You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265417885

Impact of Advance Rate on Entrapment Risk of a Double-Shielded TBM in


Squeezing Ground

Article  in  Rock Mechanics and Rock Engineering · May 2014


DOI: 10.1007/s00603-014-0645-2

CITATIONS READS

12 255

3 authors:

Rohola Hasanpour Jamal I Rostami


39 PUBLICATIONS   149 CITATIONS    Colorado School of Mines
145 PUBLICATIONS   1,580 CITATIONS   
SEE PROFILE
SEE PROFILE

Giovanni Barla
Politecnico di Torino
303 PUBLICATIONS   1,704 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Numerical modelling of rock cutting using a monowire system View project

Rock characterization project sponsored bo NIOSH. View project

All content following this page was uploaded by Rohola Hasanpour on 28 January 2015.

The user has requested enhancement of the downloaded file.


Rock Mech Rock Eng
DOI 10.1007/s00603-014-0645-2

ORIGINAL PAPER

Impact of Advance Rate on Entrapment Risk of a Double-Shielded


TBM in Squeezing Ground
Rohola Hasanpour • Jamal Rostami •

Giovanni Barla

Received: 28 July 2013 / Accepted: 25 August 2014


Ó Springer-Verlag Wien 2014

Abstract Shielded tunnel boring machines (TBMs) can including the magnitude of displacement and contact forces
get stuck in squeezing ground due to excessive tunnel on shields and ground pressure on segmental lining versus
convergence under high in situ stress. This typically time for different advance rates.
coincides with extended machine stoppages, when the
ground has sufficient time to undergo substantial dis- Keywords Double-shielded TBM  Squeezing ground 
placements. Excessive convergence of the ground beyond 3D numerical simulation  Creep  Advance rate  TBM
the designated overboring means ground pressure against entrapment
the shield and high shield frictional resistance that, in some
cases, cannot be overcome by the TBM thrust system. This List of symbols
leads to machine entrapment in the ground, which causes A Creep model parameter
significant delays and requires labor-intensive and risky Ai Shield surface
operations of manual excavation to release the machine. To AR Machine advance rate
evaluate the impact of the time factor on the possibility of c Cohesion
machine entrapment, a comprehensive 3D finite difference CPOW Power-law visco-plastic model
simulation of a double-shielded TBM in squeezing ground CVISC Burger-creep visco-plastic model
was performed. The modeling allowed for observation of D Tunnel diameter
the impact of the tunnel advance rate on the possibility of eij Deviatoric strain
machine entrapment in squeezing ground. For this purpose, evol Volumetric strain
the model included rock mass properties related to creep in E Modulus of elasticity
severe squeezing conditions. This paper offers an overview f Failure criterion
of the modeling results for a given set of rock mass and Ff Required maximum thrust force
TBM parameters, as well as lining characteristics, FN Maximum cutterhead thrust
g Plastic potential
G Shear modulus
R. Hasanpour (&)
GK Kelvin shear modulus
Department of Mining Engineering, Hacettepe University,
06800 Beytepe, Ankara, Turkey GM Maxwell shear modulus
e-mail: roha93@gmail.com kn Normal stiffness
ks Shear stiffness
J. Rostami
kt Tensile stiffness
Department of Energy and Mineral Engineering, Pennsylvania K
State University, University Park, PA, USA Kelvin components
e-mail: rostami@psu.edu K Bulk modulus
K0 In situ stress ratio
G. Barla
LDP Longitudinal displacement profile
Politecnico di Torino, Corso Duca degli Abruzzi 24,
10129 Turin, Italy LCFP Longitudinal contact force profile
e-mail: giovanni.barla@polito.it LPP Longitudinal ground pressure profile

