International Journal of Impact Engineering 59 (2013) 18e28

Contents lists available at SciVerse ScienceDirect

International Journal of Impact Engineering

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

Conclusions from experimental testing of blast resistance of FRC and

RC bridge decks
Marek Foglar*, Martin Kovar
Czech Technical University in Prague, Faculty of Civil Engineering, Prague, Czech Republic

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

Article history: According to recent publications, from 2005 to 2008 there were more than 13,000 terrorist attacks
Received 1 July 2012 around the world, which took more than 73,000 human lives. The attacks were targeted mainly on the
Received in revised form technical and civic infrastructure, such as governmental buildings and bridges, etc. Due to improved
27 January 2013
ductility, fiber-reinforced concrete (FRC) shows better performance under blast and impact loading
Accepted 18 March 2013
compared to conventionally reinforced concrete. Field tests of FRC and reinforced concrete specimens
Available online 1 April 2013
were performed in cooperation with the Czech Army corps and Police of the Czech Republic in the
military training area Boletice. The tests were performed using real scale reinforced concrete precast
Keywords:
Blast resistance
slabs with varying fiber content and concrete strength class and 25 kg of TNT charges placed in a distance
Fiber-reinforced concrete from the slab for better simulation of real in-situ conditions. This paper presents conclusions from two
High performance concrete sets of tests and results of their numerical evaluation.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction concrete strength class and 25 kg of TNT charges placed in a dis-

Recent terrorist attacks in Moscow (2011), Stockholm (2010),
London (2005) and Madrid (2004) etc. show great vulnerability of
civil and transport infrastructure to this kind of threat. Attacks on
structures like airports, railway and subway stations, bridges and
governmental buildings can cause great casualties. These casualties
multiply when the explosion causes the collapse of the entire
structure.
One of the ways of improving the blast performance of civil and
transport infrastructures is the use of progressive materials like
fiber-reinforced polymer composites [1]. When building new
structures the desired structural ductility can be achieved by using
plastic fibers in the concrete mix. A bigger increase of blast and
impact performance can be achieved with the use of ultra-high
performance fiber-reinforced concrete or engineered cementitious
composites [2,3].
This paper presents conclusions from two sets of tests and re-
sults of their numerical evaluation. The tests were performed in
cooperation with the Czech Army corps and Police of the Czech
Republic at the military training area Boletice using real scale
reinforced concrete precast slabs with varying fiber content and

* Corresponding author. CVUT
v Praze, Thákurova 7, 166 29 Prague, Praha 6,
Czech Republic. Tel.: þ420 224 354 627; fax: þ420 233 335 797.
E-mail address: marek.foglar@fsv.cvut.cz (M. Foglar).
tance from the slab for better simulation of real in-situ conditions.
The outline of the research program was based on a large study
of up-to-date references. It was found that, except from some
extraordinary experiments [4e6], blast resistance experiments are
performed on relatively small specimens whose dimensions rarely
exceed 2 m in one dimension [23e28]. The reason is quite simple:
blast experiments are very demanding in the means of logistics,
costs and safety measures. Due to these limitations, the charges
used do not correspond in dimensions to the charges which can be
transported by an individual on-foot or by the means of public
transport on a bridge.
Ref. [4] focused on blast performance of a concrete vehicle
barrier used for providing offset from endangered structures. It
compared various fiber contents and materials (carbon fiber-rein-
forced concrete (CFRC), nylon fiber-reinforced concrete (NFRC),
both at 1.5% fiber content, synthetic/steel fiber mix (SS) (high fiber
volume 5% and low fiber volume 3.8%)). The damage was much
smaller than at the compared RC specimens and decreased with the
increased fiber content, the increase stopped at fiber volume 3.8%.
Even 1.5% fibers showed a beneficiary effect on the blast perfor-
mance and dimensions of the ejected debris.
Ref. [23] compared the blast damage of RC and high strength
steel fiber concrete (HSSFC) specimens on simply supported slabs
(1
1.3
0.1 m). HSSFC performed better on both surfaces.
Ref. [24] focused on the use of carbon fiber-reinforced polymer
(CFRP) strips on one/both surfaces of an RC slab (1.2
1.2
0.9 m).
The CFRP strips placed on the soffit of the specimens did not show a

