Surface Settlement Analysis

Surface settlement analysis
and prediction during construction of

the Toulon Underground Crossing's
northern tunnel
Jean-François SERRATRICE Jean-Pierre MAGNAN
Laboratoire régional des Ponts et Chaussées Laboratoire Central des Ponts et Chaussées
d’Aix-en-Provence
5-36
PP .
Introduction
4412 -
The excavation of shallow tunnels causes surface deformations in the host soil or rock mass. These
RÉF .
deformations are primarily represented by settlements, which while not uniform are concentrated in
2002 -
alignment with the underground cavity in the form of a basin. The problem becomes particularly acute
in cities, where these settlements affect all components making up the urban fabric, whether they be
MARS - AVRIL
buildings, civil engineering structures, roads or utility networks. Deeper beneath the surface, other struc-
tures may also be located within the zone of influence of the tunnel construction project: tunnels, espe-
C HAUSSÉES - 237 -
cially when built in pairs with interaction between the two; but more typically underground structures
(galleries, collector systems, etc.). Deep foundations, which have not necessarily been designed to with-
stand nearby tunneling work, must also not be overlooked.
Excavation of the northern tunnel of the Toulon underground crossing, which lasted from 1994 to 2000,
ET
caused a number of soil surface settlements, the analysis of which is the focus of the present article. The
P ONTS
real-time analysis of these settlements, conducted in order to respond to the well-founded interrogations
DES LABORATOIRES DES
of both project engineers and the local population, was oriented around integrating the most recent
advances in this field. Special focus was placed on identifying an analytical model capable of producing
reliable and coherent extrapolations of measurement results. For this reason, the present article devotes
considerable attention to a review of settlement computation models.
Following a broad overview of the problems raised by surface settlements within an urban setting and
B ULLETIN
the analytical methods proposed in the literature, we will present the site of the Toulon underground
crossing's northern tunnel, then the two predictive methods implemented in real time during monitoring
of this tunnel from 1998 to 2000 and finally the conclusions of the analysis on soil surface settlements
5
along the tunnel.
Surface settlements during tunnel excavation
In his report on the current state of knowledge presented to the International Conference on Soil
Mechanics and Foundations Engineering held in Mexico City in 1979, Peck commented on "the pressing
need for efficient resources that allow estimating the extent and type of movements and disturbances
related to tunneling or trenches in urban areas, as only such estimations ensure a rational basis for
decision-making" (Peck, 1979). In essence, "the design of these projects requires being able to: assess
the feasibility of using one construction method over another, estimate the settlements or other
movements generated in the enclosing rock mass and neighboring structures, and provide the final
structure with sufficient strength and appropriate stiffness or flexibility".
Since that time, major progress has been made from a technological standpoint in the methods for build-
ing shallow tunnels in technically-difficult urban sites. Many underground projects have been con-
structed in France and throughout the world (road tunnels, rail tunnels, subway systems, galleries, sewer
collector networks, etc.). In many cases, these works take place either in ground with a cover of Quater-
nary Period deposits or in the altered fringe of the underlying rock masses. These poor quality, hetero-
geneous soils often prove to be aquiferous. Construction methods have continuously been improved in
order to better confront the multitude of difficulties invariably encountered during construction of such
shallow facilities in this type of soil.
In conjunction with this progress, sizable efforts have been devoted to studying surface movements at
the time of tunnel excavation, with corresponding modeling approaches, as well as to compiling the feed-
back such studies have generated. The AFTES recommendations (1995) relative to settlements due to
excavation work for underground structures are intended to provide a more global perspective, "with
respect to both the soil-structure interaction phenomena in effect during underground excavation and the
means necessary for their evaluation, measurement, prevention and eventual remedy, without overlook-
ing the inherent contractual ramifications".
The bibliographical entries included at the end of this article pertain to experimental data on surface set-
5-36
tlement collected at the time of shallow underground works carried out in urban areas and then published
in the literature. This analysis has been produced subsequent to many syntheses previously written on
PP .
the topic (Peck, 1969; Cording and Hansmire, 1977; Rowe and Kack, 1983; Schlosser et al., 1985; Oteo,
4412 -
1993).
RÉF .
2002 -
The underlying mechanisms

MARS - AVRIL
Surface settlements result from a dual ground-structure interaction mechanism: one interaction between
the tunnel and the ground, and the other between the ground and the adjacent built environment. Near
the tunnel face, the displacement field created by the excavation is three-dimensional: added to ground
C HAUSSÉES - 237 -
convergence within the plane orthogonal to the tunnel axis (crown settlement, convergence, raft raising)
are ground extrusion displacements in the tunnel axis direction heading towards the tunnel face. In
general, displacement vectors are directed to the active excavation region, which in light of the quasi-
continuous progression of the tunneling works imparts a certain evolution in the displacement field: at a
point located in a given section of rock mass, the direction and amplitude of the soil displacement vector
ET
vary as the tunnel face approaches, both at the time of its passage and then as it moves further from the
P ONTS
considered section. The tunnel-ground interaction is thus reflected by: sidewall convergence, extrusion
of the tunnel face, and forces in the supports. Considerations related to the role of water may be added
to this list (flow, pore pressure, pressure on the support, etc.). Total convergence (i.e. convergence both
in front and in back of the tunnel face) plus extrusion compose what is referred to as "volume losses".
These various elements have been combined in Figure 1, which depicts a three-dimensional section of a
tunnel with construction advancing towards the right-hand side of the figure and lit from the inside. This
figure clearly shows:
B ULLETIN
¾ convergence and extrusion of the tunnel face,

¾ displacements within the ground,
6
¾ the surface settlement trough, graduated using contour lines.
Total basin
y
3
2
1
1. Tunnel face extrusion

2. Convergence Fig. 1 - Soil deformations and surface
3. Displacement in the ground settlements during excavation of a
shallow tunnel: Definition of axes
The primary set of factors to be considered in the control of ground deformations and their surface effects
consist of the following:
¾ the selected tunnel construction method, which may enable limiting underground volume losses or
may even entirely avoid them, and ensuring structural stability;
¾ ground behaviour, which often proves distinctly non-linear within the given range of deformations
and low stresses and the response of which depends heavily upon its initial state;
¾ ground heterogeneities with, in general, a layer exhibiting weaker characteristics at the surface (in
which most foundations are located) and, in the case of Toulon, a very large degree of heterogeneity at
greater depths along the tunnel alignment;
¾ the type and state of of surrounding buildings, which must be appraised prior to the project and then
5-36
monitored throughout tunnel construction.
PP .
In the case of shallow tunnels, these displacements propagate within the ground as the tunnel face
4412 -
advances and provoke settlements and horizontal displacements at the surface. Such horizontal
displacements tend to "follow" the tunnel face, with changing directions when advancing, as revealed by
RÉF .
the two test sections of the northern tunnel, Chalucet and Dumont d'Urville (Serratrice, 1999). Surface
2002 -
subsidence therefore depends not only on the ground around the tunnel, but on the state and behaviour
of the overlying layers, as well. The geometry of shallow tunnels, with low depth to diameter ratio,
MARS - AVRIL
reduces the effects of transversal and longitudinal arching which develop naturally at depth. The risk of
instability rises in consequence (weak confinement, uncontained plasticity) and displacement amplitude
increases. As a result, within deformable or cracked, low strength and heterogeneous ground, the support C HAUSSÉES - 237 -
is set up as early as possible and as closely as possible to the tunnel face, or even ahead of the tunnel face
(case of pre-support), in order to avoid instabilities hazardous for worksite personnel and adversely
affecting the structure prior to construction of the definitive lining. In an urban area however, while
tunnel safety must be considered a priority, limiting surface settlements to acceptable levels is also
ascribed high importance and may necessitate support adaptation along with a carefully-designed
construction method.
ET
P ONTS
The interaction between the ground and the surface built environment is reflected by settlements,
differential settlements and horizontal displacements, which submit structures to loads (distortional,
compressive, extensional) capable of causing them to crack or even fail. At greater depths, soil
deformations may also undermine the equilibrium of neighboring installations (deep foundations,
underground structures, etc.).
Transversal cross-section of the surface settlement trough
Following passage of the face of a shallow tunnel within a rock mass cross-section, the surface
B ULLETIN
settlement trough assumes the shape of a bell-shaped curve centered on the tunnel axis, which since the
work of Schmidt (1969) and Peck (1969) has been widely described in the form of a Gaussian curve:
–y2
s ( y ) = s o exp  -------- (1)
 2i 2  7
where y is the horizontal distance to the tunnel axial plane, so the (maximum) settlement on the axis and
i a parameter characteristic of the trough expanse (or relative depth). This curve displays the following
characteristics:
¾ inflection point for y = i s(i) = 0.61 so p(i) = s’(i) = 0.61 so/i
¾ maximum curvature for y = i 3 s ( i 3 ) = 0.22s o
¾ unit volume (per meter on the tunnel axis) Vs = 2πis o = 2.5is o