123
R. Hasanpour et al.

M Maxwell components shielded TBM, D = 4.50 m) (Farrokh and Rostami 2009),


MC Mohr Coulomb and the Yindaruqin Irrigation Project (China, double-
n Creep model parameter shielded TBM, D = 5.54 m) (Ramoni and Anagnostou
N Number of contact points 2010) are examples of cases where the TBM became
p0 In situ stress trapped in squeezing ground during a 1-week holiday stop
r Shield radius or during an extended maintenance stop. This suggests that
R Tunnel radius maintaining a high daily advance rate and reducing
Si,j Deviatoric stress downtime may have a positive effect in avoiding TBM
TBM Tunnel boring machine entrapment (Ramoni and Anagnostou 2010).
b Reduction coefficient Standstills are unfavorable also with respect to cutter-
e_ Strain rate head operation. Depending on the ground rheological
gK Kelvin viscosity behavior, high ground pressures acting against the cutter-
gM Maxwell viscosity head or an extremely high extrusion rate of the core may
w Dilatancy angle develop. In this respect, the Gilgel Gibe II Tunnel (Ethio-
DZmin Smallest width of an adjoining zone pia, double-shielded TBM, D = 6.98 m) is a case history
in the normal direction that can be used as an example (Barla 2010; Ramoni and
/ Friction angle Anagnostou 2010). Furthermore, an open-face TBM was
c Unit weight also reported to be trapped in the Yacambú-Quibor Tunnel
l Skin friction coefficient in Perù. The tunnel was excavated in silicified and gra-
m Poisson’s ratio phitic phyllites with creep behavior and bored at depths of
r Deviatoric stress up to 1,270 m below the surface, and extreme squeezing
r0 Volumetric stress problems were encountered. In this project, delay in
r1 Major principal stress excavation because of holidays and maintenance of the
r3 Minor principal stress machine led to machine entrapment and loss of the TBM
rh Horizontal stress (Hoek and Guevara 2009).
rt Tensile strength As stated by the International Society for Rock
rv Vertical stress Mechanics (ISRM), squeezing behavior is the time-
dependent large deformation related to the progressive
yielding, which occurs around the tunnel and is essentially
associated with creep, caused by exceeding a threshold
shear stress. Deformation may terminate during construc-
1 Introduction tion or continue over a long period of time (Barla 2001). In
engineering practice, the difficulties in dealing with
Tunnel excavation by a shielded tunnel boring machine squeezing conditions are connected to: (1) the evaluation of
(TBM) is a cyclic process between mining and lining the time-dependent characteristics of the rock mass by
installation or regripping, but, relative to the gradual means of laboratory or in situ tests, (2) the use of an
movement of the ground, it can be considered as a continuous appropriate constitutive model, and (3) the choice of a
movement of the face. This is true unless a major delay, such suitable excavation and support system (Debernardi and
as long holidays, weekends, and extended repair or mainte- Barla 2009; Barla et al. 2010).
nance stoppages, and other issues during operation are For the design of mechanized tunneling in squeezing
experienced. The entrapment of shielded TBMs can occur in conditions, the complex interaction between the rock mass,
squeezing ground due to excessive tunnel convergence the tunnel machine, its system components, and the tunnel
during such extended stoppages. This shows the importance support has to be analyzed in detail and 3D models
of the ‘‘time’’ factor in tunneling operations, which should be including all of these components are better suited to cor-
considered for evaluating the stability conditions and rectly simulate this interplay and avoid the errors intro-
designing the support system due to the considerable amount duced with the assumption of plane strain conditions
of deformation and contact pressure which may develop over (Cantieni and Anagnostou 2009). It is noted that axisym-
time (Ramoni and Anagnostou 2010). metric models can also be used to avoid such errors, as
There are many cases of shielded TBMs being trapped described by Ramoni and Anagnostou (2010).
in the ground when there was a slowdown in operation or a There are many studies in the literature that are related
standstill in the TBM drive. The Nuovo Canale Val Viola to the numerical analyses of mechanized tunneling in
(Italy, double-shielded TBM, D = 3.60 m) (Ramoni and squeezing conditions. As reported by Ramoni and Anag-
Anagnostou 2010), the Ghomroud Tunnel (Iran, double- nostou (2011a, b), Lombardi (1981) presented the first

123
Impact of Advance Rate on Entrapment Risk

results of 3D numerical modeling and investigated the account by Sterpi and Gioda (2007), who highlighted the
impact of the advance rate on the lining loading for the fundamental effect of creep, as well as by Einstein and
simplified model of a lining that starts to become loaded Bobet (1997) and Ramoni and Anagnostou (2007), who
40 m behind the face. Furthermore, Lombardi and Panciera studied the consolidation processes associated with the
(1997) utilized numerical analyses to evaluate the feasi- development and subsequent dissipation of excess pore
bility of a double-shielded TBM drive for the Guadiaro- pressures around the tunnel in a low-permeability water-
Majaceite Tunnel (Spain, D = 4.88 m). The model incor- bearing ground.
porates advance rate and time-dependent ground behavior A recent description of the ‘‘steady state method’’
in the simulations, but the analysis was not fully 3D and (including its further development for poro-elasto-plastic
did not consider the interactions between the rock mass and materials) and numerical comparisons with the step-by-
the TBM components. step simulation of an advancing tunnel can be found in
Similar numerical studies were performed by Shalabi Cantieni and Anagnostou (2009). Step-by-step simulations
(2005), who carried out a back analysis of the creep of tunnel excavation have also been performed by Schmitt
deformations and pressures of the Stillwater Tunnel (USA, (2009), who studied the behavior of single-shielded TBMs.
D = 3.06 m) by assuming the tunnel as being lined up to The model offers valuable insights into the effects of non-
the face. The developed model did not consider the shield uniform convergence and of non-hydrostatic shield and
in the numerical computations. In addition, Amberg (2009) lining loading.
and Lombardi et al. (2009) simulated the shield by As a follow-up of the work above, this paper is intended
applying a support pressure of 1 MPa at the face and at the to describe a comprehensive 3D model of a shielded TBM
excavation boundary around the shield. Their model ana- which has been developed with the FLAC3D code (Itasca
lyzed the impact of the advance drainage on the ground FLAC3D Manual 2006), so that all the properties of the
behavior for the excavation of the service tunnel of the main TBM components can be used as variables at each
planned Gibraltar Strait Tunnel between Morocco and step of the analysis. The 3D model has the advantages of
Spain (D = 6.50 m). numerical analysis and the ability to model complex
Numerical studies that evaluate the stresses and defor- ground behavior, without the disadvantages of interjected
mations of the shield structure of the single-shielded TBM inaccuracies of modeling the tunneling process in 2D
of the Hallandsås Tunnel (Sweden, D = 10.70 m) were models, including the limitation of the axisymmetric
performed by Wittke et al. (2007). The analysis considers models.
seepage flow, but the shield is modeled by introducing the The numerical modeling discussed builds upon the
a priori assumption that the ground closes the steering gap computational model of Zhao et al. (2012) and Hasanpour
at a distance of 4 m behind the working face. et al. (2014) by taking creep into consideration. Further-
Ramoni and Anagnostou (2006) employed axisymmetric more, the 3D modeling uses finite difference analysis with
numerical models in order to investigate the effects of thrust the capability to allow for large strains in the numerical
force, overboring, shield length, and skin friction coefficient computation. Also, the analysis is based on simple and
between the shield and the ground with respect to the prob- accurate configurations of the shield components that are
lem of shield jamming. Ramoni and Anagnostou (2007, used in practice.
2008) created the model by implementing the stress-point The finite difference models can be applied in various
algorithm in accordance with the so-called ‘‘steady state rock mass conditions, including hard, weak, and interme-
method’’, a numerical procedure for solving problems with diate rock masses. In addition, different shield types and
constant conditions in the tunneling direction by considering correct dimensions of the shield components can be mod-
a reference frame, which is fixed to the advancing tunnel eled by small changes in the input data. The effects of
face. A recent description of the computational method and backfilling materials can also be taken into account by
numerical comparisons with the step-by-step simulation of considering the geometry of the tunnel and machine, as
an advancing tunnel can be found in Cantieni and Anag- well as time-dependent properties of the backfill grout.
nostou (2009). Various TBM configurations can be simulated with
Fully 3D numerical simulations of shielded TBMs have respect to the type of TBM only by changing the values
been presented by Graziani et al. (2007), Sterpi and Gioda assigned to each component. For example, a single-shiel-
(2007), Schmitt (2009), Pellet et al. (2009), Einstein and ded TBM can be simulated by modeling one part of the
Bobet (1997), Ramoni and Anagnostou (2007), and Zhao shield and changing the grouting material properties.
et al. (2012). The model developed by Graziani et al. Elimination of the shield elements by advancing the face
(2007) was a 3D model that considered creep effects for the and relocation of the shield can be used to simulate
planned Brenner Base Tunnel (Austria, double-shielded machine advance in the tunnel and extrusion of the seg-
TBM, D = 11.00 m). Time effects were also taken into ments from the tail shield (Hasanpour et al. 2014).

