Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions.

Bergen, Norway

Evaluation and analysis of laboratory tests of bolt-anchored,

steel-fibre-reinforced shotcrete linings.
L. Ahmed
KTH Royal Institute of Technology, Stockholm, Sweden
A. Sjölander
KTH Royal Institute of Technology, Stockholm, Sweden
A. Ansell
KTH Royal Institute of Technology, Stockholm, Sweden

ABSTRACT: Results from laboratory tests on statically loaded bolt-anchored, steel-fibre-reinforced

shotcrete linings in interaction with rock are here evaluated using a 2D finite element model.
Calculations are made to determine the state of stress in the rock-shotcrete interface near the rock
joints. Plane-stress elements are used with a non-linear material model, capable of describing
cracking and de-bonding during loading. The simulated crack position and force-displacement curves
are compared with laboratory test results. Since most construction work in underground hard rock
involves the use of explosives for excavation work, dynamic load cases are also analysed and
compared to results from previous research on vibration resistance of shotcrete.

shotcrete and a granite slab. The tests indicated

1 INTRODUCTION
Shotcrete is widely used for tunnelling works as
a support element to make the rock carry its
inherent loads. Shotcrete is projected
pneumatically onto a rock surface immediately
after excavation, giving the rock support to
stabilise itself around the opening. The ability of
the shotcrete to adhere to the rock surface
depends on the bond strength between shotcrete
and rock. The rock support is generally
designed for static loading conditions, but
during the excavation work of tunnels and
underground spaces the openings are also
subjected to dynamic loads, for example from
detonation of explosives.
There are few studies that measured the bond
strength, e.g. Saiang et al. (2005) and Bryne
(2014), and most of these have been performed
under static loads. However, in studies by
Ahmed (2012) and Ansell (1999) was bond
strength measured while exposing shotcrete to
impact-type vibration.
To obtain a measure on the bond strength,
destructive tests such as for example pull-out
tests have been performed, e.g. by Bryne et al.
(2011) who measured the bond between young
1
a relatively rapid development of the bond
strength which reached 1.0 MPa after 24 hours.
Saiang et al. (2005) investigated the strength
and stiffness of the interface by considering the
load at which two jointed pieces came apart as
the bond strength. An average bond strength of
0.56 MPa was obtained. However, some other
important methods have also been developed,
for example non-destructive evaluation methods
based on the impact-echo method that can be
used to detect partial de-bonding or voids
between shotcrete and rock (Song et al., 2009).
Hahn and Holmgren (1979) suggested that
the bond strength on good quality granite should
be set to 0.5–1.0 MPa at 28 days. Low bond
strength is assumed in cases where significant
fractures or other planes of weakness exist
parallel to the shotcrete rock interface (Barrett
and McCreath, 1995) or when the rock surface
is not initially cleaned. The bond can be
improved by first washing the rock using high-
pressure water. The influences of this surface
treatment, together with effects from the
shrinkage and hardening of shotcrete on the
bond strength have been investigated by
Malmgren and Nordlund (2008). It was
concluded that in that case using water-jet
scaling gave a higher bond strength of 0.68
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway
MPa, compared with 0.33 MPa for normal
treatment.
However, the modes of shotcrete failure
using punching block tests to simulate loads
applied on a shotcrete lining were studied by
Holmgren (1987). In this test, it was indicated
that the direct shear failure tends to occur when
the bond of the shotcrete to the rock is good,
while flexural and punching shear failure occur
when the bond is poor.
As previously mentioned, due to the
detonation of the explosives during the
excavation works, the openings are subjected to
dynamic loads. These detonations lead to stress
waves that propagate through the rock and may
cause severe damage to installations and support
elements such as shotcrete. However, attempts
to characterise the vibrations that occur along
tunnel walls during excavation blasting has been
performed by Ansell (2004), Reidarman and
Nyberg (2000), Ahmed and Ansell (2012) and
Nyberg et al. (2008).
Due to practical difficulties in performing
such kind of experiments, simulation models are
developed through the use of the finite element
(FE) method. These numerical models have
been used to predict the blasting vibration limits
that would not cause damage to the shotcrete on
the rock surface. For example, the performance
of shotcrete exposed to high magnitudes of
vibration was investigated by Ahmed (2015),
Ahmed and Ansell (2014), and Ahmed et al.
(2012), to identify safe distances for
underground and tunnelling construction, using
numerical analyses and comparison with
measurements and observations. FE models
have been developed for dynamic analysis, with
elastic material formulations used for shotcrete
since the strains reached do not exceed the
elastic material strength in the measurements
