>68 Technology Innovation in Underground Construction

Chapter 13

Safety assessment during construction

of shotcrete tunnel shells using
micromechanical material models

Stefan Scheiner, Bernhard

Pichler, Christian Hellmich &
Josef Eberhardsteiner
Institute for Mechantes of
Materials .id Structures,
Vienna University of
Technology, Vienna, Austria

ABSTRACT
Knowledge of the stresses in shotcrete tunnel shells
is of great importance for assessing their safety
against severe cracking or failure. Estimation of
these stresses from optical measurements of 3D
shell displacements requires shotcrete material
models capable to deal with variations of the water-
cement and the aggregate-cement ratio. This is the
motivation for employing two representative
volume elements within a continuum microme-
chanics framework: One of them relates to cement
paste (with a spherical material phase representing
clinker, needle-shaped hydrate phases with
isotropically distributed spatial orientations, a
spherical water phase, and a spherical air phase,
with all phases being in direct mutual interaction),
whereas the second one relates to shotcrete (with a
spherical aggregate phase, embedded into a matrx
phase made up by cement paste). Elasticity
homogenization follows self-consistent schemes (at
the cement paste level) and Mori-Tanaka estimates
(at the shotcrete level). Stress peaks in the
hydrates related to quasi-brittle material failure are
>68 Technology Innovation in Underground Construction
estimated by second-order stress averages over
hydrates, derived from the elastic energy stored in
the RVE. These higher-order stress averages permit
upsealing from the hydrate strength to the
shotcrete strength. Experimental data from
resonant frequeney tests, ultrasonics tests,
adiabatic tests, uniaxial compression tests, and
nanoindentation tests suggest that early-age
(evolving) shotcrete elasticity and strength can be
reasonably well predicted from mixture-and
hydration-independent mechanical properties of
aggregates, clinker, hydrates, water, and air, and
from the strength properties of the hydrates.
Notably, the model-predicted final strength (at
completed hydration) almost perfectly follows the
famous Feret formula. At the structural level, the
micromechanics model, when combined whit 3D
displacement measurements of the tunnel shell,
allows for assessing the safety of the
shotcrete tunnel shell as a function of the
shotcrete mix (defined in terms of the water-
cemertt ratio and of the aggregate-cement
ratio). Related micromechanics-based hybrid
analyses of shotcrete tunnel shells predict
that structural safety is rather sensitive to
changes in the water-cement ratio, whereas
standard-type variations in the aggregate-
cement ratio, resulting from rebound during
shotcreting, have virtually no influence on the
overall structural safety.

13.1 INTRODUCTION
If a high degree of flexibility is required
during the tunnel excavation process (e.g. in
difficult ground conditions or in urban reas),
the New Austrian Tunneling Method (NATM),
pioneered by (Rabcewicz 1948), is well suited.
When driving tunnels according to the NATM,
after excavation of a cross-section of a
tunnel, shotcrete is sprayed onto the tunnel
walls, constituting a thin and flexible shell.
Accordingly, the excavation mode (type of
cross section, sequence of excavation steps,
time span between excavation and
shotcreting) and the shotcrete mixture
characteristics (such as water<.-nt ratio
wlc and aggregate-cement ratio ale) are key
>68 Technology Innovation in Underground Construction
design parameters in NATM-tunneling: for
example, too stiff shells or shells put up too
early, trying to resrrict stress redistributions
in the ground caused by the excavation, may
bedestroyed, whereas too soft shells or shells
put up too late may not be able to prevent
loss of sta-bility of the opening (Pacher
1964). The stability of the opening is often
improved by reinforcing the sutrounding
ground by bolts. The success of the NATM is
primarily based on the in situ realization of
its concept by expenenced engineers. This
reali-zation is supported by elabrate
measuiement systems on site [see, e.g.
(Rabcewicz et al, 1972)]. These systems have
been continuously improved, and nowadays,
3D ptica! displacement monitoring of the
tunnel shell (Schubert and Steindorfer 1996;
Steindrfer et al. 1995) is the golden
standard. From changes of characteristic
trend parameters, extracted from 3D
displacement ineasurements, it is possible to
estmate changes in the geological structure
of the soil or rock (Schubert et al. 2002).
Besides the monitoring of variations in the
geological conditions, knowledge on the
srresses in the tunnel shell is of great
importance, as to assess its safety against
severe cracking or failure. Thereforc, the
shell forces can be derived from quasi-
instantaneous deformations of holes cored
out of the tunnel shell (Celestino et al. 1997).
Since this semi-destructive method is
restricted in applicability and versatil-it>, it
turned out to be more benefioial to back-
calculate, on the basis of elabrate shotcrete
material models, shotcrete stresses from
shotcrete strains (Schubert 19SS).
Preferentially, the strains are gained from the
aforementioned 3D displacement meas-
urements, as proposed in (Rokahr and
Zachow 1997) on the basis of relative move-
ments of pairs of vsasuremenr points. At a
more elabrate state, the strain fields can be
estimated on the basis of a hybrid method
(Hellmich 1999; Hellmich et al. 2001), in
which displacement vector fields are
approximated from measured displacement
vectors at discrete points of the tunnel shell
>68 Technology Innovation in Underground Construction
and in which these fields are prescribed as
boundary vales for a three-dimensional
Finite Element structural model of the tunnel
shell. This type of method has also been
extended towards segmented tunnei linings
(Lackner et al. 2002; Macht et al. 2000), and
use of approximations from thin shell theory
(Lackner et al. 2006) turned out to be beneficial for
day-to-day use of the hybrid method in engineering
practice (Brandmer et al. 2007).
Underlying shotcrete models need to reasonably
represent the creep behavior of the material
(Lechner et al. 2001; Meschke 1996; Rokahr and
Lux 1987; Schubert 1988), and consideration of
hydration-induced, thermal, and chemical strains
further improves the reliablity of shell forc
estimation (Hellmich et al. 1999b; Hellmich et al.
1999a; Lechner et al. 2001; Sercombe et al. 2000).
All aforementioned material models rely on
material properties, which are shotcrete mixture-
specific. Henee, any change in mixture (e.g.
variations in water-cement ratio, as often
encountered in situ) can only be considered if
additional experiments (related to strength, creep,
and shrinkage) are performed on samples with the
modified shotcrete composition. Praeti-cally, this is
often unfeasible so that the engineers on site
usually agree on a "A-pical" shotcrete for a tunnel
track. Clearly, this situation is unsatisfactory. As a
remedy, the shotcrete composition needs to be
incorporated into the shotcrete material models,
within a micromechanical framework: This was
recently shown by (Hellmich and Mang 2005) for
the case of elasticity following earlier work on
concrete by (Bernard et al. 2003) and on bone by
(Hellmich and Ulm 2002). While elasticity is an
impor-tant factor for attracting stresses iato the
tunnel shell, strength is the key factor to
understand whether these stresses significantly
compromise the shells safety. Also the strength
evolution is influenced by the w/c and alo ratios
(Neville 1981), and this here motivares us to
present recent developments on a
micromechanical model for shotcrete elasticity and
strength, and to employ this new rriodel for a
hybrid analysis of an NATM-tunnel shell, (see
Pichler et al. 2008; Pichler et al. 2009).
>68 Technology Innovation in Underground Construction
13.2 MODELING CEMENTITIOUS MATERIALS IN THE
FRAMEWORK OF CONTINUUM MICROMECHANICS

