ARTICLE IN PRESS

International Journal of Impact Engineering 31 (2005) 843–860

Experiments and numerical analyses of blast and fragment

impacts on concrete
Joosef Lepp.anen*
Department of Structural Engineering and Mechanics, Concrete Structures, Chalmers University of Technology,
.
SE-412 96 Goteborg, Sweden
Received 11 November 2003; received in revised form 20 April 2004; accepted 23 April 2004
Available online 8 June 2004

Abstract

Concrete structures are commonly used as protective structures. An important issue is how the blast
wave and fragment impacts from an explosion affect the concrete. It is well known that the fragments
penetrate or even perforate the structure. Moreover, spalling occurs in the impact zone and scabbing may
occur on the reverse side of a wall that receives an impact. However, knowledge of how the blast wave and
fragment impacts influence the material properties of concrete is quite limited. Experiments and numerical
analyses were carried out to examine the extent to which the concrete, at various distances, is affected by
the blast wave and fragment impacts. The fragments, which were spherical, were shot against thick concrete
blocks by using the explosives octol or hexotol; the fragment velocity was approximately 1650 m/s. After
the concrete blocks were shot, the depths of penetration and spalling were measured. Next, the concrete
blocks were cut into halves, and the global macro-cracking could be observed. To study how the material
properties of concrete were influenced, uniaxial compressive and splitting tensile tests were carried out on
cylinders drilled from selected positions in the block. Furthermore, specimens from the blocks were thin-
ground to facilitate analysing the micro-cracking with a microscope. The experiments and numerical
analyses presented here showed that the damage in the concrete, from the blast wave and fragment impacts,
is localized in the impact zone. The concrete below this zone, at a depth of approximately twice the depth of
the maximum penetration, was hardly affected at all by the blast wave and fragment impacts. This indicates
that it is possible to distinguish between the global load effects and the local damage effects that are caused
by the fragment impacts. Consequently, it may be possible to separate the loads, at the design stage, from a
blast wave and fragment impacts.
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Concrete; Fragment impacts; Blast wave; Experiment; Numerical analysis

*Fax: +46-(0)31-772-2260.
E-mail address: [email protected] (J. Lepp.anen).

0734-743X/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijimpeng.2004.04.012
 ARTICLE IN PRESS

844 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

1. Introduction

Since massive concrete structures withstand blast waves and fragment impacts effectively, they
are often used as protective structures. According to the Swedish Shelter Regulations [1], a shelter
shall withstand: the effect of a pressure wave corresponding to that produced by a 250-kg GP
bomb with 50% by weight TNT that bursts freely outside at a distance of 5.0 m from the outside
of the shelter during free pressure release. Furthermore, according to these regulations, the shelter
shall withstand the effect of fragment impacts from the same type of bomb. In design, normally,
the thickness of concrete is dimensioned to withstand the fragment impacts; a static load, with a
dynamic increase factor, approximates the blast load.
Experiments show that a concrete structure exposed to a combination of blast wave and
fragments collapses more easily than one that is exposed only to a blast wave or fragment impacts,
see Forse! n and Edin [2]. The impulse density of fragment impacts is usually much lower that of a
blast wave; therefore, the slightly increased impulse density observed for the combined blast wave
and fragment impacts cannot explain the greater damage to the structure. It is believed that the
reason for the increased damage is a combination of the spalling effect and the increased impulse
density [2]. Forse! n and Nordstrom . [3] experimented with combined blast wave and fragment
impacts against concrete slabs. They showed that a very good estimation of the deflection can be
obtained, by taking into account that the resistance of the slab is decreased by the fragment
impacts and by adding the fragment impulse density to the positive impulse density of the blast
wave. The pressure–time history was rearranged to a triangular shape by using the maximum
pressure from the blast wave.
Earlier experiments have dealt with the structural level: beams, walls or even entire structures
were analysed [2,3] and Nordstrom . [4]. Nevertheless, the effect of the blast wave and fragment
impacts on the concrete material properties is not known in detail. The study reported in this
paper aims to add to the knowledge of how the damage to concrete is affected by the blast wave
and fragment impacts. The spherical fragments were shot by a detonation, with either octol or
hexotol, against thick non-reinforced concrete blocks. However, in a real bomb detonation, the
fragments are not spherical, and the structure is reinforced. The purpose of the simplifications in
the experimental set-up was to have as few uncertain parameters as possible. With improved
understanding of how the concrete material properties are influenced by the loading, other
parameters can be added, for example, irregular fragments, reinforced concrete, structural
elements, or even reinforced concrete structures.
The depth of penetration and cratering were measured after the concrete blocks were shot.
Next, to study the damage, the blocks were cut into halves so that global macro-cracking could be
observed. Uniaxial compressive and splitting tensile tests were made on drilled cylinders to study
the change in strength. Furthermore, thin-ground specimens were prepared from the blocks, to
facilitate analysis of micro-cracking with microscope.
Another aim was to investigate how numerical methods can simulate the experiments with the
loading of combined blast wave and fragment impacts; the experimental set-up was chosen in
order to have clear boundary conditions for the numerical analyses presented in this paper. In the
literature several papers deal with numerical analyses of projectile penetration, as in Clegg et al.
[5], Johnson and Beissel [6], Johnson et al. [7], Lepp.anen [8], Scheffler and Zukas [9], Zukas and
Scheffler [10]. Numerical analyses of a blast wave against concrete structures were carried out by
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 845