(Reidarman and Nyberg, 2000 and Nyberg et
al., 2008). By using two-dimensional elastic FE
models, it is possible to describe the
propagation of the waves through the rock mass,
from the centre of the explosion to the reflection
at the shotcrete-rock interface. Preliminary
recommendations for practical use are given by
Ahmed (2015).
As a step towards studying the consequences
of crack degeneration and bond failure, the
presented models in the following allows for
nonlinear analysis of shotcrete. A material
2
model based on fracture mechanics is used as a
constitutive model for shotcrete.
The objective of this work was to develop
and verify a two-dimensional FE model of a
shotcrete lining subjected to static and dynamic
loading. With this model, the stresses that occur
at the shotcrete-rock interface can be simulated
for comparison with laboratory measurements.
After verification, the model has been used to
identify limits for vibration damage through a
series of calculations with increasing load level.
2 PREVIOUS EXPERIMENTAL
INVESTIGATION
A laboratory investigation of the modes of
shotcrete failure using punching block tests has
been performed by Holmgren (1987). In these
tests, the shotcrete was subjected to a punching
load from a single block. The set-up is shown in
Figure 1, with a test rig that consists of three
large granite blocks fixed to a steel stand. The
shotcrete is applied on the rock blocks surfaces.
Supports for the fibre-reinforced shotcrete layer
were provided by two rock anchors, one on each
side of the punching block.
This loading case simulates the rock-
shotcrete interface near the rock joints. Tests
were performed with two sizes of round stiff
washers at the end of the rock bolt, flexible
washers and rock bolts without washers. For the
following numerical analyses, the results for
160 mm diameter stiff washers (see Figure 2) is
chosen for comparison and calibration of the
models.
Figure 1. Test rig for main tests (Holmgren, 1987).
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway
Spherical plate f 160 mm t = 12 mm
CL
Figure 2. Bolt-anchored shotcrete linings (Holmgren,
1987).
3 FINITE ELEMENT MODEL
A two-dimensional numerical model is
implemented using the FE program
Abaqus/Standard and its Explicit solver. The
geometry of the model is the same as used by
Holmgren (1987), see Figure 2, with 1.2 m in
depth. The model consists of shotcrete, rock and
steel bolts, see Figure 3, and is discretized with
first-order 4-node plane stress elements of type
“CPS4R”. The element size is 0.01×0.01 m2.
The steel bolts are tied to the surface of the
shotcrete lining whereas the interaction between
the shotcrete and the rock is constrained using
surface-to-surface bonding. The bolts and the
rock blocks at the sides are restricted in x- and
y- directions using displacement boundary
conditions.
The FE numerical simulations presented in
this study have been performed based on the
properties of the material used in the preformed
test by Holmgren (1987).
3.1 Mechanical properties
Based on control tests that were performed by
Holmgren (1987), the mechanical properties of
the shotcrete with the volume fraction of 1.4%
Bolt
Boundary conditions
Constraint in x-and y-direction
Figure 3. Finite element model of the test.
3
of steel fibre are given in Table 1. For this mix
steel fibre with end hooks and a diameter of
0.35mm and a length of 35-mm were used
(Holmgren, 1987).
Table 1 Mechanical properties of the shotcrete
Properties
Compressive strength 78 MPa
Tensile strength 5.0 MPa
Adhesive strength 0.81MPa
The rock properties are here chosen to
represent hard granite type Scandinavian
bedrock which typically has a density of 2500
kg/m3, a modulus of elasticity of 70 GPa and a
Poisson’s ratio that is 0.15 (Ansell 2004). The
density of 7800 kg/m3 and modulus of elasticity
of 200 MPa are assumed to represent the steel
bolt properties.
3.2 Damage criteria of shotcrete
Data on testing of the flexural strength are given
by Holmgren (1987). A shotcrete beam shows
softening behaviour after reaching the tensile
strength. The softening behaviour is due to the
formation of a fracture zone within the shotcrete
matrix where the material progressively softens
due to the formation of micro-cracks. The
softening curve is defined in terms of the
fracture energy Gf and the tensile strength ft, and
several researchers have discussed its shape.
Yoo et al. (2015) employed a trilinear curve
characterised by three linear segments, as shown
in Figure 4. This model has been found to give
results that are in good agreement with
experimental measurements (Yoo et al. 2015).
Therefore, this representation was adopted for
the present study.
Shotcrete
Rock
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway
Figure 4. Conceptual view of modelled trilinear (Yoo et
al., 2015).
The damage criteria of steel fibre reinforced
shotcrete used in this paper assume that the
initiation of damage occurs when loading
conditions produce effective inelastic strains
that exceed a certain threshold value leading to
cracking. It can be seen from Figure 4 that after
reaching the tensile strength (ft) the descending
branch of the softening curve, which represents
the inelastic behaviour due to the micro-
cracking, passes through two points at a strength