13.2.1 Fundamentis of micromechanics -

Representative volume element (RVE)
In continuum micromechanics (Hill 1963; Suquet
1997; Zaoui 1997; Zaoui 2002), a.material is
understood as a macro-homogeueous but micro-
heterogenecnis body fill-ng a representative
volume eiement (RVE) with characteristic length i, i
d, d stand-ing for the characteristic length of
inhomogeneities within the RVE, and l< L, L
standing for the characteristic lengths of geometry
or loading of a structure built up by the material
defined on the RVE. In general, the microstructure
within an RVE is too complicated to describe it in
full detail. Therefore, quasi-homogeneous
subdomains with known physical quantities (such
as volume fracturas or elastic properties) are rea-
sonably chosen. They are called material phases.
The homogenized mechanical behavior of che
overall material, i.e., the relation between
homogeneous deformations acting on the
boundary of the RVE and reslting (average)
stresses, or the ultmate stresses sustainable by
the RVE, can then be estimated from the
mechanical behavior of the aforementioned
homogeneous phases representing the
inhomogeneities within the RVE), their dosages
within the RVE, their characteristic shapes, and
their interactions. If a single phase exhibits a
heterogeneous microstructure itself, its mechanical
 >68 Technology Innovation in Underground Construction

behavior can be estimated by introduction of an

RVE within this phase, with dimensons < d,
comprising again smaller phases with
characteristic length d2 l2, and so on, leading
to a multistep homogenization scheme.

13.2.2
Micromechanical
representation of
cementitious
materials
For cementitious materials (such as shotcrete),
we employ two RVEs: The first one relates to
cement paste (with phases representing
clinker, water, hydrates, and air), and the
second one to shotcrete (with phases
representing cement paste and aggregates),
see Fig. 13.1.
The RVE of cement paste, with characteristic
length - = 0.25-0.50 mm, see Fig. 13.1(a),
consists of four different material phases with
characteristic dimensions d = 1-50 microns
(see also Fig. 13.2): (i) a spherical clinker
phase, (ii) needle-shaped

(a) Homogenization step I: cement paste (b) Homogenization step II: shotcrete

homogeuized material and zero volume fraccin - -

 >68 Technology Innovation in Underground Construction

figure 13.1 Micromechanical representation of shotcrete microstructure through a two-step hom

>68 Technology Innovation in Underground Construction

hydrate phases with isotropically

distributed spatial orientations, (iii) a
spherical water phase, and (iv) a
spherical air phase. The RVE of
shotcrete, in turn, with the
characteristic length c = 7-10 cm, see
Fig. 13.1(b), consists of material
phases with characteristic dimensions
dsc = 1-15 mm: (i) cement paste matrix,
and (ii) aggregate inclusions.

13.2.3 Elasticity and strength

of cementitious materials
13.2.3.1 Morphological
representation of hydration
producs in cement paste
Several micromechanical concepts
(Sanahuja and Dormieux 2005;
Heukamp et al. 2005; Barthlmy and
Dormieux 2004; Lemarchand et al.
2002) have been put for-ward in the
context of strength of cementitious
materials. All these concepts are based
on elastoplasticity theory: At the level
of concrete, they employ rigid,
spherical aggre gate inclusions
(potentially coated by a cohesive-
frictional interfacial transition zone),
being embedded in a Drucker-Prager-
type cement paste matrix; and one
model pro poses to represent cement
>68 Technology Innovation in Underground Construction

paste as rigid cement inclusin coated

by von Mi ses-type hydrates, cast,
together with spherical pores, into a
self-consistent matrix. The focus of
these important works is of rather
qualitative nature, discussing e.g. the
eff ect of aggregate volume fractions on
overall properties. Convineing
experimental validation of these
models has not been accomplished so
far.
The micromechanical strength models
proposed in (Pichler et al. 2008; Pichler
et al. 2009) resolve the cementitious
materials down to the level of single
hydrates (set Fig. 13.2 for microscopic
images of hydrates) within the cement
paste. Thereby, the hydrates are
represented as needle-shaped particles with
isotropic orientation distri-"bution.
This innovative investment into a more
realistic morphological representation
of hydrates allows for use of a simple
strength law at the microscale of the
individual hydrates. Instead of
employing raTher sophisticated
elastoplastic developments, the
proposed models are based on an
elastic limit criterion for the deviatoric
stress peaks within The hydrates. In the
foliowing, it is shown that this simple
criterion has high potental for
explaining the microscopic origin of the
failure of cementitious materi als. This
is discussed fi rst for cement paste, and
then for quasi-brittle shotcrete (repre -
sented as a cement paste matrix with
aggregate inclusions).