123
R. Hasanpour et al.

The contact between the shield and rock mass has been microfractures. If the applied stress is reduced to zero
modeled by using interface elements on both tunnel and during the primary creep stage, the deformation will
shield boundaries. This is accomplished while considering eventually decrease to zero (Fig. 1a) (Fjaer et al. 2008).
the gap between the ground and the shield by accounting In the next stage, the deformation rate is constant. This
for a non-uniform overcut around the shielded TBM. is called steady state (or secondary) creep. If the applied
However, given that the numerical formulation used is stress is reduced to zero during this stage, the deformation
based on the large strain assumption, for preventing the will not vanish completely. Steady-state creep thus implies
penetration of the rock mass into the shield elements due to a permanent deformation of the material.
large displacements in squeezing ground, an algorithm was Finally, the deformation rate may increase with time.
used to control the ground displacement at the contact This is called accelerating (or tertiary) creep. This stage
surfaces. For this purpose, a FISH code was developed in leads to rapid change in the deformation rate, and, ulti-
FLAC3D that monitors all displacements with respect to mately, to failure. The process may be associated with a
non-uniform overcut at each solving step of the numerical rapid spreading of ‘‘unstable’’ fractures (Fjaer et al. 2008).
analysis. The actual creep behavior of rock depends on the
The increase of the gap due to the conical shape of the magnitude of the applied stress. For low or moderate
shield (stepwise reduction of the diameter of the rear shield stresses, the material may virtually stabilize after a period
in comparison to the front shield) is considered. This of transient creep. For high stresses, the material may
property distinguishes the model from other 3D models rapidly run through all three stages of creep and, finally,
that have been developed for the numerical simulation of fail. The intermediate stress regime, where the material
shielded TBMs in the past (Hasanpour et al. 2014). fully develops each stage of creep, may be small and hard
Moreover, the model developed for this study is capable of to find in practice (Fig. 1b).
using Mohr–Coulomb or Hoek–Brown failure criteria as The time scale of a creep stage may vary over a wide
input data depending on field measurements and ground range: in some cases, it lasts for minutes, in other cases for
conditions. years. Creep is a molecular process, and the time scale
This paper will also offer an overview of the issues depends on temperature; the process generally speeds up
related to the creep behavior of rock with an emphasis on with increasing temperature (Farmer 1983). The fact that
the Burger-creep visco-plastic model (CVISC model), as even steady-state creep eventually leads to failure means
available in the FLAC3D code. This is followed by a dis- that a rock which is loaded to a level somewhat below its
cussion on the modeling procedure implemented in order to ultimate strength may fail after some time, if the load is
simulate the tunneling operation using a double-shielded maintained. This effectively reduces the long-term uniaxial
TBM. strength to typically 50–70 % of the ultimate strength
The results obtained show that modeling can be used to (Farmer 1983).
estimate tunnel convergence during TBM excavation, The CVISC model as available in FLAC3D has been
compare the longitudinal and sectional ground pressures used in this paper to simulate the time-dependent behavior.
for different advance rates, and predict the magnitude of CVISC is an analogical model which couples, in series, the
the contact forces on the shields in squeezing conditions, so Burger visco-elastic model (i.e., Kelvin and Maxwell
as to estimate the frictional force between the rock and the models in series) with a plastic flow rule, based on the
shield and, thus, the required machine thrust to move the Mohr–Coulomb yield criterion. The CVISC model is
machine forward. characterized by a visco-elasto-plastic deviatoric behavior
and an elasto-plastic volumetric behavior.
The deviatoric and volumetric behavior are schemati-
2 Creep Behavior cally illustrated in Fig. 2, where a Kelvin unit characterized
by its shear modulus GK and viscosity gK, a Maxwell unit
Creep is a time-dependent deformation that may occur in characterized by its shear modulus GM and viscosity gM,
materials under constant stress. Creep originates from vi- and a Mohr–Coulomb plastic unit characterized by its
sco-elastic effects in the solid framework; thus, creep, cohesion c, friction angle /, and dilation angle w are
unlike consolidation, may occur in both dry and saturated connected in series and subjected to a certain deviatoric
rock conditions. There are three stages of creep following a loading jointly.
change in the stress state. In this model, the visco-elastic strains are deviatoric and
First, there is a region where the rate of the time- depend only on the deviatoric stress Si,j; instead, the plastic
dependent deformation decreases with time. This is called strains are both deviatoric and volumetric and depend on
transient (or primary) creep. The process may be associated rij in accordance with the chosen flow rule (Bonini et al.
with minor spreading at the decaying rate of ‘‘stable’’ 2009).