level of (f1) and (f2), as shown in Figure 4.
Based on the evaulation between the
experimental and numerical results, the location
of (f1) and (f2), can be asummed as 14%, and
15% reduction in tensile strength, where the
crack-opening width is 0.025 mm and
0.090 mm, respectively. The point (wc)
represents the stress-free crack width equal to
4 mm. The above tensile strength values with
respect to the displacement are summarised in
Table 2.
Table 2. Shotcrete damage criterion.
Tensile strength (MPa) Crack width (mm)
5.00 0.000
4.30 0.025
4.23 0.090
0.05 4.000
3.3 Damage criteria of bond failure
For this test series, the adhesive strength of
0.81 MPa was measured by Holmgren (1987).
For the following numerical examples, this
value is set corresponding to 28 days old
4
shotcrete. There is no known investigation on
the adhesive strength in shear between shotcrete
and rock, but tests of the adhesive bond between
old concrete and shotcrete are presented by
Silfwerbrand (1992). These results are within 3–
4 MPa of shear bond strength and should be
compared to 1–2 MPa from corresponding pull-
out testing. Based on these results, it can be
assumed that the shear bond strength is twice
the normal strength, and here assumed to be
1.62 MPa for the numerical examples.
4 NUMERICAL ANALYSES
Two numerical analyses were performed;
verification the FE model followed by an
investigation of vibration damage. In the first
part, a comparison was made between FE
results and experimental observations. The
following section 5 consists of studying the
effect of dynamic load on the bond between the
shotcrete and rock when subjected to impact
load.
4.1 Material verification
A verification of the suggested material
properties is done by comparing numerical
results with corresponding experimental results
given by Holmgren (1987). The model consists
of a concrete prism of steel fibre reinforced
concrete and is loading with two separate point
loads, as shown in Figure 5. The test setup used
by Holmgren (1987) is chosen for the
comparison, and the parameters in the
constitutive material model, therefore,
correspond to the experimental data, see
Table 1. The geometry of the concrete prism is
82×88×800 mm3.
250 mm 250 mm 250 mm
750 mm
800 mm
Figure 5. The configuration of the finite element model.
Considering the damage criteria in Table 2,
the simulated tensile stress and crack width are
presented in Figure 6. The results are
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway
constructed from the element when loading
conditions exceed a certain threshold value of
5.0 MPa, i.e. point (a) in Figure 7.
Figure 6. Simulated tensile stress versus crack width of
steel fibre reinforced concrete.
In Figure 7, the crack patterns of the beam
are also shown. It can be seen that the failure is
initiated at a flexural crack in the area between
the two point loads. Similar observations were
found by Holmgren (1987), with 1.5% volume
fraction of steel fibres. The simulated results
show that the failure cracks are located within
the middle third of the beam length. This
location is in good agreement with the position
of experimentally observed cracks and indicates
that the failures normally occurred at a flexural
crack for a beam with 1.5% volume fraction of
steel fibre, whereas for 2.0% steel fibres a
domination of shear stresses is observed
(Holmgren 1987).
One comparison between the behaviour of
the FE model and experimental prism
(Holmgren 1987) is given in Figure 8. The
simulated results show less softening than in the
experimental results due to a local failure
occurred in one crack, point (a) in Figure 7.
Point a
Figure 7. The inelastic strain of steel fibre reinforced
concrete beam.
5
Figure 8. Force- displacement curves.
4.2 Model verification
A verification of the suggested FE model is
done by comparing numerical results with
corresponding experimental results. As
previously described the test setup used by
Holmgren (1987) is chosen for the comparison,
see Figures 1 and 2. The material parameters
used in the simulations are previously
summarised in Table 1 and verified in the
previous section.
This FE model examines the debonding
behaviour at the shotcrete-rock interface,
predicted in both Abaqus/Standard and
Abaqus/Explicit (Simulia, 2017). The results
from Abaqus/Standard are compared with the
experimental results by Holmgren (1987). The
results were calculated by using cohesive and
bond damage contact and are presented in
Figure 9. It can be seen that there are good
agreement between the experimental and
simulated responses when the force and
displacement are normalised in relation to their
respective maximum values.
A local view of the finite element model
where the inelastic strain occurs is presented in
Figure 10. When the displacement was applied
upwards on the movable block, bending cracks
developed at the joints between this and the
fixed blocks and at the section where the rock
bolts are introduced, i.e. at the sections where
the maximum bending moment occurred. The
cracks are indicated with red areas showing high
inelastic strain values while the blue is the area
where inelastic straining occurred. The crack
pattern is in good agreement with the test
observation (Holmgren, 1987).
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway

5 DYNAMIC ANALYSIS

To increase the knowledge of the performance

of shotcrete-rock interfaces exposed to vibration
in poor rock condition, a modification of the
load type for the previous model is here
presented. In the real case, during the
excavation work, the rock area around the blast
hole after the detonation contains cracks. Thus,
planes of weakness exist parallel to the
shotcrete-rock interface. The fractured rock is
here represented as a movable block and the
detonation is introduced as velocity boundary
Figure 9. Normalised force-displacement curves. conditions under this block. The interaction
between the movable rock block and the
From Figure 10 it can be seen that the bond shotcrete is modelled using a surface to surface
crack spreads along the rock-shotcrete interface interface.
from the joint between the movable and fixed
blocks to the rock-bolt section. Figure 11 shows
the contact damage criterion versus time of the
analysis in different positions, at nodes 1-4. It
can be seen that at a certain load level, the bond
crack formed in the shotcrete at the surface of
the outer block, i.e. at node 1 in Figure 11, and
that the bond crack propagated along the
shotcrete-rock interface; i.e. through node 2-3.
The load remained nearly constant during the
crack propagation until a flexural crack
appeared in the shotcrete layer.

Bolt

Shotcrete
4• 3• 2• 1• Figure 12. The maximum tensile stress versus particle
velocity.

Bolt Rock blocks

Rock blocks
Applied displacement
Figure 10. A local view of the finite element model, Shotcrete
showing inelastic strain.

Velocity boundary conditions

Figure 13. The local view of the finite element model

showing inelastic strain when applying particle velocity
of 300 mm/s.

Although the contact damage criterion is well

verified in Abaqus/Standard analysis, it is
Figure 11. Contact damage criterion of the shotcrete-rock
difficult to define a reliable contact damage
interface. criterion for the bond between shotcrete and
6
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway
rock in a dynamic analysis. For this, the general
contact algorithm in Abaqus/Explicit is used
which includes the option of tangential and
normal behaviours, as a simple way to define
contact in the model.
The response of the shotcrete in a tunnel is
also investigated for various blast load values.
As previously mentioned, the vibrations are
induced by applying a particle velocity under
the movable block as a velocity boundary
condition. In this paper, the dynamic excitation
is described as a cosine-pulse velocity with the
frequency of 1000 Hz, see also Ahmed (2012).
The response of the shotcrete for various
particle velocities is shown in Figure 12. The
maximum tensile stress versus the particle
velocity is presented in this figure. It can be
seen that the concrete damage was initiated at a
velocity of more than 300 mm/s, where the
maximum tensile stress reaches 5 MPa. Under
the blasting loads, the shotcrete stress along the
lining length increased, giving stress
concentrations around the movable block and
the bolts, see Figure 13. It can be seen that the
model with dynamic load gives less cracking
compared with the results from a static load, see
Figure 10.
6 CONCLUSION
The evaluation of the presented FE models and
proposed material models was made in two
steps, with a focus on describing the behaviour
of shotcrete on hard rock. First, the results from
laboratory tests on statically loaded bolt-
anchored, steel-fibre reinforced shotcrete linings
in interaction with rock are evaluated using a
two-dimensional finite element model. The
numerical simulations, conducted especially to
study the behaviour of shotcrete under statically
load, demonstrated a good agreement compared
with laboratory measurements, but with
overestimated values. It is observed that the
two-dimensional plane stress models used give
higher load level and lower displacement than
the measured results. In the experiment, the
steel bolts are situated only in the middle of the
shotcrete lining, constraining a circular area of
0.02 m2 around each bolt. However, in the two-
dimensional FE model, the steel bolts extract to
1.2 m depth, constraining an area of 0.192 m2.
7
Calculated response of the shotcrete
subjected to blasting vibrations is also
presented. It is observed that using the simple
contact algorithm in Abaqus/Explicit, including
the options of tangential and normal behaviours
gives different crack patterns compared with the
statically loaded model. From the view of
tensile stresses in the shotcrete, the damage
vibration is limited to a particle velocity of
300 mm/s, which is in good agreement with
previously observed and measured results
(Ahmed, 2015 and Ansell, 2004).
The following conclusions are drawn for the
further studies:
• By comparing the numerical results with
experimental data, it is demonstrated that
the two-dimensional FE analysis based on
fracture mechanism provides a good
description of the force-displacement
behaviour.
• Two-dimensional plane stress and plane
strain models give similar results; both
results are overestimated due to that the
model is over-constrained at the steel bolts
area.
• To represent the interaction between
shotcrete and rock under a blasting load, a
cohesive element may be chosen to
simulate the bond damage at the shotcrete-
rock interface.
ACKNOWLEDGEMENTS
The authors would like to acknowledge
Professor Jonas Holmgren for performing the
testing at the division of Concrete Structures
within the KTH school of Architecture and the
Built Environment.
REFERENCES
Ahmed, L. 2012. Models for analysis of shotcrete on rock
exposed to blasting. Thesis (Licentiate in Concrete
Structures), KTH Royal Institute of Technology,
Stockholm, 92 p.
Ahmed, L. 2015. Models for analysis of young cast and
sprayed concrete subjected to impact-type loads.
Thesis (PhD in Concrete Structures), KTH Royal
Institute of Technology, Stockholm, 68 p.
 Proceedings of the World Tunnel Congress 2017 – Surface challenges – Underground solutions. Bergen, Norway

