to the construction of a 10 m diameter
shallow tunnel, underneath a masonry building in London (Osborne et al., 1997; Mair, 1998). The tunneling
works were performed in a layer of London clay, overlain with water-bearing gravel. Prior to construction, a
shaft was excavated next to the building, and used to install a network of horizontal pipes within the clay layer,
at some depth between the building foundations and the tunnel crown. The building was equipped with shallow
and deep settlement markers to allow ground movements to be monitored during each construction stage.
Construction involved two excavation steps: a 5.75 m diameter pilot gallery was first excavated and lined, and
then enlarged to 10 m.
Because of the large size of the opening in comparison to its depth, it was anticipated that large settlements
could take place during the enlargement stage. As a result, settlements were continuously monitored during

construction, and grouting activated through the already installed pipes, so as to counteract observed movements.
Grouting was performed using the “tube-à-manchette”, technique to allow accurate grout placement above the

tunnel crown to be achieved. This procedure allowed the tunneling works to proceed successfully, with deep
settlements being kept lower than 20 mm, whereas up to 90 mm of vertical displacements had been recorded
above the tunnel crown.

Figure 7 : Compensation grouting scheme (after Obsborne et al., 1997)

This overview of ground conditioning cases allows some appreciation to be made of the broad scope of
application of these techniques in underground construction. Based on observed practices, the following points
should be emphasized in view of achieving appropriate design specifications, as well as satisfactory field
performances:
(1) Obtain sufficient knowledge of the geotechnical and environmental conditions, as well as of the
accessibility of ground improvement equipment to the site;
(2) Identify correctly the ultimate purpose of the conditioning works: mechanical improvement and/or water-
tightness;
(3) Account for local technical practices and capabilities, as the success in such work largely depends on the
contractor’s skills;
(4) Adjust design to local practices and establish the cost/time schedule accordingly;
(5) Adapt the contract type in view of the actual ground conditioning purposes.

3.0 GEOLOGICAL AND GEOTECHNICAL INVESTIGATIONS

Geological and geotechnical investigations provide early information on the tunnel feasibility and on the
ground characteristics to be used for design. Some general guidelines for planning and performing geotechnical
investigations for tunneling projects were presented by Parker (1999), who emphasized the need for increased
geotechnical investigations at the early stage of a project, in view of the amount of construction cost overrun
attributed to unexpected ground conditions.
Practical guidelines for the preparation of geotechnical reports for contract documents related to underground
construction projects were produced by the Underground Technology Research Council of the ASCE (1997c).
Additional insights into the parameters to be collected at the different stages of a tunneling project were provided
by the AFTES (Guillaume et al., 1994 & 1999). The information on the ground conditions to be expected along

a tunneling project can also be improved by conducting additional investigations in the course of construction.
Modern technologies have allowed advances to be made both in terms of processing the information
collected during the different stages of investigations and of the capability for performing specific ground
investigations during construction. Some of these advances are commented in the present section. They mainly
refer to the developments of geological models and the increased use of geophysical methods in underground
projects.

Deep settlement
indicators
Shaft
Tubes à manchette
Thames gravel
London clay
Tunnel
Building

3.1 Geological Models
The potential use of Geotechnical Information Systems for tunneling projects was discussed by
Maurenbrecher and Herbscheb (1994), who presented a model developed in the late 1980s in Amsterdam,
Netherlands, to store the information retrieved from geotechnical investigations completed in a district of the
city. A database, termed INGEOBASE, was designed to store data from borings, static penetrometer tests and
hydro-geological investigations performed over a 4 square kilometer area. Computerized data storage of
available geotechnical information allowed some mapping to be performed of this area, according to specific
pre-established geotechnical parameters.
The authors mentioned the potential use of this model for a planned metro rail project in the same district of
Amsterdam. Such device should help identify geological difficulties along a tunneling project, and provide early
information for performing risk analyses and making decisions in terms of project relocation or ancillary
construction measures to be introduced to cope with any major anticipated difficulty.
In a similar effort, Gaudin et al. (1997) commented on the advances allowed in geotechnical investigations by
three-dimensional geological modeling, on the basis of the experience gained over the past years, with the

software, EARTH-VISION. The primary purpose of this software is to allow all data obtained from geological
and geotechnical investigations to be collected and stored within the same electronic base. These include surface
geology, photo-geology, thermographical data, numerical earth models, satellite views, boring logs, geophysical
records. In a second stage of analysis, this information is processed to prepare a three-dimensional geological
model of the site.
One essential feature of this system is its capability to be updated in view of additional data that are collected
in the course of the project. The model can be used to prepare different graphical representations of the expected
geology, check for the consistency of all available data, and identify areas where additional investigations should
be conducted. A typical three-dimensional output obtained from this model is shown in Figure 8 (after Gaudin et
al., 1997). It refers to one section of the Hurtières motorway tunnel, in the French Alps, where a collapse was
experienced during construction.


Figure 8 : Three-dimensional modeling of the glacial gorge found on the Hurtières tunnel (Gaudin et al., 1997)