123
Impact of Advance Rate on Entrapment Risk

Fig. 1 a Strain versus time for


a creeping material. b The
development of creep for
different values of the applied
stress (Fjaer et al. 2008)

the 3D isometric view of the model, where the horizontal


(x), longitudinal (y), and vertical (z) directions are 75, 100,
and 150 m, respectively. The 3D model comprises a total
of 87,000 elements and 88,236 nodes. The model is very
finely discretized near the excavation perimeter and within
the lining. The tunnel is assumed to be in squeezing rock
with characteristics back-calculated from the observed
behavior of the ground based on field measurements.
Figure 3b shows the schematic view of the simulated
double-shielded TBM arrangement to reflect the true
geometry of the machine and reduced shield diameter in
stages, and also to represent the eccentric overcut in the
tunnel, with maximum overcut at the crown. The double-
shielded TBM arrangement used in the simulation was
taken from the study by Zhao et al. (2012). However, the
excavation diameter was modified and applied in accor-
dance with the studies performed on the Saint Martin La
Porte access adit along the Lyon–Turin Base Tunnel (Barla
Fig. 2 Schematical representation of the Burger-creep visco-plastic
model (CVISC) model: a deviatoric behavior and b volumetric et al. 2010).
behavior (Bonini et al. 2009) The discretization of the numerical model is shown in
Fig. 4. Additionally, Table 1 contains the main specifica-
3 Numerical Model tions of the double-shielded TBM used in the model. The
shield, segmental lining, and annular gap backfill were
In order to simulate a mechanized tunneling process by a considered to behave as linearly elastic material, with the
double-shielded TBM, various 3D models were developed pertinent properties listed in Table 2. In the modeling of
in FLAC3D. A parametric study was carried out to deter- the shield skin, the total weight of the TBM was applied by
mine the required distance from the tunnel face to the edge a normal stress to the lower 45° area of the shield which
of the final segmental ring to prevent the edge effect on was in contact with the tunnel floor. The thrust force was
displacement and stress magnitudes of the numerical model applied to the excavation face by using an equivalent
in the advancing direction of the tunnel. The initial results normal stress.
indicated that a distance larger than 2.5 times the tunnel It is noted that the design of the double shield consid-
diameter to the face (in hard rock) and larger than 4 times ered in this study is not the standard design for double-
the tunnel diameter (in weak rock) is necessary to prevent shielded TBMs. The shield geometry is a special solution
edge effects (Hasanpour et al. 2014). for TBM tunneling in squeezing ground, based on the
The 3D block model and relevant dimensions were modified design after Concilia and Grandori (2004), as
selected and implemented based on the Lyon–Turin Base reported in Ramoni and Anagnostou (2010).
Tunnel data (Barla et al. 2010; Bonini and Barla 2012), as In this study, the contact between the cutterhead and the
shown in Fig. 3a. The screenshot shows a cross-section of rock mass, as well as the contact between the shields and

123
R. Hasanpour et al.

Fig. 3 a Geometric dimensions of the numerical model of a double-shielded tunnel boring machine (TBM) (Hasanpour et al. 2014). b Schematic
drawing of the modified double-shielded TBM in squeezing rock (from Zhao et al. 2012, modified by the authors)

Fig. 4 Numerical model for the simulation of tunneling with a double-shielded TBM: a complete model and b description of the model
(Hasanpour et al. 2014)

Table 1 Geometric dimensions for the double-shielded tunnel boring Table 2 Mechanical properties of the double-shielded TBM
machine (TBM) components components
Double-shielded TBM components Unit Value Material Unit Shield Segmental Soft Hard
properties lining backfilling backfilling
Cutterhead length m 0.75
Front shield length m 5 Elastic GPa 200 36 0.5 1.0
modulus, E
Rear shield length m 6
Poisson’s – 0.3 0.2 0.3 0.3
Boring diameter m 12.34
ratio, m
Front shield diameter m 12.13
Unit weight, c kN/ 76 30 24 24
Rear shield diameter m 12.07 m3
Shield thickness cm 3
Lining segment width m 2
Lining segment thickness cm 45 shield and the excavation surroundings (and also between
the segmental lining and the backfill).
It is noted that the use of very high stiffness values for
interface elements in FLAC3D could lead to very slow
the rock mass, has been modeled by using interface ele-
response and solution convergence, and produce numerical
ments on both the tunnel and shield boundaries, and by
errors related to the computer’s precision. A good rule of
considering the gap between these components according
thumb is that kn and ks should be selected to be ten times
to a non-uniform overcut in the shielded TBM. Cross-
the equivalent stiffness of the softer neighboring zone,
sections of the front shield and the rear shield are illustrated
which is given by:
in Fig. 5. Normal, shear, and tensile stiffness values (kn, ks,
and kt, respectively) were assigned to the interface ele- K þ 4=3G
ð1Þ
ments for the simulation of the interaction between the DZmin