Johansson [11], Krauthammer [12], Krauthammer and Otani [13]. Numerical analyses of single
fragment impacts were made by Ag ( a( rdh and Laine [14]; Papados [15] conducted numerical
analyses of multiple fragment impacts.
In this paper numerical analyses of combined blast wave and fragment impacts were carried
out, and the software used was AUTODYN [16]; this is a general-purpose program for solving a
variety of non-linear problems in dynamics. The phenomena studied with this type of program
can be characterized as highly time dependent with both geometric and material non-linearities.
The code, which combines finite difference, finite volume, and finite element techniques, is known
as a hydro-code.

2. Experimental set-up

The aim of the experiments was to study the damage caused by a blast wave and fragment
impacts at various depths in concrete blocks. The charges and the dimensions of the concrete
blocks were chosen to obtain a damage level high enough so that drilling of cylinder specimens
would be possible. After casting, the concrete blocks were turned upside down, and the fragments
were shot against the finer surface (at the bottom of the cast concrete). The dimensions of the
blocks were 750  750  500 mm3, and the fragments were spherical with a radius of 4 mm; the
impact velocity was around 1650 m/s. The test set-up is shown in Fig. 1. A total of six blocks were
shot with fragments. To vary the fragment area density, the charges were fired from four heights
above the concrete blocks.
The charges were made by gluing bearing balls onto a convex end of a cylinder. Two similar
explosives were used, octol and hexotol, both with a weight of 1.3 kg. A total of seven charges
were fired, five with octol and two with hexotol, and the height was varied between 0.6 and 1.0 m.
The fragment velocity was measured with an accelerometer and impact sensor. Although,
single fragments were shot against concrete blocks, the results discussed in this paper are limited
to the multi-fragment impacts. Information about the single fragment shots may be found in
Lepp.anen [17].

Charge
Fragments

Accelerometer
Wood fibre sheets h

Specimen Impact
sensor

Fig. 1. Experimental test set-up.

 ARTICLE IN PRESS

846 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

Table 1
Fragment impact velocities and experimental set up
Charge nr. Block nr. Velocity (m/s) Explosive material Height (m) Frame
1a 1 1450 Octol 1.0 No
2 1 1680 Octol 1.0 No
3 2 1660 Octol 0.8 No
4 3 1650 Hexotol 0.6 Yes
5 4 1650 Octol 0.7 Yes
6b 5 — Octol 0.7 Yes
7b 6 — Hexotol 0.6 Yes
a
Four wood fibre sheets with a total thickness of 51 mm were placed below the charge.
b
For charges six and seven, the velocity could not be registered; the signal that registered the velocity did not work.