0734-743X/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijimpeng.2013.03.008
 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28
Fig. 1. Layout of the experiments.
great beneficial effect. The CFRP strips placed on both specimen
surfaces showed a great beneficial effect.
Ref. [25] focused on the impact resistance of a fiber RC slab
(1
1
0.05 m) against a low velocity projectile caused by a 43 kg
hammer falling from 4 m. It studied the effect of various types and
dosages of fibers (polyolefin, polyvinyl alcohol (PVA) and steel, all at
0.5%, 1% and 2%). The RC comparative specimen was completely
destroyed during the tests, while the steel fiber-reinforced slabs
performed well.
Ref. [26] focused on the use of FRC in the pavements used for
vertical take-off and landing (VTOL). The 5% of steel fibers increased
the fire resistance of the concrete slab.
Refs. [2,3] focused on high performance concrete fiber-rein-
forced concrete (HPFRC).
Ref. [2] describes dynamic behavior of HPFRC, the experiments
were focused on the dynamic bending and the dynamic shear in-
crease factor (DIF). Bending tests simulated distant explosion
loading and impact & projectile loading, shear tests simulated near-
face explosions. The specimens were subjected to three-point
bending by a 30.1 kg steel hammer. The steel fiber content was
1.5%, 2.0% and 6%. The DIF decreased with the increasing fiber
content.
Ref. [3] focused on the blast performance of an ultra-high per-
formance fiber concrete (UHPFC) slab, the reinforced ultra-high
performance fiber concrete (RUHPFC) and an RC slab externally
retrofitted with fiber-reinforced polymer (FRP) plates. The speci-
mens were double sided supported slabs (2
1
0.1 m) with a
span of 1.8 m. The blast resistance of UHPFC and normal reinforced
Fig. 2. Time-plot of the blast overpressure wave, blast No. 3.
19
concrete was the same, the FRP plates added on the loaded side
decreased the damage of the slab.
Refs. [5,10,27] focused on improving the blast resistance of RC
structures by external retrofitting.
Ref. [27] analyzed a two-way RC slab (2.64
2.64
0.076 m)
subjected to blast loading. The slab was retrofitted with CFRP strips
(0.46 m wide, 0.584 thick) and loaded with three blasts (0.45, 56.7
and 453.4 kg of TNT). The use of CFRP strips at the loaded surface
decreased the deflection by 40%, the CFRP strips at both surfaces
reduced deflection by 72%. The author studied the change of natural
frequencies caused by the blast loading. The retrofitted specimen
showed lower reduction of natural frequencies which denotes a
higher residual structural stiffness.
Ref. [5] focused on the use of aluminum foam as a coating of the
blast adjacent concrete surface. It tested the effect on RC slabs with
a span of 3 m. The specimens were loaded with 1000 kg TNT blast at
a 20 m distance. The aluminum foam was able to absorb a large
portion of the blast energy and thusly decrease the damage of the
RC slab.
Ref. [28] focused on the use of glass fiber-reinforced polymers
(GFRP) in improving the blast resistance of concrete. Within the
experiments, it used a simply supported RC slab (1
1
0.07 m)
loaded with 33.4 kg ANFO (82% TNT equivalent). The surface
adjacent to the blast was retrofitted with GFRP strips (0.5 m wide)
in both dimensions. The use of GFRP decreased the deflection from
12 to 8 mm.
Ref. [1] focused mainly on the state-of-the-art blast-retrofitting
of masonry structures.
Due to the recent geo-political situation, the improvement of
resistance of civil and transport infrastructure to extreme loadings
like blast or impact has become a significant issue in structural
engineering. Fiber-reinforced concrete shows better performance
under blast and impact loading compared to conventionally rein-
forced concrete due to improved ductility. The insertion of fibers
also decreases the volume of debris which is ejected into the air
from the soffit of a concrete specimen and thusly reduces fatalities
and injuries. The authors believe that this experimental study of
blast performance of FRC and RC specimens will be a valuable
contribution to this topic.
2. Experimental testing of blast resistance of FRC and RC
bridges
2.1. Specimens and materials
The dimensions of the specimens were designed to scale of a
small span bridge of concrete slabs, 6 m long, 1.5 m wide and 0.3 m
thick.
20 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28

Table 1
Blast overpressures, comparison of calculated and measures values.

Blast No. 3 Sensor Sensor Sensor Sensor

No. 1 No. 2 No. 3 No. 4
Weight of the charge (kg) 25 25 25 25
Distance (m) 15 20 25 30
Blast overpressure (kPa) 49.2 18.7 13.8 15.5
Impulse of the blast overpressure 159.6 129.2 127.5 89.9
(kPa ms)
Calculated blast overpressure (kPa) 44.8 26.4 18.5 14.0
Measured/calculated overpressure 110% 71% 75% 110%

