Tunnelling and Underground Space Technology 59 (2016) 221–229

Contents lists available at ScienceDirect

Tunnelling and Underground Space Technology

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

Effect of fibres on early age cracking of concrete tunnel lining.

Part II: Numerical simulations
M. Briffaut a,b,⇑, F. Benboudjema c, L. D’Aloia d
a
Univ. Grenoble Alpes, 3SR, F-38000 Grenoble, France
b
CNRS, 3SR, F-38000 Grenoble, France
c
LMT-Cachan, ENS-Cachan, CNRS, Université Paris Saclay, France
d
CETU Tunnel Study Centre (CETU) French Ministry of Ecology, 25 Avenue François Mitterrand, 69674 Bron Cedex, France

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

Article history: The early-age cracking of concrete structures increases permeability and diffusivity and moreover
Received 18 February 2015 accelerates the penetration of liquid, gas and aggressive agents. Consequently, the serviceability of these
Received in revised form 29 April 2016 structures could be reduced drastically. Early-age cracking might be due to external loading, but also to
Accepted 31 July 2016
the internal or external restraint resulting from autogenous, drying and thermal shrinkage. This study
Available online 18 August 2016
focuses more specifically on these latter phenomena.
In the first part of this study (see effect of fibres on early-age cracking of concrete tunnel lining - Part I:
Keywords:
Laboratory testing), ring tests were performed to investigate the sensitivity of concrete to cracking due to
Shrinkage
Cracking
both shrinkage strain and type of fibre (two organic fibres and one steel fibre were studied).
Ring tests simulations Ring test results were then used to validate the capacity of a chemo-thermo-viscoelastic damage model
Fibre-reinforced concrete aimed at reproducing the complex behaviour of fibre-reinforced concrete subjected to restrained
Concrete tunnel lining shrinkage through identifying the material parameters with standardised tests. The numerical
simulations conducted on a real tunnel lining show that for the studied geometries and concrete
mixtures, thermal shrinkage constitutes the major phenomenon capable of causing early-age transverse
cracks and moreover crack opening is highly dependent on the type of reinforcement. Modifications to
both fibre type and lining thickness may serve to avoid the onset of transverse cracks.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction cracks to appear. In addition, temperature and relative humidity

Early-age cracking may occur in massive concrete structures
due to strains generated by internal and external restraints. The
chemical reactions between cement and water are exothermic
and result in thermal dilatation followed by thermal shrinkage
(temperature may exceed 60 °C in massive concrete structures).
Moreover, during hydration reactions, capillary pressures are
induced by water consumption and external drying. Consequently,
autogenous and drying shrinkage occurs. Regardless of their origin,
when autogenous, thermal and/or drying strains are restrained due
to boundary conditions and gradients, stresses are generated
(Gawin et al., 2006; Benboudjema and Torrenti, 2008; Azenha
et al., 2009; Briffaut et al., 2011; Klemczak and Knoppik-Wróbel,
2015; El Bitouri et al., 2016). In the case of external restraint (foun-
dations, rock formation for tunnels, a cold joint), the maximum
concrete tensile strength may be reached, thus causing transverse
gradients increase during hydration and drying, which in turn
leads to self-equilibrated stresses and concrete skin cracking. The
serviceability of massive structures, such as tanks, bridges, nuclear
containment and tunnels, may be reduced due to the penetration
of aggressive agents like carbon dioxide, sulphate and chloride
ions. Moreover, cracking may induce a significant increase in con-
crete permeability and become responsible for water inflows in the
tunnel when no seal has been installed behind the concrete lining
(according to the RATP,1 40% of Paris subway stations are subjected
to this phenomenon). For this particular construction method, a
thick lining is supposed to ensure sealing of the structure.
To avoid cracking or at least limit crack opening, the tunnel lin-
ing could be reinforced by a welded steel mesh with a limited
structural role. Nevertheless, positioning this reinforcement is a
time-consuming process and the concrete cover may not fulfil its
function, thus leading to disorders associated with early corrosion.

⇑ Corresponding author at: Univ. Grenoble Alpes, 3SR, F-38000 Grenoble, France. 1
Régie Autonome des Transports Parisiens (public transit operator for the Paris
E-mail address: matthieu.briffaut@3sr-grenoble.fr (M. Briffaut). Metropolitan Area, France).