The first two charges were fired against block one. For the first shot, four sheets of wood fibre
were placed below the charge to reduce the velocity. The fragment impact velocity for this shot
was 1450 m/s. To increase the impact velocity, the wood fibre sheets were removed for the
remaining shots. For charge two, the velocity was raised to 1680 m/s. In block two (charge three)
the damage at the edge was high. To facilitate drilling cylinders from the blocks (as described in
Section 3.2), a steel frame was used to reduce the damage at the edges for the remaining shots. The
steel frame was made of a 10 mm thick L-profile that covered 70 mm of the concrete edges. The
fragment impact velocities and the experimental set-up for the seven charges fired are shown in
Table 1.

3. Experimental results

After shooting fragments at the concrete blocks, photographs were taken; the depth of
penetration, crater depth and diameter were measured. To study how the material properties were
changed, drilled cylinders with dimensions of f50  100 mm2 were used for uniaxial compressive
and tensile tests. Furthermore, the concrete blocks were cut into halves so that the macro-cracking
could be examined. To analyse micro-cracking, thin-ground sections were taken from the blocks;
using a microscope, the cracks could be observed.

3.1. Photographing the concrete blocks

Five concrete blocks (numbered from two to six), after being struck by fragments, were
photographed as shown in Fig. 2. The cracks were marked to improve the visualization. The
depth of penetration varied between 30 and 50 mm, and the crater diameter varied between 45 and
60 mm for the concrete blocks. To study the macro-cracking in the blocks, they were cut into
halves and the crack pattern was examined, see Fig. 3 with the crack patterns marked. All five
blocks had similar overall crack patterns, with clear spalling in the impact zone; the depth of the
damage zone was approximately 50 mm. At the boundaries of the blocks, a global crack pattern
developed; this is caused by reflections of the stress wave generated by the impulse from the blast
and fragment impacts.
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 847

Fig. 2. Top view of concrete blocks two to six after being shot with fragments.

To study the change in material properties, uniaxial compressive and splitting tensile tests were
conducted on cylinders drilled from the concrete blocks. However, during drilling, most of the
cylinders broke into two pieces. The broken cylinders were measured and drawn, as shown in
Fig. 4. For the cylinders drilled from the middle section, the micro-cracks could not be seen from
the cross section (compare Fig. 3). Still, the cylinders broke, due to micro-cracking. Cylinders
 ARTICLE IN PRESS

848 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

Fig. 3. Macro-crack patterns in the cross section of concrete blocks (after cutting into halves).

Fig. 4. Crack plane drawn from drilled cylinders.

were also drilled from a reference block that was not subjected to any loading; these cylinders
could be drilled out in whole pieces.

3.2. Uniaxial compressive tests

From the drilled cylinders, f50  100 mm2 specimens were sawed out at various heights and
smoothed for uniaxial compressive tests. Cylinders were drilled in two directions: horizontally
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 849

500

400

Height [mm]
300 Block 3,
horizontal
References
200

100

0
0 10 20 30
Compressive strength [MPa]

Fig. 5. Uniaxial compressive tests on cylinders drilled horizontally from concrete block three. Marked circles show
where the cylinders were drilled.

(perpendicular to the direction of the fragment impacts) and vertically. In concrete block three,
the first one analysed, the cylinders were drilled horizontally at heights of 50, 125, 200, 275, 350
and 425 mm measured from the side of the block opposite the surface struck by the fragment. The
cylinders drilled at a level of 425 mm were so severely damaged that none of them could be used
for the uniaxial tests. In the uniaxial compressive test on cylinders drilled in the horizontal
direction, the strength was hardly affected at all by the fragment impacts, as shown in Fig. 5.
In the vertical direction, f50  100 mm2 specimen were saved out from the drilled cylinders at
heights of 55, 90, 190, 260, 290 and 360 mm, as shown in Fig. 6. The compressive strength was
very little affected at a depth of 140 mm below the fragment impacts. Instead, the uniaxial
compressive tests showed somewhat higher strength than in the reference cylinders (concrete
block that was not exposed to any loading). However, the specimen that was drilled near the
global crack pattern at the block edges had much less strength. For the specimen drilled from the
middle section, where micro-cracking occurred, the strength was reduced up to one-third of the
uniaxial compressive strength of cylinders drilled from the reference block.
To verify the results of the compressive test on block three, additional specimens were taken
from drilled cylinders from blocks four and block six. Three specimens were taken out at various
heights, all from the centre of the blocks; the area is marked in Fig. 7, as well as the results from
these tests. The overall response was very similar to that of block three. Below the spalling zone, at
a depth of 140 mm below the surface of the fragment impacts, the strength of concrete was slightly
higher than in test samples taken from a reference block. At the bottom of the blocks, the strength
was less.