13.2.3.2 Strength of cement paste

>68 Technology Innovation in Underground Construction
If cement paste is subjected to uniaxial
compression, the observed axial stress-
strain behavior shows three regimes:
1 Initially, the material is linear
elastic [see ascending branch in
Fig. 13.3 (a)];
2 in the vicinity of the compressive
strength, axiai#tams increase
overlinearly with increasing axial
stresses [see nonlinear part of the
curve in Fig. 13.3 (a)]; and
3 once the peak load is reached,
cement paste exhibits brittle failure:
under stress-control, the material
exhibits instant loss of its load
carrying capacity (Harsh et al. 1990).
0.061
elasiic limit4of0.002
cement pasee

modeling approach proposed herein

4 0.00
0.001 0.002 compressive axial0.003
0.0G0 strairt [-] 0.000 3
5 Figure 13.3 Typica! stress-strain diagrams of
cement paste subjected to uniaxial
compression: (a) experimenta! data from
(Psakhie et al. 2001), (b) modeling approach
proposed by (Pichler et al. 2009).

These experimental observations motivated to

approximate the constitutive behavior of cement
paste as linear elastic, right up to (brittle) failure,
see Fig. 13.3 (b). Consequently, strength of
cement paste can be studied in the framework of
elastic limit analysis. Since cement paste is a
microheterogeneous material, the macroscopic
elastic limit of cement paste corresponds to the
elastic limit of one o the material phases: In the
present case, it is the elastic limit of the
hydrates. In more detail, each hydrate is
assumed to behave linear elastically as long as
deviatoric stress peaks remain below a critical
valu (the hydrate-specific elastic limit).
>68 Technology Innovation in Underground Construction
Reaching, due to increase of macroscopic
uniaxial compression, this critical valu in the
most heavily stressed regin o the hydrate
phase corresponds to the macroscopic elastic
limit of cement paste, which is associated with
overall failure of the material. In order to
estmate stress peaks within micron-sized
needle-shaped hydrates, which are part of
millimeter-sized voluntes of cement paste
subjected to uniaxial compression, this model
relies on quad-ratic stress averages over
orientation-specific subvolumes of the hydrate
phase, for details see (Pichler et al. 2009).
Predictive capabiiities of this model are
summarized in Section 13.3.

13.2.3.3 Strength of shotcrete

If shotcrete is subjected to uniaxial compression,
the axial stress-strain behavior shows three
regimes which are qualitatively similar to those
observed in cement paste, see the above listing at
the beginning of Section 13.2.3.2. The pre-peak
nonlineari-ties of cement paste (regime 2),
however, are by far less pronounced than the ones
o shotcrete. The latter are commonly explained by
successive bond failure oceurring in the so-called
"interfacial transition zone (TZ)" between
aggregates and the sur-rounding cement paste
matrix, see, e.g. (Neville 1981; Hofstetter and Mang
1995). Once the peak load is reached, a
second failure mechanism is observed: cracks start
to propgate also through the cement paste matrix,
see, e.g. (Neville 1981; Hofstetter and Mang 1995).
These experimental observations have rvv direct
implications on the mode of modeling performed by
(Pichler et al. 2008): since modeling of (uniaxial
compressive) shotcrete strength (bcing related
to peak loads in stress-strain diagrams) is the
current topic of interest, rather than the dcscriprion
of pre-peak nonlinearities, the micromechanical
formularion was not complicated throgh
introduction of ITZs, but clearly focussed on the
onset of cracking in cement paste, by use of an
elastic limit criterion related to the hydrates as
weakest elements of cement paste. Regarding ihc
challenge to estimate stress peaks within micron-
 >68 Technology Innovation in Underground Construction
sized hydrates, which are part of decimeter-sizcd
volumes of shotcrete subjected to uniaxial
compression, the concept of quadratic stress
averages was extended to inulti-scale materials
(Pichler et ai. 2008). Predictive capabilities of the
related model are summarized in the next Section.

13.3 EXPERIMENTAL VALIDATION OF

MICROMECHANICS-BASED MATERIAL MODELS

The micromechanical model presented in Section

13.2, based on the universal phase properties o Ta
ble 13.1 (Pichler et al. 2008; Pichler et a!. 2009),
will be fed with shotcrete mixture - and hydration
kinetics-specific input data concerning material
composition, i.e. with experimental vales for the
volume fractions of air, water, clinker, hidrates,
cement paste, and aggregate: fgr, fH fdht, f^d, f and fa
. Por these input data, the model delivers
predictions fot mixture- and hydration-specific
shotcrete stiffnesses and strengths. Comparison of
these predictions to corre.sponding experi-mentally
derived vales allovvs lor assessing the predictive
capabilities of the model.

6 13.3.i Mixture-dependent shotcrete

The volume fractions inside an RVE o cement pasre depend on the degree of hydration q which
composition

7
8 with the mass densities of clinker, water,
and hydrates, pdh, ph0, a.n:phyd, following from
(Acker 2001), see also (Pichler et al. 2009): pdin =
3.15 kg/dm3, pHJ= 1 kg/dm3, and phyd = 2.073
kg/dm3. The creation of air voids filling far stems
>68 Technology Innovation in Underground Construction
from the fact that hydration producs occupy a
smaller volume than their reactants, see e.g.
(Acker and Ulm 2001). In contrast to the volume
fractions within the RVE, the volume fractions of
cement paste and aggregates do not change
during hydration of the material. They can be
determned from the water-cement ratio {wlc),
the aggregate-cement ratio (ale), and the mass
densities of aggregates, water, and clinker, p .
pHl0, and pchn (Acker 2001; Pichler et al. 2009),
through (Bernard et al. 2003)

9
10 with pagg = 2.5 kg/dm3. The evolution of the
degree of hydration is determned by means
of adiabatic tests where the measured
accumulated hydration heat is consid-ered
a s a n indicator for the hydration ptogress of
the investigated shotcrete specimen (Ulm
and CoussyT996; Hellmich 1999).
11
12 13.3.2 Experimental validation on
cement paste level
13 Young's moduli, shear moduli, and the
compressive strengths of cement paste are
estimated by the self-consistent
#.omogenization step depicted in Fig. 13.1(a),
for mixture- and hydration degree-specific
volume fractions (13.1)-(13.5), on the basis
of universal phase stiffnesses (Table 13.1)
and strength vales.
14 The microelastic model of cement paste is
validated by comparing model-predicted to
experimentally obtained dynamic Young's
moduli and dynamic shear moduli, for wlc
ratios ranging from 0.35 to 0.60 (Sun et al.
2005). The agreement between model
predictions and corresponding experimental
tesults, obtained under drained conditions, is
satisfactory, see Figs. 13.4(a)-(c) and (Pichler
et al. 2009).
15
 >68 Technology Innovation in Underground Construction
16 (b)
17 (d)
uniaxial compressive strength of drained18
wlc = 0.60 shotcrete [MPa]