¾ relative volume (with respect to the tunnel section area)vs = Vs/(πR2)
where R is the excavated radius of the tunnel and p the maximum slope of the transversal trough,
obtained at the inflection point of abscissa i. As an example, for so = 16.5 mm and i = 10 m, a result of
p(i) = 0.001 = 1 mm/m is derived.
Ground losses
It is normal to seek a correlation between ground displacements at the tunnel level and measured surface
displacements. This leads to developing the notion of ground (volume) loss, i.e. the portion of ground
volume that either converges through the theoretical tunnel cross-section or is extruded at the tunnel face
(Peck, 1969).
Habib (1998) proposed an order-of-magnitude evaluation of the elementary ground loss by
distinguishing the main causes behind the development of a surface subsidence trough. For a shallow
tunnel with a horseshoe-shaped section, these causes would be:
¾ contraction of the lining at the time of loading,
¾ settlement or punching of the sidewalls,
¾ bending of the sidewalls submitted to the action of horizontal forces,
¾ swelling of the raft,
¾ swelling of the tunnel face (soil extrusion).
5-36
Habib concluded that "the rapid installation of a strong raft to close the tunnel lining ring and prevent the
PP .
raft from swelling, while blocking convergence between sidewalls and perhaps limiting the punching of
4412 -
sidewalls, represents the most effective method by virtue of acting on the mechanisms that cause the
largest contractions in the tunnel section". He also noted that "the role of tunnel face bolting is not to
RÉF .
limit elastic deformations, but primarily to reinforce the earthen core lying in front of the tunnel, in order
2002 -
to ensure the best possible level of pre-confinement and avoid plastic extrusion of the tunnel face, which
would generate both surface subsidence and subsequent disorders in the support". This point of view
MARS - AVRIL
reflects the three-dimensional and evolving nature of the mechanisms that take place during the
excavation and support of a shallow tunnel and may lead to tunnel face instability.
The AFTES recommendations (1995) called for inventorying the sources of settlement related to con-
C HAUSSÉES - 237 -
struction activity and distinguished the case of sequential excavation-type works (the focus herein) from
that of works performed using a tunneling machine. The recommendations emphasize the importance of
reducing, to the maximum extent possible, underground deformations as a means of limiting surface
settlements. Since the construction method has been established in order to achieve this objective, both
the quality of its implementation and its capacity to adapt to specific situations also contribute to
ET
accomplishing the task. For the purpose of setting an order of magnitude, allowable surface deformations
P ONTS
associated with underground works in urban areas lie at a distortion level of approximately 1/1000
(0.1%). This deformation range remains quite small and does not correspond with either the plastic
regime in most soils or cracked rocks. In contrast, average deformation levels at the periphery of the
structure display a higher order of magnitude (1%).
This quick analysis of deformation mechanisms at the tunnel face has enabled perceiving the evolution
in tunnel construction methods, for which sizable progress has been achieved from a technological
standpoint: pressurized tunnel face shields, allowing for tunnel excavation in weaker soils and aquifers;
B ULLETIN
pre-support techniques (bolting, pre-arches); ground treatment techniques (drainage, bolting, injections);
and compensation injections. In parallel with these advances, considerable effort has been devoted to:
studying surface movements during tunneling in urban settings, modeling such movements (by
8 accounting, to the greatest extent possible, for both construction methods and phasing); and compiling
feedback from investigative work.
The ground loss VL represents the portion of ground volume that converges within the excavated or
extruded tunnel section at the tunnel face. With respect to a tunnel section of diameter D, this relative
volume loss is equal to:
4V L
v L = ---------- . (2)
πD 2
The lost underground volume proves complicated to evaluate for the following reasons:
¾ ground pre-convergence (ahead of the tunnel face), which remains difficult to measure;
¾ the impossibility of conducting convergence measurements prior to installation of the support, which
means that part of the deformations is not being detected;
¾ the difficulty of performing convergence measurements from the raft;
¾ the presence of extrusion movements at the tunnel face in the tunnel direction;
¾ contact flaws between the ground and the support.
Despite lying beyond the bounds of precise measurement, the ground loss still constitutes a key index in
the expression of surface settlements. Its amplitude depends primarily on the soil type, the excavation
and supporting method, and the quality of execution. Cording and Hansmire (1977) provided several
estimations of these relative volume losses (ranging between 0.2% and 12%). Values capable of reaching
3% were recorded in the soft marine clays of Singapore with either earth-pressure or air-pressure
tunneling machines (Shirlaw and Doran, 1988). On the other hand, losses of 0.2% could be observed
using an earth-pressure shield through the gravel underneath the Tokyo groundwater table (Kanayasu et
al., 1995). For tunnels dug through London clay, these losses are typically on the order of 1 to 2% for
shield tunneling (O’Reilly and New, 1982). Tunnels built in this clay with a shotcrete support also lead
to volume losses of the same magnitude, in the 1-1.5% range (New and Bowers, 1994). Similar results
were obtained during construction of the Jubilee Line Extension (Mair et al., 1996). Moreover, it should
be pointed out that the ground disturbance caused by pressurized tunneling shields serves to heavily
reduce these ground losses, and in some instances even transform them into volume "gains".
For undrained clayey masses, which undergo deformation at constant volume (i.e. the speed of
excavation does not lend sufficient time for the soil to consolidate), the lost underground volume VL may
5-36
be assumed equal to the unit volume of the surface settlement trough Vs:
PP .
Vs = VL
4412 -
Mair et al. (1993) then deduced the following relation:
RÉF .
2πD 2 vL D2 vL D 2
2002 -
s o = ------------------ v L ≈ 0.313 ------------- ≈ 0.313 ------------- (3)
8i i Kh
MARS - AVRIL
which allows correlating settlement on the axis with the lost underground volume, in knowing the value
of either i (parameter of the settlement trough Gaussian curve) or K (coefficient depending on the ground
type) and h (tunnel axis depth). C HAUSSÉES - 237 -
The prediction of the shape of vertical displacements within the ground is critical whenever a tunnel is
built beneath undergrounded structural elements (basements, deep foundations, existing underground
facilities). It can be stated that the transversal profile of vertical settlements at each depth z is also
represented by a Gaussian curve. From an analysis of structural data and physical centrifuge models,
Mair et al. (1993) demonstrated that the value of i decreases with z and proposed a formulation for
ET
settlements within the axial plane of the tunnel:

P ONTS
1.25v L R 2
s y = 0 ( z ) = -----------------------------------------------------------
- (4)
 z
0.175 + 0.325 1 – --- h
 h
The approach that consists of correlating underground deformations with surface deformations, which
would seem entirely natural at first glance, is really only valid for undrained clayey soils (i.e. zero
volume variation). Such an approach winds up being limited by its empirical aspect as well as by:
B ULLETIN
¾ imprecise knowledge of underground volume losses via pre-convergence and extrusion;

¾ lack of complete understanding of the mechanisms by which deformations are transmitted towards
the surface; recall that such mechanisms depend on: the initial (geomechanical and hydraulic) state of 9
the ground, ground heterogeneity (superimposed soil or rock layers), its anisotropy and the various
effects of consolidation, creep and progressive deformation. Historically speaking, the notion of ground
loss has nonetheless contributed to inciting construction method improvements.
The approach proposed by Rowe and Kack (1983) is based on the concept of soil intrusion created by
the difference between excavated volume and volume occupied by the structure once the lining has been
installed; it is applied in particular to shield tunneling in soils where the void left between the shield and
the lining ring extrados gradually closes as the tunneling machine advances. In more general terms, this
notion may be used to take account of three-dimensional effects in plane strain computations, along the
same lines as the convergence-confinement method (Leca, 1997). It may also be employed to incorporate
other sources of ground deformation (soil intrusions at the tunnel face, quality of execution, etc.), with
each effect being reflected by an adjustment in the void volume (displaced earth) introduced at the tunnel
crown.
Limitation of settlements
The AFTES recommendations (1995) place considerable emphasis on presenting the methods imple-
mented in practice to limit surface settlements, by means of distinguishing the following:
¾ improvement in general project conditions,
¾ improvement in soil behavior,
¾ improvement in the built environment,
¾ improvements in sequential excavation methods (tunnel face support, pre-support),
¾ improvements in tunneling machine excavation methods.
Oteo (1993) also highlighted the various methods used for limiting surface settlements and provided sev-
eral pertinent examples.
Problems specific to tunnel doubling and tunnel junctions

Double tunnels are often built for subway systems, railroads, highways and collection systems. The deci-
5-36
sion of whether to build a single tunnel with a large cross-section or two smaller-section tunnels takes
on major importance and widely exceeds the scope of geotechnical considerations alone.
PP .
4412 -
The literature provides extensive coverage of surface settlements produced following the construction of
a second tunnel. In general, these settlements are greater than the sum of the settlements induced by
RÉF .
building two identical isolated tunnels: the excavation of the second structure generates a trough the
2002 -
depth of which is larger than the sum of the individual troughs of two independent structures and which
is wider and located outward in the direction of the existing facility (Peck, 1969; Cording and Hansmire,
MARS - AVRIL
1977; Ou et al., 1998). Settlements depend on the geometrical configuration of the structures involved
(and especially on the distance between tunnels, in comparison with their depth). Should consolidation
resume, it often proceeds more quickly above the existing structure due to the prior rearrangement of soil
C HAUSSÉES - 237 -
caused by the construction activity. The same would apply to progressive deformations. It is therefore
essential for the first tunnel to be well built, by means of minimizing soil deformations; otherwise,
irregular and unpredictable movements may occur above the first tunnel and, ultimately, above the
second as well (Peck, 1969).
Reinforcement of the core in front of the first tunnel face serves not only to limit both surface settlements
ET
and eventual disorders on the support behind the tunnel face, but also to better control the ground
P ONTS
deformations around the structure and the movements due to the excavation of the second tube. Beyond
the compression of the pillar located between the two tunnels, additional settlements stem from
interference between the deformed zones of each tunnel. Reliance on compensation injections
sometimes proves necessary for superimposed tunnels or tunnels located very close to one another (Oteo,
1993; Mair et al., 1996).
An analogous situation is encountered at the junction of two separate faces of the same tunnel or at the
periphery of a shaft, another gallery or an end trench. Upon approaching a zone that has become more
B ULLETIN
deformable by the presence of a preexisting cavity in front of the core and by non-linearities in both
ground behaviour and the ground-support interaction, additional settlements are to be expected in the
absence of ground reinforcement. As opposed to double tunnels, for which excavation interference tends
10 to lie within a transversal plane, junction interference is more likely to take place in the longitudinal
profile.
The junction of faces F1 and F2 of the Toulon underground crossing's northern tunnel, which occurred
on March 13, 2000, did not dramatically accentuate surface settlement, given the specific geological con-
text of this zone and thanks to the prior introduction of both pre-support and support reinforcement.
Nonetheless, the evolution in surface settlements was modified with an amplification of preliminary set-
tlements in front of the tunnel faces, to the detriment of the delayed settlements behind the faces (which
constitutes a sign of the two faces interfering with one another).
The special case of reconnaissance galleries dug preliminarily may also, to a certain extent, be associated
with this type of situation.
Furthermore, the effects of reconnaissance galleries excavated for shallow tunnel projects are worth
mentioning: both their construction method (by small sections) and temporary support are capable of
causing unacceptable deterioration to the ground in the space occupied by the future structure
(progressive deformations). Put otherwise, the presence of a reconnaissance gallery may deteriorate the
core ahead of the tunnel face and thus produce the opposite effect of a pre-support. In terms of
construction speed, this situation is not comparable with that of a pilot gallery according to the divided-
section methods in that such a gallery is associated with the normal progression of excavation works, as
opposed to a reconnaissance gallery, which is often dug well in advance of actual tunneling startup.
Computation of settlement trough

The analytical or numerical computation models for surface settlements of the ground above a tunnel
generally simplify the geometry and mechanical properties of the soil and phasing of the works, yet they
do provide keys to analyzing measurements as well as selecting the shapes of surface settlement profiles
in both the longitudinal and transversal directions.
The analytical (two-dimensional) solutions to the theory of elasticity allow easily performing the para-
metric analyses to yield a simplified project representation. At the dawn of the 20th century, much work
was devoted to computing the redistribution of stresses caused by the introduction of a discontinuity into
an elastic medium. The problem of the perforated plate loaded by a uniform pressure was studied by
5-36
Kirsch (1898) for a circular hole (plane stress). His solution may be extended by superposition first to
the case of biaxial loading and then to the case of a deep small-diameter tunnel (plane strain). The case
PP .
of a hole featuring an elliptical section was also investigated.
4412 -
Jeffery (1920) transcribed the elasticity equations into bipolar coordinates (network of coaxial circles),
RÉF .
which allows applying the solution to various types of problems, such as the semi-infinite plate
2002 -
containing a hole close to the free boundary and submitted to an interior pressure. This solution results
in applying the stress computation in the vicinity of the hole when a rivet is forced into place.
MARS - AVRIL
Mindlin (1940) computed the stresses present around a shallow circular tunnel placed into a gravitational
semi-infinite elastic medium. The solution is expressed by means of an infinite bipolar coordinate series
(Jeffery, 1920) for various initial stress states. The solution shows that if the tunnel is located at a depth C HAUSSÉES - 237 -
of more than four diameters, stress distribution in the ground is approximated by the Kirsch solution.
Mindlin's solution correctly incorporates the effect, in plane strain, of excavated ground weight into the
post-excavation ground equilibrium.
The case of three-dimensional cavities was also examined (Terzaghi and Richart, 1952). The set of these
two- and three-dimensional solutions was sought over the elastic range, which in practical terms limited
ET
P ONTS
the scope of investigation to cavities displaying simple geometrical shapes within non-gravitational
media. For more complex shapes, since analytical methods are no longer applicable, solutions were
sought using physical models. In this vein, photo-elasticity methods were often used to determine the
stresses around tunnels with differently-shaped cross-sections. The analyses conducted by Terzaghi and
Richart (1952) concluded that the greatest indeterminate feature on final equilibrium stems from uncer-
tainty over the initial stress state in the ground prior to cavity excavation.
Sagaseta (1987) presented a solution for computing the deformations induced by excavation of a hori-
zontal circular tunnel in a homogeneous, elastic and incompressible ground. The corresponding lost
B ULLETIN
volume, equal to the settlement trough volume, is concentrated on the tunnel axis.
Verruijt and Booker (1996) extended this solution by introducing an ordinary Poisson's ratio, which
served to authorize volumic deformations in the soil. The authors assumed a convergence mechanism 11
along with an ovalization mechanism as part of the tunneling works.
Loganathan and Poulos (1998) integrated non-linear terms into these formulae in order to better repre-
sent the displacement field both within the rock mass and at the surface. These expressions were then
utilized, as discussed further below, to interpret data from the Chalucet and Dumont d'Urville test blocks
(Serratrice, 1999), which were set up in two sections of the northern tunnel. The detailed expressions of
soil displacement according to these various theories are provided in Appendix A.
Oteo and Moya (1979) proposed the following relation for estimating settlement above the tunnel:
ψ ( 0,85 – ν )γD 2
s arch = --------------------------------------- (5)
E
where γ is the weight density of the ground, D the tunnel diameter, E the Young's modulus of the ground,
ν its Poisson's ratio and ψ a factor correlated with the speed of action of the support (ψ = 1 if the support
is not active, where ψ generally lies between 0.25 and 0.5). This relation reveals the role of Poisson's
ratio, the decrease of which results in increasing the crown settlement. Low Poisson's ratio values are not
conducive to arch effects. At a larger scale, cracked rock masses most likely display small ν values.
This type of relation may be extended in order to compute maximum surface settlement in the following
form (AFTES, 1995):
kλD 2
s o = -------------- (6)
E
This expression is noteworthy for its ability to highlight the influence of each main parameter, i.e.:
¾ excavation cross-section (D2),
¾ rock mass deformability (E),
¾ the method and quality of execution (λ, rate of confinement removal),
¾ the experience-related factor (k).
5-36
Excavation of Toulon's northern tunnel