http://dx.doi.org/10.1016/j.tust.2016.08.001
0886-7798/Ó 2016 Elsevier Ltd. All rights reserved.
222 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229
Companies responsible for tunnel projects are therefore interested
in substituting a fibre reinforcement (either metal or polypropy-
lene) for this welded anti-crack mesh, while keeping in mind that
polypropylene microfibres also prevent concrete from spalling in
the case of fire. For hardened concrete, the use of macrofibres in
concrete tunnel lining for the purpose of replacing structural rein-
forcements has already been studied (Nanakorn and Horii, 1996;
Buratti et al., 2011, 2013; Chiaia et al., 2009).
The objective of this study is to investigate the effect of fibres on
early-age cracking, more specifically the cracking due to thermal
shrinkage. No individual study has in fact been found in the liter-
ature by the authors, whereas drying and autogenous shrinkage
restraints have been widely investigated, especially on shotcrete
(e.g. Malmgren et al., 2005; Ansell, 2010; Bryne et al., 2014a).
In the second part, numerical simulations will be performed in
order to study the impact of fibres on the shrinkage cracking of a
concrete tunnel lining. The chemo-thermo-mechanical analysis of
the behaviour of concrete at early ages has been investigated in-
depth on simple structures based on either empirical formulae
(Xiang et al., 2005; Jeon et al., 2008) or a viscoelastic damage
model (Benboudjema and Torrenti, 2008; Briffaut et al., 2011;
Buffo-Lacarriere et al., 2011). However, studies focusing on the
effect of both fibres and structural geometry (in the case of tunnel
lining) are still lacking. After a short description of the model used
in this study, the material parameters will be identified on exper-
imental results, presented in the part I. The ring test will then be
introduced to validate the ability of the model to reproduce the
early-age behaviour of FRC. Lastly, drying, thermal and mechanical
simulations of the concrete lining of an actual tunnel will be pre-
sented, by highlighting the influence of each phenomenon (ther-
mal, autogenous and drying shrinkage) and then describing in
greater detail the influence of both fibres and cross-section
thickness.
2. Numerical modelling of the frc behaviour
2.1. Thermo-hydro-chemo-mechanical modelling
At early age, several phenomena occur simultaneously. These
phenomena must be taken into account in order to provide a rele-
vant analysis of the thermal active ring test results and highlight
the effect of fibres. The thermo-hydro-chemo-mechanical model
used is a viscoelastic model with coupling between creep and dam-
age. It has been described in further detail in de Sa et al. (2008),
Benboudjema and Torrenti (2008) and Briffaut et al. (2011). Only
the main features of this numerical model will be provided
hereafter.
The evolution of hydration is successfully modelled by intro-
ducing a chemical affinity (Ulm and Coussy, 1998) and then con-
sidering that the chemical reactions are thermo-activated
according to Arrhenius’ Law (Regourd and Gauthier, 1980). The
temperature evolution is derived by the energy balance equation,
which includes the heat release due to hydration reactions, as well
as by assuming that the thermal boundary conditions are convec-
tive. Since drying is very slow (up to 106 times slower than heat
transfer), no coupling between hydration and drying has been
considered.
The autogenous shrinkage strain and thermal strain (respec-
tively eau and eth) are assumed to evolve linearly with the degree
of hydration (Ulm and Coussy, 1998; Mounanga et al., 2006) and
with the coefficient of thermal expansion considered as a constant
(Laplante and Boulay, 1994). The drying shrinkage strain (eds) is
calculated using the approach proposed in Baroghel-Bouny et al.
(1999), which is based on the mechanics of porous media. Drying
creep strain edc is modelled using the approach proposed by
Bažant and Chern (1985). During hydration, the mechanical beha-
viour of concrete is modelled by an isotropic elastic damage model
coupled with creep. Such a set-up proves to be sufficient for pre-
dicting cracking due to restrained strain (De Sa et al., 2008;
Briffaut et al., 2011) since the stresses generated are mainly tensile
ones. Both Young’s modulus E and tensile strength ft evolve with
increases in the degree of hydration (De Schutter, 1999), leading
to a continuously increasing tensile strain threshold with hydra-
tion (Benboudjema and Torrenti, 2008).
The relationship between apparent stresses r, effective stresses
~ , damage D, elastic stiffness tensor E, elastic strains ee , basic creep
r
strains ebc (see thereafter), total strains e and previously defined
strains is expressed as follows:
r ¼ ð1  DÞr~ ¼ ð1  DÞEðnÞe_ e
¼ ð1  DÞEðnÞðe_  e_ bc  e_ au  e_ th Þ ð1Þ
where D is given by the following law inspired by Mazars’ proposal
(Mazars, 1986):
j0
D¼1 ½ð1 þ At Þ expðBt ^eÞAt expð2Bt ^eÞ ð2Þ
^e
where At (set equal to 1) and Bt are material parameters that control
the softening branch of the stress-strain curve in tension. Strain
softening induces inherent mesh dependence and leads to failure
without any energy dissipation (Pijaudier-Cabot and Bažant,
1987). In order to dissipate the same amount of energy after mesh
refinement, when strains have localised in a row of finite elements,
a characteristic length lc relative to mesh size is introduced (Rots,
1988; Cervera and Chiumenti, 2006). The fracture energy also
depends on the degree of hydration (De Schutter and Taerwe, 1997).
Lastly, the evolution of Bt parameter with respect to the degree
of hydration is expressed as:
ð1 þ At =2Þ
Bt ðnÞ ¼ Gft ðnÞ f t ðnÞ
ð3Þ
lc f t ðnÞ
 2EðnÞ
where n is the degree of hydration, lc the characteristic length of the
finite element, ft the tensile strength, E the Young’s modulus, and Gft
the fracture energy.
High stress levels or post-peak loading lead to nonlinear creep
strains in both compression and tension (which may induce fail-
ure). According to Mazzotti and Savoia (2003), creep strains ebc
are partially included in the expression of the equivalent strain
defined by Mazars (1986).
To reproduce the reversible component of basic creep without
increasing the number of model parameters, Kelvin-Voigt units
are used (see, for instance, Benboudjema and Torrenti, 2008). The
stiffness parameter for each unit is calculated with the equation
proposed by De Schutter (1999) and slightly modified thereafter
(Benboudjema and Torrenti, 2008) to take temperature effects into