wlc = 0.35 18
wlc = 0.50
model predictions: seaied conditions drained conditions

0 2.5 5 7.5 10 12.5 dynamic Young's modulus

01020 30[GPa] (c)
'experimental measurements

model predictions: sealed conditions drained conditions

/ model predictions: 11 sealed conditions drained conditions
0 5 10 15 20 25 dynamic Young's modulus [GPa]
0 2.5 5 7.5 10 2.5 15 17.5 dynamic Young's modulus [GPa]
18 fi gure 13.4 Comparison of model
predictions with experimental
data characterized by diff erent wlc
ratios (wlc = 0.35, 0.50, 0.60): (a)-
(c) dynamic shear modulus versus
dynamic Young's modulus, and (d)
uniaxial compressive strength of
drained cement paste.
19
20 By comparing model-
predicted to experimentally
obtained vales for the
uniaxial strength of cement
pastes with different WLC ratios,
a prediction accuracy,
quantified by a squared
correlation coefficient r2 =
97%, is obtained, see Fig.
13.4(d). This cor-roborates the
assumption of the single
strength-type valu for
hydrates, C& = 26 MPa (Pichler
et al. 2009).
21
22 13.3.3 Experimental validation on
shotcrete level
>68 Technology Innovation in Underground Construction
23 A second homogenization
step, resulting in a two-step
homogenization scheme, is
necessary to predict
stiffnesses and strengths of
shotcrete, see Fig. 13.1(b).
24 For the sake of experimental model
validation, the model-predicted Young's
odulus of shotcrete is compared to
corresponding experimental vales of
(Lafarge 002) who subjected a shotcrete
characterized by wlc = 0.48 and ale =
5.3, to reso-ant frequeney tests, see Fig.
13.5(a). The agreement between-model
predictions and xperiments is excellent
for sealed conditions (underlined by a
mean relative error f 1.0% and a
corresponding standard deviation of
6.5%), confirming findings in Hellmich
and Mang 2005) that the tests of (Lafarge
2002) are rather characterized y sealed
than by drained conditions. Model
predictions related to sealed conditions
orrelate to experimental results by r 1 =
98.8%, see Fig. 13.5(a),
25 Experimental validation of the
micromechanics-based homogenization of
shot-rete strength is carried out
according to the experiments of (Lafarge
1997) and (Pillar 002): The applied
mefhodology of experimental strength
determination (Hilti gun nd penetrometer
tests) suggests that model predictions
referring to drained condi-ions are cioser
to the experimental findings than the
ones referring to sealed con-itions
(Pichler et al. 2008). The agreement
between model predictions (related to
rained conditions) and experimental data,
characterized by wlc = 0.5, ale = 3.8
Pillar 2002), and by wlc = 0.4, ale =
3.94 (Lafarge 1997), is quantified by a
mean elative prediccin error of -5.0%
and by a related standard deviation o
19.4%. The elatve errors constitute
upper bounds (Pichler et al. 2008),
suggesting that the corre-rtion between
>68 Technology Innovation in Underground Construction
model predictions and experimental
vales of r 2 = 95% is satisfactory, ee Fig.
13.5(b).
26
27 13.4 MICROMECHANICS-BASED
CHARACTERIZATION OF SHOTCRETE:
INFLUENCE OF WATER-CEMENT AND
AGGREGATE-CEMENT RATIOS ON
ELASTICITY AND STRENGTH EVOLUTIONS
 >68 Technology Innovation in Underground Construction
28 the experimentally validated
micromechanics model (see Section 13.2)
allows for pre-sicting hydration degree-
dependent evolutions of Young's
modulus, Poisson's ratio, and he uniaxial