PP .
4412 -
Background
RÉF .
The Toulon underground crossing project comprises two road tunnels designed to ensure continuity
between the A 50 motorway to the west of Toulon (heading towards Marseille) and the A 57 to the east
2002 -
(towards Nice). The northern structure spans a total of approximately 3000 m and is excavated as a
MARS - AVRIL
tunnel over a length of 1800 m within its central part. It crosses the city of Toulon at an average depth of
35 m and passes underneath a dense and older built-up area (see Fig. 2). To both the west and east, the
tunnel extremities are composed of 600-m long openings and covered trenches, respectively. These
structures were built at the beginning of the 1990's. The northern tunnel was excavated from both the
C HAUSSÉES - 237 -
Western opening (via a shaft dug in 1993) and the Eastern opening: excavation work commenced in
December 1994, with tunnel face F1 advancing from the Western opening in an eastward direction,
tunnel face F2 advancing from the shaft towards the Western opening, and tunnel face F3 advancing from
the shaft towards the Eastern opening. Face F3 reached the Eastern opening in March 1996; however,
during this same period, a ground surface collapse occurring near face F2 interrupted the works.
ET
P ONTS
Excavation resumed in February 1998 on faces F1 and F2 with an improved mode of support.
Following the resumption of works in February 1998, two major families of profiles were implemented:
Profiles PN3 with tunnel face bolts, divergent bolts, pre-arches, trusses and shotcrete lining; and Profiles
PN5 (or PN6) with tunnel face bolts, divergent bolts, shotcrete, trusses and shotcrete lining. The junction
of faces F1 and F2 took place on March 13, 2000 at PM 1291. Figure 3 presents a PN3-type tunnel
section. Average tunneling speed typically lies in the neighborhood of 0.8 m/day.
As regards typical section cuts, the section excavated for the northern tunnel amounts to an area of
B ULLETIN
106 m2, hence an average radius of 5.5 m. The structure reaches a maximum depth of 42 m (at the center
of the tunnel's typical section) at approximately position reference -22 m NGF heading towards
PM 1200. An expanded cross-section (excavated section of 140 m2) was dug between PM 1180 and
12 PM 1220 to accommodate a garage. Table I indicates the key dates during the northern tunnel excavation
project.
Radier de roulement
Chalucet Hospital Radier de roulement 3
RadierRue Chalucet et pieds de cintres
en contrevoûte Pied de cintre
Cintres Avenue Vauban
Rue Dumont d’Urville
Avenue Carnot Rue Chabannes
Western opening
Place Mazarin
Boulevard de Tessé
Fabié Bridge
Marchand Bridge
Origin 891800 98200
PM 485 PM 1291
Junction F1 F2
Scale (m)
Avenue Colbert Rue Fabié Central shaft
0 50 100
PM 1950
Eastern opening
Tunnel graduated with a 25-m increment
PM = metric point, distance measured from the beginning of the alignment
PM 2330
Fig. 2 - Layout plan for the Toulon underground crossing’s northern tunnel
Axial plane of the tunnel

Pre-arches Divergent bolts
4,35m
Divergent bolts (18 m)
Phase n-2
Phase n-1
Phase n
Frontal bolting
5-36
Frontal bolts
Tunnel axis
PP .
4412 -
Rolling raft
RÉF .
Rolling raft 3m
Counter-arched raft and truss footings
2002 -
Truss footing
Trusses
b. Transversal section
MARS - AVRIL
a. Longitudinal section
Fig. 3 - Standard tunnel section

C HAUSSÉES - 237 -
The tunnel site: Geotechnical context

The northern tunnel site consists of both older ground formations heavily influenced by the orogenesis
of Primary Era tectonics and Permian and Trias Epoch deposits, with the following stratigraphic
ET
P ONTS
sequence:
¾ Quaternary: colluvium, alluvium and backfill;
¾ Keuper: mottled clays with the interspersion of cellular dolomite and patches of gypsum;
¾ Muschelkalk: dolomitic limestone, clear marl interspersed with gypsum;
¾ Permian Epoch: sandstone, red pelites;
¾ Stephanian Epoch: "poudingues" (cemented shingles) and black schist (coal);
¾ Bedrock: quartzo-phyllades.
B ULLETIN
From a tectonic standpoint, the tunnel is located at the northern boundary of the Cape Sicié thrust zone
(first thrust nappe discovered by M. Bertrand in 1887). This nappe is composed of heavily-folded Pri-
mary Era deposits, with the direction of this thrust being SW-to-NW. The bedrock of this nappe consists
of coal-bearing deposits lying in a pit that collapsed at the end of the Hercynian Period. In assuming its 13
present position, this nappe caused spalling of the more recent deposits (Permian and Trias). It must also
TABLE I
Key dates in the northern tunnel construction project
Date Face F1 Face F2 Face F3
PM Location PM Location PM Location
Dec. 6, 1994 1 947,5 Central shaft

(heading west)
Jan. 10, 1995 1 966 Central shaft
(heading east)
April 10, 1995 485 Western opening
Aug. 1, 1995 1 801,5 Profile change
Jan. 3, 1996 1 725 Marchand Bridge
Mar. 6, 1996 2 332 Eastern opening
Mar. 13, 1996 1 667,5 F2 shutdown
June 24, 1996 693 Avenue Carnot
Nov. 12, 1996 769.5 F1 shutdown
Feb. 6, 1998 775.5 F1 resumption
Feb. 12, 1998 1 708 F2 resumption
July 29, 1998 863 Chalucet P1 block
Sept. 10, 1998 889 Chalucet P2 block
Sept. 21, 1998 895 Rue Chalucet
Nov. 5, 1998 1 664 Profile change
Jan. 15, 1999 1 615 Fabié Bridge
Mar. 17, 1999 1 575 Boulevard de Tessé
April 8, 1999 1 560 Rue Truguet
April 21, 1999 1,039 Avenue Vauban
Aug. 4, 1999 1,140 Rue Dum.d’Urville
5-36
Aug. 11, 1999 1 487 Profile change

Aug. 17, 1999 1,150 Dum.d’Urville block
PP .
Aug. 27, 1999 1 475 Boulevard de Tessé

4412 -
Sept. 3, 1999 1 469 Rue d’Entrechaus

Sept. 17, 1999 1,171.5 Garage entrance
RÉF .
Nov. 24, 1999 1 391 Place Mazarin

2002 -
Dec. 7, 1999 1,221.5 Garage end

Jan. 18, 2000 1,254 Rue Chabannes
MARS - AVRIL
Mar. 13, 2000 1,291 F1-F2 junction 1 291 F1-F2 junction

C HAUSSÉES - 237 -
be noted that to the north of Toulon, displacements pursued a course in the opposite direction, i.e. from
north to south (the Mont-Faron structure). These movements date back to the end of the Lutetian Period.
Sliding along the faults oriented 25° north cut all of these fragments into strips. The gypsum was locally
dissolved, thereby creating not only voids but, above all, a subsidence crossed by the tunnel that wound
up causing numerous technical difficulties. Figure 4 displays the simplified geological section cut for the
ET
tunnel.
P ONTS
At the time of the design studies conducted in coordination with the University of Marseille, the magni-
tude of the Cape Sicié thrust was no longer recognized. Convergence of the facies of differently-aged
rocks led to a partially inaccurate geotechnical section and in particular to an underassessment of crack-
ing in the deposits. The absence of outcroppings within the urbanized area lies at the origin of this gap
in knowledge since it is difficult to determine the complexity of cracking by means of core sampling.
This observation underscores once again the necessity of not limiting the extent of geological investiga-
B ULLETIN
tions in the immediate vicinity of the studied structure. Within these heterogeneous formations at the
level of the samples, our means of investigation are insufficient to establish the scale as of which the
ground may be considered homogeneous, as regards mechanical behaviour. For this tunnel, the corre-
14 sponding scale is most likely on the order of ten meters, hence greater than the dimension of the actual
structure.
Dumont d’Urville
Central shaft
Entrechaus
Marchand
Western opening
Chabannes
Mazarin
Chalucet
Fabié
Vauban
Carnot
20
10
0
Altitude (m)
-10
-20
-30
-40
Quaternary Period cover layer (Würm)

Trias Epoch (conglomerate, limestone, gypsum, marls, clays)
Permian Epoch (sandstone, pelites)
Stephanian Epoch (quartz schists from prior coal-rich deposits)
Pre-Stephanian bedrock (quartzo-phyllades, arcose, quartzites)
Fig. 4 - Geological structure diagram of the Toulon sub-soil along the tunnel alignment
Uncertainties regarding site geology affected: the organization of the site investigations, the choice of
the northern tunnel construction methods, the implementation of such methods, the structural behaviour
predictions, and Toulon city center impact monitoring. The focus, during both project development and
construction in this complex geological medium, consisted of reconciling the difficulties inherent not
only in shallow tunnel construction, but especially in their construction on an urban site. From a geotech-
nical perspective, the project was confronted with a dual soil-structure interaction problem: the tunnel-
soil interaction and ensuing issues related to excavation, support and tunnel stability; and the ground-
5-36
surface buildings interaction, with primary emphasis on limiting deformations (and settlement, in partic-
PP .
ular).
4412 -
The prediction of the behaviour of the entire tunnel project, including ground and buildings, was thus
RÉF .
heavily emphasized during the design phase. For the northern tunnel, it must be pointed out that all geo-
technical predictive approaches remained, in part, fruitless even though some progress had been
2002 -
achieved in the area of surface settlement monitoring and prediction. The progress made in tunnel con-
struction methods does however merit recognition: these methods ultimately allowed completing the
MARS - AVRIL
northern tunnel excavation.
Oddly, from the standpoint of surface settlements, the largest deformations seem to appear within the C HAUSSÉES - 237 -
formations (in the most highly-tectonic zones) and not at the contact between formations (Carnot,
Vauban, Mazarin, Fabié-Marchand).
The geomechanical state of the ground at its various points constitutes an unknown. It seems that the
compactness-cracking process played a major role in the ground behaviour and was responsible for part
of the surface settlements generated by northern tunnel excavation. It has also been suggested to consider
ET
the role played by the mechanical anisotropy of fractured formations. The initial stress state however
P ONTS
remains an unknown.
At the time this article was published, few references were available in the literature on Toulon's nortern
tunnel excavation works. Some information has been published in the journal entitled "Tunnels et
ouvrages souterrains" ("Tunnels and Underground Structures") (Durand, 1991; Levêque, 1995) and a
paper was delivered at the AFTES international symposium (Dubois et al., 1999).
Measurement of surface settlements along the tunnel alignment