3.3. Uniaxial splitting tensile tests

From the drilled cylinders specimens were sawed out at various heights and smoothed also for
uniaxial splitting tensile tests. For these, the orientation when testing is important, as shown in
 ARTICLE IN PRESS

850 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

500

400
a a

Block 3,

Height [mm]
300
a horizontal

References
200
b

100

0
0 10 20 30
Compressive strength [MPa]

Fig. 6. Uniaxial compressive tests for cylinders drilled vertically in concrete block three. The cylinders are drilled from
marked areas. Results at different heights correspond to the centre of gravity for each cylinder. (a) The specimens were
taken out near the edge. (b) A crack was visible in the specimen before the test.

500

400
Block 4,
Height [mm]

vertical
300
Block 6,
vertical
200
References

100

0
0 10 20 30
Compressive strength [MPa]

Fig. 7. Uniaxial compressive tests for vertically drilled cylinders: concrete blocks 4 and 6. The cylinders are drilled from
marked areas.

Fig. 8: if a specimen is loaded where there are cracks parallel to the loading, the strength is
reduced, but if it is loaded perpendicular to the cracks, the strength is hardly affected. The first
concrete block analysed was number three. In the tests, cylinders that were drilled vertically
showed very little change in strength, as shown in Fig. 9. This is due to the effect of the crack
orientation, as already noted; the main global crack pattern was perpendicular to the direction of
the drilling.
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 851

σ1 σ2

σ1 < σ2

σ1 σ2

Fig. 8. Influence of the strength due to orientation of a crack.

500

400
Height [mm]

300 Block 3,
vertical

200 References

100

0
0 1 2 3 4
Splitting tensile strength (Mpa)

Fig. 9. Uniaxial splitting tensile tests, vertically drilled cylinders, block three.

For horizontally drilled cylinders, as for the compressive tests, the splitting tensile strength was
affected very little by the blast wave and fragment impacts at a depth of 150 mm from the top
surface. At the lower levels, where the global crack plane was in the cross section, the strength was
reduced in most of the specimens. The results of the splitting tensile tests are shown in Fig. 10.
However, for block three, the crack orientation was not registered when the splitting tensile tests
were carried out.
To study the effect of the crack orientation in block six, specimens were marked before drilling,
so that the main orientation of the micro-cracks was known in the splitting tensile tests. Although
the number of tests was limited, some inferences could be made from them. At a level of 50 mm
from the bottom, in the tests with the cracks orientated perpendicular to the loading, the average
splitting tensile strength was 2.8 MPa. The average strength was 2.5 MPa, for the test at the same
level, with the cracks orientated parallel to the direction of the loading. The average strength for
the references was 2.9 MPa. The results are shown in Fig. 11.
Specimens from block four were also tested. For these tests the crack was orientated parallel to
the loading. The results were similar to those for block six; the strength was hardly affected in the
 ARTICLE IN PRESS

852 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

500

400

Block 3,

Height [mm]
300
vertical

References
200

100

0
0 1 2 3 4
Splitting tensile strength (Mpa)

Fig. 10. Uniaxial splitting tensile tests, horizontally drilled cylinders, block three.

500

400
perpendicular a
Height [mm]

300 parallelb
References
200

100

0
0 1 2 3 4
Splitting tensile strength (Mpa)

Fig. 11. Uniaxial splitting tensile tests, horizontally drilled cylinders, block six. (a) Crack orientated perpendicular to
the direction of the loading. (b) Crack orientated parallel to the direction of the loading.

upper zone (below the fragment impacts), except for one specimen, see Fig. 12. However, for the
specimens taken out at the bottom, the strength was lower. At a level 50 mm above the bottom,
the splitting tensile strength was 2.4 MPa, which is approximately the same as for block six.