The tunnel consists of two tubes, and was excavated using conventional methods in rock. Failure occurred at
one particular location where a glacial gorge, filled with weak water-bearing materials was encountered. This

geological accident, which had not been identified at the design stage, caused construction to be halted several
months, to allow additional investigations to be undertaken and remedial action to be taken (Hingant, 1999).
The three-dimensional output presented in Figure 8 was obtained by combining data collected from 63
investigation borings and 104 additional bore-holes, that were completed as part of the grouting remedial works.
This model allowed a more accurate evaluation to be made of the extent of encountered weak zone, and helped
in determining the ancillary works that had to be performed before construction could resume. Additional
experience of three-dimensional geological modeling was reported by Houlding (1995), with reference to works
planned in heterogeneous variable grounds, as part of the Taïpei metro extension project.
These few examples of recent developments related to ground modeling illustrate some of the potentials
offered by these techniques, which should probably be used on a more regular basis to assist in planning and
conducting tunneling projects in the future.
One major difficulty in tunnel construction relates to the length of the projects, which requires that geological

conditions be interpolated on the basis of relatively limited information. Improvements could be expected in this
respect by both increasing the amount of geotechnical investigations and making the best possible use of all
available information. The development of decision aid systems, based on the statistical evaluation of geological
data (Sinfield and Einstein, 1996) could assist the engineer in achieving a more accurate appreciation of
construction hazards related to geological uncertainties.
A software, termed DAT (Decision Aids for Tunneling), was developed by Einstein et al. (1992) to provide
some evaluation of the implications, in terms of tunnel construction cost and duration, of geological hazards.
This model makes use of probabilistic methods to account for geological and construction uncertainties in the
simulation of some anticipated project characteristics and construction sequences. It can help identify additional
investigations that would be required to reduce cost uncertainties, and can be updated in the course of the project,
and the cost and duration estimates adjusted accordingly.
This model has been used for the St-Gothard and Lötschberg major railway tunnel projects in Switzerland, to
assist in the cost and duration evaluation process. One contribution of the model was to demonstrate that a
drastic reduction in the cost uncertainties related to the construction of the St-Gothard tunnel project could be
anticipated, after a more accurate knowledge of the geology found in the difficult Piora area had been
established. The model also allowed to confirm, on the basis of all available information obtained after the
required complementary investigations had been completed, that cost uncertainties would remain relatively low
(in the order of ± 7%) for both projects and that the error on construction duration estimates would be less than
one year (Dudt and Descoeudres, 1998).

3.2 Geophysical Investigations
Improved knowledge of geological conditions can also be obtained from the combined use of geotechnical
and geophysical means of investigations. Significant progress has been made over the past years in the
applications of the latter approach to tunneling projects, both at the design stage and during construction.
Corbetta and Lantier (1994) reported on an experimental use of the “electrical cylinder” method, as part of the
geological investigations completed for the construction of a water main in Bordeaux, France. This tunnel had to
be excavated with an EPB shield in karstic grounds.
The “electrical cylinder” technique consists in measuring the ground resistivity from a pre-drilled borehole. It
was used on this site to assist in locating existing voids or pockets of weaker materials within the ground before
proceeding with excavation. The method is designed to identify contrasts in ground properties in a 5 m deep area

around the borehole, and can be implemented from the soil surface or through sub-horizontal boreholes driven
from the front during construction.
Other similar attempts have been made more recently to use different geophysical techniques at the face of
advancing tunnels, including seismic investigation techniques (Bousquet-Jacq and de Sloovere, 1999) and
geological radar surveying. In particular, an extensive radar surveying program was performed during the
construction of the EOLE urban train tunnel in Paris, using boreholes driven through the face of a slurry shield
(Pierson d’Autrey et al., 1995). This tunnel was excavated in heterogeneous materials, and data obtained from
the radar investigations used to identify areas of weak grounds where ancillary action had to be taken.
These experiments were completed as part of a nationwide research program, termed EUPALINOS-2000,
which aimed at improving technologies for managing shield tunneling in heterogeneous grounds. The feasibility
of conducting radar investigations through the face of the machine was clearly demonstrated, but questions
remained as to the possibility of extending the method to various ground conditions and the time efficiency of
the process, as construction would need to be halted for the tests to be performed.

Further work is underway to analyze potential improvements in that sense, and investigate the benefits of
combining different geophysical methods at the tunnel face. Another concept, consisting in using vibrations
emitted by the TBM during construction, was also tested, as part of this research program and lead to promising
results. Such approach could contribute to reducing investigation times, as it would require no construction
stoppages to be performed.
An experiment using a similar principle was completed during the construction of the second Heinenoord
tunnel, south of Rotterdam, Netherlands Swinnen et al. (1999). This tunnel consists of a double tube excavated
with a TBM, at shallow depth in soft soils. The experiment was intended to analyze the possible use of shear
waves emitted by the TBM to improve the knowledge of ground conditions to be found at the face. It was
demonstrated that the source for the shear waves would be located at the hydraulic jacks used to advance the
shield, and that these waves could be recorded at the ground surface. A second experiment was carried out after
completion of the first tube, to check for possible uses of shear waves emitted by the TBM in ground
investigation.
Geophones were installed along the liner of the second tube under construction, in an attempt to record shear
waves emitted during the drive that would be reflected by the already built tube. The experiment showed clear
evidence of such reflections, which tended to confirm the potential use of vibrations induced by the TBM to

better check for anomalies in the surrounding ground. Further work would however be required before this
technique could be introduced for operational use. These should primarily aim at improving the characterization
of the uncontrolled source waves, and locating receivers in such a way that geological conditions could be
investigated in the ground ahead of the tunnel.
Another approach based on the analysis of seismic signals was proposed by Neil et al. (1999) who developed
a three-dimensional ground mapping system, termed ROCKVISION3D, for detecting ground anomalies (voids,
hard inclusions,…). The system makes use of seismic tomographies to identify anomalies or geological
boundaries that would be expected to act as reflectors to the seismic waves. Data recorded during the
investigations are processed to generate a ground model; this allows comparisons to be made between the
modeled seismic response and the recorded travel times. The model parameters (wave velocity, distance to
anomaly) are then varied until the predicted response matches that recorded in the field.
The authors discussed the potential application of this technique for investigating the ground ahead of TBM
driven tunnels in hard rock (Figure 9). In that case the disk cutters would be used as seismic sources and arrays
of receivers would be installed at the TBM tailpiece to record the source signals and reflected waves. This
system would have the capability of producing real time mapping of the ground ahead of the tunnel face.
Questions would however seem to remain as to the required time for installing the receivers behind the TBM and
the generalized use of the system for the variety of ground conditions that could be encountered when tunneling
in difficult ground.