B ULLETIN
The monitoring system used for surface settlements is composed of a set of marks (nearly 750 in all)
affixed to buildings located within a bandwidth of about 100 m on both sides of the northern tunnel
alignment. Some marks were fastened into the ground, and additional ones were introduced over the 15
course of the project in order to complete the system locally. This set-up provides the basic elements for
tracking surface settlements. Measurements were performed level by level during the tunnel excavation
(from 1994 to 1996 and again between 1998 and 2000), within the boundaries corresponding to the
progression of tunnel faces on at least a daily basis. The surveying of each group of marks is conducted
along a specific plane-table traversing, the closing of which yields a quality index of the measurement
reading.
The analysis of these series of measures reveals a measurement noise along with very momentary
variations, which often prove difficult to explain for reasons other than fastening flaws or movements at
the point of monitoring system location. Measurement noise can be estimated at ± 0.7 mm on average.
Several marks exhibited more significant noise. Observations also included fluctuations in ground level,
apparently independent of tunnel construction, with an amplitude on the order of a few millimeters.
Figure 5 shows the position of survey marks within the alignment zones (at place Mazarin, PM1275-
1425), where faces F1 and F2 merge.
5-36
PP .
4412 -
RÉF .
2002 -
MARS - AVRIL
Fig. 5 - Position of the measurement marks within a zone on the alignment (place Mazarin)
C HAUSSÉES - 237 -
Use of measurement data

Surface settlements have been monitored closely. A first daily analysis focused on the evolution in
settlement of each mark with respect to time as well as to the tunnel face advance. This analysis included
examination of settlement speeds as a function of time and settlement gradients as a function of tunneling
ET
P ONTS
progress, an inventory of settlement jumps for a given group of marks (in relation to the traverse
followed during surveying), and differential settlements when several marks were available on a single
building. Various approaches were envisaged in an attempt to set speed limits or thresholds or to
determine trend curves that would allow predicting short-term surface settlements. The main difficulty
had to do with surface settlement variability along the alignment.
The monitoring campaign proceeded by investigating the evolution in each mark settlement as a function
of either time or tunneling progress in those regions most immediately concerned by the works. Yet
measurements may also be used for obtaining both transversal settlement troughs and longitudinal
B ULLETIN
settlement profiles as a function of tunneling progress. These approaches are faced, among other things,
with the measurement noise and ground heterogeneity problems. Therefore, a number of criticisms have
been directed at these methods:
16
¾ measurement noise is sizable and as its amplitude increases, predictions become less precise;
¾ concern has been raised over the representativeness of measurements conducted on marks affixed to
buildings, as opposed to those conducted on the benchmarks set into the soil; experience however has
shown that the responses are comparable;
¾ variability in behaviour of the rock mass, including the surface layer, complicates the use of the
measurements in terms of thresholds, ranges, speeds, etc.
A more general format for interpreting surface settlement measurements consists of drawing curves of
equal settlement contours and their evolution as a three-dimensional trough. "Total troughs" represent
the (total) settlement observed on a given date, whereas "instantaneous troughs" reflect the variation in
settlement over a given period, which corresponds to a certain length of advanced tunnel face.
Construction of the three-dimensional troughs, whether total or instantaneous, is performed as follows.
The surface marks form a network of points with known coordinates (x,y,z) and on which a nappe
composed of triangles may be based. This type of nappe may be depicted in a perspective view in the
form of a surface or be graduated into contour curves. The representation of three-dimensional troughs
is obtained by substituting for each mark either settlements s or settlement differences ∆s in the place of
z. Examples will be provided further below.
The representation of settlement troughs by means of a nappe of triangular facets exhibits a coarse
appearance by virtue of both irregularity in the position of surface leveling blocks and settlement
measurement noise. However, this representation and the accompanying linear interpolation in contour
curves serve to restore settlements measured on the surface with no other form of computation or
interpretation (just factual data).
To pursue interpretation, the total and instantaneous troughs (for a 15-m tunneling advance) are
superimposed onto the plane views, with: the tunnel projection at a 10-m increment until reaching the
current tunnel face, the segment of the last 15 m of excavation, the city blocks of buildings on the
surface, and the grid network (at a 50-m increment):
¾ total troughs are represented by colored surfaces, extending from blue to red at an increment of 2 mm;
¾ instantaneous troughs are represented by contour lines every 5 mm (red) and every millimeter
(yellow); the "0" line (zero settlements) is often poorly defined due to measurement noise.
5-36
The troughs were built by consecutive 15-m segments. Some of them (for a 15-m tunneling advance)
PP .
were established using a smaller increment in order to closely track each and every surface settlement
4412 -
evolution.
These three-dimensional troughs serve to complete the s(x,y,ηf) representation of surface settlements per
RÉF .
test mark; they enable clearly distinguishing the evolution in settlements with excavation progress,
2002 -
especially in front of the tunnel faces, and then revealing the influence of both the ground structure and
lateral and longitudinal heterogeneities, which are difficult to observe on either the longitudinal or cross-
MARS - AVRIL
sectional profiles. A comparison of the maximum amplitude of instantaneous troughs ∆s15 m with the
final settlement sfinal in those regions where settlements had been stabilized also makes it possible to
derive an empirical prediction method, which will be discussed below. C HAUSSÉES - 237 -
Two methods for predicting settlements as the tunnel face advances

Conventional surface settlement prediction methods are based on the surface trough shape within a
cross-sectional profile. The trough exhibits the outline of a Gaussian curve (see Equation 1). In reality,
ET
surface deformation mechanisms are expressed in three dimensions, as a function of tunnel face
P ONTS
progress. During construction of the northern tunnel, it proved useful to ascertain settlement predictions
based on observed settlements in front of the tunnel face. Two methods were developed as of 1998 (see
Appendix B): one relying on a three-dimensional formulation that allows describing both the transversal
trough and longitudinal trough (featuring an expression in front of the tunnel face); and a so-called
instantaneous trough method, according to which the settlement increment produced by each
incremental tunnel face advance is examined.
Transversal and longitudinal troughs: The longitudinal profile method

B ULLETIN
Figure 6 shows an example of a transversal settlement trough (Profile 2 of the Chalucet block, PM 889).
The points represent final settlement measured after passage of the tunnel face. The curve indicates the
settlement computed using the expression published by Loganathan and Poulos (1998), in which the only 17
free parameter is the settlement amplitude at the center of the trough (i.e. 28.5 mm in this example).
Horizontal distance to the axis (m)
-50 -30 -10 10 30 50
0
-5 Chalucet Profile 2 PM 889
-10
-15
-20 Computed settlement
-25
-30
Measured settlement Fig. 6 - Settlements observed in the plane

-35 transversal to the tunnel (Profile 2 of the
Settlement (mm) Chalucet block)
Figure 7 presents the settlement of a block positioned at PM 862 vs. advance of tunnel face F1. The curve
identifies the longitudinal profile of the settlement, with the tunnel progressing towards the left. The
vertical arrow denotes the passage of the tunnel face at the level of the measurement mark. Data are
generated progressively as the tunnel face advances, beginning on the left-hand side of the graph. The
method consists of calibrating the model on the initial settlement measurements, at the fringe of the
longitudinal trough, and then predicting final settlement on the basis of asymptote amplitude. The model
can be enhanced over time as new measurement data are input, with predictions becoming increasingly
precise.
5-36
0
PP .
4412 -
Model s = s(x, y, ηf)

Measurements
RÉF .
Fringe of the
2002 -
longitudinal basin
-5
Settlement (mm)
MARS - AVRIL
Passage of tunnel face at PM 862

C HAUSSÉES - 237 -
-10
EC1
Marks Excavation direction
EC1 c
-15
ET
P ONTS
800 825 850 875 900 925 950

Tunneling advance (m)
Fig. 7 - Settlements caused at PM 862 by the advance in tunnel excavation
This method was implemented between 1998 and 2000 for tunnel faces F1 and F2; afterwards, it was
extended to all data available for the entire project dating back to 1995, i.e. to some 700 marks set up on
the surface, whether they be near the northern tunnel axis or laterally offset. The method was applied on
a weekly basis for the marks raised within the perimeter of influence of faces F1 and F2.
B ULLETIN
Figure 8 provides an example of settlement curves for four blocks vs. both time and tunnel face advance
(blocks EC1 to EC3 at PM 862 and block EC4 at PM 888). The curve of tunnel face F1 advance is shown
18 on Figure 8a, which combines excavation cycles and downtime periods (curve entitled "Tunnel face F1
advance"), measured settlements (points with different symbols), and the model (drawn in a solid line).
Tunnel face position (PM)
Settlement (mm)
1000
0
Measurements
-5
950
Model results
-10
EC1
900
-15
EC2
-20
EC3
850
-25 Advance of tunnel face F1 EC4
-30 800
04/06/1998 24/07/1998 12/09/1998 01/11/1998 21/12/1998

Date
8a - Advance of the tunnel face and evolution in settlements vs. time
-5 Measurements
5-36
Model results
PP .
-10
Settlement (mm)
4412 -
EC1
-15
RÉF .
EC2
2002 -
-20 Passage of tunnel face at PM 862 EC3
MARS - AVRIL
-25
Passage of tunnel face at PM 888 EC4
-30 C HAUSSÉES - 237 -

800 825 850 875 900 925 950 975
Tunneling advance (m)
8b - Evolution in settlement as the tunnel advances

ET
Fig. 8 - Settlements of four survey marks (EC1-EC4) as a function of both time and tunnel face advance
P ONTS
The same curves are presented in Figure 8b as a function of tunneling progress. The difficulty of model
calibration in certain cases (e.g. during the settlement jump that appeared in Block EC2 measurements
around September 12, 1998) can be easily observed.
B ULLETIN
Figure 9 depicts the settlements of five marks aligned on the tunnel axis spaced roughly twenty meters
apart. Final settlement is not the same at every point of the longitudinal profile due to ground
heterogeneity. The tunneling advance graph clearly indicates this heterogeneity, with final amplitude
varying by up to 100% over a distance of 80 m (see Fig. 9b). All in all, the model is quite well calibrated 19
on the settlement measurements.
1100
0 Measurements
Model results
-10 1050
Tunneling advance F1 (PM)

Settlement (mm)
-20 3252
1000
3253
-30 Avancement
3254
950
-40 3256
3255
-50 900
01/11/98 01/12/98 31/12/98 30/01/99 01/03/99 31/03/99 30/04/99

Date
9a - Advance of the tunnel face and evolution in settlements vs. time
0
Measurements
Model results
5-36
-10
PP .
4412 -
Passage of tunnel face

Settlement (mm)
at PM 939 (3252)
-20 3252
RÉF .