3.4. Thin-ground sections

Thin grinding, a precise method to localize the micro-cracking in a material, is commonly used
in geological studies, see Kim and McCarter [18]. Here, to study the micro-cracking in the
concrete below the impact zone further, a vertically drilled cylinder at the mid section was
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 853

500

400

Height [mm]
300 parallel

200 References

100

0
0 1 2 3 4
Splitting tensile strength (Mpa)

Fig. 12. Uniaxial splitting tensile tests, horizontally drilled cylinders, block four.

thin-ground to a rectangular section of 90  50 mm2, and a thickness of 25 mm. The section was
thin-grounded from two opposite sides from one of the drilled cylinders at a depth between 80 and
170 mm below the surface of block three. And as a reference, a thin-ground section was also taken
from a block that was not subjected to any loading; this section was 90  50 mm2, and had a
thickness of 25 mm. Examples of photographs of the thin-ground sections were taken with a amera
(placed inside the microscope), as shown in Fig. 13. To make the micro-cracks easier to see,
polarized light was used when the photographs were taken. Also before grinding, the specimens
were impregnated with fluorescent penetrant, to make the cracks clearer. The thin-ground sections
from the upper zone showed that micro-cracking occurred at a depth of approximately 120 mm
below the surface of the fragment impact. The width of the micro-cracks is up to approximately
0.02 mm. These results verify well the uniaxial tests from drilled cylinders. The concrete strength
was not affected at a depth of 150 mm below the surface. Furthermore, cylinders could not have
been drilled 125 mm below the surface; at this level, there was micro-cracking.

4. Numerical analyses of the experiments

4.1. The numerical model

A combination of experiments and numerical methods is a powerful tool for detailed analyses.
The Lagrangian method has been used in AUTODYN for the analyses. The governing equations
in AUTODYN are the Rankine-Hugoniot equations: the conservation of mass, momentum and
energy. To complete the description of the continuum, two additional relations describing the
material behaviour are needed (besides the load and boundary conditions): first the equation of
state (EOS), and second a constitutive model. The EOS relates the pressure to the density; the one
used in the numerical analyses was chosen from the AUTODYN material library, since no three-
axial material tests were conducted during the experiments. The constitutive model used was the
RHT one in AUTODYN, developed by Riedel [19]. The model, which consists of three yield
 ARTICLE IN PRESS

854 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

Fig. 13. Photographs of thin-ground sections at three depths (block 3, and reference). Numbers refer to depth below
the surface.

Yield strength,Y

Failure Surface

Elastic Limit Surface

Residual Strength

Pressure, p

Fig. 14. The RHT constitutive model used for concrete [19].

surfaces, as shown in Fig. 14, includes pressure hardening, strain hardening, strain rate hardening,
third-invariant dependence for compressive and tensile meridians, and a damage model for strain
softening. For a detailed description of the material model, see AUTODYN [16] and Riedel [19].
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 855

Table 2
Material parameters for the concrete
Parameter Value
Young’s modulus 20.7 GPa
Compressive cylinder strength 31.2 MPa
Splitting tensile strength 3.16 MPa
Tensile strength 2.84 MPa
Density 2225 kg/m3

Several methods to calculate the failure surface have been published, for example, Attard and
Setunge [20] and Imran and Pantazopoulou [21]. However, in the analyses presented here, the
failure surface and the residual strength are determined by a model proposed in Ref. [20]. The
failure surface is determined for confined concrete and static load. To describe the failure surface
and the residual strength, the uniaxial compressive cylinder strength and Young’s modulus form
the input to the model.
The maximum aggregate size used in the concrete mix was 8 mm. From standard uniaxial
cylinder tests, the compressive cylinder strength of the concrete was determined; the same was
done for the splitting tensile strength. The tensile strength was calculated according to CEB-FIB
Model Code 1990 [22]. To determine Young’s modulus, another standard test was conducted. All
test results and detailed input for the analyses in AUTODYN are published in Lepp.anen [17]; here
the material parameters for concrete are shown in Table 2.

4.2. The blast wave

The charge used to shoot the fragments into concrete block three (charge four), was 1.3 kg
hexotol; it was fired from a height of 0.6 m above the concrete block. In the experiments the
pressure was not measured. The pressure–time history from the blast wave was estimated by
ConWep [23], see Fig. 15. The arrival time for the blast wave was estimated to be approximately
0.2 ms. However, in the analyses, the pressure was applied to the surface of the concrete block at
time zero; this pressure was simplified to a piecewise fall from 25.26 MPa to zero, with a positive
duration time of 0.42 ms. Another approximation in the analyses was that the same pressure was
applied to the whole surface: in reality, the pressure varies on the surface depending on the
distance from the charge and angle.