TopRock drillhole
Probe – 2-in dia
Disc cutter
Direct waves generated
By disc cutter
Geological
anomaly


Figure 9 : Ground mapping ahead of a TBM driven tunnel in hard rock (after Neil et al., 1999)

Seismic techniques, with calibrated signals, have also been developed to investigate the ground at the tunnel
face, using sources installed along the tunnel walls (Sattel et al., 1996) or within the head of the TBM. The latter
was implemented on the shield designed for the 4
th
Elbe crossing in Hamburg, with a theoretical investigation
range of 50 m ahead of the tunnel face (Herrencknecht, 2000).
Other tunneling applications of geophysical methods include the increased use of these techniques as part of
preliminary investigations. New fields of application could also be envisaged in this area, in association with
recent developments in directional drilling technologies. Directional drilling techniques, which were originally
issued from the petroleum industry, have been increasingly used over the past ten years for trenchless pipe
installations and could contribute to the development of geological investigation methodologies for tunneling
(Mermet et al., 1997).
This is of particular relevance for long mountain tunnels, with limited access to the site, and where long
vertical boreholes would otherwise be required to investigate existing ground conditions at the planned tunnel
level. Such investigations are currently underway, as part of the preliminary studies of the Lyons-Turin Rail
Link, between France and Italy, in the Alps.

4.0 DESIGN METHODS

In his state-of-the-art report on deep excavation and tunneling in soft ground, presented at the 7
th
International
Conference on Soil Mechanics and Foundation Engineering, Peck (1969) introduced three main issues to be
addressed for the design of soft ground tunnels:
- stability of the opening during construction, with particular attention to tunnel face stability;
- evaluation of the ground movements induced by tunneling and of the incidence of shallow underground
works on surface settlements;
- design of the tunnel liner system to be installed to ensure the short and long term stability of the structure.
These three aspects are reviewed in the following sections, with reference to the concepts developed by

Peck (1969) and to theoretical and practical contributions published over the past thirty years in this area.
Design principles for bored tunnels -with particular attention to urban tunneling - were recently discussed in a
comprehensive review by Mair and Taylor (1997), which should be referred to for detailed background on some
of the issues presented hereafter. Additional insights into recent developments related to modeling and prediction
techniques are also provided in papers presented at the International Symposium on the Geotechnical Aspects of
Underground Construction in Soft Ground (Leca, 1996).

4.1 Tunnel Face Stability
Several approaches have been developed, over the past thirty years, to analyze the face stability of tunnels
constructed in soft ground. Most methods refer to the case of a circular tunnel constructed in homogeneous soil
(Figure 10). The main parameters involved in the face stability are:
- the tunnel diameter, D and depth to axis, H (or depth of cover C = H-D/2);
- the soil unit weight, γ and pressure of overburden (or existing building), σ
S
;
- the support pressure to be applied at the tunnel face (if required), σ
T
;
- the soil shear strength.
The shear strength can be characterized by the soil cohesion, c’ and friction angle, ϕ’ or, for tunnels in clays
and other non-pervious soils, the undrained shear strength, s
u
.

Figure 10 : Model for tunnel face stability analysis
H
C
D
σ

S
γ
σ


T

4.1.1 Undrained Stability in Cohesive Ground
Early works on face stability were concerned with tunnels constructed in clays (Broms and Bennermark,
1967; Peck, 1969). They allowed a stability criterion to be established, on the basis of the consideration of an
overload factor, N,

u
TS
s
H
N
σ
γ
σ
−+
=

(1)

with instability occurring for values of the overload factor in excess of 5 to 7. Studies completed in the 1970s
and 1980s at Cambridge University, in UK allowed a more accurate view of the influence of some parameters on
tunnel face stability to be obtained (Mair, 1979; Schofield, 1980; Davis et al., 1980). These works showed that
the limiting value, N* of the overload factor would be dependent on the depth of cover to diameter ratio, C/D,
with the criterion N* = 5-7 being obtained for C/D values in the order of 1.5.

This result is illustrated in Figure 11 (after Mair, 1979; Kimura and Mair, 1981) where design curves for N*,
derived from centrifuge model tests, are plotted as a function of the depth of cover to diameter ratio, C/D. This
Figure also introduces the influence of the length of unlined tunnel, P behind the face, often found when
tunneling with conventional methods. The plots obtained for different values of the ratio, P/D provide an
indication on how face stability can deteriorate when long stretches of tunnel are left unlined behind the heading.

Figure 11 : Critical overload factor, N* versus depth ratio, C/D (after Mair, 1979; Kimura and Mair, 1981)

Works completed at Cambridge University also allowed to evidence the influence of the ratio,
γ
D/s
u
, which
should be considered for characterizing the local stability of large diameter tunnels. These conclusions were
mainly derived from the results of model tests completed on the Cambridge centrifuge, as well as theoretical
work based on the yield analysis principles (Salençon, 1990).