2002 -
3253
at PM 958 (3253)
-30
MARS - AVRIL

at PM 975 (3254) 3254
-40
C HAUSSÉES - 237 -
at PM 996 (3255) 3256

at PM 1025 (3256) 3255
-50
900 950 1000 1050 1100

Advance of tunnel face F1 (PM)
ET
P ONTS
9b - Evolution in settlement as the tunnel advances

Fig. 9 - Settlements of survey marks aligned on the tunnel axis
Figure 10 compares measured final settlement with the model prediction s(x,y,ηf) derived on February
26, 1999 when tunnel face F1 was located at PM 1000.5.
B ULLETIN
The instantaneous trough method
20 On the perspective drawing shown in Figure 11, the most recent segment of excavated tunnel
immediately in back of the tunnel face is considered and represented in light green. For the northern
0 Instantaneous trough
Model predictions
Minimum settlement Total trough
Maximum settlement
-25
Settlement (mm)
-50 Final settlement
Position of tunnel face F1

PM 1000.5
on February 26, 1999
-75
950 975 1000 1025 1050
PM (m) Most recently-excavated segment
Fig. 10 - Comparison of predicted and measured Fig. 11 - Instantaneous trough for a 15-meter advance
settlements in tunneling excavation
tunnel, a segment length of 15 meters was chosen. The green gridlines represent the total surface trough
for the current tunnel face position. The instantaneous trough obtained during excavation of the 15-m
segment is displayed in the contour curves superimposed on this three-dimensional view. This trough is
closed in back, as opposed to the total settlement trough. The sum of all instantaneous troughs along the
tunnel alignment yields the total trough.
It turns out that the amplitude (or surface area, or volume) of the instantaneous trough is just about
proportional to the final settlement. In knowing the amplitude of the instantaneous trough for the last
increment of the excavated tunnel, it is possible to predict the final settlement; this step entails a certain
experience with the model (specific on-site calibration). Furthermore, it is possible to predict
heterogeneity on a longitudinal profile. If the trough nadir is located ahead of the tunnel face, the tunnel
5-36
is heading towards a softer, and hence more dangerous, ground zone. In contrast, if the instantaneous
PP .
trough nadir is behind the tunnel face, then excavation is heading towards a harder ground zone, less
4412 -
deformable and thus less of a concern.
The instantaneous trough method was introduced over 1999-2000 for the two faces F1 and F2 and in
RÉF .
particular at the junction of the two faces in March 2000.
2002 -
Figure 12a shows a total trough during tunnel advance. On this perspective drawing, the surface city
blocks of buildings are visible in brown along with the surface projection of the tunnel (using a 10-m
MARS - AVRIL
increment, graduated scale of green), with the tunnel face on the considered date. The total trough, open
onto the backside, is graduated in color, ranging from blue to red.
C HAUSSÉES - 237 -
On Figure 12b for this same view, the total trough has been represented in gray. The most recently-
excavated segment is identified by the green rectangle projected onto the surface. The instantaneous
trough is depicted by its contour curves graduated every 5 mm in red and every millimeter in yellow.
Within this part of the tunnel alignment, the trough is located ahead of the tunnel face, which is
associated with the maximum observed settlement.
ET
P ONTS
Settlement prediction
To summarize the method applied to the Toulon site, the amplitude s15 of the instantaneous trough in
front of the tunnel face determines, by means of correlation, the final settlement sfinal. The shape of the
instantaneous trough enables predicting the incidence of ground heterogeneities in front of the tunnel
face.
Figure 13 exhibits the correlation that ties the amplitude of the instantaneous trough, for a 15-m advance,
to the final settlement measured more than 150 m behind the tunnel face. Each point represents a 15-m
B ULLETIN
segment of the northern tunnel. The final settlement appears even larger as instantaneous settlement
increases. Should the instantaneous trough exceed 15 mm, a final settlement of over 50 mm would have
to be expected. On the other hand, for an instantaneous settlement of 5 mm, the final settlement will not
surpass 20 mm. The red line, with a slope of 2.5, lends an idea of the ratio of instantaneous settlement to 21
final settlement.
Total settlement from 0 to 50 mm
(curve spacing of 2 mm)
Scale
100 m
Îlots de bâtiments
City blocks of buildings
12a - Total trough
Instantaneous trough:
settlement from 0 to 14 mm
(curve spacing of 1 mm)
Scale
100 m
5-36
PP .
4412 -
RÉF .
2002 -
MARS - AVRIL
City blocks of buildings
12b - Instantaneous trough

C HAUSSÉES - 237 -
Fig. 12 - Instantaneous and total troughs within a section of the alignment (PM 1015.5)
This correlation has been refined over time, with model experience being acquired in real time for the
ET
P ONTS
segments further away from the tunnel faces.

The figures are defined with respect to the types of ground encountered: no direct correlation would
appear between ground types crossed by the tunnel and final settlement.
Figure 14 presents the prediction result and compares the final settlement measured along the northern
tunnel using the longitudinal profile, as represented by filled circles for the period 1998-2000 and empty
circles for the period prior to 1996 (except for the March 1996 collapse zone).
The longitudinal profile of the final settlement appears in the form of a succession of dips and peaks, in
B ULLETIN
relation to ground heterogeneity. The prediction made upon tunneling advances was framed by both an
upper limit, as represented by empty squares, and a lower limit marked by the filled squares. The predic-
tion proved to be accurate during the period when the instantaneous trough method had been imple-
22 mented, especially at the junction of faces F1 and F2, where the two faces act simultaneously on the sur-
face settlements.
PM 550 800 Bedrock 1
0
PM 800 940 Permian
PM 940 1030 Permian Gypsum
-20
PM 1030 1150 Permian
PM 1150 1310 Schists
Final settlement (mm)
-40
PM 1310 1550 Bedrock 2
-60 PM 1550 1600 Fabié Schists
PM 1600 1760 Fabié Permian
-80 PM 1760 1950 Trias
Slope 2.5
-100
-120
-140 Fig. 13 - Comparison of

-60 -50 -40 -30 -20 -10 0 "instantaneous" and final measured
Settlement for a 15-m tunnel advance (mm) settlements
0
1998 - 2000
F1 F2
15
Chalucet
Settlement (mm)
de Tessé
Fabié - Marchand
30
Garage
5-36
Mazarin
Junction
PM 1291
PP .
45
4412 -
Carnot
Lower prediction limit

Vauban
Upper prediction limit
RÉF .
Final settlement 1995-1996
2002 -
Final settlement 1998-2000
60
400 600 800 1000 1200 1400 1600 1800 2000
MARS - AVRIL
PM positions of tunnel faces F1 and F2 (m)
Fig. 14 - Measured and computed final settlement along the tunnel alignment
C HAUSSÉES - 237 -
Final settlement troughs

The final settlement was represented in the form of isovalue maps, which clearly reveal the dips and
peaks of the final settlement trough along the northern tunnel. These outlines coincide well with the
ground structure encountered underground.
ET
P ONTS
For the purpose of example, Figure 15 shows the final settlement map west of the northern tunnel. The
equal settlement curves are depicted in contour curves spaced every 10 mm in red and every 2 mm in
yellow. The presence of two special troughs may be observed (Carnot and Vauban) along with a peak
(Chalucet). The grid network is spaced at a 50-m increment, which demonstrates that the lateral exten-
sion in some areas exceeds 50 m. The Carnot trough perfectly coincides with the orientation of a fault
structure encountered during tunnel excavation in 1996. The Vauban trough turns out to be the deepest
of the entire alignment, with a final settlement of 50 mm (not including the 1996 collapse zone).
The central zone (see Fig. 16) is delimited by the Mazarin trough, where F1 and F2 merged in March
B ULLETIN
2000, as well as by the Dumont d'Urville peak, where a 140-m2 section garage was built. This finding
proves the lack of any obvious correlation between underground and surface.
The eastern zone map (see Fig. 17) only indicates the settlements produced since resumption of works 23
in 1998 and neglects the 1996 collapse between the Marchand and Fabié bridges.
Chalucet Hospital Rue Chalucet
10
10
40
F1 20 20 10
30 10 30
Western opening
20 40
10 20 30
10
Avenue Carnot
Avenue Vauban
Origin 891800 98200
100 m Tunnel graduated every 25 m Settlements: every 2 mm
Fig. 15 - Final settlements on the northern tunnel alignment between the Western opening and avenue
Vauban
Avenue Vauban Rue Dumont d’Urville
10 Rue Chabannes
Place Mazarin
40 30 20
10
20 20
10
10 10
20
20 10
5-36
100 m 10
PP .
4412 -
Tunnel graduated every 25 m

Settlements: every 2 mm
Avenue Colbert
RÉF .
2002 -
Rue d’Entrechaus
Fig. 16 - Final settlements on the northern tunnel alignment between avenue Vauban and rue d’Entrechaus
MARS - AVRIL
C HAUSSÉES - 237 -
20
Fabié Bridge
10
20 Marchand Bridge
ET
10
10
P ONTS
10
Rue d’Entrechaus
Rue Truguet
Rue Fabié
100 m
Puits central
F2
B ULLETIN
Settlements: every 2 mm
Tunnel graduated every 25 m
The settlements indicated are those subsequent to the 1998 resumption of works.
24
Fig. 17 - Final settlements on the northern tunnel alignment between rue d’Entrechaus and Marchand Bridge
Conclusion
The analysis of settlements observed in the city of Toulon during excavation of the underground cross-
ing's northern tunnel, between 1998 and 2000, was conducted by making use of two surface settlement
prediction methods for an advancing tunnel face. These methods, derived as the project progressed, are:
the longitudinal profile method, and the instantaneous trough method.
The development of both these methods has relied on an extensive bibliographical search focused on
methods for computing such settlements. This bibliographical review has exposed the influence of the
various factors contributing to settlement amplitude and geometry, within the scope of works carried out
in urban areas. Among these factors, special emphasis must be put on the role of the geometry of the
structure (its diameter and depth), which at the beginning was considered as primarily determinant of the
shape of both surface settlement transversal troughs and their associated differential settlements. In
reality, the transmission of deformations from underground towards the surface also depends upon
ground behaviour, ground heterogeneity, the initial ground state, water effects and deferred effects.
These factors specific to the ground are obviously accompanied not only by impacts due to the choice of
tunnel construction method, construction quality and speed of execution (normal speed given the ground
behaviour - consolidation, creep, progressive deformations, etc.), but also by the capacity to adapt to
difficulties ultimately encountered along the project alignment (heterogeneities).
The diversity of the formations encountered within the Trias, Permian and ante-Permian substrata by the
Toulon northern tunnel as well as their heterogeneity had been acknowledged as of the first geological
and geotechnical site investigations. This particularity of the ground was then confirmed during tunnel
excavation. The ground is heterogeneous at all scales, by virtue of the juxtaposition of the formations
crossed in reverse order, except to the east of the tunnel alignment. In these rocky or soft rock materials,
followed by highly-tectonic materials, the level of heterogeneity is observed inside the formations
themselves, with the presence of zones of faults and breccia, outlier, tamping clusters and dense
cracking. From a geotechnical perspective, the dispersion in soil properties is equal for all deposited
formations.
5-36
The two methods developed for predicting surface settlements of an advancing tunnel face are three-
dimensional and incorporate the main characteristics of surface settlements (amplitude, slope,
PP .
differential settlements, transient state during the advance, etc.). Research is currently underway to
4412 -
specify the theoretical bases for these methods and correlate the settlements observed with the properties
RÉF .
of soils found on-site. Geotechnical concerns dictate the necessity of being able to perform this type of
prediction within a project framework, on the basis of geological and geotechnical reconnaissance of the
2002 -
rock mass.
MARS - AVRIL
This study also reminds that soil heterogeneities along the tunnel alignment heighten difficulties during
tunnel construction at shallow depths in urban areas. The methods that enable reacting and adapting the
most quickly are consequently highly attractive. In order to anticipate eventual difficulties, construction C HAUSSÉES - 237 -
operations (which are to remain based on extensive knowledge of both the ground formations and
implemented techniques) only make use of rather limited warning indices. Against this backdrop, the
monitoring of underground and surface deformations proves to be of vital importance. On the surface,
the tracking of settlement dynamics on a longitudinal profile (or even better in three dimensions) during
tunnel face progression provides a particularly insightful means of observation.
ET
From a very general standpoint, a rather direct relationship has appeared between surface settlements and
P ONTS
ground structure, as represented by tectonic zones located at the core of the various formations. The
nature of the interspersed rocks and the tunnel construction methods employed would seem to be of
lesser importance. This intentionally global view implies the hypothesis of strong influence being
exerted by the initial geomechanical state of the rock mass; in practice however, the characterization of
this state, for each homogeneous tunnel segment, is not accessible.
B ULLETIN
Acknowledgments. The monitoring of measured settlements during construction of

the Toulon underground crossing's northern tunnel was carried out on behalf of the
Var Department Public Works Office, this project's sponsor and architect. The authors
are grateful to the Department for having authorized publication of the results 25
obtained.
APPENDIX A
Settlement computation formulae
This appendix presents the methods for computing soil surface settlements above a tunnel under construction;
these methods have been published by:
¾ Sagaseta (1987),
¾ Verruijt and Booker (1996) and Verruijt (1997),
¾ Loganathan and Poulos (1998).
Also discussed will be the work conducted during the design phase of the Toulon underground crossing's
northern tunnel (Serratrice, 2001).
Sagaseta's solution (1987)

In order to determine the displacement field within an isotropic, homogeneous and incompressible semi-infi-
nite elastic medium, Sagaseta (1987) proposed a solution based on the effects of a loss (respectively, an
increase) in soil mass volume caused at a single point by the excavation of a shallow tunnel or by the extraction
of sheet piles (respectively, pile driving or solid injection). This solution enables evaluating the components of
the three-dimensional surface displacement field:
VL 1
s x = ------- ----
2π ω
V
s y = – ------L- -----------------  1 + ----
y y (A.1)
2π y 2 + z 2  ω
o
5-36
PP .
2
where ω = x 2 + y 2 + z o and zo is the tunnel axis depth. The origin of the coordinate system is located on
4412 -
the surface in alignment with the tunnel face and (x,y,z) are the distances in the longitudinal, transversal and
vertical directions (heading downwards), respectively (see Fig. 1). The displacements sx, sy and sz are oriented
RÉF .
in the direction of the axes, either towards the front of the tunnel, namely towards the exterior from the axis or
2002 -
towards the bottom. VL represents the underground volume loss per unit length, which is assumed to be con-
centrated on the tunnel axis. This solution allows generating the three-dimensional maps of surface settle-
MARS - AVRIL
ments, horizontal displacements and, by means of derivation, slopes.