Five specimens were made in total, of which three of them were

made of C30/37 grade concrete (fc,cyl ¼ 30 MPa) (specimen No. 1, 2
and 5), two of C55/67 grade concrete (fc,cyl ¼ 55 MPa) (No. 3 and 4).
Polypropylene 54 mm long synthetic fibers (ft ¼ 620e758 MPa)
were used in three of the specimens. The fiber dosage was as fol-
lows: specimen No. 1 0, No. 2 4.5 kg/m3 (0.5% by volume), No. 3 0,
No. 4 4.5 kg/m3 and No. 5 9 kg/m3 (1% by volume). The dosage of the Fig. 3. Damaged specimen No. 1 after blast e top view.
fibers was kept as low as it could be achieved on site.
The specimens were designed to fulfill the bridge loading ac-
cording to the EN 1991-2 design code. All specimens were rein- practice target area (900 m above sea level) which is now used for
forced by conventional reinforcement (B500B according to the dismantling obsolete ammunition and ammunition from WW2
EN 1992-1-1 design code, fyk ¼ 500 MPa, ft ¼ 600 MPa, which is still found in the Czech Republic during construction work.
Es ¼ 210,000 MPa, r1 ¼ 0.5%) at both surfaces, longitudinally by The slabs were placed on timber posts which were fixed in
11B16 mm reinforcing bars (every 140 mm), transversely by position by steel tubes. The timber posts were placed on dug-in
B10 mm (every 150 mm). The shear reinforcement was provided rock bed to minimize the settlement. The settlement caused by
by B8 mm links (9 pcs/m2). the first experiment of both sets was about 5 mm, the settlement
caused by the following experiments was below this value. The
effect of the stiffness of the supports was studied during the nu-
2.2. Layout of the experiments
merical modeling and proved itself to play an insignificant role, i.e.
it did not considerably influence the response of the specimen.
The experiments were carried out at the military training area
The ground under the slabs was removed; 5 m in diameter and a
Boletice in cooperation with the Czech Army corps and Police of the
1 m deep excavation was prepared under the slabs so that the re-
Czech Republic. The blasts were performed on the former artillery
sults of the experiments would not be influenced by a rebound of
the pressure wave.
Table 2 The 25 kg TNT charges were placed on steel “chairs” (3B10 mm
Comparison of blast performance of studied specimen. links, 3 mm thick steel plate) in the middle of the slabs. The “chairs”
Specimen No. 1 2 3 4 5 provided a standoff of 450 mm from the slab. This value was chosen
Concrete C30/37 C30/37 C55/67 C55/67 C30/37
as the most usual height of car trunk. The charge had a form of a
Fibers e 4.5 kg/m3 e 4.5 kg/m3 9.0 kg/m3 prism (w0.3
w0.3
0.2 m).
Puncture e top surface 0.43 m2 0.26 m2 0.02 m2 e e The charges were covered by a woolen blanket. According to the
Concrete spalling 2.35 m2 1.89 m2 1.51 m2 0.73 m2 0.61 m2 cooperating pyrotechnic, the blanket would concentrate the blast
(soffit) e <concrete
cover
Concrete spalling 1.71 m2 1.09 m2 1.2 m2 0.44 m2 0.37 m2
(soffit) e >concrete
cover
Concrete spalling 0.43 m2 0.26 m2 0.89 m2 0.68 m2 0.66 m2
(top surface) e
<concrete cover
Concrete spalling 0.43 m2 0.26 m2 0.29 m2 0 0.08 m2
(top surface) e
>concrete cover
Concrete spalling 0.52 m2 0.05 m2 0.08 m2 0 0
(left side) e <concrete
cover
Concrete spalling 0.35 m2 0 0.02 m2 0 0
(left side) e >concrete
cover
Concrete spalling 0.34 m2 0.16 m2 0.08 m2 0 0
(right side) e <concrete
cover
Concrete spalling 0.23 m2 0.11 m2 0.02 m2 0 0
(right side) e >concrete
cover
Volume of crushed 0.23 m3 0.15 m3 0.20 m3 0.05 m3 0.06 m3
concrete
Permanent deflection 0.31 m 0.37 m 0.28 m 0.30 m 0.26 m
Fig. 4. Damaged specimen No. 1 after blast e bottom view.
 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28 21

Fig. 7. Damaged specimen No. 3 after blast e top view.

Fig. 5. Damaged specimen No. 2 after blast e top view.

2.3. Measurements of blast wave overpressure

During the first set of experiments, the overpressure in the blast

wave by up to 15%. The woolen blanket represents the camouflage,
wave was recorded by four pressure sensors PCBÒ-type ICP 137A23,
as the charge would not be discovered by a routine police road
the signal was converted on a four-channel oscilloscope
check when transported by car or the effect of a backpack when
TektronixÒ-type TDS3014B; the bus was located w60 m far away in
transported on-foot.
a former gun bank.
The burlap was not incorporated in the model. Several ap-
The sensors were dug in to the ground level, in order to mini-
proaches were tried but without satisfactory results. Several tests
mize the possibility of damage by flying debris, but keeping the
were performed on site with 5 kg charge of TNT placed on the
reading optimal. This arrangement minimizes the effect of the
ground but no increase of the force of the charge measured by
multi-reflections from the ground which would occur when placing
means of the dimensions of the crater was found. Unfortunately the
the sensors on posts.
financial resources did not enable further verification in the lab.
The sensors were placed on a helix with the center at the
The charges were fired remotely by radio impulse. The layout of
location of the charge in distances of 15, 20, 25 and 30 m. The
the experiments can be seen in Fig. 1.
recorded value of the overpressure at the front of the blast over-
The response of the slab was recorded by a high-speed camera
pressure wave corresponded to Ref. [7] Eg. (1):
and further analyzed using picture-processing methods. The cam-
era was located w25 m from the specimen on an earth dam; it was 93:2 383 1275
protected from the flying debris by a 40 mm thick glass and a DPS ¼ þ 2 þ 3 ; (1)
Z Z Z
timber slab.
The field conditions did not enable the location of any instru- where
mentation on the specimens or in their closer vicinity.
The power supply was provided by mobile source at the location Z ¼ the reduced distance,
of the oscilloscope (see below).