account (Ladaoui et al., 2012). According to Bažant and Prasannan
(1989), the characteristic time of each chain should not be a
parameter defining the evolution laws of creep strains. They
proposed a ratio of 10 between the characteristic times of two
consecutive chains, which enables avoiding the risk of solving
an ill-conditioned problem. To retrieve irreversible creep strains
after unloading, an ageing dashpot placed in series with the
Kelvin-Voigt chains has also been employed to model basic creep
(Briffaut et al., 2012).
2.2. Simulation of thermal active ring test results
A global strategy has been implemented to identify and validate
the entire set of parameters. Standardised tests, such as semi-
adiabatic tests, compressive tests, splitting tests and autogenous
shrinkage tests, make it possible to identify some of the key model
 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229
parameters (Tables A1 and A2 display the parameter values and
corresponding standardised test run). Experimental results are
presented in the part I. The other parameters are calculated from
results in the literature and adapted for the studied concrete. In
these tables, the parameters deduced from tests conducted on
the tunnel lining concrete mixture as well as those deduced from
other studies performed by the authors or found in the literature
are clearly indicated. For instance, viscoelastic parameters have
not been identified on creep tests but instead calculated by multi-
plying the values from another concrete mixture (studied in
Briffaut et al., 2012) by the Young’s modulus ratio of both con-
cretes. Creep properties are indeed highly correlated with Young’s
modulus (ACI Committee 209, 2008). Moreover, it is assumed that
the amplitude is the same in compression and tension, given the
lack of consensus in the literature (see Hilaire et al., 2014 for a lit-
erature review). According to the authors (see Briffaut et al., 2011),
the effect of creep on cracking prediction is expected to be
considerable.
Furthermore, the capacity of this model to reproduce the beha-
viour of fibre-reinforced concrete at an early age subjected to
restrained thermal shrinkage is measured by the active ring test
simulations.
Since macroscopic numerical simulations will be undertaken,
the fibres cannot be meshed. The fibre effect is taken into account
by increasing the global tensile strength in accordance with exper-
imental results (see Part I). The low fibre content does not lead to
hardening in the post-peak behaviour, which helps justify our
approach. The mesh used is given in Fig. 1a (only one-fourth of
the ring).
An example of damage pattern is shown in Fig. 1b.
The results of the numerical simulations of the thermal active
ring test are displayed in Fig. 2.
The good agreement between experimental and numerical
results suggest that the effect of fibres on the tensile creep beha-
viour is rather small, which contradicts the observations recorded
by (Bissonnette and Pigeon, 1995) and (Lamour et al., 2006), who
indicated that de-bonding between the fibres and the cement paste
matrix could lead to an apparent tensile creep strain. For the tun-
nel simulations therefore, only the tensile strength and fracture
energy have been modified to take the type of fibre into account.
3. Numerical simulation of the tunnel lining
For the numerical simulation of the tunnel, the mesh was gen-
erated from tunnel drawings for the construction site underlying
this study (Fig. 3). Three types of materials with different proper-
ties were taken into account: the surrounding limestone rock
Face 2
Holes for
fluid
circulaon
Brass
Fig. 1. Mesh used in the simulation of the active ring test (a). Example of concrete damage pattern (b).
223
(5-m thickness); concrete lining (0.5-m thickness); and shotcrete
(0.2-m thickness).
The concrete lining is cast in place by 10-m long blocks. The
variations in shotcrete thickness (‘‘fish scale” pattern) are due to
the tunnel boring method employed (i.e. the cutting machine).
Drying shrinkage (approx. 3 weeks between shotcrete spraying
and tunnel lining casting) could lead to shotcrete cracking
(Ansell, 2010; Bryne et al., 2014a) or de-bonding (Malmgren
et al., 2005). Nevertheless, the behaviour of rock and shotcrete is
assumed to be elastic (Young’s modulus values of 20 and 35 GPa,
respectively) and adhesion is considered to be perfect (both
between rock and shotcrete and between shotcrete and concrete).
While the adhesion of shotcrete subjected to tension has in fact
been studied (Malmgren et al., 2005; Bryne et al., 2014b), the shear
strength of the interface between materials (concrete, shotcrete
and rocks) remains an open issue. It should therefore be kept in
mind that in our simulation, the restraint on the concrete tunnel
lining has probably been overestimated, leading in turn to an over-
estimation of the damage pattern (pessimistic prediction).
A 5-m thick rock has been chosen to ensure that the mechanical
and thermal concrete boundary conditions are representative of
the surrounding rock. Calculations with a 20-m rock thickness
have actually been performed according to an elastic analysis
(see Fig. 4). Results show that the difference in longitudinal stress
calculation in the concrete lining is less than 1% (along the x-axis in
Fig. 3).
It is worth noting that no external load is applied on the tunnel
since the purpose of the tunnel lining is to prevent water inflow;
moreover, ultimate tunnel convergence will require more than a
few weeks. Lastly, thermal simulations have also been run with
various rock thicknesses, in revealing that 5 m is sufficient for esti-
mating temperature evolution (with the max. temperature differ-
ence less than 0.1 °C).
3.1. Thermo-chemical results
An ambient temperature of 20 °C has been chosen and held con-
stant over time. This assumption is consistent with actual temper-
ature values measured in the tunnel. In the simulation, the
temperature exchange between the studied tunnel lining block
and the concrete previously cast and hardened has been neglected.
The convective heat transfer coefficient adopted takes into account
the steel formwork, which had been removed after 24 h. The tem-
perature field at 24 h is shown in Fig. 5.
In this case, the relatively moderate maximum temperature is
due to the partial replacement of clinker by blast furnace slag in
the cement. An increase of just 25 °C has been recorded. For mas-
sive concrete structures, limiting the maximum temperature is a
Face 1
Concrete
224 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229