123
Impact of Advance Rate on Entrapment Risk

shield for a double-shielded TBM. The total number of


solving steps was set up based on the operational mode of
the TBM in squeezing ground with respect to the machine
advance rate. A total of 31 excavation steps were simu-
lated, consisting of the initial undisturbed ground step and
30 excavation steps (each excavation step was to be 1 m in
length). As illustrated in Fig. 5, 30 steps are sufficient in
order to avoid boundary effects. Moreover, 15 steps are
necessary to build up the computational model. On the
other hand, due to the short-term analysis performed, the
Fig. 5 Cross-section of a double-shielded TBM at the front shield final value of the ground pressure is not reached at the end
and the rear shield to reflect the true geometry of the shield and of the calculations.
overcut (Zhao et al. 2012)
For creep runs, a time step in the computer code rep-
resents real time, while in static analyses, the time step is
where K and G are the bulk and shear moduli, respectively, used as a virtual value for stepping the steady-state con-
and DZmin is the smallest width of an adjacent zone in the dition. For creep simulation in FLAC, a time step depends
normal direction. on the TBM advance rate. Furthermore, FLAC allows the
It should be reminded that, since there is a difference in user to adopt a time step for time-dependent phenomena
zone density between two adjacent grids (shield and rock such as creep analysis. The constitutive laws for creep
mass), the interface should be attached to the grid with the make use of the time step in the equations, so the time step
greater density. Also, the size of the interface elements may affect the response. For the CVISC model, the max-
should always be equal to or smaller than the target faces imum creep time step is calculated as follows:
with which they are expected to come into contact. If this  K M
cr g g
requirement is not met in discretization, the interface ele- Dtmax ¼ min K ; M ð2Þ
ments should be subdivided into smaller elements. On the G G
other hand, interface elements should be limited to grid where the superscripts K and M refer respectively to Kel-
surfaces that will actually come into contact with another vin and Maxwell properties.
grid (Itasca FLAC3D Manual 2006). In the numerical analysis for tunneling with a double-
The numerical formulation used in this study is based on shielded TBM, each solving time step has been adjusted
the large strain assumption, but, sometimes, unforeseen according to the advance rates. For example, when the
errors such as the penetration of rock mass into shield advance rate is 12 m/day, for 1 m of excavation (which is
elements occurred during the numerical calculations. equal to one step), 2 h elapsed. Therefore, the solving step
Therefore, to avoid the problems related to large dis- or iteration time is adjusted for 2 h or 120 min. The units
placements in squeezing ground, the method of displace- of creep parameters such as viscosities change from MPa-
ment control has been applied to contact surfaces. For this year to MPa-min to coordinate with each other.
purpose, a FISH code was developed in FLAC3D to mon- The modeling excavation stages are defined as follows
itor and control all displacements with respect to non- (Fig. 6):
uniform overcut at each solving step of the numerical
analysis (Hasanpour et al. 2014). 1. In the first stage, the initial in situ stresses are
This means that, when contact between the ground and implemented and the correct distribution of stresses
shields takes place, the code stops the displacement of the is applied to the rock mass model.
interface elements for each solving step. Therefore, it 2. In the second stage, tunnel boring starts excavating
prevents penetration of the ground into the shield that into the face by making a circular shape impression
leads to unforeseen runtime errors. Increasing of the gap into it; the cutterhead is activated, the thrust force and
due to the conical shape of the shield is considered in the TBM weight are applied to the tunnel face and invert,
code as part of the longitudinal profile of the double- respectively, and are maintained for all stages of
shielded TBM. excavation. The contact analysis between the cutter-
head and tunnel face is performed in this stage. The
maximum cutterhead thrust force has been set to 11
4 Modeling the Excavation Process MN for the given machine size and ground properties.
3. In the third stage, the front shield moves forward.
The excavation stages were simulated based on the con- Numerical results are examined for the evaluation of
struction design of the cutterhead, front shield, and rear probable contact between the ground and the front

123
R. Hasanpour et al.

Fig. 6 Double-shielded TBM


tunneling. Stages of numerical
simulation

shield. Also, the entrapment risks are analyzed when 6. In the sixth stage, the segmental ring is subjected to
contact occurs between the walls and the shield. ground loading; this is assumed to start from the third
4. In the fourth stage, the rear shield is activated by segment behind the machine. Moreover, the injection
considering its length. The analysis of the numerical of backfill into the annular space between the rock
results for this stage is the same as that for the third mass and lining by using the soft grout is applied. This
stage. also allows for the simulation of injecting pea gravel.
5. In the fifth stage, installation of the segmental linings is 7. In the seventh stage, hardening of the backfill in the
implemented inside the rear shield. annular space is activated.

123
Impact of Advance Rate on Entrapment Risk

Table 3 Rock mass parameters (Barla et al. 2010) following to a non-uniform gap between the shields and the
Rock mass parameters Unit Value
bored tunnel walls.