4.1.2 Stability in Sands and Other Frictional Materials
A similar approach was used more recently to analyze the case of tunnels excavated in sandy grounds, which
had gained interest because of the number of tunnels constructed with pressurized shields in pervious water-
bearing soils. A method was proposed by Leca and Dormieux (1990) to estimate the support pressure,
σ
T
to be
applied at the tunnel face, using a three dimensional failure mechanism, as described in Figure 12 This
mechanism involved the rigid body motion of two conical blocks (labeled ① and ②, with velocities V
1
and V
2
in

Figure 12). For dry cohesionless soils, the limiting support pressure,
σ
T
*, derived from this mechanism, can be
expressed as,

D
SST
γ
ασασ
γ
*
+=

(2)

where
α
S
and
α
γ
are weighing factors which depend on the soil friction angle,
ϕ
’ and the depth of cover to tunnel
diameter ration, C/D, and can be obtained from pre-established charts. A more general expression, using the
same weighing factors,
α
S
and

α
γ
, was also proposed for the case of cohesive-frictional soils.
This approach was checked against model tests performed with Fontainebleau sand in the centrifuge of the
Laboratoire Central des Ponts et Chaussées (LCPC) in Nantes, France (Chambon and Corté, 1994) and lead to
values of the limiting support pressure in close agreement with experimental data. Recent developments allowed
this method to be extended to tunnels excavated in waterbearing soils (Leca et al., 1997). In that case, water
0
2
4
6
8
10
1 1.5 2 2.5 3 3.5
C/D
P/D oo
P/D = 2.0
P/D = 1.0
P/D = 0.5
P/D = 0

effects are introduced as additional loads, which magnitudes are derived from Finite Element seepage analyses
(Atwa and Leca, 1994).


Figure 12 : Failure Mechanism considered by Leca and Dormieux (1990)

Similarly to the approach proposed by Davis et al. (1980) for clays, this method presents the advantage of
taking full account of the main factorrs involved in the front stability, and more importantly of the three-
dimensional geometry found at the tunnel face. This latter parameter should be viewed as an essential feature of

the evaluation of the ground response at the tunnel face, as demonstrated by several analytical as well as
numerical works (Davis et al., 1980; Chaffois, 1985). The same requirement applies to seepage analyses: three-
dimensional numerical computations of seepage towards the tunnel face indicate a high concentration of
hydraulic gradients around the tunnel face (Figure 13) which would not be properly evaluated with a two-
dimensional model (Leca et al., 1993).
Figure 13 : Hydraulic gradients computed at the tunnel face

Another approach, taking full account of the three-dimensional geometry at the tunnel face, was proposed by
Anagnostou and Kovári (1996a), using limit equilibrium principles. This latter method is based on the
consideration of a failure mechanism initially introduced by Horn (1961), and consisting of the rigid body
motion of prismatic blocks (Figure 14). The stability analysis is run in effective stresses, with due account of
water loads, and provides an estimate of the support pressure to be used at the face of a shield driven tunnel. One
key computational parameter is the magnitude of horizontal stresses acting upon the boundaries of the prismatic
blocks, which needs to be empirically established.
Both slurry and EPB shields can be considered, and charts are provided to allow for parameters such as head
losses within the excavation chamber or slurry penetration into the face, to be accounted for. A comparison was
made between this method and that proposed by Leca and Dormieux (1990) on the basis of centrifuge model
tests published by Chambon and Corté (1994), and lead to similar estimates of the limiting support pressure σ
T
*
(Anagnostou and Kovári, 1996b) for conditions used in the tests.
One difficulty, when using most existing approaches for evaluating the stability of the tunnel face, is that they
were primarily developed for tunnels excavated in homogeneous materials. Care must be taken when analyzing
heterogeneous conditions, as often found in the construction of shallow tunnels in soft grounds.
H
C
D
σ

S

γ
σ

T
1
2
V
1
V
2

An interesting case study in this respect was provided by the construction of section 1 of the Lille metro in
France (Leblais et al., 1996). This tunnel was excavated using a 7.65 m diameter EPB shield, in Flandres clay,
with a depth of cover in the range 8.7-21.8 m. At one location along the tunnel alignment, a face collapse
occurred and a sinkhole, with an average diameter of 8 m, formed at the ground surface.

A
F
K
G
D
C
J
N
M
L
H
E
B
ω

Figure 14 : Three-dimensional failure mechanism considered by Anagnostou and Kovári (1996a)

Figure 15 illustrates the soil profile found at this location, which consists of 5.8 m of silty and sandy loess, 4.8
m of clayey and sandy silts, and Flandres clay. The upper one meter of the Flandres clay layer is weathered. The
Flandres clay is a stiff clay, with an undrained shear strength at this location in the order of 130 kPa. When
failure occurred, the shield was operated with a face pressure of 200 kPa, which means that the overload factor
could be estimated to be in the order of N=2.5, i.e. significantly lower than values classically associated with
failure (N<5-7).

H =16.30 m
L = 8.50 m
Silty-sandy Loess
Clayey-sandy Silts
Weathered Clay
Flandres Clay
Shield
Theoritical failure mechanism


Figure 15 : Geotechnical profile found on section 1 of the Lille metro (after Leblais et al., 1996)

Table 3 summarizes the results of stability analyses completed in an attempt to explain the observed tunnel
face collapse: the limiting overload factor, N* was estimated using Davis et al. (1980) approach, with three
different assumptions for the clay cover, C over the tunnel crown. The first analysis assumed that the tunnel had
been excavated in a homogeneous layer of clay (C/D=1.6). The two latter cases took account of the actual extent
of the clay layer, with (C/D=0.10) and without (C/D= 0.25) allowance for some weathering in the upper one
meter. In both cases the ground above the clay layer was conservatively assumed to act as a surcharge.