Ansell, A. 1999. Dynamically loaded rock reinforcement. Silfwerbrand, J. 1992. The Influence of Traffic-induced
Thesis (PhD in Concrete Structures), KTH Royal Vibrations on the Bond between Old and New
Institute of Technology, Stockholm, 158 p. Concrete. Stockholm: KTH Structural Mechanics and
Ansell, A. 2004. In situ testing of young shotcrete Engineering.
subjected to vibrations from blasting. Tunnelling and Simulia. 2017. Abaqus/Explicit FE analysis program.
Underground Space Technology 19, pp.587–596. Dassault Systèmes. http://www.3ds.com/products-
Ahmed, L, Ansell, A, Malm, R. 2012. Finite element services/ simulia/products/abaqus/.
Song, K., Cho, G. 2009. Bonding state evaluation of
simulation of shotcrete exposed to underground
tunnel shotcrete applied onto hard rocks using the
explosions. Nordic Concrete Research 45, pp. 59-74. impact-echo method. NDT&E International Journal,
Ahmed, L, Ansell, A. 2012. Laboratory investigation of 42, pp. 487-500.
stress waves in young shotcrete on rock. Magazine of
Concrete Research 64(10) pp. 899-908.
Ahmed, L, Ansell, A. 2014. Vibration vulnerability of
shotcrete on tunnel walls during construction blasting.
Tunnelling and Underground Space Technology 42,
pp. 105–11.
Barrett, S, McCreath, D. 1995. Shotcrete support design
in blocky ground: Towards a deterministic approach.
Tunnelling and Underground Space Technology, 10,
pp. 79–89.
Bryne, LE, Holmgren, J, Ansell, A. 2011. Experimental
investigation of the bond strength between rock and
hardening sprayed concrete. Proceedings of the Sixth
International Symposium on Sprayed Concrete,
Norwegian Concrete Society, Tromsø .
Bryne, L.E. 2014. Time dependent material properties of
shotcrete. Thesis (PhD in Concrete Structures), KTH
Royal Institute of Technology, Stockholm, 79 p.
Hahn, T, Holmgren, J. 1979. Adhesion of shotcrete to
various types of rock surfaces and its influence on the
strengthening function of shotcrete when applied on
hard jointed rock. In: Proceedings of the 4th
International Congress on Rock Mechanics.
Montreux: International Society for Rock Mechanics;
pp. 431–439.
Holmgren, J. 1987. Bolt-anchored, steel- fiber-reinforced
shotcrete linings. Tunnelling and Underground Space
Technology, 2, pp. 319–333.
Malmgren, L, Nordlund, E. 2008. Interaction of shotcrete
with rock and rock bolts – A numerical study.
International Journal of Rock Mechanics and Mining
Sciences, 45, pp. 538–553.
Nyberg, U, Harefjord, L, Bergman, B, Christiansson, R.
2008. Monitoring of vibrations during blasting of the
APSE. SKB Report R-05-27, Sweden.
Reidarman, L, Nyberg, U. 2000. Blast vibrations in the
Southern Link tunnel - Importance for fresh shotcrete?
(in Swedish: Vibrationer bakom front vid
tunneldrivning i Södra Länken - Betydelse för
nysprutad betong?). Stockholm: Rock Engineering
Research Foundation. SveBeFo-report 51.
Saiang, D., Malmgren, L., Nordlund, E. 2005. Laboratory
tests on shotcrete- rock joints in direct shear, tension
and compression. Rock Mechanics and Rock
Engineering 38, pp. 275–297.