Elastic modulus, E GPa 0.942 6.1 Double-Shielded TBM Time-Dependent


Poisson’s ratio, m – 0.25 Excavation
Cohesion, c MPa 0.61
Friction angle, / ° 28 Figure 7 shows the longitudinal displacement profile
Dilatancy angle, w ° 8 (LDP) and the longitudinal contact force profile (LCFP) at
the tunnel crown based on the CVISC model. The TBM
advance rate (AR) is taken to be equal to 12 m/day. As
illustrated in the figure, the gap between the front shield
Table 4 Creep constitutive parameters (Barla et al. 2010)
and ground is closed at the end of the shield, behind the
CVISC model Unit Value cutterhead, as the ground gradually converges and comes
into contact with the machine along the shield. The closure
Maxwell shear modulus, GM MPa 566
of the gap between the front shield and the ground occurs
Maxwell viscosity, gM MPa-year 27.98
approximately 4.0 m behind the face, and the ground starts
Kelvin viscosity, gK MPa-year 4.26
to load the front shield with a magnitude of up to 6.8 MN.
Kelvin shear modulus, GK MPa 498.1
Due to the conical shape of the shields, the contact stress
Tensile strength, rt MPa 8.5e-3
between the ground and the rear shield is initially zero.
Then, the ground comes into contact with the rear shield at
a distance of 9 m from the face and loads it up to 4.7 MN.
The model is set up such that, when a segmental ring is It is noted that the load values become maximum at one
extruded from the tail shield, the material property of the location (for example, in Fig. 7, the loads are calculated
rear shield is replaced by the material property of the soft above the front shield and along the tunnel crown, or in a
filling for the annular space in the area of the first two 12 o’clock position). The contact forces at any point are
segmental rings. It should also be noted that, in modeling calculated based on the estimated contact pressure from the
the lining, the joints between segments and adjacent rings ground for an increment of 1 m at the given point (1 m is
are not considered, and the ring is modeled as a thick wall one excavation step).
cylinder. The corresponding longitudinal ground pressure profile
(LPP) at the tunnel crown is illustrated in Fig. 8. For this
purpose, an area between 11 and 1 o’clock along the tunnel
5 Modeling the Ground Time-Dependent Behavior crown is selected. A stress redistribution due to tunnel
excavation occurs and the ground pressure decreases from
The in situ stresses are taken to vary linearly with depth, the in situ stress value of 13.8 to 0.2 MPa in the area of the
and the tunnel is assumed to be at a depth of nearly 600 m. front shield. When contact between the front shield and the
The ratio between the horizontal and vertical stress com- tunnel wall starts, the shield is shown to support the
ponents (K0 = rh/rv) in the ground is assumed to be equal ground. With the tunnel face advance, the ground pressure
to 1. The rock mass is to follow an elastic perfectly plastic increases up to 3.1 MPa.
behavior according to the Mohr–Coulomb failure criterion, Due to the conical shape of the rear shield (a stepwise
with the parameters as summarized in Table 3. reduction in the shield diameter is assumed), the ground
The CVISC model is applied for describing the tunnel pressure decreases to below 0.2 MPa behind the front
time-dependent response associated with severely squeez- shield. Contact between the rear shield and the ground
ing conditions. The creep parameters of the CVISC model takes place with the advancing of the tunnel face, and the
were derived from Barla et al. (2010). Table 4 summarizes ground pressure at the crown increases to 2.7 MPa. The
the creep parameters. installation of the segmental lining and the subsequent
application of backfilling allows for contact between the
ground and the lining through the backfill. Therefore, due
6 Results of the Numerical Analysis to the reaction of the segmental lining, the internal pressure
grows as the tunnel face advances.
The results of the numerical analysis include the contact Plots for both the LDP and LCFP at the tunnel sidewall
forces between the ground and the machine components, as are shown in Fig. 9. It is noted that the contact between the
well as the history of the principal stresses in the ground, cutterhead and the ground started in the last solving time
for different advance rates. Consideration is given in the step at the sidewall, with minimal forces being applied to

123
R. Hasanpour et al.

Fig. 7 Longitudinal
displacement profile (LDP) at
the tunnel boundary and
longitudinal contact force
profile (LCFP) on the shields at
the crown along the tunnel for
an advance rate
(AR) = 12 m/day

Fig. 8 Longitudinal ground


pressure profile (LPP) at the
tunnel crown for
AR = 12 m/day

the cutterhead. Also noted is a smaller gap between the contact between the rock and the shield as a consequence
ground and the TBM shield at the sidewall, because of a of the gradual movement of the ground due to creep. This
non-uniform overcut around the shield. The gap is maxi- results in closure of the gaps and loading of the shield, and,
mum at the crown and gradually decreases to zero at the hence, higher frictional forces. Contact between the ground
invert, where the machine rests against the floor. and the shields occurs at the sidewall soon after the
Therefore, closure of the gap occurs faster at the lower advance of the machine behind the face. The contact forces
parts of the shield as boring proceeds. On the other hand, a on both the front and the rear shields increase to 10.5 and
slowdown in TBM advance may cause extended areas of 10.4 MN, respectively.

123
Impact of Advance Rate on Entrapment Risk

Fig. 9 LDP at the tunnel


boundary and LCFP at the
tunnel sidewall (spring-line
level) for AR = 12 m/day

Fig. 10 LPP at the tunnel


sidewall (spring-line level) for
AR = 12 m/day

The LPP along the tunnel sidewall is shown in Fig. 10. multiplying the integral of the total contact force obtained
A redistribution of stresses at the sidewall due to some from the numerical analysis by the skin friction coefficient
loading and unloading process similar to the mechanism at l and the reduction coefficient b (i.e., ratio of the shield
the crown is observed as discussed for Fig. 7. radius r to the tunnel radius R). The values of b for the
The contact forces between the ground and the front front and rear shields are calculated to be 0.983 and 0.978,
shield, as well as between the ground and the rear shield, respectively.
are shown in Fig. 11. The required thrust force to over- Then, the required maximum thrust force is obtained by
come friction and drive the TBM forward is calculated by the following relationship:

123
R. Hasanpour et al.

Fig. 11 Contact force distribution between the ground and shields for AR = 12 m/day

Fig. 12 LCFP for different


advance rates (3, 6, 12, and
24 m/day): a tunnel crown and
b tunnel sidewall

123
Impact of Advance Rate on Entrapment Risk

Fig. 13 Load distribution (maximum values) on the TBM components for different advance rates

X
N For the rear shield, the maximum cutterhead thrust is not
Ff ¼ b  l Fi ð3Þ taken into account: F = Ff = 47.9 MN.
i¼1

where N is the number of contact points (contact nodes) on 6.2 Effect of Advance Rate
the shield surface (Zhao et al. 2012). The skin friction
coefficient l is assumed to be 0.40 for restart after the To evaluate the effect of different advance rates on the rock
installation of the segmental lining and for the average mass behavior as well as loadings on the machine com-
friction coefficient in this mode. ponents, the LCFP for the contact forces on the shield at
For the front shield, the maximum total thrust force by the the crown and sidewall are plotted in Fig. 12 for different
thrust cylinders is the sum of the maximum cutterhead thrust advance rates (3, 6, 12, and 24 m/day).
FN and the required thrust to overcome friction Ff, as follows: As shown in Fig. 12a, the front shield is loaded to 8.5
For the front shield: F = FN ? Ff = 11 ? 88.3 = 99.3 MN at the crown, with higher values when the advance rate
MN. is 3 m/day compared to 6.6 MN for 24 m/day, a difference

123
R. Hasanpour et al.

Fig. 14 Required thrust force versus different advance rates

of 1.85 MN. This difference is 4.03 MN for the rear shield.