Table 3 : Limiting overload factor estimates (after Leblais et al., 1996)
Assumed ground conditions C/D N*
Homogeneous clay layer 1.60 6.0
Clay layer limited to the extent of the Flandres clay 0.25 2.8
Clay layer limited to the extent of the unweathered Flandres clay 0.10 2.2

The results shown in Table 3 indicate that no tunnel face collapse could have been anticipated with the
assumption that the tunnel would be driven in a homogeneous layer of clay. Conversely the consideration of the
actual extent of the clay layer lead to estimates of the limiting overload factor in the same order of that computed
where failure occurred. It is also worth mentioning that the extent of the sinkhole observed at the ground surface
was similar to that derived from the theoretical failure mechanism as proposed by Davis et al. (1980).
Another difficulty, when using existing methods for tunnel face stability analyses relates to the restrictive
assumptions they imply, as these essentially apply to circular shield driven tunnels, where stability is achieved by
means of the supporting action a pressure, σ
T
applied at the front. Such assumptions were found appropriate in
view of the number of tunnels excavated in weak grounds with pressurized shields.
With recent advances in conventional tunneling, large size tunnels can now be excavated full-face, in soft
grounds. In such cases, face support is not provided by a fluid or earth pressure but through the action of a
reinforcing system, which primarily consists of fiberglass bolts. Such configuration is not accounted for by
design methods developed for shield driven tunnels. Some attempts have been made to extend the use of existing
solutions for shields to the face stability of tunnels excavated by means of conventional methods, with bolt
support. These usually introduce an equivalent support pressure, or equivalent soil cohesion (Grasso et al., 1991),
to account for the supporting action of the bolts.
More recently, the approach proposed by Leca and Dormieux (1990) was extended to the case of
conventional tunnels constructed in soft ground, with bolt reinforcement at the face (Leca et al., 1997). These
developments also included the supporting action of a “pre-vault”, as used with the precutting method
(Figure 16). The “pre-vault” was modeled as a rigid boundary that restrained the possibility for failure to develop
above the tunnel face. Theoretical estimates derived from this approach showed that the “pre-vault” would have

little impact on the face stability, and this was confirmed by model tests performed on the LCPC centrifuge
(Skiker, 1995). The tests also allowed to evidence the restraining effect of the “pre-vault” with respect to ground
deformations around the opening (Skiker et al., 1994).

γ
V
ϕ
'

Figure 16 : Failure Mechanism with a “pre-vault” associated with face bolting

Typical failure mechanisms observed in centrifuge tests, with and without a “pre-vault”, are illustrated in
Figure 17. They show that, with no “pre-vault”, the failed area would tend to propagate towards the soil surface,
whereas it was restricted to the ground at the front where a “pre-vault” of sufficient length was used. As a result,
it was concluded that the main benefit of the “pre-vault” would be to help control ground movements around the
face and, in turn, tunneling induced settlements.
The bolting effect can be introduced in the stability analysis through additional loads provided by bolts
intersected by the failure envelope (Leca et al., 1997). The contribution of face bolting to the stability of tunnel
headings is well evidenced by the model. Model tests performed in the LCPC centrifuge also allowed some
experimental quantification of the bolting effect to be obtained (Al-Hallak, 1999).

The tests were analyzed by means of a three-dimensional Finite Element model, using the computer code
CESAR-LCPC, and this approach was found to provide a reasonable representation of experimental results (El-
Hallak, 1999). In this study, the individual action of every singular bolt was considered separately and allowed
for in the model. Another approach for introducing the bolting effect in Finite Element analyses consists in
replacing the bolted ground by an equivalent homogeneous material. Promising results have been obtained in
applying this technique to the analysis of bolt reinforced tunnel sections (Greuell, 1993; Greuell et al., 1994).

Figure 17 : Failure mechanisms observed in centrifuge model tests

Advances in computer systems have allowed complex conditions to be accurately modeled by means of
numerical methods, thus providing new approaches for characterizing the stability of shallow tunnels constructed
in soft ground. Sloan and Assadi (1997) developed a Finite Element model for the evaluation of the two-
dimensional stability of tunnel sections excavated in clays. One interesting feature of this approach was to
introduce the influence of an increase in undrained shear strength with depth, which can usually not be
accounted for with analytical solutions.
A similar approach was used by Antao (1997), on the basis of works performed over the past twenty-five
years at the LCPC in France (Friaâ, 1978; Frémond and Friaâ, 1979; Guennouni and Le Tallec, 1982). It is based
on cinematic solutions derived from visco-plastic Finite Element analyses of the ground response to tunneling.
This numerical method was introduced by Jiang (1992), in a subroutine, termed LIMI, of the Finite Element
program CESAR-LCPC; it was primarily intended to provide solutions for the stability of earth structures, and
then extended to the analysis of shallow tunnels constructed in soft ground (Antao et al., 1997).
The model allows failure mechanisms to be automatically generated through an optimizing process of
numerically obtained cinematic solutions. It was applied to the stability of shallow tunnels and provided results
in close agreement with published analytical solutions for both the two-dimensional tunnel section and three-
dimensional tunnel heading cases (Antao, 1997).
Unlike analytical methods, this approach does not require any pre-established failure mechanism to be
considered. It also presents the advantage of allowing a more accurate representation to be made of the actual
geometry and geotechnical conditions found on the site. Improvements would however be needed in terms of
required computer time, to allow a more general use of such tools. Work is currently underway at the LCPC to
increase the numerical efficiency of the model and extend its application to a broader range of geotechnical
conditions.

4.2 Evaluation of Tunneling Induced Ground Movements
4.2.1 Surface Settlements
The construction of a tunnel produces some deformation of the surrounding ground, which could result in
surface settlements when the excavation is performed at shallow depth. In that case, a settlement trough is
formed at the ground surface (Attewell et al., 1986), and tends to propagate together with the advance of the
tunnel heading (Figure 18).
Based on observations made on several tunneling projects, Peck (1969) proposed that the settlement trough

produced at the ground surface could be characterized by means of a reversed error function curve as shown in
Figure 19.

With this assumption, the surface settlement, s(x) observed at a distance, x from the tunnel center-plane can
be computed using the following mathematical expression:

()








−=
2
2
max
2
exp.
i
x
sxs


(3)

where s
max

is the maximum surface settlement produced above the tunnel crown, and i the distance to centerplane
of the inflexion point in the reversed error function curve.