400 400
Numerical strains (without fibers) Numerical strains (with PMiF)
350 350

Experimental strains (without fibers) 300 Experimental strains (PMiF)

300
Ring brass strains (µm/m)

Ring brass strains (µm/m)

250
250
200
200
150
150
100
100
50

50 0
0 10 20 30 40 50 60 70 80 90 100
0 -50
Time (hours)
0 10 20 30 40 50 60 70 80 90 100
Time (hours)

(a) (b)

400 400
Numerical strains (PMaF) Numerical strains (with MF)
350 350
Experimental strains (PMaF) Experimental strains (with MF)
300
Ring brass strains (µm/m)

300
Ring brass strains (µm/m)

250
250
200
200
150
150
100
100
50

50 0
0 10 20 30 40 50 60 70 80 90 100 110 120
0 -50
0 10 20 30 40 50 60 70 80 90 100 Time (hours)
Time (hours)
(c) (d)
Fig. 2. Comparison between experimental and numerical results: without fibres REF (a), FRC-PMiF-1.8 (b), with FRC-PMaF (c), with FRC-MF (d).

1 -1
6.3m 2 Temperature probe
Error (%) with respect to simulation

(for maturometry)
with a rock thickness of 20m

-2
3.7m 3
y -3

x -4
Rock thickness = 1m
10m Rock thickness = 2m
z Rock thickness = 5m
-5 Rock thickness = 10m
Fig. 3. Tunnel mesh: the limestone rock (1), the shotcrete (2) and the concrete Rock thickness = 15m
lining (3).
-6
0 0.1 0.2 0.3 0.4 0.5
concrete lining depth (m)
critical issue with respect not only to thermo-mechanical strains
(when restrained and in considering gradient effects), but also to Fig. 4. Influence of rock thickness on elastic stress in the tunnel lining.
the risk of delayed ettringite formation (DEF), especially under
wet conditions and in the presence of groundwater. Due to the
variability in concrete tunnel lining thickness, it can be remarked To the extent that the concept of maturity has been introduced
that the temperature field is not homogenous along the x-axis on the construction site in order to optimize the time of formwork
(see Fig. 3). removal, concrete temperature measurements are available (see
 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229 225

45

40

35

Temperature [°C]
30

25

20

In situ data
Fig. 5. Temperature field after 24 h.
model
15
0 50 100 150 200 250 300
Fig. 3 for the location of the temperature probe). Fig. 6 shows the
results of both temperature measurements and the numerical Time [h]
simulation.
Fig. 6. In situ temperature of concrete lining and numerical simulation.

3.2. Mechanical results
3.2.1. Determination of the relative contribution of each phenomenon
Even though the ring test simulations are in good agreement
with experimental data, several key features regarding concrete
behaviour and numerical issues at the structural scale could not
be calibrated by simulation of the laboratory device. To obtain a
cracking pattern close to what might be observed in actual struc-
tures, the simulated construction of a fibre-reinforced concrete
tunnel lining should take size effects into account. In the present
case, a decrease in the tensile strength for large structures
(L’Hermite, 1973; van Vliet and van Mier, 2000) and a regulariza-
tion (dissipation of the same energy quantity during cracking
regardless of mesh size (Hillerborg et al., 1976; Matallah and La
Borderie, 2009)) have been introduced. It is worth noting that for
fibre-reinforced concrete, size effects tend to decrease with an
increase in fibre content (Li et al., 1998). Hence, since the studied
fibre contents remain below 1% by volume, the influence of fibres
on size effects is assumed to be negligible.
The simulated cracking pattern resembles what is typically
observed inside tunnels by the CETU (French Tunnel Study Centre),
which supervises the majority of tunnel inspections in France. This
pattern is mainly due to the restraint of thermal expansion and
autogenous shrinkage by the surrounding limestone rock.
Comparing damage patterns with and without taking creep into
consideration (Fig. 7) serves to confirm that creep at an early age
has a favourable effect and must be taken into account in the anal-
ysis of numerical results. It can also be observed in this Figure that
thermal shrinkage is the primary phenomenon leading to trans-
verse cracks. Drying only really generates damage on the concrete