At a large distance behind the tunnel face (x → -∞), the displacement field is planar, i.e.:
sx ( x = –∞ ) = 0
C HAUSSÉES - 237 -
VL y
s y ( x = – ∞ ) = – ------- ----------------- (A.2)
π y 2 + z2
o
VL zo
ET
s z ( x = – ∞ ) = ------- -----------------
P ONTS
π y2 + z2
o
The maximum horizontal displacement (at a distance h from the axis), the position of the settlement trough
inflection point and the maximum settlement on the axis are given by:
max VL
sy = s y ( x = – ∞ , y = h ) = – ---------
-
2πh
h i h
i = ------- ≈ 0,56h ---- ≈ 1,12 ----
B ULLETIN
(A.3)
3 R D
max V-
so = sy = s z ( x = – ∞ , y = h ) = -----
26 πh
Solutions found by Verruijt and Booker (1996) and by Verruijt (1997)
Verruijt and Booker (1996) superimposed a convergence mechanism and an ovalization mechanism for the wall
of a circular tunnel bored into a semi-infinite elastic soil mass. The relative radial and uniform displacement is
characterized by:
ε = ∆R/R 2ε = ∆V/Vo (A.4)
where ∆R is the radial displacement, R the tunnel radius, Vo the unit volume and ∆V the loss in unit volume
(with ∆V/Vo being the relative volume loss). Ovalization is then characterized by:
δ = ∆Rv/∆Rh (A.5)
i.e. the ratios of displacement to vertical tunnel diameter and displacement to horizontal tunnel diameter. This
ovalization may result from a special state of the initial stresses or from any other reason, such as the presence
of plastic deformations above the tunnel, thereby causing larger crown displacements, or the installation of a
non-homogeneous support (e.g. difference introduced between the sides and crown).
The tunnel axis is located at a depth h with respect to the surface. The approximated solution proposed by Ver-
ruijt and Booker (1996) can be expressed as follows, in terms of displacements and within the Oxz coordinate
system, where z is vertical descendant:
(A.6a)
5-36
(A.6b)
PP .
4412 -
with:
RÉF .
2 2
z1 = z – h z2 = z + h r 1 = y2 + z 1 2 r 2 = y 2 + z22
2002 -
m = 1 / (1-2ν) k = ν / (1-ν)
MARS - AVRIL
and ν is the Poisson's ratio. The trough-shaped surface settlement is obtained from the above expressions tak-
ing z = 0:
C HAUSSÉES - 237 -
(A.7)
Integrating this relation from -∞ to +∞ yields the settlement trough volume:

ET
P ONTS
(A.8)
For ν = 0.5 (incompressible soil mass), Sagaseta’s formula (1987) is obtained:

A = 2 επR2 (A.9)
The two terms in the expression of uz(y, z = 0) represent the convergence effect controlled by parameter ε and
the ovalization effect controlled by parameter δ, respectively. The first term has the shape of a flared bell curve,
whereas the second is more tightly compressed around the origin. This second term allows approximating the
B ULLETIN
shape of the settlement troughs actually observed. Parameters ε and δ represent the underground volume
losses. The benefit of this approach in fact is to consider the parameters as model input parameters and to
directly correlate underground deformations with surface deformations (Sagaseta, 1998). Its drawback how-
ever is to overweight the hypothesis of soil mass homogeneity. Another advantage lies in the fact that the solu- 27
tion is given for all values of Poisson's ratio ν and not just for an incompressible material (ν = 0.5). It should
be pointed out that Poisson's ratio ν appears in the formula, but Young's modulus E does not. The Young's
modulus is already incorporated into parameters ε and δ. Stress components may be drawn from this displace-
ment field (Verruijt and Booker, 1996).
Sagaseta (1998) provides the expressions of ε and δ in elasticity:
(A.10a)
(A.10b)
with:
G: shear modulus G = E / 2(1+ν), E: Young's modulus,
ν: Poisson's ratio,
Ko: coefficient of earth pressure at rest,
po: initial vertical stress: σvo = po, σho = Ko po.
These relations highlight the physical significance of the two parameters ε and δ with respect to both elasticity
parameters and initial stress state.
These analytical solutions are obtained for a linear elastic material and a shallow circular tunnel in plane defor-
mation within a homogeneous soil mass. Verruijt and Booker (1996) indicated that such solutions only repre-
sent a crude schematic of actual soil and rock behaviour, yet still enabled analyzing the characteristics of the
stress and strain fields within the soil mass and on the surface.
Loganathan and Poulos' solution (1998)

5-36
PP .
Rowe and Kack (1983) defined the ring void parameter as the equivalent two-dimensional representation of a
void formed around the tunnel, which combines the effects of:
4412 -
¾ elastoplastic and three-dimensional deformations of the tunnel face,

RÉF .
¾ over-excavation of the soil mass within the tunnel project site due to convergence in front of the tunnel face
(volume loss),
2002 -
¾ the void left between the ground and the tunnel lining, and in particular that created by tunneling machines.
MARS - AVRIL
Estimating the ring void entails summing three terms (Lee et al., 1992). The first term represents the actual
opening, i.e. the difference between the apron diameter and the outer lining diameter in the case of excavation
using a tunneling machine. The second term is introduced in order to take the elastoplastic deformations of the
tunnel face into account; this term serves to measure three-dimensional effects at the tunnel face. The last term
C HAUSSÉES - 237 -
allows including the quality of the project execution phase.

Loganathan and Poulos (1998) proposed a novel approach by virtue of defining volume losses (under und-
rained soil deformation conditions) on the basis of a ring void parameter that describes the oval shape of tunnel
wall convergence. This notion extends the initial notion proposed by Lo and Rowe (1982), Rowe and Kack
(1983) and Lee et al. (1992); it relies primarily on the concept of soil influx generated by the difference between
ET
excavated volume and volume occupied by the structure once the lining has been installed (Leca, 1997). This
P ONTS
approach would be applicable to shield-tunneling projects. The opening left between the shield apron and the
lining ring extrados closes in back of the tunneling machine and induces deformations within the enclosing soil
mass.
Loganathan and Poulos (1998) defined underground volume loss as a function of the ring void parameter g by
the following:
π ( R + g ⁄ 2 ) 2 – πR 2 4gR + g 2
ε o = v L = ----------------------------------------------- = ---------------------- (A.11)
πR 2 4R 2
B ULLETIN
In practice, the radial displacement at the wall surrounding the tunnel is not uniform and instead displays an
oval shape, which depends on the type of construction, the support settlement and the elastoplastic deforma-
28 tions within the soil mass. The vertical displacement of the wedge is larger than that of the raft (for a circular
tunnel, except in the case of swelling). In order to account for this particular mode of non-uniform deformation
both around the tunnel and within the soil mass, Loganathan and Poulos (1998) proposed weighting the
expression of εo by an exponential function:
ε yz = ε o C exp ( – Ay 2 – Bz 2 ) (A.12)
Constants A, B and C are deduced from assessments regarding the form of both the surface and sidewise dis-
placements. The authors ultimately derived the following expression:
4gR + g 2 1,38y 2 0,69z 2

ε yz = ---------------------- exp – -------------------- – ---------------- (A.13)
4R 2 ( h + R )2 h2
Put otherwise, εyz may be considered as the contribution of the displacement of point (y,z) to the tunnel wall
volume loss.
The non-linearity introduced by Loganathan and Poulos (1998) only applies to the convergence term and not
to the ovalization term proposed by Verruijt and Booker (1996). The computation is then carried out by replac-
ing ε with the expression of εyz (18) in the set of equations in Verruijt and Booker (1996) and by setting δ= 0.
Loganathan and Poulos (1998) proposed the following solution for settlements along a cross-sectional profile
at a given spot height z:
(A.14)
More specifically, the shape of the surface settlement trough is given by the relation:
(A.15)
5-36
PP .
4412 -
This expression ascribes a narrow shape to the surface settlement trough, without requiring an ovalization
term. Along a vertical line (the lateral inclinometric tube), horizontal displacements are expressed as follows:
RÉF .
2002 -
(A.16)
MARS - AVRIL
Moreover, the method may be used only for instantaneous deformations (and not for consolidation or creep). C HAUSSÉES - 237 -
Loganathan and Poulos (1998) applied this method to four cases of instrumented tunnels, for which data was
available in the literature. The proposed analytical solution provides a powerful tool for predicting vertical and
horizontal displacements both within the soil mass and on the surface. The only soil parameter needed to be
known is Poisson's ratio ν, which may be set equal to 0.5 in the case of a clayey mass under undrained condi-
tions. The influence of soil type along with the three-dimensional effects of excavation have been integrated into
the method by means of the ring void parameter g. Among the variety of cases studied, the method yields
ET
poorer results for soil masses composed of heterogeneous layers and for sand-filled masses. In the case of
P ONTS
heterogeneous profiles, an equivalent parameter g must be evaluated.