Z ¼ R=W 1=3 ; (2)

Fig. 6. Damaged specimen No. 2 after blast e bottom view. Fig. 8. Damaged specimen No. 3 after blast e bottom view.
22 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28
Fig. 9. Damaged specimen No. 4 after blast e top view.
R ¼ the distance from the blast epicenter (m)
W ¼ the weight of the charge in kg of (TNT).
Fig. 2 and Table 1 show the results of the measurements and its
comparison to calculated values. Taking into account the field
conditions of the experiments, the agreement of the calculated and
measured values is excellent.
2.4. Results of the experiments
The experiments were focused on the effect of fibers, concrete
compressive strength and its combination on the blast performance
of concrete. Performance is understood as the dimensions of
puncture and spalling of the concrete.
The volume of the spalling was obtained by picture-processing
methods: the depth of the spalling was measured on site using a
grid of reference points to calibrate the off-site evaluation by means
of a 3D model. Furthermore, the concrete cover (55 mm thick
concrete cover was used for concrete bridges according to present
European design codes (European standard EN 1992-1-1 and 1992-
2)) was divided into 10 mm thick layers which were divided into
rectangular elements (30
30 mm, dimension chosen matching to
the FEM models). The two models were overlaid, the concrete
element with spalling greater than ½ of the thickness of the layer
was considered eroded.
Fig. 10. Damaged specimen No. 4 after blast e bottom view.
Fig. 11. Damaged specimen No. 5 after blast e top view.
The volume of punctures is taken as a volume of the body of
elements with spalling equal to the thickness of the specimen.
Therefore, the volume of the cone at the sides of the puncture body
is attributed to the volume of spalling.
The differences in puncture and spalling of concrete on the soffit
of the slabs can be found in Table 2. In this section, the findings
presented in Table 2 are described in detail.
The specimen No. 1, C30/37, without fibers, is regarded as the
reference specimen. This specimen is the one most damaged by the
blast loading. The area of the puncture is 0.43 m2, volume 0.12 m3,
which represents 4.4% of the total volume of the specimen. Total
volume of the damaged concrete (puncture þ spalling) is 0.23 m3,
which represents 8.5% of the total volume of the specimen. The
sides of the specimen were damaged severely. The deflection was
295 mm on the left side and 310 mm on the right side of the
specimen. The shape of deflection was similar to deflection from
uniformly distributed loading.
The specimen No. 2, C30/37, 4.5 kg/m3 PP fibers was less
damaged. The area of the puncture is 0.26 m2, volume 0.08 m3,
which represents 2.9% of the total volume of the specimen. Total
volume of the damaged concrete (puncture þ spalling) was 0.15 m3,
which represents 5.6% of the total volume of the specimen. The area
of the puncture was reduced by 40% in comparison to specimen No.
1, the total volume of damaged concrete was reduced by 35% in
comparison to specimen No. 1. The left side of specimen No. 2 was
not damaged, the damage of the right side was reduced by more
than 50%. The deflection was 365 mm on the left side and 380 mm
Fig. 12. Damaged specimen No. 5 after blast e bottom view.
 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28
Fig. 13. The results of the experiments plotted on the spall and breach prediction curves of UFC 3-340-02.
Fig. 14. Set-up of the FE model.
on the right side of the specimen. The shape of deflection was
similar to deflection from point loading in the mid-span of the
specimen. The deflection was increased by 19% in comparison to
specimen No. 1.
Specimen No. 5, C30/37, 9 kg/m3 PP fibers was less damaged. The
total volume of the damaged concrete (puncture þ spalling) was
0.06 m3, which represents 2.2% of the total volume of the specimen.
There was no puncture, the total volume of damaged concrete was
reduced by 75% in comparison to specimen No. 1. The sides of
Fig. 15. Top view of the FRC specimen after the blast (t ¼ 1 ms).
23
specimen No. 5 were not damaged. The deflection was 255 mm on
the left side and 265 mm on the right side of the specimen. The
shape of deflection was similar to deflection from point loading in
the mid-span of the specimen. The deflection was decreased by 16%
in comparison to specimen No. 1.
Specimen No. 3, C55/67, without fibers was less damaged. The
area of the puncture is 0.02 m2, volume 0.01 m3, which represents
0.4% of the total volume of the specimen. The total volume of the
damaged concrete (puncture þ spalling) is 0.2 m3, which repre-
sents 7.4% of the total volume of the specimen. The area of puncture
was reduced by 95% in comparison to specimen No. 1, the crushed
concrete remained within the reinforcement matrix, the total vol-
ume of damaged concrete was reduced by 15% in comparison to
specimen No. 1. The damage to the sides was reduced by 85%. The
deflection was 285 mm on the left side and 290 mm on the right
side of the specimen. The shape of deflection was similar to
deflection from uniformly distributed loading. The deflection was
increased by 10% in comparison to specimen No. 1.
Specimen No. 4, C55/67, 4.5 kg/m3 PP fibers was less damaged;
its damage was similar to specimen No. 5. The total volume of the
damaged concrete (puncture þ spalling) is 0.05 m3, which repre-
sents 1.9% of the total volume of the specimen. There was no
puncture, the total volume of damaged concrete was reduced by
22% in comparison to specimen No. 1. The sides of specimen No. 5
were not damaged. The deflection was 300 mm on the left side and
305 mm on the right side of the specimen. The shape of deflection
was similar to the deflection from point loading in the mid-span of
the specimen. The deflection was increased by 15% in comparison
to specimen No. 1.
24 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28

Fig. 16. Comparison of the numerical modeling and the experiment C30/37 e no fibers top surface of the specimen.

Fig. 17. Comparison of the numerical modeling and the experiment C30/37 e no fibers, soffit of the specimen.