Extent of surface
settlement trough
S
max
z
y
x
H

Figure 18 : Three-dimensional distribution of tunneling induced settlements (after Attewell et al., 1986)

Equation (3) allows an expression of the volume of settlement trough, V
s
per unit length of tunnel to be
obtained:
max
2
siV
s
π
=


(4)

Surface Settlement, s
Distance to tunnel center-line, x

i
Smax
x

Figure 19 : Surface Settlement Trough (after Peck, 1969)


As a result, the settlement trough induced by tunneling can be entirely characterized by means of two
parameters:
- the volume of settlement per unit length of tunnel, Vs and
- the distance to centerplane of the settlement curve’s inflexion point, i.
Based on these considerations, semi-empirical methods were proposed by several authors to evaluate the
magnitude and distribution of tunneling induced settlements; these include works by Peck (1969), Cording and
Hansmire (1975), Clough and Schmidt (1981), Fujita (1981), O’Reilly and New (1982), Attewell et al. (1986),
Rankin (1988), Uriel and Sagaseta (1989).
The distribution parameter, i is mainly dependent on the depth to tunnel axis, H and nature of ground
conditions. A recent review of existing correlations by Mair and Taylor (1997) concluded that this parameter
could be reasonably estimated, using the following expression:

HKi
.
=


(5)

with the coefficient, K being a function of the ground type. The settlement trough tends to be of broader extent in
clays than in sands, with K values typically in the range 0.4-0.6 for tunnels in clays and 0.25-0.45 in sands. It
must be emphasized that these estimates were derived from observations made in mostly homogeneous grounds,
which means that some adjustments would be required when estimating tunneling induced settlements in

heterogeneous soils (Mair and Taylor, 1997).
The volume of settlement trough, V
s
is generally more difficult to evaluate, as this parameter is highly
dependent on construction methods as well as workmanship. This parameter is usually compared to the volume
of ground loss produced at tunnel level, and expressed as a percentage, V
l
of the theoretical volume of excavated
ground. For a circular tunnel of diameter, D, this leads to:

4
.
.
2
D
VV
ls
π
=


(6)

with V
l
being expressed in percent.
Sources for ground loss were analyzed by Cording and Hansmire (1981) for the shield driven tunnel case,
which can be considered as the most complex as far as the ground response to tunneling is concerned.
Four major contributions to ground loss can usually be identified (Figure 20):
(1) face intake due to stress relief associated with ground excavation;

(2) displacements along the shield: these included deformations induced by shear stresses along the shield,
and inward displacements due to deviations of the machine, as well as actions taken to ease shield
advance (overcutting, conical shaped skin);
(3) ground movement into the tail gap (that forms as a result of the shield outer diameter being larger than
the lined tunnel section);
(4) liner deformations (usually of limited magnitude for reinforced concrete segments).
Depending on the characteristics of the soils at and over tunnel level, ground losses could propagate fully or
partly towards the ground surface.
Tunneling in dense dilating sands usually results in surface settlements, which are lower than the cumulated
volume of ground losses produced at tunnel level. Conversely surface settlements in loose or compressible soils
could lead to larger amounts of settlements at the ground surface. The immediate ground response to tunneling in
saturated clays can be assumed to take place with no volume change, in which case ground losses produced at
tunnel level tend to result in an equal amount of settlement at the ground surface. Additional settlements could
take place in the longer run with cohesive soils, because of consolidation effects associated with pore water
pressure changes induced around the tunnel during construction.
Some correlations have been proposed (Clough and Schmidt, 1981; Attewell and Yeates, 1984) to provide an
order of magnitude of the amount of tunneling induced settlements. Based on a semi-empirical approach, Clough
and Schmidt (1981) established that the volume of settlement trough for tunnels in clays should be related to the
overload factor N, with values typically in the range 0-1% of the excavated volume, for overload factors lower
than 2, and up to 10% of the excavated volume, for overload factors in the range 2-4. Other correlations were
proposed for tunnels in sands by Attewell and Yeates (1984).

Such correlations can provide an indication of the maximum amount of settlement that could be induced by
tunneling works, but do not account for the potential for failure to take place due to accidental losses of face
support during construction. Conversely, advances in the tunneling industry, particularly in the field of shield
tunneling, have allowed significant improvements to be made in terms of ground motion control, which should
allow settlements to be kept under significantly lower levels than predicted by existing semi-empirical estimates.

Face
intake

Movement
into the tail gap
Surface settlements
Displacements
along the shield
Shield
Liner segments


Figure 20 : Contribution to ground losses around a shield driven tunnel

A review by Mair (1996) of recent case histories lead to the following conclusions (Mair and Taylor, 1997):
(1) open face tunneling in stiff clays usually results in V
l
values in the range 1-2 %;
(2) closed face (EPP or slurry) shields allow settlements to be kept under relatively low levels, with V
l
being
usually lower than 0.5 % in sands and in the order of 1 to 2 % in soft clays;
(3) larger amounts of settlements could result from the use of shields, including of the pressurized EPB or
slurry type, in mixed face conditions;
(4) conventional tunneling with sprayed concrete liner can provide satisfactory ground motion control (e.g.
V
l
values in the range 0.5-1.5 % were recorded on recent tunneling projects in London clay).
Among the case histories considered by Mair (1996), particular attention is due to the Cairo metro Line 2
project, in Egypt (Ata, 1996) which was excavated with a 9.48 m diameter slurry shield in the Nile alluvium,
under the water-table. Ground losses on this project were in the range 0.2-1.0 %, with an average of 0.5 %. This
value should be compared to measurements taken during the construction of the Villejust High Speed Rail tunnel
(TGV), in France (Leblais and Bochon, 1991), which was excavated by the same contractor as in Cairo, using a