Damage field due to both thermal and Damage field due to both thermal and
autogenous shrinkages after 360 hours (in autogenous shrinkage after 360 hours
considering creep) (in neglecting creep)

Damage field due to thermal shrinkage after Damage field due to drying shrinkage
360 hours after 1,000 days

Fig. 7. Damage field due to different phenomena in the case of the reference mixture (REF).
226 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229
skin as expected, but this type of cracking has little effect on the
permeability of the concrete lining itself.
3.2.2. The influence of the type of fibre
The damage field, and hence the calculation of crack opening,
will strongly depend on the type of fibre. This dependence is due
to changes in the tensile strength induced by adding fibres, as well
as to the post-peak behaviour, which reveals a more ductile mate-
rial. Since the post-peak behaviour is heavily modified by the pres-
ence of fibres, a determination of toughness (i.e. fracture energy) is
not straightforward. For a high fibre content, the softening phase
(which is characteristic of a plain concrete) is in fact replaced by
a hardening behaviour. In this case, several methods for determin-
ing concrete fracture energy have been proposed, but no consensus
could be reached for any one of them (e.g. Banthia and Mindess,
2004). The fracture energy concept can be contested by virtue of
corresponding to the fracture energy of concrete and the debond-
ing / frictional resistance of fibres. However, debating this topic
is not the aim of the present paper. Besides, the fibre contents cho-
sen in this study are relatively low and even though fracture
energy increases, the concept remains relevant. Finally, fracture
energy is expected to impact only crack openings and spacing,
but not the prediction of crack occurence. In the case of steel fibre
reinforcements, the fracture energy can overcome 1000 J m2 (Bei-
xing et al., 2004) and is strongly dependent on fibre content (Barros
and Sena Cruz, 2001). The values extracted from the literature for
this study are reported in Table 1.
The damage fields obtained are displayed in Fig. 8 for the vari-
ous types of fibre. As discussed in Section 3.2.1, drying shrinkage
Table 1
Fracture energy values used in the numerical simulations.
Fracture energy (J m2)
REF 100
FRC-PMiF-0.9 230
FRC-PMaF 280
FRC-MF 1200
Damage field without fibres (REF)
Damage field with FRC- PMaF
Fig. 8. Damage field for the different types of fibre (after 360 h).
has not been taken into account. It can be observed that the dam-
age pattern with PMiF resembles that obtained without fibres,
whereas with PMaF, the damage seems to be less. Within the
framework of this study, PMiF, which are usually used to prevent
cracking due to plastic shrinkage, i.e. shrinkage at a very early
age (before 48 h), appear to be irrelevant in preventing or signifi-
cantly reducing structural early-age cracking. The largest effect is
found with steel fibres, whereby both the number of damaged
areas and the magnitude of damage are reduced.
3.2.3. The influence of lining thickness and rock/concrete interface
The initial mesh was designed from tunnel drawings, and the
changes in material behaviour induced by adding fibres have been
studied. Another approach to limiting or preventing cracking might
consist of modifying the tunnel lining geometry. Since thermal
shrinkage is the main phenomenon leading to transverse cracks,
a decrease in tunnel lining thickness will serve to limit the thermal
evolution, hence thermal cracking. Moreover, in a real configura-
tion, the variable thickness of concrete lining generates a non-
uniform stress and then a preferential crack location. To study
the effect of stress concentration, simulations with a smooth and
regularly-shaped interface between the shotcrete and concrete lin-
ing have been run. The resulting damage fields are presented in
Fig. 9.
3.2.4. Impact of cracking on permeability
The damage pattern analysis is interesting yet it remains qual-
itative. Since the surrounding rock is often self-sufficient to sustain
loading (essentially a dead load in the present case), the thick con-
crete lining mainly plays the role of sealant, insofar as no specific
system is typically installed by the RATP in Paris subway tunnels.
To propose a quantitative analysis of the results obtained in this
study, cracks and then flows through cracks have been calculated.
The cracking is mainly due to mode 1 since the stress is primarily
tensile one; also, since a regularized local damage model has been
used, the crack is concentrated in one row of the element.
The crack opening is calculated by numerically extracting the
Damage field with FRC –PmiF -1.8
Damage field with FRC - MF
 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229 227

Damage field without fibres (REF) Damage field with a smooth interface
between lining and shotcrete

Damage field with a 30-cm concrete

thickness
Fig. 9. Damage field for different geometrical configurations (after 360 h) for the reference mixture (REF).