4.3. Description of the mesh and boundary conditions

To analyse penetration in concrete, a very fine mesh must be used, see Refs. [8] or [10]. The
usual method is to refine the mesh, and compare the coarse mesh with the refined one until the
results differ only negligibly. 2-D analyses were carried out first to determine the mesh size. For
the 2-D analyses only one fragment were shot against the concrete block. The same cross section
was used in the 2-D analyses as for the 3-D analyses, as shown in Fig. 16. Quadratic elements were
used and the number of elements in the plane was varied. Three different mesh sizes were used.
The coarse mesh had 50  50, the medium had 100  100 and the fine mesh had 200  200
 ARTICLE IN PRESS

856 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

25000
Pressure from
ConWep [23]

20000 Simplified relation

Pressure [kPa] used in the analyses

15000

10000

5000

0
0.0 0.1 0.2 0.3 0.4 0.5
Time [ms]

Fig. 15. Pressure–time history from the blast wave; 1.3 kg hexotol.

Cross section

Fragments

Blast wave

Plane view
57 elements

100 elements

100 elements

Fig. 16. Numerical mesh. One quarter of the block is modelled.

 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 857

elements in the plane. The difference in the spalling between the coarse mesh and the medium
mesh was 29% and the difference was 4% between the medium mesh and the fine mesh. The
medium mesh was chosen for the 3-D analyses. The final 3-D mesh that was used in the analyses is
shown in Fig. 16. One quarter of a block (with symmetry boundaries) was modelled; it consisted
of 100  100  57 elements. For the first 100 mm below the surface of the fragment impacts, 27
fixed size elements were used, and for the 400 mm below this zone, 30 elements were used. The size
of the elements increases, the closer to the bottom they are.
The arrival time for the fragments was calculated as 0.36 ms (based on an average fragment
velocity of 1650 m/s and a height of 0.6 m). The blast wave hits the concrete block before the
fragments do; the difference between the arrival times is approximately 0.16 ms. In the analyses
the blast wave was applied at time zero, and the fragments were placed 0.264 m above the concrete
block. This corresponds to a time difference of 0.16 ms, when the fragments strike at a velocity of
1650 m/s. The location of the fragments in the plane were determined from the experiments, see
Fig. 2. The analyses were simplified by designating the angle of impact as 90 . Another
simplification was that all of the fragments arrived simultaneously. Furthermore, in the
experiment the blocks were placed on a concrete floor, and the stress wave can propagate to it.
Whereas in the analyses, at the bottom of the concrete block, both free and fixed boundary
conditions in the direction of the loading were used, and in both analyses the energy remains in
the block.

4.4. Results from the numerical analyses

For concrete block three, three types of analyses were carried out. In the first, only the fragment
impacts were taken into account, while in the second and third, both combined blast wave and
fragment impacts were analysed, with free and with fixed boundary conditions. In these analyses,
the spalling zone was well simulated. A comparison of experiments and an analysis that takes into
account only the fragment impacts, viewed from above, is shown in Fig. 17. The results of all of
the analyses, first with fragment impacts and then with combined blast wave and fragment
impacts, were very similar, viewed from above. However, inside the concrete block below the
spalling zone, the blast wave was found to cause more damage in the concrete than in analyses

Fig. 17. Comparison of experiments and a contour damage plot from the numerical analysis, as seen from above.
 ARTICLE IN PRESS

858 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

Fig. 18. Comparison of experiments and contour damage plots from the numerical analyses, cross section. The variable
damage monitors the cracking strain. Damage is defined to be 0.0 for intact cell and 1.0 for fully failed cell.