The extent of damage decreased with increased fiber content.
Specimens with high fiber content or medium fiber content and
high compressive strength did not experience puncture.
The concrete spalling at the sides of the specimens is caused by
their small width e the blast overpressure wave passes around and
causes surface tensile stresses. This spalling cannot occur with
wider specimens, the slab behavior is dominant.
The effect of blast loading on top and bottom surfaces of the
specimens can be found in Figs. 3e12.
The results of the experiments are plotted on the spall and
breach prediction curves of Ref. [21] in Fig. 13. All the results of
specimens made of concrete C30/37 are plotted under the breach
curve, although the specimen with 9 kg/m3 of PP fibers did not
experience any breach, only spalling occurred. Also the results of
the specimens made of concrete C55/67 are plotted under the
breach curve, although the specimen with 9 kg/m3 of PP fibers did
not experience any breach, only spalling occurred.
Following can be stated:
- the RC specimens agree with the spall and breach prediction
curves according to UFC 3-340-02,
- the spall and breach prediction curves according to UFC 3-340-
02 are not suitable for FRC.
3. Numerical modeling
A numerical model of the experiment was prepared for the
purpose of the evaluation of the experiments. The process of model
set-up is described in the following text. The model was calibrated
according to the outcomes of the experiments described in the
previous section.
3.1. Numerical modeling
The LS-DYNA solver is used for non-linear analysis of fast dy-
namic phenomena like blast or impact. Within the calculation, the
FEM mesh can adapt by deleting elements whose resistance was
depleted; these FEM elements “erode” [8].
The model is composed of several parts. The air provides
boundaries of the model, the air is modeled in a form of a prism
with dimensions 7
7.5
6 m; the explosive (e.g. TNT) transfers
the energy from the blast to FE elements of the air, where the blast

Fig. 18. Comparison of two different concretes (30
50 MPa compressive strength).
 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28 25

wave propagates. The concrete specimen is modeled by solids, for

example of the set-up of the model, see Fig. 14. The specimen is
located 0.7 m from the bottom of the “air cube”. The dimension of
the “air cube” was chosen so that by the time the rebound of the
blast overpressure wave reached its sides the damage of the spec-
imen has already worn off.
The air is modeled using the 009-Null material and forms the
undeformable FE network. The concrete specimen is modeled by
the 159_CSCM_Concrete material model (brittle material model
with damage). The explosive is modeled by the 008-High explosive
burn material model. The blast overpressure is calculated by the
JWL equation of state (EOS):



u u uE
p ¼ A 1 eR1 V þ B 1  eR2 V þ : (3)
R1 V R2 V V

Fig. 20. Stressestrain diagram of a material with damage according to Ref. [10].
3.2. 3D model

The 3D reinforced concrete model was prepared using real di-

mensions of longitudinal and transverse reinforcement including
links (steel reinforcement modeled as beam elements).
The mesh size was chosen 30 mm for concrete and reinforce-
ment and 50 mm for air [9]. The convergence tests were performed
according to Refs. [14,15] for the MAT159_CSCM material model. It
was found, that usage of approximately 100,000 elements is suffi-
cient for the analyzed specimen. The same size of elements
(30
30
30 mm) was used in Ref. [4] for an analysis of a similar
specimen. The size of the elements was chosen to provide at least
two elements in the concrete cover which was necessary for the
analysis.
The use of smaller elements both for concrete and air increased
the calculation time enormously without corresponding increase in
the quality of the results. The used element size proved itself to be
the most efficient.
The first elements erode when the blast overpressure wave
reaches the surface under the hypocenter at 0.25 ms, see Fig. 15. The
results show good agreement with the experiments for reinforced
concrete specimens, see Figs. 16 and 17.
The effect of the compressive strength of concrete and strain-
rate effect was studied. The Smaller damage of specimens made of
concrete with higher compressive strength is shown in Fig. 18.
A numerical study on the size of the elements was performed
according to Ref. [9]. The study showed that the blast wave prop-
agation in the material is properly controlled by the “Hourglass
control” function of the LS-DYNA solver. This function is used
within the model.
3.3. Material model of fiber-reinforced concrete
The behavior of fiber-reinforced concrete (FRC) subjected to
blast load is quite different from the behavior of plain (respec-
tively reinforced) concrete specimens. With plain concrete,
the specimen fails in tension after reaching its tensile strength.
With fiber concrete, the stiffness of the specimen decreases after
the tensile strength is reached (strain softening). The fibers take
over the tensile stress during strain softening and the material
performs with a residual tensile strength (approximately 1/3 of
the original tensile strength according to fiber type, dosage
and geometry [10e13]). The difference between stressestrain
diagrams of plain concrete and fiber concrete can be seen in
Fig. 19.
The material model of plain concrete MAT159_CSCM (material
model with damage and plasticity [14,15]) used in LS-DYNA solver
has to be recalibrated to perform according to the presented
description, i.e. to provide ductile behavior. For the analytical
description of the MAT159_CSCM material model, see Fig. 20.
The key to MAT159_CSCM material model recalibration lies in
tuning its unloading part to describe behavior of fiber concrete
while the loading part can remains unchanged [16,17] (the differ-
ence in the compressive strength of plain concrete and fiber con-
crete can be neglected). The instrument for tuning the unloading
part of the diagram is changing the fracture energy Gf of the
material.