These relations have been used to interpret data from both the Chalucet and Dumont-d’Urville blocks: vertical
displacements either at the surface or belowground measured using extensometers, and displacements mea-
sured along a vertical inclinometric tube (Serratrice, 1999).
Finite element numerical computations

B ULLETIN
Numerical computations using the finite element method have been employed by several authors in order to
complete the analytical developments or establish direct relations between soil properties, tunnel geometrical
characteristics (depth and diameter) and surface settlements (Sagaseta and Oteo, 1974; Oteo and Moya, 1979;
Sagaseta, 1987; Oteo, 1993; Verruijt, 1997). 29
Sagaseta and Oteo (1974) deduced the following relations from both finite element and linear elastic computa-
tions:
max ( 0,85 ( – ν ) )γD 2

sy = s o = ------------------------------------- (A.17)
E
max γ D 2 ( 0,27 ( – 0,3ν ) ) max

sy = ------------------------------------------------ ≈ 0,3s z (A.18)
E
max max
where s z is the maximum settlement (on the axis), s y the maximum horizontal displacement (at the
trough inflection point), D the tunnel diameter, γ the volumic weight of the ground, E the Young's modulus of
the ground and ν its Poisson's ratio. In these computations, the influence of tunnel depth h is very slight.
Along the same lines, parametric computations were carried out in order to correlate the computed surface set-
tlements with tunnel geometrical characteristics and soil mechanics properties (Serratrice, 2001). This
approach is limited to the case of a circular-section horizontal tunnel excavated within an isotropic and homo-
geneous semi-infinite elastic mass, considered in plane deformation (two-dimensional computation). The fol-
lowing parameters have been included in the formulation: tunnel axis depth h, tunnel radius R, weight density
γ of the soil, Young's modulus E of the soil, Poisson's ratio ν of the soil, the coefficient of earth pressure at rest
Ko, and the confinement-removal rate λ. The finite element computations are conducted using the CESAR-LCPC
computational software, and in particular the LINE (elasticity) and MCNL (elastoplasticity) modules (Mestat,
1994). In the case of elastoplastic computations, the level of cohesion assigned to the soil is such that soil mass
behavior remains within the elastic domain. The MCNL module thereby enables introducing confinement loss
via the coefficient λ.
The parametric study has focused on various radius-depth (R,h) couples. Mesh dimensions were chosen in
order to maintain a constant width-to-height ratio. A sizable soil thickness was left underneath the raft.
Table A.1 lists the set of selected (R,h) couples.
5-36
TABLE A.1
The (R,h) couples used in the parametric study
PP .
4412 -
R (m) h (m)
RÉF .
2 10 20
2002 -
5 10 15 25 45 100
10 20 30 50
MARS - AVRIL
The values of the soil mechanics parameters are provided in Table A.2.
C HAUSSÉES - 237 -
TABLE A.2
Computation parameters used for the parametric study
Parameter Values used in combination

ET
γ (kN/m3)
P ONTS
16 20 24
E (MPa) 10 20 50 100
ν (-) 0 0.2 0.499
Ko (-) 0 0.5 2
l 0.2 0.5 1
B ULLETIN
Boundary conditions of the computation domain are as follows:

¾ zero vertical displacement on the horizontal boundary over the entire depth,
30 ¾ zero horizontal displacement on the vertical boundaries on the sides.
The primary difficulty encountered pertains to the shape of the transversal settlement trough computed at the
surface. In some geometrical configurations, a rising of the ground would appear at a lateral distance of approx-
imately one diameter from the tunnel axis. This rising will be denoted sb. Also appearing in these configurations
are nonzero lateral settlements "out to infinity", from a lateral perspective, which will be denoted si. As far as
this rising is concerned, such an anomaly (qualified as the "bubble" effect) becomes more significant as the soil
mass thickness under the raft increases; it reflects the limits of linear elasticity for this type of study.
The computed values of so, sb and si were used in the aim of identifying a relationship between so and the var-
ious geometrical parameters (R,h,z), elastic parameters (E,ν) and state parameters (γ,Ko,λ). For so, these com-
putations led to the following approximate relation:
(A.19)
Settlement on the axis decreases whenever both E and ν rise. Settlement increases with γ, λ and R2 (tunnel
cross-section). The influence of the h/R ratio is combined with that of Ko. For higher values of Ko, the ground
rises above the tunnel axis. As Ko decreases, settlement increases. The sensitivity of so to Ko grows with tunnel
diameter. The value of Ko yielding a zero settlement increases with tunnel depth.
Berest et al. (2000) set out to compare long-term ground surface subsidence above a tunnel with instantaneous
subsidence at the time of excavation over the linear elasticity range. Computations are performed using finite
elements on meshes with different geometries, with realistic mechanical characteristics for the soil. When the
base of the mesh is located at a great depth, the deformable layer included between the tunnel and the mesh
base deforms upon exposure to the effect of ground loss caused by underground excavation, which in turn
causes the soil to rise at the surface. This surface swelling becomes predominant with respect to the excava-
tion- induced settlement yet remains imperceptible provided the non-deformable limit of the mesh lies near the
tunnel raft. The authors conclude that on the basis of numerical results obtained on the realistic set of examples,
5-36
the eventual detection of major surface displacements would most likely reflect irreversible deformations (onset
of plastification or the effect of soil behaviour non-linearities).
PP .
4412 -
RÉF .
2002 -
MARS - AVRIL
C HAUSSÉES - 237 -
ET
P ONTS
B ULLETIN
31
APPENDIX B
Methods for the early prediction of surface settlements
The longitudinal profile method

Developing a method for the early prediction of surface settlements could not be easily envisaged on the basis
of transversal troughs alone. It seemed important to take advantage of the observation points layout, which
depends on the tunnel alignment, in order to develop a longitudinal profile method. To accomplish this, it was
necessary to choose an a priori shape of the longitudinal profile of settlements above the tunnel. The selected
approach (Serratrice, 1999) consisted of extending the two-dimensional and transversal approach derived by
Loganathan and Poulos (1998) to three dimensions. A multiplier term, analogous to that appearing in the two-
dimensional approach, was introduced to represent the longitudinal profile of surface settlements as a function
of tunneling advance. Dubois and Jassionnesse (1997) also proposed an analytical method based on the lon-
gitudinal profile of surface settlements. Pursuing the orientation adopted by Sagaseta (1987), they postulated
that at a given point on the surface, the portion of settlement induced by the excavation of an underground seg-
ment (source) is proportional to the deformations caused within the segment and, conversely, proportional to
the square of the distance separating this source from the particular point. Total surface settlement is thereby
expressed as the sum of contributions from all excavated segments along the tunnel, thereby generating an
expression of settlement as a function of tunneling advance (which may be calibrated on experimental data).
The circular tunnel axis of radius R is located at depth h. Tunnel excavation at coordinate point (X,Y,Z) in the
soil mass produces a settlement s. A system of local coordinates (x,y,Z), connected to the tunnel face plane, is
used: x represents the abscissa in the excavation direction beginning at the tunnel face, and y the transversal
horizontal distance of the point with respect to the tunnel axis. Depth Z is measured from the natural ground
surface. The tunnel face is identified by its curvilinear abscissa ηf along the project alignment. Similarly, the
point (X,Y,Z) may be identified by its curvilinear abscissa ηi and distance y.
It is acknowledged that the surface settlement is equal to the product of two terms representing the transversal
trough shape fy(y) and the longitudinal trough shape fx(x,), respectively:
5-36
(B.1)
PP .
4412 -
Tunnel axis
x0
RÉF .
x
Gaussian
2002 -
curve
MARS - AVRIL
Tunnel face plane 0

y
Gaussian
curve
C HAUSSÉES - 237 -
Fig. B.1 - Transversal and longitudinal trough shapes for the computation
ET
P ONTS
In the transversal direction, the shape of the settlement trough follows a Gaussian curve centered on the tunnel
axis and modified by Loganathan and Poulos (1998), as indicated in Appendix A:
4cR 2 h by 2 
f y ( y ) = ----------------- exp  – -------------------- (B.2)
2
h +y 2  ( h + R ) 2
where b and c are two parameters. Parameter c corresponds, using the notation adopted by Loganathan and
Poulos (1998), to the expression:
( 1 – ν ) ( 4gR + g 2 )
B ULLETIN
c = --------------------------------------------- (B.3)
4R 2
where g is the ring void parameter (see Equations A.11 and A.15). As stipulated by Loganathan and Poulos
32 (1998), parameter b is considered fixed: b = 1.38.
According to the longitudinal axis of the tunnel, the shape of the settlement is an inverted semi-Gaussian curve
centered on the extremity of the longitudinal trough at abscissa point xo, with the following equation:
h2 a ( x – xo )2
x ≤ xo f x ( x, η face ) = 1 – ---------------------------------- exp – ------------------------
- (B.4a)
h2 + ( x – xo )2 ( h – R )2
x ≥ xo f x ( x, η face ) = 0 (B.4b)
The two parameters of this equation are a and xo. The choice of a Gaussian curve centered on the extremity of
the longitudinal trough rather than on the vertical line of the tunnel face was made in order to distance the inflec-
tion point from the tunnel face and better reproduce the experimental curves.
The function presented in (B.1) is an even function of y. The transversal trough is thus centered on the tunnel
axis and symmetrical. On the axis, f y ( y = 0 ) = 4R 2 c ⁄ h . Far from the axis, for very large values of y,
f y ( y → ∞ ) = 0. To take an eventual dissymmetry of the trough into account, additional parameters may be
introduced, yet such has not been the case within the scope of the Toulon study.
The expression for settlement winds up being, for x < xo:
4cR 2 h  h2 a ( x – xo )2  by 2
s ( x, y, η face ) = s o + ----------------------  1 – ---------------------------------- exp – ------------------------
-  exp – -------------------- (B.5)
(h + y )
2 2 h + ( x – xo )
2 2 (h + R)  2 ( h + R)2
In all, the model contains seven parameters (R, h, so, xo, a, b and c), including two geometrical parameters R
and h and a parameter so that represents the zero offset between the settlement coordinate system set up in
1991/1992 and the initiation of tunnel excavation within the zone.
As stipulated by Loganathan and Poulos (1998), parameter b is to be taken as fixed with b = 1.38. Parameter a
is also considered fixed: a = 0.25. Moreover, since R and h are known, four of the seven parameters cited above
may be considered as fixed (R, h, a and b), which leaves three parameters (so, xo and c) to be determined from
the leveling data for each block.
5-36
The expression in (B.5) enables extrapolating the measurements conducted in the vicinity of abscissa point xo
PP .
and thereby predicting the final (asymptotic) settlement based on the first signs of tunnel face settlement of the
4412 -
longitudinal trough. For a given block, the uncertainty associated with the settlement prediction naturally dimin-
ishes as the monitoring duration increases during advance of the tunnel face. In practice, this prediction method
RÉF .
proves efficient for a large number of blocks (even during periods of excavation downtime and resumption).
2002 -
Nonetheless, certain difficulties arise when early settlements occur in front of the normal tunnel face zone of
influence due to: seasonal variations in the spot height of the ground, consolidation phenomena, intersection
MARS - AVRIL
of the faces, etc. In addition, this method does not incorporate long-term deformations (creep).
C HAUSSÉES - 237 -
The instantaneous trough method
The representation of settlement differences associated with a given excavation advance distance has resulted
in the notion of instantaneous troughs for a 15-m tunnel face advance. The representation of total settlement
or of settlement differences in the form of maps provides a three-dimensional image that encompasses both
the longitudinal profile and cross-sectional profiles of the settlement. This ascribed length of 15 m was selected
ET
for the northern tunnel based on correspondence with the observed mode of settlements: with a shorter dis-
P ONTS
tance, the instantaneous troughs would be difficult to distinguish from the measurement noise; whereas with
a longer distance, the analysis would span an excessive period and lose its predictive capability.
The relation between instantaneous trough and longitudinal settlement profile has been illustrated in Figure B.2,
whereby the total longitudinal trough is the sum of the instantaneous troughs from previous excavation phases.
Modeling the instantaneous trough geometry may be performed using functions of various forms: normal,
beta, gamma or uniform distributions, either symmetrical or not, especially when approaching a heterogeneous
zone. An assessment of values computed for the Toulon project reveal that:
B ULLETIN
¾ if the instantaneous trough extends further towards the front, the tunnel is heading into a zone of softer
rock, and settlements will be more pronounced (e.g. the Vauban sector);
¾ if the instantaneous trough extends less towards the front (and more towards the back), the tunnel is head-
ing into a less deformable zone (e.g. the Dumont d’Urville sector). 33
Instantaneous settlement trough associated
with the excavation of Section i - 1
Initial ground level

i-1
i
i-2 i-1
Settlement
i-3 i-2 i-1 i
15m 15m 15m 15m

Fig. B.2 - Principle behind the instantaneous trough method
The prediction of final settlement amplitude was carried out in the example presented in the body of this article
by use of both the experimental correlation between maximum amplitude s15 of the instantaneous settlement
at the center of the trough (for a 15-m tunnel face advance) and the stabilized settlement observed at the back
sfinal, which yields a three-dimensional generalization of the longitudinal profile approach in the axial plane of
the tunnel. It may also be envisaged to incorporate trough volumes (i.e. instantaneous volume plus the final
total volume by segment), rather than settlements.
In conclusion, analytical methods are overly simplistic for proceeding with computations other than those nec-
essary in the pre-sizing project stage, yet such methods may be commended for highlighting the role of the
primary parameters entering into the surface settlement problem. They were successfully used for interpreting
test block measurements on the northern tunnel and for deriving a method for the early prediction of surface
settlements.
5-36
PP .
4412 -
__________________________________________ REFERENCES __________________________________

RÉF .
AFTES (Association française pour les travaux en souterrains), Texte des recommandations relatives aux tasse-
2002 -
ments liés au creusement des ouvrages en souterrain, Travaux et ouvrages souterrains, 132, 1995, pp. 373-395.
BÉREST P., GHOREYCHI M., HABIB P., Affaissement en surface lors du creusement d’un souterrain, Revue
MARS - AVRIL
française de géotechnique, 92, 2000, pp. 41-47.