gure 13.5 Comparison of model predictions with experimental data characterized by wlc = 0.48
compressive strength of different
shotcrete mixtures as functions of their
wlc and ale ratios, see Figs. 13.6 and 13.7.
We observe that Poisson's ratio increases or
decreases with increasing water-cement
ratio, for drained or sealed conditions,
respec-tively, see Figs. 13.6(b) and 13.7(c,d).
Sealed conditions, as a rule, lead to higher
stiff-ness and strength vales when
compared to drained conditions, but this
difference becomes very small for complete
hydration ( t ) > 1), see Figs. 13.6 and 13.7.
We also observe that both Young's modulus
and the uniaxial compressive strength
decrease with increasing water-cement ratio,
whereby the percental decrease of Young's
modulus is smaller than that of the
compressive sttength, see Figs. 13.6(a) and
(c). The loss in final strength (at completed
hydration) with increasing water-cement
ratio almost perfectly follows Feret's famous
empirical relationship (Feret 1892), stating
that the final strength [L"lt ( t = )] is
proportional to the square of a ratio between
the volumes of cement, water, and air
contained in a material volume of concrete,
29
30
>68 Technology Innovation in Underground Construction
31 where P is a factor of proportionality. This
relationship has proven remarkable use-fulness,
and it is widely used for mix designs in the
cement and concrete industry: The match
between this relationship (with P = 197.2 MPa)
and the model-predicted strength at completed
hydration, see Fig. 13.8, further corroborares the
relevance of our model, in addition to the
experimental evidence given in Section 13.3 and
in ear-lier publications (Pichler et al. 2009; Pichler
et al. 2008).
32 Remarkably, Feret's formula (13.6) does
not include the aggregate volume, whereas our
model directly accounts for the influence of the
aggregate-cement ratio. However, increase of
the ale ratio from 3.J to 5.0 results in an increase
of Young's modulus of only up to 10%, see Figs.
13.7(a) and (b), whereas Poisson's ratio and the
uniaxial compressive strength are virtually
unaffected by such variations of the shotcrete
mixture, see Figs. 13.7(c)-(.f). It is concluded that
for typical shoteretes used in NATM tunneling,
the ale ratio plays a minor role in derermining
the pverall mechanical properties - and this is
beneficial to the reliability of structural
computations as will be detalled in the next
subsection.
33
34 13.5 CONTINUUM
MICROMECHANIC S-BASED
SAFETY ASSESSMENT OF
NATM TUNNEL SHELLS
>68 Technology Innovation in Underground Construction
chemomechanical phenomena in shotcrete,
especially concerning autogeneous shrink-age
and creep, are considered macroscopically: In
detall, the thermochemomechanical material
law proposed by (Hellmich et al.
1999b; Sercombe et al. 2000; Lechner
et al. 2001) is employed, considering
the relations shown in Fig. 13.6 for
aging elastic ity and strength. The
remaining material functions (for
chemical affi nity, creep, and
shrinkage) are given in (Lechner et
al. 2001), whereby short-term creep
is considered according to (Macht et
al. 2001). For simulations based on
the hybrid method, displacements
measured at km 156.990 of the
Sieberg tunnel (Upper Austria) are
prescribed as boundary conditions for
a three-dimensional Finite Eiement
model of the tunnel shell as shown in
Fig. 13.9, compare (Hellmich 1999;
Hellmich et al. 1999; Hellmjch et al.
2001). Simulation results are
illustrated in terms of the so-called
level of loading which can be inter-
preted as the degree of utilization.
The latter is defi ned as the-ratio of
the loading (stress) over loading
capacity (strength) of the shotcrete
tunnel shell. More specifi -cally, it is -
on the basis of a Drucker-Prager
failure surface calibrated for uniaxial
and biaxial compressive failure. In
the present evaluation we focus on
the average nature of L over the
tunnel shell thickness h, with r and <p
as polar coordinates defi ning
(macroscopic) positions within a
circular shell segment, and we
desgnate the mximum valu of C((p,i)
in the tunnel shell, m a x (), as the
"level of loading" in Figs. 13.10(a)
and (b).
36
>68 Technology Innovation in Underground Construction
37 13.5.1 Water-cement
ratio-dependence of structural
safety
38 Three shotcrete mixtures are
investigated: wlc = 0.40 (mix I), wlc
= 0.50 (mix II), and wlc = 0.60 (mix
III), each with ale = 5, see Fig.
13.6. The simulation results show
that throughout the observed
loading phase the resulting level
of loading of the tunnel shell is
decisively infl uenced by the wlc
ratio, see Fig. 13.10. On the other
hand, the devia-tions between the
levis of loading predicted for
sealed conditions in the shotcrete,
see Fig. 13.10(a), only negligibly
diff er from those predicted for
drained conditions in the
shotcrete, see Fig. 13-10(b). Still,
we note that, in principal, sealed
conditions result in slightly lower
levis of loading as compared to
drained conditions.
39
40 13 3.5.2 Aggregate-
cement ratio-dependence of
structural safety
41 During shotcreting, shotcrete
constituents may detach from the
sprayed material,
>68 Technology Innovation in Underground Construction

which is referred to as rebound.

Assuming rebound to concern
aggregates only (mdin =
const., mH20 = const.), the ale ratio
decreases with respect to the
targeted ale ratio,
denoted as (alc)hrget, namely: ale =
(alc)target - A(alc)rAam. Considering that
the targeted
mass of shotcrete, is composed of
the mass of clinke r, wate r, and
aggregates,
42
43
(a) (b)

degree of hydration {-]

44
45
 >68 Technology Innovation in Underground Construction

46
Safety assessment during construction of

47 Figure 13.6 Micromechanics-based input

for hybrid analyses of Section 13.5:
Evolutions of (a) Young's modulus E, (b)
Poisson's ratio V, and (c) uniaxial
compressive strength f respectively, over
the hydration degree ; diagrams refer to
shotcrete with ale = 5, three different wlc
ratios [wlc = 040, 0 50, 0.60), under drained
as well as under sealed conditions.
48
49 And that the shotcrete mass remaining
on the tunnel wall, m _. is related to m : by
und R,
50
51 allows, through substitution of. (13.8) into
(13.9), for est i marin of the actual ale ratio
function of the rebound R, the water cement
ratio wlc, and the targeted aggre--cement
ratio [alc)irs:,,,

52 ale = (alc)tam - R( + w/e + (a/e)tarJ. (13.10)

53 for wet shotcreting, as has been used in the
Sieberg tunnel, the rebound hardly exceeds
> (Hague 2001)1. In order to elucdate the
dependence o the structural safety ebound-
related variations of the ale ratio, eight
different mixes are investigated
54
55 Technology Innovation in
Underground Construction
56 (<0
57
(b)
0.4

o 0.3

0.2 0.4 0.6 0.8 degree of hydration [-]

0.2 0.4 0.6 0.8 degree o hydration [-]
0.2 0.4 0.6 0.8 degree of hydration [-]
[~]
 >68 Technology Innovation in Underground Construction
58

I 0.2 0.4 0.6 0.8 degree of hydration 0.2[-] 0.4 0.6 degree o hydration [-]
drained, a/c == 3.5 drained, a/c - 5.0 sealed, a/c = 4.5
drained, a/c - ---------drained, a/c - 4.5
4.0 ---------sealed. a/c = 4.0

59 sealed, a/c -*'

3.5
278 Technology Innovation n Underground Construction
60 13.7
Micromechani
cs- based
input for
hybrid
analyse s of
Section i3.5:
Evolutions of
(a.b) Young's
modulus .
(c,d)
Poisson's
ratio v; and
(e,f) uniaxial
compressive
strength fd
respectiveiy,
over the
hydration
degree g.
diagrams
refer to
shotcrete
with wlc =
0.40 (a,c,e)
and w/c =
0.50 (b,d,f),
four diff erent
ale ratios
(ale = 3.5,
4.0, 4.5, 5.0),
under drained
as well as
under sealed
conditions.
61
62 : Table 13.2 for
corresponding
rebounds as
(ALE) LV TL = 5.0):
W/C - 0.40 and A/C
= 3.5 x IV), W/C =
0.40 and ALE =
4.0 (mix V), WLC
= 0*40 and ALE =
4.5 (mix VI), WLC
= 0 and ALE = 5.0
(mix VII), WLC =
0.50 and ALE =
3.5 (mix VIII),
WLC = 0.50 and
ALE = (mix IX),
WLC = 0.50 and
278 Technology Innovation n Underground Construction
ALE = 4.5 (mix
X), as vveli as
WLC = 0.50 and
ALE = 5.0 x XI),
see Fig. 1.3.7
for the
mictomechanics-
based material
properties.
278 Technology Innovation n Underground Construction
uniaxial compressive strength of drained shotcrete [MPa]

010?030 40
Feret predictions
63 Hg
urc
13.
8
Co
mp
aris
on
of
mic
ro
me
cha
nic
s-
bas
ed
mo
del
pre
dict
ion
s
wit
h
pre
dict
ion
s of
Fer
et's
for
mul
a
(6)
spe
cifi
cd
for
P=
278 Technology Innovation n Underground Construction
197
.2
MP
a.