Fig. 19. Stressestrain diagram of plain and fiber-reinforced concrete in tension. Fig. 21. Layout of the calibration experiments.
26 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28

Fig. 22. The dynamic increase factor from the calibration experiments.

Fig. 23. Forceedeflection diagrams, experiment vs. its numerical evaluation.

The fracture energy Gf can be expressed as an integral of the area fc,cyl ¼ 30 MPa) for various loading rates: the usual 0.2 mm/min, and
below the stressestrain diagram: the accelerated 1 mm/min, 2 mm/min, 4 mm/min and 6 mm/min.
The peak value of the force in the forceedeflection diagram
ZN increased with the loading rate. Since the plastic fibers do not in-
Gf ¼ ð1  dÞfct dx; (4) fluence the compressive strength and tensile strength before crack
opening, the values could be compared with the referenced curves
0
for plain concrete under tension [29,30]. The tests were performed
where: d is the damage (from 0 to 1) and fct is the tensile strength. to verify this assumption.
According to this approach, an integral of the area below the The DIF is included in the material models used by the LS-DYNA
stressestrain diagram is used for the verification of the value of the solver (e.g. MAT159_CSCM). The material model used in the nu-
fracture energy. merical evaluation does the scaling dependent on the loading rate.
The FRC material model was calibrated according to specially The fracture energy increases with the loading rate as well [20]. The
designed experiments (detailed description in Ref. [18]). The ex- calibration four-point bending tests were modeled with the loading
periments studied the strain-rate effect of FRC specimens rate of 0.2 mm/min (3 ¼ 8.696$106) which can be approximately
(700
150
150 mm, C30/37, fc,cyl ¼ 30 MPa) subjected to various regarded as static loading.
speeds of loading under 4 points bending load tests (see Fig. 21). The calibration models were prepared using LS-DYNA for plain
The concrete mix properties and fiber dosage were the same as in concrete and both dosages of PP fibers. The fracture energy in
the field experiments. The finite elements size in the numerical tension of the material model MAT159_CSCM of plain concrete was
model used for evaluation of the experiments was chosen the same
as in the full scale 3D models described in the preceding para-
graphs, as the blast modeling is highly scale-sensitive. Table 3
Facture energies for different strength classes of concrete and various fiber dosages.
The experiments were performed to verify the behavior of fiber-
cement composites subjected to various loading rates: the (for Specimen The area under Fracture energy
FRC testing) usual 0.2 mm/min, and the accelerated 1 mm/min, the force/deflection (normalized to
diagram [N m] the CS area) [N/m]
2 mm/min, 4 mm/min and 6 mm/min. The tensile strength,
LS-DYNA e C30/37 plain 5.6 248.9
compressive strength and modulus of elasticity of concrete increase
concrete
with the loading rate [19]. The ratio between the static values Calibration experiment 62.0 2755.6
(loading rate ¼ 1$106) and the values obtained at higher loading C30/37 e 4.5 kg/m3
rates is called the dynamic increase factor ¼ DIF, and it is used for LS-DYNA e FRC 4.5 kg/m3 68.6 3048.9
modeling fast dynamic phenomenae, like blast and impact or other Calibration experiment 109.3 4857.8
C30/37 e 9 kg/m3
extreme loadings. The values of DIF in Fig. 22 were obtained for LS-DYNA e FRC 9 kg/m3 102.9 4573.3
FRC specimens as described (700
150
150 mm, C30/37,
 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28
Fig. 24. Comparison of the numerical modeling and the experiment C30/37 e 4.5 kg/m3 fibers top surface of the specimen.
modified to obtain similar fracture energy both from the experi-
ments and the numerical evaluation.
For comparison of the forceedeflection (Fed) diagrams of plain
concrete, 4.5 kg/m3 and 9 kg/m3 PP fibers FRC both experimental
and numerical, see Fig. 23. The area of the forceedeflection diagram
was limited to the deflection of 8 mm.
The fracture energy of concrete using MAT159_CSCM material
model can be only modified by changing the slope of the linear
strain softening curve. The area under the bi-linear stressestrain
diagram of concrete was sought to be equal to the area of the forcee
deflection diagram of FRC from the calibration experiments.
Because the forceedeflection diagram of FRC is more complicated,
the agreement is limited, as can be seen in Fig. 23.
According to the RILEM recommendations [22], the standard
method for evaluation of fracture energy of concrete is based on a
three-point bending test. The fracture energy Gf [N/m] is defined as
the work AF [N m] of the force P [N] causing deflection dmax [m]:
Z
dmax
AF ¼ P dd; (5)
0
applied on a unit of area BW [m2]:
AF
GF ¼ ; (6)
BW
where B [m] is the width of the specimen and W [m] is the height of
the crack.
The four-point bending test is better for fiber concrete because
the location of the crack is not localized in the middle of the
specimen but in the weakest spot according to the distribution of
fibers.
Fig. 25. Comparison of the numerical modeling and the experiment C30/37 e 4.5 kg/m3 fibers, soffit of the specimen.
27
The obtained values of the fracture energy for various concrete