displacement gap between undamaged areas. For more complex
cracking, processing methods have recently been developed to
calculate the crack opening from damage fields (see, for instance,
Matallah et al., 2010; Dufour et al., 2008, 2012). These methods
required to remove the delayed strains, but they do not take into
consideration how the coupling between cracking and creep must
be handled. An example of this analysis of the effect of fibres is
presented in Fig. 10. In this example, it can be remarked that PMiF
decrease the crack opening, given that the displacement gap is
narrowed. Moreover, macroscopic fibres (PMaF and MF) enable
avoiding the appearance of macroscopic transverse cracks.
-4
x 10
1 that cracking is mainly due to a differential thermal shrinkage.
Displacement (m)
Table 2 indicates the decrease (or increase) of each simulated
situation in terms of flow through the cracks. The flow is calculated
using Poiseuille’s Equation. The reference flow has been calculated
by considering thermal shrinkage, autogenous shrinkage and
creep, and the same applies for the concrete mixture without
fibres.
4. Conclusion
The initial objective of this study was to investigate the influ-
ence of various types of fibre on the sensibility to cracking due to
the restraint of thermal and autogenous shrinkages. Thermo-
chemo-mechanical simulations showed that the heat released by
cement hydration may lead to early-age cracking and moreover
The effect of autogenous shrinkage (for the studied concrete mix-
ture) is relatively small. The drying shrinkage effect can be consid-

0.5
ered negligible due to the substantial thickness of the concrete
lining: only surface cracks are to be expected. Besides, the pre-
0 dicted damage is found to be greatly affected by creep. The calcu-
lation of crack opening has highlighted the effect of fibres in
-0.5
displaying a slight decrease in the global permeability of the con-
crete lining. Macroscopic fibres also seem capable of preventing
the appearance of macroscopic transverse cracks. Even though it
-1 Without fibers
might seem counterintuitive, decreasing the concrete tunnel lining
FRC - MF
FRC - PMaF thickness serves to decrease water inflow into the tunnel because
FRC - PMiF - 1.8
-1.5 thermal shrinkage is reduced.
1 2 3 4 5 6 7 8 9 10 In offering perspective, numerical simulations will be per-
concrete lining abscissa (m) formed in order to extend this analysis. First of all, long-term
numerical simulations will allow studying the impact of drying
Fig. 10. Effect of fibre reinforcement on longitudinal displacement on sidewalls. shrinkage. Afterwards, the use of a transfer model that takes into

Table 2
Influence of involved phenomena, type of reinforcement and tunnel geometry on the global flow through cracks.

Without creep REF FRC - PMaF FRC - MF FRC - PMiF - 1.8 kg/m3 Low lining thickness (30 cm) Smooth interface
Q/Qref 1.12 1.00 0.00 0.00 0.62 0.49 0.44
228 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229

Table A1 References
set of thermal concrete parameters used in the simulations.