64

65 Figur
e
13.9
H/brid
metho
d for
determ
ination
of the
level of
loading
278 Technology Innovation n Underground Construction
from
prescri
ption
of
measur
ed
displac
ements
on a
three-
dimens
ional
Finicc
Eiemen
t
model
of the
tunnel
shell.
66
67 In contrast to the
wlc ratio, the
(effective) ale ratio
has no significant
Lnfluence of
loading of the
shotcrete tunnel
shell, see Fig.
13.11. This is
because the
rease-reiated
increase of Young's
modulus (see Fig.
13.7), letting
expect ?ation
degrees in the
tunnel shell, is
compensated by
stress
redistributions
278 Technology Innovation n Underground Construction
68

69 Figure 13.11
Evolution of
the level of
loading Cmlx as
278 Technology Innovation n Underground Construction

400 600 time [h]

function of the
time aftcr
fa) sealed conditions

(b) drained conditions

installation of
the top
heading for (a)
sealed
conditions.
70
71 wie = 0.40
278 Technology Innovation n Underground Construction
72
73 w/c = 0.S0
74
75 5.0 4.5 4.0
3.5
76 R = 0.0% R =
7.8% R = 15.6% R
= 23.4%
77 R = 0.0% R =
7.7% R = 15.4% R
= 23.l%
 78

79 in the tunnel shell. Henee, the level of

loading shows no significan! dependence on
the (actual) a/c ratio for the investigated
shotcrete mixtures. This structural behavior
can be considered as beneficial: Our
calculations suggest that such changes of the
(actual) a/c ratio because of shotcrete
rebound do not compromise the structural
safety of the tunnel shell. Since the
investigated shotcrete mixes exhibit a
mximum rebound of 23,4% (see Table 13.2),
thus comprising common rebounds in wet
shotcreting, the insenstity of the loading
level of the tunnel shell with respect to
aggregate rebound is a robust featuxe of the
NATM. It suggests that careful in sit.it
monitoring of the w/c ratio, as compared to
the ale ratio, is much more critical.
80

81 13.6 CONCLUSION
82 Many tunnels are constructed with the New
Austrian Tunneling Method (NATM). Combining tn-
situ data stemming from 3D optieal displacement
monitoring of NATM tunnel shells with realistic
material models for shotcrete "Hellmich 1999;
Hellmich et al. 1999; Hellmich and Mang 2005;
Hellmich et al. 2001; Hellmich et al. 199.9b;
Lechner et al. 2001; Brandtner et al. 2007) allows
engineers to look incide the tunnel shell.
Comparing interna! forces with the currenr
strength of the (hydrating; shotcrete results in
quantitative safety assessments. This hybrid
method reduces risks and ensures not only
structural integrity of the ning and safety of the
tunneling crew, but also economic construction. In
NATM tunneling. shotcrete composition is adapted
to changing needs. When employing macroscopic
material models, every change of shotcrete mix
requires an additional, expensive, and time-
consuming series of mechanical tests. We have
developed new micromechanics models which
economically account for the impact of the
shotcrete composition on the elasticity and
strength properties of the material. On this basis,
we have shown the pdtentially major influence of
the shotcrete composition (in particular of the
water-cement ratio) on the forces induced in a
NATM-tunnel shell. Henee, currently developed
micromechanical models are paving the way
towards on-site systems for real-time safety
assessment in Underground Construction with
strongly varying shotcrete mixture characteristics.
To further elucidare this role, it is highjV desir-able
to exterid the mieromechanical description of
shotcrete towards consideration of both creep and
shrinkage. As regards shotcrete creep, a
microviscoelastic model has already been
developed (Scheiner and Hellmich 2009), aliowing
for determina-tion of the macroscopic viscoelastic
material behavior of shotcrete on the basis of
elastic and viscoelastic properties of the shocrete
constituents on the microscopic scale.
Implementation of this model in the herein presented
analyses is topic of ongo-ing research. Micromechanics-
based modeling of shotcrete shrinkage, on the other
hand, requiring introduction of a poromechanical
framework, is a topic of future research. Another open
issue relates to the question whether high levis of load-
ing (such as those encountered in Figs. 13.9(b) and (c)
1120 hours after shell nstalla-tion) might be
overestimattons as a result of enforcing Cl-continuity of
displacements between shell components installed at
different time instants, see (Hellmich et a!. 1999;
Hellmich et al. 2001) for details. This underlines that
further improvement of data anatysis during NATM-
tunneling calis for even more refined applied mechanics
tools, both at the material level (microstcuctural level =
shotcrete) and at the (macro-) structure level (=tunnel
shell).
83
84
85
86
87
88
89
90
91
92
93
94
95
96
 97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118