strength classes and fiber dosages can be seen in Table 3; the left
column presents the value of the area under the force/displacement
diagram, the right column presents the value of the fracture energy,
i.e. the value from the left column normalized on a unit of area.
The input number for the tensile fracture energy in the MAT_159
material model is scaled according to the ratio of the fracture en-
ergy of FRC and plain concrete as presented in Table 3.
The calibrated material model was used in the 3D models of
specimens No. 1, 2 and 5 described in the previous chapter. For an
example of the results of the modeling, see Figs. 24 and 25. Due to
the limited extent of the experiments, the values of fracture energy
for C55/67 concrete with/without fibers could not be obtained.
The results obtained from numerical modeling of FRC sub-
jected to blast loading, the method of incorporating the effect of
fibers on the blast performance of concrete based on increasing
the fracture energy was compared to the results of the experi-
ments. The area of specimen puncture and concrete spalling at
both surfaces was used as the scale of the agreement, see Table 4.
The agreement is very good, the method will be used for further
numerical investigations.
3.4. Conclusions from the numerical modeling
The numerical modeling of reinforced concrete specimens
showed great agreement with the experiments. A method of
incorporating the effect of fibers on the blast performance of con-
crete based on increasing the fracture energy was used for the fiber-
reinforced specimens. It was based on supplementary experiments
performed on smaller specimens and showed good results. This
method will be generally used for further numerical investigations
after calibration for other concrete strength classes and fiber
contents.
28 M. Foglar, M. Kovar / International Journal of Impact Engineering 59 (2013) 18e28
Table 4
Damage of the specimens expressed in the extent of concrete spalling; comparison
of the calculated and measured values.
Specimen No. 1 2
Concrete C30/37 C30/37
Fibers e 4.5 kg/m3
Puncture e FEM model 0.41 m2 0.19 m2
Puncture e experiment 0.43 m2 0.26 m2
Puncture e agreement 95% 73%
Concrete spalling (soffit) e FEM model 2.40 m2 1.43 m2
Concrete spalling (soffit) e experiment 2.07 m2 1.78 m2
Concrete spalling (soffit) e agreement 86% 80%
Concrete spalling (top surface) e FEM model 0.65 m2 1.33 m2
Concrete spalling (top surface) e experiment 0.53 m2 0.96 m2
Concrete spalling (top surface) e agreement 82% 72%
4. Summary and conclusions
The paper presented results of the field experiments focused on
the effect of fibers, concrete compressive strength and its combi-
nation on blast performance of concrete. The tests were performed
using real scale precast slabs (0.3
1.5
6 m) and 25 kg of TNT
charges placed in a distance from the slab.
The experiments showed the beneficial effect of added fibers on
blast performance of the specimens. The volume of the debris was
reduced, the character of the specimen failure was changed from
brittle to ductile. The debris was more concentrated by the fibers, it
can be assumed, that it could not fly very far due its higher mass.
The experiments also showed the beneficial effect of increased
concrete compressive strength on the blast performance of the
specimens. The specimens from high performance concrete were
less damaged, the deflection was reduced; the area of puncture was
reduced. The shape of the top/bottom spalling held its shape, but its
area was reduced.
Combination of fibers and increased compressive strength
proved itself to be very effective for improving the blast perfor-
mance. The high performance concrete (C55/67) specimen with
4.5 kg/m3 (0.5% in volume) fiber content performed in a similar way
to the regular concrete specimen (C30/37) with double fiber con-
tent (9 kg/m3, 1% in volume), which is already high to be achieved
on site. The double fiber content provided the same increase of the
blast performance as the double compressive strength.
The results of the experiments are plotted on the spall and
breach prediction curves of Ref. [21], the RC specimens show
agreement with the spall and breach prediction curves, the spall
and breach prediction curves are not suitable for FRC.
Debris of crushed concrete on the bottom surface of RC structural
elements is ejected into the surroundings at very high speed and
behaves like gun-shot projectiles. It can cause severe injuries to
persons that are in the distance where they are no longer endan-
gered by the primary effect of the blast overpressure wave. Reduc-
tion of the volume of debris by fiber insertion represents a significant
improvement in overall blast safety of concrete structures.
Acknowledgments
The support of the Czech Science Foundation project no. 103/09/
2071 and 13-30441S is gratefully acknowledged.
References
[1] Buchan PA, Chen JF. Blast resistance of FRP composites and polymer
strengthened concrete and masonry structures e a state-of-the-art review.
Composites: Part B 2007;38:509e22.
[2] Millard SG, Molyneaux TCK, Barnett SJ, Gao X. Dynamic enhancement of
blast resistant ultra high performance fiber-reinforced concrete under