Figure 12b shows this behavior at the sidewall. It is con-
cluded that the entrapment of shielded TBMs can occur in
squeezing ground during extended machine downtime or
when lower advance rates are experienced.

6.3 Evaluation of TBM Entrapment Potential

Figure 13 shows the magnitude of the ground pressure


around the machine for various advance rates at the cut-
terhead, end of the front shield, and end of the rear shield.
The maximum ground pressure occurs at the invert. This is
due to the non-uniform overcut, which causes immediate
contact between the machine and the ground at the invert Fig. 15 Ground pressure around the lining: a distribution on the ring
and gradual contact and increasing of the ground pressure segment and b average ground pressure on the last set of segment
rings versus TBM daily advance rate
on various points towards the crown and the back of the
machine.
The weight of the machine accounts for additional load Also, the role of the advance rate on the loading of the
at the invert near the tunnel face. On the other hand, as also lining is shown to be significant. As clearly demonstrated
shown in Fig. 13, the impact of the advance rate on the on the same plot of Fig. 15a, for the advance rate of 24 m/
applied pressure from the ground to the cutterhead and the day, the last set of segmental rings installed experiences
front shield is not significant, but is more pronounced on about 2.5 MPa less than when the advance rate is 3 m/day.
the rear shield, where a lower load is anticipated at high This allows the grout to harden and be able to take higher
advance rates. loads at a larger distances from the machine and the seg-
For assessing the shield entrapment risk and predicting ments to take lower loads initially and hold their position.
the thrust force, the maximum thrust force required is Figure 15b shows the diagram that can be used for the
calculated for different advance rates, as illustrated in prediction of ground pressure at the boundary of the seg-
Fig. 14. mental ring at variable advance rates. According to this
diagram, the average pressure on the lining depends on the
6.4 Effect of Advance Rate on the Loading advance rates. By developing such diagrams for a specific
of Segmental Linings tunnel, one can calculate the pressure distribution on the
lining for a given advance rate and optimize the design of
Figure 15 illustrates the ground pressure distribution on the the lining system accordingly.
lining. In particular, Fig. 15a shows that, after the instal- It should be noted that the results given in Fig. 15 are for
lation of the segmental lining, the ground pressure is short-term conditions, when the lining is located 30 m
transferred uniformly to the ring due to the backfill, thus behind the face or, in other words, for the first group of
providing evidence of the importance of the backfill for segmental rings behind the shield. Thus, a further analysis
achieving a uniform redistribution of stress around the is needed for the long-term conditions in order to assess the
lining. ground pressure distribution on the segmental lining for

123
Impact of Advance Rate on Entrapment Risk

design purposes. This refers to the full loading conditions evaluation of machine entrapment potentials in various
of the segment as time elapses and an equilibrium condi- ground types could be prepared by systematic sensitivity
tion is reached between the in situ stresses in the ground analyses.
and the segment. One should always be aware of the uncertainties
The results of the numerical simulations performed regarding the geological conditions, the in situ stresses, and
illustrate the importance of the advance rate parameter on rock mass properties when assessing the potential of shield
the prediction of the TBM entrapment potential. One entrapment in a given underground project. Also, the lim-
should, however, be aware of the rather simple mathe- itation of the numerical simulation when modeling the
matical formulation of the CVISC model used and of its ground behavior and the step-by-step nature of the solu-
inherent limitations in the simulation of the time-dependent tions should be added to the list of uncertainties when using
response of a tunnel (Barla et al. 2010). The model does not this approach for design purposes.
predict the observed deformation when the tunnel exhibits
a gradual decrease in the rate of convergence, reaching a Acknowledgments The authors gratefully acknowledge the finan-
cial support of the Scientific and Technological Research Council of
near-stable condition. Turkey (TÜBİTAK) under Project No. MAG-114M568.