Parameter Value Unit Measured Standardised test used Azenha, M., Faria, R., Ferreira, D., 2009. Identification of early-age concrete
temperatures and strains: monitoring and numerical simulation. Cem. Concr.
for this
Compos. 31, 369–378.
study
ACI Committee 209, 08, 2008. Guide for Modeling and Calculating Shrinkage and
Ea 45,729 J mol1 Calculated from concrete Creep in Hardened Concrete, Technical Report no. 209. 2R-08. American
mix (Schindler, 2004) Concrete Institute.
a 64.4 (–) X Semi adiabatic test Ansell, A., 2010. Investigation of shrinkage cracking in shotcrete on tunnel drains.
b 18,042 (–) X Semi adiabatic test Tunn. Undergr. Space Technol. 25, 607–613.
c 94,620 (–) X Semi adiabatic test Banthia, N., Mindess, S., 2004. Toughness characterization of fibre reinforced
concrete: which standard to use? ASTM. J. Test. Eval. 32 (2), 138–139.
d 215,819 (–) X Semi adiabatic test
Baroghel-Bouny, V., Mainguy, M., Lassabatere, T., Coussy, O., 1999. Characterization
e 280,339 (–) X Semi adiabatic test
and identification of equilibrium and transfer moisture properties for ordinary
f 208,172 (–) X Semi adiabatic test and high-performance cementitious materials. Cem. Concr. Res. 29, 1225–1238.
g 67,901 (–) X Semi adiabatic test Barros, J.A.O., Sena Cruz, J.M., 2001. Fracture energy of steel fibre-reinforced
C 2400 kJ K1 m3 Calculated from concrete. Mech. Compos. Mater. Struct. 8 (1), 29–45.
concrete mix Bažant, Z.P., Chern, J.C., 1985. Concrete creep at variable humidity: constitutive law
k 2.8 W m1 K1 Typical value and mechanism. Mater. Struct. 18, 1–20.
L 117,840 kJ m3 X Semi adiabatic test Bažant, Z.P., Prasannan, S., 1989. Solidification theory for concrete creep I.
h 8 W m2 K1 Typical value Formulation. J. Eng. Mech. 115 (8), 1691–1703.
Bei-xing, L., Ming-xiang, C., Fang, C., Lu-ping, L., 2004. The mechanical properties of
polypropylene fibre reinforced concrete. J. Wuhan Univ. Technol.-Mater. Sci. Ed.
19 (3), 68–71.
Benboudjema, F., Torrenti, J.M., 2008. Early-age behavior of concrete nuclear
containments. Nucl. Eng. Des. 238 (10), 2495–2506.
Table A2 Bissonnette, B., Pigeon, M., 1995. Tensile creep at early ages of ordinary silica fume
set of mechanical concrete parameters used in the simulations. and fibre reinforced concretes. Cem. Concr. Res. 25 (5), 1075–1085.
Boulay, C., 2007. Développement d’un dispositif de mesure du retrait endogène d’un
Parameter Value Unit Measured Standardised test béton au jeune âge, Huitième édition des Journées Scientifiques du
in this used Regroupement Francophone pour la Recherche et la Formation sur le béton
study (RF)2B, Montréal, Canada, 5–6 juillet 2007 (in French).
Briffaut, M., Benboudjema, F., Torrenti, J.M., Nahas, G., 2011. Numerical analysis of
j 53 lm m1 X Autogenous the thermal active restrained shrinkage ring test to study the early age behavior
shrinkage tests BT of massive concrete structures. Eng. Struct. 33 (4), 1390–1401.
Jade; (Boulay, 2007) Briffaut, M., Benboudjema, F., Torrenti, J.-M., Nahas, G., 2012. Concrete early age
a 7.5 lm m1 K1 X Free expansion basic creep: experiments and test of rheological modelling approaches. Constr.
measurement on Build. Mater. 36, 373–380.
hydrated concrete Bryne, L.E., Ansell, A., Holmgren, J., 2014a. Investigation of restrained shrinkage
between 20 and 40 °C cracking in partially fixed shotcrete linings. Tunn. Undergr. Space Technol. 42,
aE 0.62 (–) Typical value 136–143.
E1 45 GPa X Compressive tests Bryne, L.E., Ansell, A., Holmgren, J., 2014b. Laboratory testing of early age bond
aft 1 (–) Typical value strength of shotcrete on hard rock. Tunn. Undergr. Space Technol. 41, 113–119.
Buffo-Lacarriere, L., Sellier, A., Turatsinze, A., 2011. Finite element modelling of
ft1 3.6–4.78 MPa X Splitting test
hardening concrete: application to the prediction of early age cracking for
n0 0.216 (–) X Compressive tests
massive reinforced structures. Mater. Struct. 44 (10), 1821–1835.
n1 0.78 (–) X Calculated from W/C Buratti, N., Mazzotti, C., Savoia, M., 2011. Post-cracking behaviour of steel and
At 1 (–) Typical value macro-synthetic fibre-reinforced concretes. Constr. Build. Mater. 25, 2713–
Bt 450–5400 (–) Calculated from Gft 2722.
d 0.46 (–) (De Schutter and Buratti, N., Ferracuti, B., Savoia, M., 2013. Concrete crack reduction in tunnel linings
Taerwe, 1997) by steel fibre-reinforced Concretes. Constr. Build. Mater. 44, 249–259.
Gft1 100–1200 N m1 Chiaia, B., Fantilli, A.P., Vallini, P., 2009. Combining fiber-reinforced concrete with
k11
bc 1125 GPa Typical value traditional reinforcement in tunnel linings. Eng. Struct. 31, 1600–1606.
k21
bc 560 GPa Typical value Cervera, M., Chiumenti, M., 2006. Mesh objective tensile cracking via a local
k31
bc 330 GPa Typical value continuum damage model and a crack tracking technique. Computat. Meth.
s11 0.1 Days Typical value Appl. Mech. Eng. 196, 304–320.
bc
s21 1. Days Typical value De Sa, C., Benboudjema, F., Thiery, M., Sicard, J., 2008. Analysis of microcracking
bc
induced by differential drying shrinkage. Cem. Concr. Compos. 30 (10), 947–
s31
bc 1.0 Days Typical value
956.
Eac 1700 J mol1 Typical value