flexural and shear loading. International Journal of Impact Engineering
2010;37(4):405e13.
[3] Wu C, Oehlers DJ, Rebentrost M, Leach J, Whittaker AS. Blast testing of ultra-
high performance fiber and FRP-retrofitted concrete slabs. Engineering
5 Structures 2009;31(9):2060e9.
C30/37 [4] Coughlin AM, Musselman ES, Schokker AJ, Linzell DG. Behavior of
portable fiber reinforced concrete vehicle barriers subject to blasts from
9 kg/m3
contact charges. International Journal of Impact Engineering 2010;37(5):
e
521e9.
e
[5] Schenker A, Anteby I, Gal E, Kivity Y, Nizri E, Sadot O, et al. Full-scale field tests
100% of concrete slabs subjected to blast loads. International Journal of Impact
1.81 m2 Engineering 2008;35(3):184e98.
1.32 m2 [6] Seible F, Hegemier G, Karbhari VM, Wolfson J, Arnett K, Conway R, et al.
73% Protection of our bridge infrastructure against man-made and natural hazards.
1.50 m2 Structure and Infrastructure Engineering 2008;4(6):415e29.
1 m2 [7] Henrych J. Blast dynamics and its use (in Czech). Praha: Academia; 1973.
61% [8] Lawrence Software Technology Corporation. LS-DYNA keyword user’s manual.
Version 971. Livermore, CA, USA: Livermore Software Technology Corpora-
tion; 2007.
[9] Nam JW, Kim J-HJ, Kim SB, Yi NH, Byun KJ. A study on mesh size dependency
of finite element blast structural analysis induced by non-uniform pressure
distribution from high explosive blast wave. KSCE Journal of Civil Engineering
2008;12(4):259e65.
[10] Bencardino F, Rizzuti L, Spadea G, Swamy RN. Experimental evaluation of fiber
reinforced concrete fracture properties. Composites Part B: Engineering
January 2010;41(1):17e24.
[11] Nakamura H, Mihashi H. Evaluation of tension softening properties of fiber
reinforced cementitious composites. Fracture Mechanics of Concrete Struc-
tures 1998;I:499e510.
[12] Mohammadi Y, Carkon-Azad R, Singh SP, Kaushik SK. Impact resistance of
steel fibrous concrete containing fibers of mixed aspect ratio. Construction
and Building Materials January 2009;23(1):183e9.
[13] Nyström U, Gylltoft K. Comparative numerical studies of projectile impacts on
plain and steel-fiber reinforced concrete. International Journal of Impact En-
gineering FebruaryeMarch 2011;38(2e3):95e105.
[14] Murray YD. User manual for LS-DYNA concrete material model 159. US
Department of Transportation; 2007. FHWA-HRT-05-062.
[15] Murray YD. Evaluation of LS-DYNA concrete material model 159. US Depart-
ment of Transportation; 2007. FHWA-HRT-05-063.
[16] Leppänen J. Concrete subjected to projectile and fragment impacts: modelling
of crack softening and strain rate dependency in tension. International Journal
of Impact Engineering November 2006;32(11):1828e41.
[17] Teng T-L, Chu Y-A, Chang F-A, Shen B-C, Cheng D-SC. Development
and validation of numerical model of steel fiber reinforced concrete for
high-velocity impact. Computational Materials Science March 2008;42(1):
90e9.
[18] Drahorád M, Foglar M, Smirinský S, Veselý M. The use of dynamic increase
factor for speed-up of the load tests of fiber reinforced concrete 2012
(in Czech). Beton TKS 1/2012. p. 74e9.
[19] Habel K, Gauvreau P. Response of ultra-high performance fiber reinforced
concrete (UHPFRC) to impact and static loading. Cement and Concrete Com-
posites November 2008;30(10):938e46.
[20] Ulfkjaer J, Labibes K, Solomos G, Albertini C. Tensile failure of normal concrete
and steel fiber reinforced concrete at high strain rates. Fracture Mechanics of
Concrete Structures 1998;I:585e97.
[21] UFC 3-340-02: structures to resist the effects of accidental explosions. U.S.
Army Corps of Engineers, Naval Facilities Engineering Command, Air Force
Civil Engineer Support Agency; 2008.
[22] Rilem Commitee FMC-50. Determination of the fracture energy of mortar and
concrete by means of the three-point-bend tests on notched beams. Materials
and Structures 1985;18:285e90.
[23] Zhou XQ, Kuznetsov VA, Hao H, Waschl J. Numerical prediction of concrete
slab response to blast loading. International Journal of Impact Engineering
October 2008;35(10):1186e200.
[24] Silva PF, Lu B. Improving the blast resistance capacity of RC slabs with inno-
vative composite materials. Composites Part B: Engineering JulyeSeptember
2007;38(5e6):523e34.
[25] Ong KCG, Basheerkhan M, Paramasivam P. Resistance of fiber concrete slabs to
low velocity projectile impact. Cement and Concrete Composites 1 December
1999;21(5e6):391e401.
[26] Robins PJ, Austin SA, Richards MR. The resistance of steel fiber concrete
to VTOL engine jet blast. Cement and Concrete Composites 1994;16(1):
57e64.
[27] Mosalam KM, Mosallam AS. Nonlinear transient analysis of reinforced con-
crete slabs subjected to blast loading and retrofitted with CFRP composites.
Composites Part B: Engineering December 2001;32(8):623e36.
[28] Nam J-W, Kim H-J, Kim S-B, Yi N-H, Kim J-HJ. Numerical evaluation of the
retrofit effectiveness for GFRP retrofitted concrete slab subjected to blast
pressure. Composite Structures April 2010;92(5):1212e22.
[29] Malvar LJ, Ross CA. Review of strain rate effects for concrete in tension. ACI
Materials Journal 1998;95(6):735e9.
[30] Tedesco JW, Ross CA. Strain-rate-dependent constitutive equations for con-
crete. Journal of Pressure Vessel Technology 1998;120(4):398e405.