CORDING E.J., HANSMIRE W.H., Les déplacements autour de tunnels en terrain tendre (traduit de l’anglais),
C HAUSSÉES - 237 -
Tunnels et ouvrages souterrains, 22, juillet-août 1977, pp. 181-192 ; 23, sept-oct. 1977, pp. 221-225 ; 24, nov.-
déc. 1977, pp. 243-250.
DUBOIS P., JASSIONNESSE C., The Toulon underground tunnel crossing. First feedback analysis carried out
using on-site measurements. Tunnels for People, Conference of the International Tunnelling Association, Vienne,
1997, pp. 157-162.
ET
DUBOIS P., CHANTRON L., DIAS D., Analyse du fonctionnement des prévoûtes en béton. Application au cas
P ONTS
du tunnel de Toulon, Journées d’études internationales de l’AFTES, Paris, 1999, pp. 73-80.
DURAND J.-P., Liaison A50/A57, traversée souterraine de Toulon, Tunnels et ouvrages souterrains, 106, 1991,
pp. 177-179.
HABIB P., Recherche achevée et recherche imprécise, Colloque mécanique et géotechnique, Laboratoire de méca-
nique des solides, Palaiseau, 1998, pp. 7-17.
JEFFERY G.B., Plane stress and plane strain in bipolar co-ordinates. Transactions of the Royal Society, London,
series A, vol. 221, 1920, pp. 265-293.
B ULLETIN
KANAYASU S., YAMAMOTO Y., KITAHARA Y., Stability of excavation face in earth pressure balanced
shield. International Symposium on Underground Construction in Soft Ground, New Delhi, 1995, pp. 265-268.
34 KIRSCH, Die Theorie der Elastizität und die Bedürfnisse der Festigkeitslehre. Veit. Ver. Deut. Ing., vol. 48, 28,
1898.
LECA E., Étude du comportement des tunnels creusés en terrains meubles. Mémoire d’habilitation à diriger des
recherches, Université des sciences et technologies de Lille. Laboratoire Central des Ponts et Chaussées, Collec-
tion Études et recherches des LPC, GT 66, 1997, 95 pages.
LEE K.M., ROWE R.K., LO K.Y., Subsidence owing to tunnelling. I: Estimation of gap parameter, Canadian Geo-
technical Journal, vol. 29, 1992, pp. 929-940.
LÉVÊQUE M.-F., traversée souterraine de Toulon. À la rencontre du tunnel nord. Tunnels et ouvrages souterrains,
132, 1995, pp. 351-354.
LO K.Y., ROWE R.K., Prediction of ground subsidence due to tunnelling in clays, Research report GEOT-10-82,
Faculty of Engineering Science, University of Western Ontario - London, Canada, 1982.
LOGANATHAN N., POULOS H.G., Analytical prediction for tunneling induced ground movements in clays,
J. Geotechnical and Geoenvironmental Engng., ASCE, vol. 124, 9, 1998, pp. 846-856.
MAIR R.J., TAYLOR R.N., BRACEGIRDLE A., Subsurface settlement profiles above tunnels in clays, Géotech-
nique, vol. 43, 2, 1993, pp. 315-320.
MAIR R.J., TAYLOR R.N., BURLAND J.B. (1996). Prediction of ground movements and assessment of risk of
building damage due to tunnelling. Proceedings, International Symposium on geotechnical aspects of underground
construction in soft ground, London, UK, 15-17 April 1996, A.A. Balkema, pp. 713-718.
MESTAT PH., Validation du progiciel CÉSAR-LCPC en comportement mécanique non linéaire. Volume 1 : Fon-
dations superficielles et tunnels, Laboratoire Central des Ponts et Chaussées, Collection Études et recherches des
LPC, série Géotechnique, GT 58, 1994, 173 pages.
MINDLIN R.D., Stress distribution around a tunnel, Trans. American Society of Civil Engineers, 105, 1940,
pp. 1117-1140.
NEW B.M., BOWERS K.H., Ground movement model validation at the Express trial tunnel, International Sym-
posium Tunneling 94, 1994, pp. 301-329.
O’REILLY M.P., NEW B.M., Settlements above tunnels in United Kingdom-their magnitude and prediction,
International Symposium, Tunneling 82, Brighton, The Institution of Mining and Metallurgy, London, 1982,
pp. 173-181.
5-36
OTEO C.S., MOYA J.F., Settlements induced by a tunnel in miocenic soft rocks of Madrid, Proceedings 4th Cong
PP .
International Society of Rock Mechanics, Montreux, vol. 1, 1979, pp. 715-721.
4412 -
OTEO C.S., Urban tunnels in hard soils. General report session 6, International Symposium on the Geotechnical
Engineering of Hard Soils - Soft Rocks, Athènes, vol. 3, 1993, pp. 2063-2098.
RÉF .
OU C.Y., HWANG R.N., LAI W.J., Surface settlement during shield tunneling at CH218 in Taipei, Canadian
2002 -
Geotechnical Journal, vol. 35, 1998, pp. 159-168.
MARS - AVRIL
PECK R.B., Deep excavations and tunneling in soft ground. State of the art. Procedings 7th ICSMFE, Mexico, State
of the Art Volume, pp. 225-290 ; Discussions. Main Session 4, vol. 3, 1969, pp. 311-375.
ROWE R.K., KACK G.J., A theoretical examination of the settlements induced by tunneling: four case histories, C HAUSSÉES - 237 -
Canadian Geotechnical Journal, vol. 20, 1983, pp. 299-314.
SAGASETA C., OTEO C., Analisis teorico de la subsidencia originada por la excavacion de tuneles, Proceedings
1st. Symposium on Tunneling, Madrid, vol. 1, paper 10, 1974.
SAGASETA C., Analysis of undrained soil deformation due to ground loss, Géotechnique, vol. 37, 3, 1987,
pp. 301-320, (Discussion par B. Schmidt, 1988 : vol. 38, 4, pp. 647-649).
ET
P ONTS
SAGASETA C., Evaluation of surface movements above tunnels. A new approach, Colloque Interaction Sols
Structures, ENPC, Paris, 1987, pp. 445-452.
SAGASETA C., Discussion de l’article de Verruijt et Booker (1996), Géotechnique, vol. 48, 5, 1998, pp. 709-713.
SCHLOSSER F., MAGNAN J.-P., HOLZ, Construction géotechnique, Proceedings 11th ICSMFE, San Francisco,
vol. 1, 1985, pp. 211-254.
SCHMIDT B., Settlements and ground movements associated with tunneling in soil, PhD thesis, University of Illi-
nois, 1969.
B ULLETIN
SCHMIDT B., Discussion of the paper by C. Sagaseta, Géotechnique, vol. 38, 4, 1988, pp. 647-649.
SERRATRICE J.-F., Suivi du plot Chalucet. Application à la prévision des tassements de surface. Tunnel de la
traversée Souterraine de Toulon. Journées de mécanique des sols et des roches des LPC, Aix en Provence, mai 35
1999, 31 pages.
SERRATRICE J.-F., Tunnels à faible profondeur. Tassements en surface. Calculs paramétriques. CETE Méditer-
ranée, LRPC d’Aix-en-Provence, Rapport de recherche FAER 1 25 31 0, juin 2001, 170 pages.
SHIRLAW J.N, DORAN S., Ground movements and settlements caused by tunneling for Singapore Mass Rapid
Transit System, International Symposium Tunneling 88, 1988, pp. 295-314.
TERZAGHI K., RICHART F.E., Stresses in rock about cavities, Géotechnique, vol. 3, 1952, pp. 57-90.
VERRUIJT A., A complex variable solution for deforming circular tunnel in elastic half-plane, International Jour-
nal for Numerical and Analytical Methods in Geomechanics, vol. 21, 1997, pp. 77-89.
VERRUIJT A., BOOKER J.R., Surface settlements due to deformation of a tunnel in an elastic half plane, Géo-
technique, vol. 46, 4, 1996, pp. 753-756 (Discussion par C. Sagaseta, 1998 : vol. 48, 5, pp. 709-713).
5-36
PP .
4412 -
RÉF .
2002 -
MARS - AVRIL
C HAUSSÉES - 237 -
ET
P ONTS
B ULLETIN
36

Surface Settlement Analysis

Uploaded by

Copyright:

Available Formats

Surface Settlement Analysis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Surface Settlement Analysis

Uploaded by

Copyright:

Available Formats

Surface settlement analysis

and prediction during construction of

The underlying mechanisms

¾ convergence and extrusion of the tunnel face,

1. Tunnel face extrusion

¾ unit volume (per meter on the tunnel axis) Vs = 2πis o = 2.5is o

settlements within the axial plane of the tunnel:

¾ imprecise knowledge of underground volume losses via pre-convergence and extrusion;

Problems specific to tunnel doubling and tunnel junctions

Computation of settlement trough

Excavation of Toulon's northern tunnel

Axial plane of the tunnel

Fig. 3 - Standard tunnel section

The tunnel site: Geotechnical context

Date Face F1 Face F2 Face F3

PM Location PM Location PM Location

Dec. 6, 1994 1 947,5 Central shaft

Aug. 11, 1999 1 487 Profile change

Aug. 27, 1999 1 475 Boulevard de Tessé

Sept. 3, 1999 1 469 Rue d’Entrechaus

Nov. 24, 1999 1 391 Place Mazarin

Dec. 7, 1999 1,221.5 Garage end

Mar. 13, 2000 1,291 F1-F2 junction 1 291 F1-F2 junction

Quaternary Period cover layer (Würm)

Measurement of surface settlements along the tunnel alignment

Use of measurement data

Two methods for predicting settlements as the tunnel face advances

Transversal and longitudinal troughs: The longitudinal profile method

-5 Chalucet Profile 2 PM 889

-20 Computed settlement

Measured settlement Fig. 6 - Settlements observed in the plane

Model s = s(x, y, ηf)

Passage of tunnel face at PM 862

800 825 850 875 900 925 950

Fig. 7 - Settlements caused at PM 862 by the advance in tunnel excavation

-25 Advance of tunnel face F1 EC4

04/06/1998 24/07/1998 12/09/1998 01/11/1998 21/12/1998

8a - Advance of the tunnel face and evolution in settlements vs. time

-30 C HAUSSÉES - 237 -

8b - Evolution in settlement as the tunnel advances

Tunneling advance F1 (PM)

01/11/98 01/12/98 31/12/98 30/01/99 01/03/99 31/03/99 30/04/99

Passage of tunnel face

Passage of tunnel face

Passage of tunnel face

at PM 996 (3255) 3256

900 950 1000 1050 1100

9b - Evolution in settlement as the tunnel advances

Fig. 9 - Settlements of survey marks aligned on the tunnel axis

The instantaneous trough method

-50 Final settlement

Position of tunnel face F1

City blocks of buildings

12a - Total trough

City blocks of buildings

12b - Instantaneous trough

segments further away from the tunnel faces.

-140 Fig. 13 - Comparison of

Lower prediction limit

________ REFERENCES