De Schutter, G., Taerwe, L., 1997. Fracture energy of concrete at early ages. Mater.
k 6.50  1013 Pa1 K1 de Sa et al. (2008)
Struct. 30 (2), 67–71.
W/C 0.44 (–) X Calculated from de Schutter, G., 1999. Degree of hydration based Kelvin model for the basic creep of
concrete mix early age concrete. Mater. Struct. 32, 260–265.
Dufour, F., Pijaudier-Cabot, G., Choinska, M., 2008. Extraction of a crack opening
from a continuous approach using regularized damage models. Comput. Concr.
5 (4), 375–388.
Dufour, F., Legrain, G., Pijaudier-Cabot, G., Huerta, A., 2012. Estimate of crack
account both water permeation into the sound material and crack opening from a 2D continuum-based FE computation. Int. J. Numer. Anal. Meth.
orientation will serve to consolidate the results obtained in this Geomech. 36, 1813–1830.
study. El Bitouri, Y., Buffo-Lacarrière, L., Sellier, A., Bourbon, X., 2016. Modelling of chemo-
mechanical behaviour of low pH concrete. Cem. Concr. Res. 81, 70–80.
Gawin, D., Pesavento, F., Schrefler, B.A., 2006. Hygro-thermo-chemo-mechanical
Acknowledgment modelling of concrete at early ages and beyond. Part I: Hydration and hygro-
thermal phenomena. Int. J. Numer. Meth. Eng. 67, 299–331.
Hilaire, A., Benboudjema, F., Darquennes, A., Berthaud, Y., Nahas, G., 2014. Modeling
The authors want to thank XELIS2 and RATP companies for basic creep in concrete at early-age under compressive and tensile loading.
fruitful discussions and given access to internal documents. Nucl. Eng. Des. 269, 222–230.
Hillerborg, A., Modeer, M., Petersson, P.E., 1976. Analysis of crack formation and
crack growth in concrete by means of fracture mechanics and finite elements.
Cem. Concr. Res. 6, 773–782.
Appendix A
Jeon, S.J., Choi, M.S., Kim, Y.J., 2008. Advanced assessment of cracking due to heat of
hydration and internal restraint. ACI Mater. J. 105 (4), 325–333.
See Tables A1 and A2. Klemczak, B., Knoppik-Wróbel, A., 2015. Reinforced concrete tank walls and bridge
abutments: early-age behaviour, analytic approaches and numerical models.
Eng. Struct. 84, 233–251.
2
Now known as the SYSTRA Agency.
 M. Briffaut et al. / Tunnelling and Underground Space Technology 59 (2016) 221–229
L’Hermite, R., 1973. Influence de la dimension absolue sur la résistance à la flexion.
Annale de l’ITBTP 30 (310), 39–41 (in French).
Ladaoui, W., Vidal, T., Sellier, A., Bourbon, X., 2012. Analysis of interactions between
damage and basic creep of HPC and HPFRC heated between 20 and 80 °C. Mater.
Struct. 46 (1–2), 13–23.
Lamour, V., Moranville, M., Chaignon, J., 2006. Effect of low volume polypropylene
fibres on early age cracking of concrete. In: ConMat’05 – 3rd International
Conference on Construction Materials: Performance, Innovations and Structural
Implications. Vancouver, Canada.
Laplante, P., Boulay, C., 1994. Evolution du coefficient de dilatation thermique du
béton en fonction de sa maturité aux tout premiers âges. Mater. Struct. 27, 596–
605 (in French).
Li, V.C., Lin, Z., Matsumoto, T., 1998. Influence of fiber bridging on structural size-
effect. Int. J. Solids Struct. 35 (31–32), 4223–4238.
Malmgren, L., Nordlund, E., Rolund, S., 2005. Adhesion strength and shrinkage of
shotcrete. Tunn. Undergr. Space Technol. 20, 33–48.
Matallah, M., La Borderie, C., 2009. Inelasticity-damage-based model for numerical
modelling of concrete cracking. Eng. Fract. Mech. 76, 1087–1108.
Matallah, M., La Borderie, C., Maurel, O., 2010. A practical method to estimate crack
openings in concrete structures. Int. J. Numer. Anal. Meth. Geomech. 34, 1615–
1633.
Mazars, J., 1986. A description of micro and macroscale damage of concrete
structures. Eng. Fract. Mech. 25 (5–6), 729–737.
229
Mazzotti, C., Savoia, M., 2003. Nonlinear creep damage model for concrete under
uniaxial compression. J. Eng. Mech. 129 (9), 1065–1075.
Mounanga, P., Baroghel-Bouny, V., Loukili, A., Khelidj, A., 2006. Autogenous
deformations of cement pastes: Part I. Temperature effects at early age and
micro–macro correlations. Cem. Concr. Res. 36, 110–122.
Nanakorn, P., Horii, H., 1996. A fracture mechanics based design method for SFRC
tunnel linings. Tunn. Undergr. Space Technol. 11 (1), 39–43.
Pijaudier-Cabot, G., Bažant, Z.P., 1987. Nonlocal damage theory. J. Eng. Mech. ASCE
113, 1512–1533.
Regourd, M., Gauthier, E., 1980. Behavior of cement submitted to accelerate
hardening. Annales de I’ITBTP 179, 65–96 (in French).
Rots, J.G., 1988. Computational modeling of concrete fracture PhD Dissertation.
Delft University of Technology, Netherlands.
Schindler, A.K., 2004. Effect of temperature on hydration of cementitious materials.
ACI Mater. J. 101, 72–81.
Ulm, F.-J., Coussy, O., 1998. Couplings in early-age concrete: from material
modelling to structural design. Int. J. Solids Struct. 35 (31–32), 4295–4311.
van Vliet, M.R.A., van Mier, J.G.M., 2000. Experimental investigation of size effect
in concrete and sandstone under uniaxial tension. Eng. Fract. Mech. 65,
165–188.
Xiang, Y., Zhang, Z., He, S., Dong, G., 2005. Thermal–mechanical analysis of a newly
cast concrete wall of a subway structure. Tunn. Undergr. Space Technol. 20,
442–451.