7 Conclusions
References
The time-dependent behavior of the ground and its impact Amberg F (2009) Numerical simulations of tunnelling in soft rock
on the loading of the shield and required thrust to move the under water pressure. In: ECCOMAS Thematic Conference on
machine forward, when tunneling with a double-shielded Computational Methods in Tunnelling (EURO:TUN 2009),
TBM, were studied in this paper. Numerical modeling was Bochum, Germany, September 2009. Aedificatio Publishers,
Freiburg, pp 353–360
used to analyze the detailed configuration of the machine Barla G (2001) Tunnelling under squeezing rock conditions. In:
and its interaction with the intruding ground in a 3D Kolymbas D (ed) Euro summerschool on tunnelling mechanics,
simulation. Innsbruck, October 08–11, pp 169–268
The most attention was directed to the possible machine Barla G (2010) Analysis of an extraordinary event of TBM
entrapment in squeezing ground conditions. In: Pilgerstorfer T
entrapment in squeezing ground, by analyzing the magni- (ed) Festschrift zum 60. Geburtstag von Wulf Schubert, Tech-
tude of the ground pressure and loading for various points nische Universitat Graz, pp 66–76
around the tunnel in both the circumferential and longitu- Barla G, Bonini M, Debernardi D (2010) Time dependent deforma-
dinal directions. The contact forces on the front and rear tions in squeezing tunnels. Int J Geoeng Case Hist 2(1):40–65
Bonini M, Barla G (2012) The Saint Martin La Porte access adit
shields that could lead to machine entrapment were (Lyon–Turin Base Tunnel) revisited. Tunn Undergr Space
evaluated. Technol 30:38–54
The calculated contact pressure between the ground and Bonini M, Debernardi D, Barla M, Barla G (2009) The mechanical
the shield and the resulting frictional forces show that the behaviour of clay shales and implications on the design of
tunnels. Rock Mech Rock Eng 42:361–388
effect of the advance rate on the ground load applied on the Cantieni L, Anagnostou G (2009) The effect of the stress path on
shields is more significant in the rear shield as compared to squeezing behavior in tunneling. Rock Mech Rock Eng
the front shield. Also, the average ground pressure on the 42(2):289–318
lining ring immediately behind the machine is shown to be Concilia M, Grandori R (2004) New Viola Water Transfer Tunnel.
Mechanized tunnelling: challenging case histories. International
highly dependent on the advance rate. congress, GEAM Turin, pp 27–34
It is concluded that, in practice, the information gained Debernardi D, Barla G (2009) New viscoplastic model for design
with the 3D simulation could be used to determine the daily analysis of tunnels in squeezing conditions. Rock Mech Rock
advance rates that would avoid TBM entrapment. It is Eng 42:259–288
Einstein HH, Bobet A (1997) Mechanized tunnelling in squeezing
recognized that the impacts of long delays and standstill rock—from basic thoughts to continuous tunneling. Tunnels for
conditions have not been fully examined. In such cases, a people, ITA World Tunnel Congress ’97, Vienna, vol 2
more advanced constitutive law would be needed in order Farmer I (1983) Engineering behaviour of rocks, 2nd edn. Chapman
to account for the long-term behavior of the ground. and Hall, London
Farrokh E, Rostami J (2009) Effect of adverse geological condition on
The results described are valid for a given set of rock TBM operation in Ghomroud tunnel conveyance project. Tunn
mass and lining parameters. The conclusions drawn can be Undergr Space Technol 24(4):436–446
generalized when considering the effects of advance rates Fjaer E, Holt RM, Horsrud P, Raaen AM, Risnes R (2008) Petroleum
on the potential for machine entrapment. However, the load related rock mechanics, 2nd edn. Elsevier, Amsterdam
Graziani A, Ribacchi R, Capata A (2007) 3D-modelling of TBM
and displacement values as indicated are valid only for the excavation in squeezing rock mass. Brenner Basistunnel und
input parameters used. In all cases, based on the 3D sim- Zulaufstrecken, Internationales symposium BBT 2007. Inns-
ulation tools as developed, generalized guidelines for the bruck University Press, Innsbruck, pp 143–151

123
R. Hasanpour et al.

Hasanpour R, Rostami J, Ünver B (2014) 3D finite difference model AFTES International Congress, Monaco, Montecarlo. Edition
for simulation of double shield TBM tunneling in squeezing specifique Limonest, pp 163–172
grounds. Tunn Undergr Space Technol 40:109–126 Ramoni M, Anagnostou G (2010) Tunnel boring machines under
Hoek E, Guevara R (2009) Overcoming squeezing in the Yacambú- squeezing conditions. Tunn Undergr Space Technol 25:139–157
Quibor tunnel, Venezuela. Rock Mech Rock Eng 42(2):389–418 Ramoni M, Anagnostou G (2011a) The interaction between shield,
Itasca FLAC3D Manual (2006) Fast Lagrangian analysis of continua ground and tunnel support in TBM tunnelling through squeezing
in 3D dimensions. User’s guide ground. Rock Mech Rock Eng 44:37–61
Lombardi G (1981) Bau von Tunneln bei grossen Verformungen des Ramoni M, Anagnostou G (2011b) The effect of consolidation on
Gebirges. In: Internationaler Kongress, Tunnel 81, Düsseldorf, TBM shield loading in water-bearing squeezing ground. Rock
Messegesellschaft mbH NOEWA Düsseldorf und Deutsche Mech Rock Eng 44:63–83
Gesellschaft für Erd- und Grundbau e.V., Essen, vol 2, Schmitt JA (2009) Spannungsverformungsverhalten des Gebirges
pp 351–384 beim Vortrieb mit Tunnelbohrmaschinen mit Schild. Heft 89,
Lombardi G, Panciera A (1997) Problems with TBM and linings in dissertation, Institut für Grundbau und Bodenmechanik, Techni-
squeezing ground. Tunnels and Tunnelling International No. 29. sche Universität Braunschweig
Miller Freeman plc, London, pp 54–56 Shalabi FI (2005) FE analysis of time-dependent behavior of
Lombardi G, Neuenschwander M, Panciera A (2009) Gibraltar tunneling in squeezing ground using two different creep models.
Tunnel Project update—the geomechanical challenges. Geo- Tunn Undergr Space Technol 20:271–279
mech Tunnel 2(5):578–590 Sterpi D, Gioda G (2007) Ground pressure and convergence for TBM
Pellet F, Roosefid M, Deleruyelle F (2009) On the 3D numerical driven tunnels in visco-plastic rocks. In: ECCOMAS Thematic
modelling of the time-dependent development of the damage Conference on Computational Methods in Tunnelling (EURO:-
zone around underground galleries during and after excavation. TUN 2007), Vienna University of Technology, pp 89–95
Tunn Undergr Space Technol 24:665–674 Wittke W, Wittke-Gattermann P, Wittke-Schmitt B (2007) TBM-
Ramoni M, Anagnostou G (2006) On the feasibility of TBM drives in heading in rock, design of the shield mantle. In: ECCOMAS
squeezing ground. Tunn Undergr Space Technol 21(3–4):262 Thematic Conference on Computational Methods in Tunnelling
Ramoni M, Anagnostou G (2007) Numerical analysis of the (EURO:TUN 2007), Vienna University of Technology, p 98
development of squeezing pressure during TBM standstills. Zhao K, Janutolo M, Barla G (2012) A completely 3D model for the
The second half century of rock mechanics, 11th Congress of the simulation of mechanized tunnel excavation. Rock Mech Rock
International Society for Rock Mechanics (ISRM), Lisbon, vol 2. Eng 45(4):475–497
Taylor & Francis Group, London, pp 963–966
Ramoni M, Anagnostou G (2008) TBM drives in squeezing rock–
shield-rock interaction. Building underground for the future.

123

View publication stats

You might also like