where only the fragment impacts where considered. A comparison of experiments and numerical
analyses of a cross section, where the fragment impacts alone, and then combined blast wave and
fragment impacts, both with free and fixed boundary were studied, see Fig. 18.
For the analyses with the fragment impacts alone, the damage inside the concrete block was
localized. When the blast wave was also included in the analyses, the damage zone was larger.
There was a minor difference in the damage inside the block by using either free or fixed
boundary. When the stress wave propagates, caused by the blast wave, it reflects at the
boundaries; if the boundary is free, the stress wave reflects as a tensile wave, and in case of fixed
boundary, it reflects as a compressive wave. However, when using fixed boundary, the block is
restrained to lift, resulting in tensile stresses at the bottom of the block. Furthermore, when the
reflected wave reaches the top surface, energy is consumed to increase the damage that is caused
by the fragments that have already struck the surface; this is further discussed in [17]. The damage
in the concrete corresponds well with the uniaxial compressive and splitting tensile tests; see
results in Fig. 10. At a level 150 mm below the impact zone, the concrete strength was not affected
by the combined blast wave and fragment impacts. At lower levels, the strength was less in the
uniaxial tests. The contour damage plots in Fig. 18 show similar behaviour.

5. Summary and conclusions

Experiments and numerical analyses presented here show that the damage from a blast wave
and fragment impacts is localized at the surface of the impact zone. Thin grinding is an accurate
method of finding micro-cracks, and in the experiments here micro-cracking occurred at a depth
 ARTICLE IN PRESS

.
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860 859

of approximately 120 mm below the surface struck by the fragments. The concrete strength below
this zone, however, was not affected at all. At the boundaries the strength was decreased by
reflections of the stress wave.
The uniaxial compressive tests showed some increase in strength below the spalling zone; this is
probably due to compaction of the concrete. In the numerical analyses, this zone was undamaged.
From the splitting tensile tests, it was also found that the strength was not affected below the
spalling zone. Furthermore, it was noted that the orientation when testing the drilled cylinders
affected their strength. If a specimen is loaded where the cracks are parallel to the loading, the
strength is decreased. The strength of the test specimens that were loaded perpendicular to the
cracks was not affected.
In the numerical analyses, the damage in the spalling zone is caused by the fragment impacts.
To capture the response of the concrete material behaviour, both fragment impacts and the blast
wave must be taken into account. The results from analyses with combined blast wave and
fragment impacts showed greater damage inside the concrete block. This indicates that it is
possible to distinguish between the global load effects and the local damage effects that are caused
by the fragment impacts. Consequently, it may be possible for designers to separate the loads
from the blast wave and fragment impacts: the structure could be analysed as a pre-damaged
structure with decreased effective depth or width, and the impulse from the fragment impacts
could be added to the impulse from the blast wave.

Acknowledgements

The research presented in this paper was financed by the Swedish Rescue Services Agency. The
author would like to thank his supervisor, the Head of the Department of Structural Engineering
and Mechanics, Professor Kent Gylltoft, and the reference group members of the project,
‘‘Dynamic behaviour of concrete structures subjected to blast and fragment impacts’’: Bjorn .
Ekengren, M.Sc., from the Swedish Rescue Services Agency, Mario Plos, Ph.D., Chalmers
University of Technology, and Morgan Johansson, Ph.D., from Reinertsen AB.

References

[1] Ekengren B. Shelter Regulations, SR—English edition, Swedish Rescue Services Board, Publication B54-168/94,
Karlstad, Sweden; 1994. 94pp.
[2] Fors!en R, Edin K. Vapenverkan mot flerv(anings betongbyggnad III: Best.amning av skador fr(an splitterladdningar
mot husfasad i skala 1:4 (in Swedish). (Weapon effects on multi-storeyed concrete structures III: Determination of
the damage from fragment impacts to fa@ades on a scale of 1:4). FOA report C 20860-2.3, Sundbyberg, Sweden:
Swedish Defence Research Agency; 1991. 140pp.
. M. Damage to Reinforced Concrete Slabs Due to the Combination of Blast and Fragment
[3] Fors!en R, Nordstrom
Loading. FOA report B 20101-2.6, Tumba, Sweden: Swedish Defence Research Agency; 1992. 12pp.
[4] Nordstrom. M. Splitterbelastning av betongplattor III: Energiupptagande form. a( ga hos armerade betongplattor
belastade med olika splittert.atheter (in Swedish). (Fragment impacts on concrete slabs III: Energy-absorbing
ability of reinforced concrete slabs loaded with various fragment impact densities). FOA report 95-000094-
7(6,2.6)—SE, Stockholm, Sweden: Swedish Defence Research Agency; 1995. 101pp.
 ARTICLE IN PRESS