9.25 m diameter shield in fine sands, with slurry pressure at the face.
Ground losses on this latter project were kept in the same range as in the Cairo case (0.22-0.90 %) above the
water table, but reached higher values, in the order of 0.77-1.32 %, in saturated areas, which could yet be
considered promising at that time (New and O’Reilly, 1991). This comparison, of two case histories of slurry
shield tunneling projects completed in similar conditions, is illustrative of the progress that has been
accomplished in the past decade: in average, ground losses were reduced by 50 %, which is considerable given
the already high standards achieved in the late 1980s.
These improvements were mainly obtained because of the experience gained in controlling ground motion
into the tailpiece gap. Settlement values in the same order as those experienced in Cairo were reported in other
recent projects such as the construction of the Lyons metro extension, in France, using a slurry shield in mixed
alluvial deposits (Ollier, 1997; Kastner et al., 1996). Based on these observations, it is proposed that a typical
ground loss of 0.5 % be used for settlement analyses in alluvial soils.
The prediction of tunneling induced surface settlements is important in view of the potential damages they
could produce to above ground structures. These are usually associated with excessive amounts of absolute and
furthermore differential settlements within the structure, which can be appreciated from the evaluation of the
magnitude and lateral distribution of surface settlements (Figure 21, after Mair et al., 1996). Care should also be
taken, in this respect, of structures located along the tunnel center-plane, that could experience some temporary
differential settlements during construction, as a result of the settlement bowl progressing above the tunnel
heading (Leblais et al., 1995 & 1999; Mair and Taylor, 1997).


4.2.2 Horizontal Surface Displacements
Damages to above ground structures could also result from horizontal ground deformations induced by
tunneling. It is usually assumed (Leblais et al., 1995 & 1999; Mair et al., 1996) that the horizontal surface
displacement, s
h
(x) at a distance, x from the tunnel center-plane can be expressed as:

() ()
xs

H
x
xs
h
=

(7)

where H is the depth to tunnel axis and s(x) the settlement at a distance, x from the tunnel center-plane. This
approximate expression, which is based on studies by Attewell (1978) and O’Reilly and New (1982) relative to
tunnels in clays, has been found to provide reasonable estimates, for practical purposes, of horizontal
deformations observed in different project conditions (Mair and Taylor, 1997).

hogging
zone
sagging
zone
building
H
L
h
L
s
i
x
h
s

Figure 21 : Deformations induced to buildings located above a tunnel (after Mair et al., 1996)

Ground motion predictions should be used in relation with the sensitivity of above ground structures to
vertical and horizontal deformations. Three damage categories are usually be considered (Leblais et al., 1995 &
1999):
(1) architectural (essentially related to visual appearance);
(2) functional (disruptive to building usage);
(3) structural (could affect the building stability).
More elaborate characterizations of damages to buildings were proposed by Burland et al. (1977) and Boscardin
and Cording (1989), on the basis of a review of several case histories. In this latter classification, damages were
quantified in terms of the estimated tensile strain produced in the building (Table 4).

Table 4 : Damage categories (after Boscardin and Cording, 1989; Mair and Taylor, 1997)
Category of damage Normal degree of severity Limiting tensile strain (%)
0 Negligible 0-0.05
1 Very slight 0.05-0.075
2 Slight 0.075-0.15
3 Moderate 0.15-0.3
4 to 5 Severe to very severe >0.3

Further works by Mair et al. (1996) allowed a method for risk assessment of tunneling induced damage to
buildings to be established. This method is based on a correlation of the category of damage to the combined

effects of the maximum vertical relative deflection and horizontal strain produced in the structure (Mair and
Taylor, 1997).

4.2.3 Subsurface Displacements
With the increasing use of soft ground tunneling in urban areas, situations more often arise where
construction may affect existing underground structures, including other tunnels, buried pipes or piled building
foundations. As a result, it becomes necessary to be able to predicting, not only the magnitude and distribution of
surface displacements, but the whole field of ground movements around the tunnel to be excavated. This issue
was addressed by Mair et al. (1993) who proposed solutions for extending existing methods for surface

settlement estimates to the prediction of tunneling induced subsurface settlements (Mair and Taylor, 1997).
Specific studies were also completed on side by side tunnels, and provided some basis for comparison with
the single tunnel case, in terms of modifications to surface settlements and/or liner loads induced on each
interacting tunnel (Cording and Hansmire, 1975; Ghaboussi and Ranken, 1977; Leca, 1989).
Some of these studies made use of numerical methods. These techniques, and particularly the Finite Element
Method, have tremendously developed over the past thirty years, and provided other efficient means of
characterizing the ground motion to tunneling. Additional insights into the overall ground response to tunneling
can now also obtained from physical modeling, primarily due to advances in centrifuge model testing.

4.2.4 Numerical Models and their contribution to the prediction of tunneling induced ground movements
A review of Finite Element modeling applied to soft ground tunneling was completed by Clough and Leca
(1989) who examined the contribution of numerical methods to the analysis of the complex soil-structure
interaction phenomena associated with the construction of tunnels in soils. The more recent advances related to
these techniques were further analyzed by Mair and Taylor (1997) and Leca and Mestat (1999). Finite Element
modeling can be considered as a powerful means of analysis, because it allows the main parameters involved in
the tunneling process to be accurately accounted for. These include:
- the actual geometry of the project (tunnel shape, size and depth, soil layering,…);
- the ground behavior (constitutive law associated with each soil layer);
- the construction sequence (a variety of loading conditions can be introduced to account for the different
construction phases associated with each excavation technique).
Additional benefits can be found in the broad variety of outputs obtained from the analyses, which include the
whole field of underground movements and surface settlements induced by tunneling, as well as liner loads to be
accounted for in the design of the tunnel liner.
Conversely, the ability for performing sophisticated numerical analyses should not elude limitations
associated with these techniques, nor should it eliminate the need for conventional methods, which are founded
and calibrated on the basis of experience and well documented evidence. Numerical methods should not be
viewed as a substitute to conventional methods, but rather as a means of filling gaps that may exist in the
conventional approaches, as well as a tool for acquiring a better understanding of the complex ground response
to tunneling.
The following limitations should be considered when using numerical methods for tunneling:

(1) the tunneling process involves a three-dimensional deformation pattern around the front, which should be
accounted for in the analysis; the ground response to tunneling can be analyzed by means of three-
dimensional Finite Element models, but this process is perceived costly and time consuming;
(2) the behavior of soils and soft rocks is complex and can hardly be accurately modeled, even with the most
elaborate constitutive models; moreover, parameters such as the soil deformation modulus, which are
essential to the output of numerical analyses, can hardly be estimated with sufficient accuracy with
existing testing techniques;
(3) the construction techniques involve complex soil-structure interaction phenomena; these should be well
appreciated and their effect accounted for in an appropriate manner in the model;
(4) the practical accuracy of existing monitoring methods can hardly provide the basis for validating some of
the most elaborate numerical tools.
Significant progress has been made over the past ten years in these fields, which has allowed some of these
difficulties to be overcome. In particular, a more general use can now be made of three-dimensional and non-
linear analyses, due to advances in numerical methods and computer power. Nevertheless, there are still
hesitations to recur to such approaches because they are often perceived as time consuming in terms of data
preparation and result processing.

As a result, the ground response to tunneling is often analyzed using two-dimensional models (tunnel cross-
section), with a simple constitutive law for the soil (linear-elastic with perfect plasticity). Techniques have been
developed to include three-dimensional effects in 2D analyses, as will be demonstrated in the next section on
liner design. These approaches have been proven to provide sufficient accuracy for practical purposes, as far as
the determination of liner loads for design is concerned (Panet and Guénot, 1982). However, limitations are
often found in applying these methods to the prediction of tunneling induced ground movements and surface
settlements.
Reviews by Eisenstein (1986), and later by Leca and Clough (1989) showed that two-dimensional models
allowed reasonable estimates of the maximum surface settlements, s
max
to be obtained, provided adequate
modulus values were used for the soil, but would generally fail to reproduce the observed distribution of surface
settlements. Moreover, the predicted settlement troughs would consistently be wider than the observed shapes

(larger i values) which is on the unsafe side, as far as potential effects on surface structures are concerned.
Such trend could conceivably reflect the inability of most soil models used in practice to account for the
strain localization effects that may be involved in the ground response to tunneling. Limitations should, however,
also be searched in other parameters such as the two-dimensional nature of current analyses. In particular, Kasali
(1981) reported a realistic representation of the ground response to shield tunneling, using a simple linear-elastic
constitutive law for the soil, in a three-dimensional model with due account of the loads associated with the
shield operation process. Ground anisotropy should also be considered as one key factor for obtaining realistic
ground deformation predictions.
A comparative study by Rowe et al. (1983) of three soil models used in the analysis of measurements made
during the construction of a tunnel in soft clayey soils showed that: (1) the simplest, isotropic elastic model
would yield much larger settlement troughs than observed; (2) allowance for some soil anisotropy in an elastic
model would allow some improvements to be obtained; and (3) allowance of both plasticity and anisotropy in
the soil’s constitutive law would lead to results in close agreement with measured ground movements.
While these conclusions provide valuable insights into the phenomena involved in the ground response to
tunneling, their impact on modeling practices remain limited at present, due to difficulties in acquiring the
necessary information for such analysis to be performed on a regular basis. Even though anisotropic constitutive
laws can be implemented in Finite Element models, the determination of anisotropic elasticity and plasticity
parameters would only be obtained through specific tests that would rarely be performed for a project.
An interesting attempt in that respect was reported by Simpson and Atkinson (1996) who used an anisotropic
model to analyze ground movements observed during the construction of the Heathrow Express trial tunnel in
London. The soil behavior was modeled using orthotropic linear elasticity with a Mohr-Coulomb cut-off. The
elastic anisotropic parameters were derived from a correlation between shear wave velocities measured in
laboratory and in situ tests, and the shear moduli required for the orthotropic model.
This procedure was successful at reproducing the observed magnitude and extent of settlement trough. The
authors reported that this approach had been investigated because previous analyses using soil non-linearity or
three-dimensional effects had failed to produce appropriate surface settlement distributions.
The ground response to tunneling may also be affected by stress induced anisotropy (Guilloux et al., 1998).
Mair and Taylor (1997) reported works by Addenbrooke (1996) on the numerical analysis of tunnels in stiff
clays, where the influence of assumed initial stresses had been considered. Different analyses were performed for
instrumented sections of the Jubilee Extension Line project in London (Standing et al., 1996), with assumed

values of the coefficient of earth pressure at rest, K
o
of 1,5. They showed that reducing the value of K
o
in the
vicinity of the tunnel, to account for stress rearrangements produced around the face during excavation, would
considerably improve the predicted distribution of surface settlements.
One major advantage of numerical studies is to allow the very complex soil structure interactions involved in
the tunneling process to be accurately investigated. An example of the potentials offered by the Finite Element
Method in addressing intricate underground construction configurations was provided by a study completed by
Higgins et al. (1996), as part of the Jubilee Line Extension project design, to analyze the influence of tunneling
and excavation works planned next to Big Ben clock tower at Westminster station in London (Figure 22).
The grounds found on the site consisted of London clay overlain with Thames gravel and made ground and
alluvium. Several construction stages were introduced in the analysis: (1) construction of two pilot tunnels within
each running tunnel area; (2) excavation of the station box with placement of temporary support, construction of
the base slab and installation of permanent wall supports; (3) enlargement of the pilot tunnels.