119 REFERENCES
120 Acker, P. (2001). Micromechanical anatysis
of creep and shrinkage mechanisms. In F.-J.
121 Ulm, Z. Bazant, and F. Wittraann (Eds.K Creep,
Shrinkage and Durability Mechanics o f
122 Concrete and Other Quasi-brittle Materials,
6th International Conference CONCREEP
123 MIT, Amsterdam, pp. 5-26. Elsevier. Acker, P.
and F.-J. Ulm (2001). Creep and shrinkage of
concrete: physical origins and practica!
124 measurements. Nuclear Engineering and
Design 203(2-3), 143-158. Arrnelin, H. and N.
Banthia (1998). Development of a general model
of aggregate rebound for
125 dry-mix shotcrete (Part II), Materials and
Structures 31(207), 195-202. Baroghel-Bouny, V. (1994).
Caractrisation des ptes de ciment et des btons-
mthodes,
126 analyse, interprtation [Characterization of cernent
pastes anu concretes - methods, analysis,
127 interpretations]. Technical report, Laboratoire
Centra! des Ponts et Chausses. Paris, France.
128 In French.
129 Barthlmy, J.-F. and L. Dormieux (2004). A
micromechanical approach to the strength criterion of
Drucker-Prager materials reinforced by rigid inclusions.
International Journal for Numrica! and Analytical
Methods in Geornechanics 28(7-8), 565-582.
130 Bernard, O., F.-J. Ulm, and E. Lemarchand (2003). A
multiscale micromechanics-hydration model for the
early-age elastic properties of cement-based materials.
Cernent and Concrete Research 33(9), 1293-1309.
131 Bjandtner, P., B, Moritz, and P. Schubert (2007). On
the challenge of evluatng stress in a shotcrete lining:
Experiences gained in applying the hybrid analysis
method. Felsbau 25(5), 93-98.
132 Celestino, T., M. de Mariano, A. Ferreira, and M.
Guimaraes (1997). Undercoring technique for stress
measurement in shotcrete linings. In J. Golser, W.
Hinkei, and W. Schubert (Eds.), Tunnels for People,
Proceedings o f the World Tunnel Congress 1997,
Vienna, Austria, pp. 59-64. Balkema, Rotterdam, The
Netherlands.
133 Feret, R. (1892). On the compaetness of bydraulic
mortars [Sur la compacite des mortiers hydrauliques].
Anuales des Ponts et Chausses 7. 5-164. in
French.
134 llague, 1. (2001). Wet shotcreting -A total system
approach, tcanuaigc.basf.com (last dpwnload: July 31,
2008).
135 Harsh, S., Z. Shen, and D. Barwin (1990). Strain-
rate sensidvc behavior of cement paste aad mortar in
compression. ACI Materials Journal 87(5), 508-516.
136 Hellmich, C. (1999). Shotcrete as part o f the
New Austrian Tunneling Method: From
Thermochemomechanical Material Modeling to
Structural Analysis and Safety Assessment o f
Tunnels. Ph.D. thesis, Vienna Unversitv f Technology,
Vienna, Austria.
137 rlellmich, C, J. Macht, and H. Mang
(1999). Ein hybrides Verfahren zur
Bestimmung der
138 Auslastung von Spritzbetonschalen [A
hybrid method for determination of the
level of
139 utilization of shotcrete shells].
Felsbau 17(5), 422425. In Germn,
rlellmich, C. and H. Mang (2005).
Shotcrete elasticity revisited in the
framework of continuum
140 micromechanics: From submicron to
meter level. Journal of Materials in Civil Engineering
141 (ASCE) 17(3), 246-256. Hellmich, C, H.
Mang, and F.-J. Ulm (2001). Hybrid method
for quantifi cation of stress
142 states in shotcrete tunnel shells:
combination of 3D in situ displacement
measurements and
 143 thermochemoplastic material law.
Computers and Structures 79(22-25), 2103-2115.
Hellmich, C. and F.- J. Ulm (2002).
Micromechanical model for ultra- structural
stiff ness of
144 mineralized tissues. Journal o f Engineering
Mechanics (ASCE) 128(8), 898-908. Hellmich, C,
F.-J. Ulm, and H. Mang (1999a). Consi stent
linearization in-Finite Eiement analysis
145 of coupled chemo-thermal problems
with exo- or endothermal reactions.
Computational
146 Mechanics 24(4), 238-244. Hellmich, C.,
F.-J. Ulm, and H. Mang (1999b). Multisurface
chemoplasticity I: material model
147 for shotcrete. Journal o f Engineering
Mechanics (ASCE) 125(6), 692-701. Heukamp, F.
H., E. Lemarchand, and F.-J. Ulm (2005). The
eff ect of interfacial properties
148 on the cohesin of highly fi lled
componte materials. International Journal o f
Solids and
149 Structures 42(1), 287-305. Hill, R .
(1963). Elastic properties of reinforced
solids. Journal o f the Mechanics.and Physics o f
150 Solids 11(5), 357-372.
151 Hofstetter, G. and H. Mang (1995).
Computational Mechanics o f Reinforced Concrete
152 Structures. Vieweg & Sohn
Verlagsgesellschaft mbH. Lackner, R ., J.
Macht, C. Hellmich, and H. Mang (2002).
Hybrid method for analysis of
153 segmented shotcrete tunnel linings.
Journal o f Geotechnical and Geoenvironmental
154 Engineering (ASCE) 128(4), 298-308.
Lackne r, R ., J. Macht, and H. Mang (2006).
Hybrid analysi s method for on-line
quantifi cation
155 of stress states in tunnel shells.
Computer Methods in Applied Mechanics and
Engineering
156 195(41-43), 5361-5378. Lafarge (1997).
Internal report, Lafarge Centre Technique
Europe Central. Commissioned by
157 the Institute for Strength of Materials,
Vienna University of Technology, Austria.
Lafarge (2002). Internal report, Lafarge
Centre Technique Europe Central.
Commissioned by
158 the Institute for Strength of Materials,
Vienna University of Technology, Austria.
Lechner, M., C. Hellmich, and H. Mang
(2001). Short-term creep of shotcrete -
159 Thermochemoplastic material
modelling and nonlinear analysis of a
laboratory test and of
160 a NATM excavation by the Finite
Eiement method. In P. Vermee r, S. Diebels,
W. Ehlers, H.
161 Herjrmann, S. Luding, and E. Ramm
(Eds.), Proceedings o f the International Symposium
162 on Discontinula Modelling o f Cohesive-Frictional
Materials - Lecture Notes in Physics,
163 pp. 47-62. Springer, Berln.