860 .
J. Leppanen / International Journal of Impact Engineering 31 (2005) 843–860

[5] Clegg RA, Sheridan J, Hayhurst CJ, Francis NJ. The application of SPH techniques in AUTODYN-2D to kinetic
energy penetrator impacts on multi-layered soil and concrete targets. Eighth International Symposium on
Interaction of the Effects of Mutions with Structures, 22–25 April 1997, Virginia, USA.
[6] Johnson GR, Beissel SR. Computed radial stresses in a concrete target penetrated by a steel projectile. Fifth
International Conference on Structures Under Shock and Impact V 1998. p. 793–806.
[7] Johnson GR, Stryk RA, Beissel SR, Holmquist TJ. An algorithm to automatically convert distorted finite elements
into meshless particles during dynamic deformation. Int J Impact Eng 2002;27(10):997–1013.
[8] Lepp.anen J. Dynamic Behaviour of Concrete Structures subjected to Blast, Fragment Impacts. Licentiate Thesis,
Department of Structural Engineering, Concrete Structures, Goteborg,. Sweden: Chalmers University of
Technology; 2002. 71pp.
[9] Scheffler DR, Zukas JA. Practical aspects of numerical simulations of dynamic events: material interfaces. Int J
Impact Eng 2000;24(8):821–42.
[10] Zukas JA, Scheffler DR. Practical aspects of numerical simulations of dynamic events: effects of meshing. Int J
Impact Eng 2000;24(9):925–45.
[11] Johansson M. Structural Behaviour in Concrete Frame Corners of Civil Defence Shelters, Non-linear Finite
Element Analyses and Experiments. DoctoralThesis, Department of Structural Engineering, Concrete Structures,
.
Goteborg, Sweden: Chalmers University of Technology; 2000. 204pp.
[12] Krauthammer T. Blast-resistant structural concrete and steel connections. Int J Impact Eng 1999;22(9-10):
887–910.
[13] Krauthammer T, Otani RK. Mesh, gravity and load effects on finite element simulations of blast loaded reinforced
concrete structures. Comput Struct 1997;63(6):1113–20.
( ardh L, Laine L. 3D FE-simulation of high-velocity fragment perforation of reinforced concrete slabs. Int J
[14] Ag(
Impact Eng 1999;22(9):911–22.
[15] Papados PP. A reinforced concrete structure under impact: response to high rate loads. Sixth International
Conference on Structures Under Shock and Impact VI 2000. p. 501–510.
[16] AUTODYN Manuals. Version 4.3, Concord, CA, USA: Century Dynamics, Inc. 2003.
[17] Lepp.anen J. Splitterbelastad betong, Experiment och numeriska analyser. Fragment impacts into concrete,
Experiments and numerical analyses. Department of Structural Engineering and Mechanics, Concrete Structures,
.
report nr. 03:6, Goteborg, Sweden: Chalmers University of Technology; 2003. 74pp [In Swedish].
[18] Kim DS, McCarter MK. Quantitative assessment of extrinsic damage in rock materials. Rock Mech Rock Eng
1998;31(1):43–62.
[19] Riedel W. Beton unter dynamischen Lasten Meso-und makromechanische Modelle und ihre Parameter. Doctoral
Thesis, der Bundeswehr Munchen, Freiburg, Germany: Institut Kurzzeitdynamik, Ernst-Mach-Institut; 2000 [in
German].
[20] Attard MM, Setunge S. Stress–strain relationship of confined and unconfined concrete. ACI Mater J
1996;93(5):432–42.
[21] Imran I, Pantazopoulou SJ. Plasticity model for concrete under triaxial compression. J Eng Mech-ASCE
2001;127(3):281–90.
[22] CEB-FIB Model Code 1990. Design Code, Lausanne, Switzerland: Thomas Telford; 1993.
[23] ConWep. Collection of conventional weapons effects calculations based on TM 5-855-1, Fundamentals of
Protective Design for Conventional Weapons, Vicksburg, USA: US Army Engineer Waterways Experiment
Station, 1992.