Lemarchand, E., F.-J. Ulm, and L. Dormieux
(2002). Eff ect of inclusions on friction
coeffi cient
164 of highly fi lled composire materials.
Journal o f Engineering Mechanics (ASCE) 128(8),
165 876-884.
166 Macht, J., C. Hellmich, R . Lackne r, W.
Schubert, and H. Mang (2000). Assessment
of a support system for squeezing rock
conditions by means of a hybrid method.
Felsbau 18(6), 9-15.
167 Macht, J., R . Lackner, C. Hellmich, and
H. Mang (2001). Shotcrete creep
properties - Review of material tests and
structural analyses of tunnels. In F.-J. Ulm,
Z. Bazant, and F. Wittmann (Eds.),
Proceedings o f the 6tb International Conference on
Creep, Shrinkage and Durability Mechanics o f Concrete
and Other Quasi-Bri'l Materials, pp. 285-300.
Elsevier Science Ltd., Amsterdam.
168 vlehlhorn, G. (1996). Der Ingenieurbau:
Grundivissen - Werkstoff e,
Elastizittstbeorie [Basic knowledge in
civil engineering: Materials and elasticity
theory], Volume 4. Ernst 8c Sohn, Berln. In Germn.
169 vleschke, G. (1996). Consideratlon of aging of
shotcrete in the context of a 3-D viscoplastic material
model. International Journal for Numerical
Methods in Engineering 39(18), 3123-3143.
170 leville, A. (1981). Propertiesvf
Concrete (3 ed.). Pitman Publishing Limited.
171 'feuffer, M. and W. Kusterle (2001). Rheology
and rebound behaviour of dry-mix shotcrete.
172 Cement and Concrete Research
31(11), 1619-1625. 'ichler, B., C. Hellmich, and J.
Eberhardsteiner (2009). Spherical and acicular
representation
173 of hydrates in a micromechanical model for
cement paste - Prediction of early-age elasticity
 174 and strength. Acta Mechanica 203(3-4), 137-
162. ichler, B., S. Scheiner, and C. Hellmich (2008). From
micron-sized needle-shaped hydrates to
175 meter-sized shotcrete tunnel shells:
micromechanical upscaling of stiffness and
strength of
176 hydrating shotcrete. Acta Geotechnica
3(4), 273-294. illar, N. (2002). Determination
of early age properties of fi bre
reinforced shotcrete to predict
177 the cracking behavior. Ph.D. thesis,
submitted to the University of New South Wales,
Prvate
178 Communication, Sydney, Australia, sakhie, S., Y.
Florie, G. Ostermeyer, S. Korostelev, A. Smolin, E. Shilko,
A. Dmitriev, S. Blatnik,
179 M. Spegel, and S. Zavsek (2001). Movable
cellular autmata method for simulating materials
180 with mesosrructure. Theoretical and Applied
Fracture Mechanics 37(1-3), 311-334. .abcewicz, L.
(1948). Patentschrift. sterreichisches Patent Nr. 165573
[Patent specification,
181 Austrian patent no. 165573). In Germn,
abcewicz, L., J. Golser, and E. Hackl (1972). Die
Bedeutung der Messung im Hohlraumbau,
182 Teil I [The significance of measurement in
tunneling]. Der Bauingenieur 47(1), 225-234.
In
183 Germn.
184 okahr, R. and K. Lux (1987). EinfluG des
rheologischen Verhaltens des Spritzbetons auf den
Ausbauwider-stand [Influence of the rheological behavior
of shotcrete on the excavation resistance]. Felsbau 5,
11-18.
185 okahr, R. and R. Zachow (1997). Ein neues
Verfahren zur tglichen Kontrolleder Auslastung einer
Spritz-betonschale [A new method for daily monitoring
of the stress intensity of a
186 , sprayed concrete lining]. Felsbau 15(6),
430-434. In Germn.
187 mahuja, J. and L. Dormieux (2005). Strength
of a porous mdium with a heterogeneous solid
188 phase. Comptes Rendus Mcanique
333(11), 818-823. :heiner, S. and C. Hellmich
(2009). Continuum microviscoelasticity model for
aging basic
189 creep of early-age concrete. Journal of
Engineering Mechanics (ASCE) 135(4), 307-323.
Lubert, P. (1988). Beitrag zum rheologischen Verhalten
von Spritzbeton [Contribotion to the
190 rheological behavior oi shotcrete]. Felsbau
6, 150-153. liubert, W. and A. Steindorfer (1996).
Selective displacement monitoring during tunnel
191 excavation. Felsbau 14(2), 93-97.
 192 hubert, W., A. Steindorfer, and E. Button
(2002). Displacement monitoring in tunnels - An
193 overview. Felsbau 20(2), 7-15. rcombe, J., C.
Hellmich, F.-J. Ulm, and H. Mang (2000). Modeling of
early-age creep of
194 shotcrete I: model and model parameters.
Journal o f Enginecr ing Mechanics
(ASCE)
195 126(3), 284-291.
196 L-indorfer, A., W. Schubert, and K.
Rabensteiner (1995). Probtemorientierte
Auswertung geotechnischer Messungen [Advanced
evaluation of geotechnica! displacement
monitoring data]. Felsbau 13(6), 386-390. In
Germn.
197 n, Z., G. Ye, and S. Shah (2005).
Microstructure and early-age properties o Portland
cement
198 paste-Effects of connectiviry of solid phases.
ACA Materials Journal 102(1), 122-129.
199...................ouquet, r. \iyj* . ^u/. .. ..
200 Springer Verag, Wien New York. Tritthart, J.
and F. HauSler (2003). Pore solution analysis of
cement pastes and nanosrructurai
201 investigations of hydrated C3S. Cement
and Concrete Research 33(7), 1063-1070.
Ulm, F.-J., G. Constantinides, andF. Heukamp
(2004). Is concrete a poromechanics material?-A
202 multiscale investigation of poroeiastic properties.
Materials and Structures/Materiaux et
203 Constmcons 37(1), 43-58. Ulm, F.-j. and O.
Coussy (19.96). Strength growth as chemo-plastic
hatdening in early age
204 concrete. Journal o f Engineering
Mechanics (ASCE) 122(12), 1123-1132.
Wesche, K. (1974). Baustff e fr tragende
Bauteile [Building materials for
structural
205 components] (3 ed.). Bauveriag, Wieshaden,
Germany. In Germn. Zaoui, A. (1997). Structural
morphology and constitutive behavior-of
microheterogeneous
206 materials. In P. Suquet (Ed.), Coniinuum
Micromechanics, Vienna, pp. 291-347. Springer.
Zaoui, A. (2002). Continuum micromechanics: Survey.
Journal o f Engineering Mechanics
207 (ASCE) 128(8), 808-816.
208