Вы находитесь на странице: 1из 17

International Journal of Impact Engineering 65 (2014) 146e162

Contents lists available at ScienceDirect

International Journal of Impact Engineering

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

Experimental and numerical investigation on a multi-layer protective

structure under the synergistic effect of blast and fragment loadings
Xiang-shao Kong*, Wei-guo Wu, Jun Li, Pan Chen, Fang Liu
Department of Structural Engineering, School of Transportation, Wuhan University of Technology, Wuhan 430063, China

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

Article history: The main function of a multi-layer protective structure of a combatant ship is to prevent the inner cabins
Received 7 March 2013 from being destroyed by anti-ship weapons. The damage effect of these weapons on ship structures
Received in revised form mainly comes from the blast wave and fragments. The motivation of this study was to investigate the
22 November 2013
synergistic effect of blast wave and fragment impact loadings on the multi-layer protective structure. A
Accepted 26 November 2013
protective structure model with four layers and a metal casing lled with TNT charge (MCTC) which was
Available online 14 December 2013
used to simulate the warhead of an anti-ship weapon were designed and manufactured. An experiment
was conducted in which the MCTC exploded inside an empty cabin of the rst layer of the multi-layer
Multi-layer protective structure
protective structure. The distribution of fragments and the equivalent bare charge of the MCTC were
Blast wave determined by a numerical method. From experimental results, the failure pattern of the multi-layer
Fragments protective structure under the synergistic effect of blast wave and fragment impact loadings was pre-
Synergistic effect sented. The synergistic effect for the stiffened plates was also presented in the experiment by comparing
Failure pattern the deformation and the rupture of the air-backed and water-backed stiffened plates. On the other hand,
the agreement between numerical results and experimental results validated the numerical method,
which enabled the numerical model to be used to predict the response of a full scale structure under
loadings of anti-ship weapons. Finally, a discussion of synergistic effects of blast and fragment loadings
on a multi-layer structure was presented and suggestions for the design of a protective structure are put
2013 Elsevier Ltd. All rights reserved.

1. Introduction the serious damage effect caused by the anti-ship missile, multi-
layer protective structures have been applied to combatant ships
The impulsive load inside a naval vessel is mainly due to an [5,6], as shown in Fig. 1(a). The main function of the multi-layer
explosion, which is clearly a major hazard that can result in severe protective structure is to prevent inner cabins from being
structural damage and loss of life. The explosion source is assumed destroyed by weapons. Generally, the multi-layer protective
to be an anti-ship missile, striking the hull just above the water line. structure includes three layers, as shown in Fig. 1(b). When the
Internal blast occurs when the hull is breached before detonation. warhead of a missile pierces the broadside plate and explodes, the
Weapons designed to explode inside the target have armor piercing empty cabins (AeC) of the rst layer supply a large space for the
or semi-armor piercing capability with delayed action detonation propagation of the shock wave. A conned explosion causes more
to maximize the caused damage [1]. Under such circumstances the severe structural damage than a similar external free-air explosion,
simultaneous effect of blast wave and fragments applied to a and this damage depends on geometrical parameters of the space
structure can cause responses more severe than the sum of the where the explosion occurs [7,8]. Since the rst-layer cabins
damage generated in the structure through the independent without any crew and important equipments have enough space in
application of the loads, particularly for the close-range internal the longitudinal direction, venting holes are arranged in the
explosions [2e4]. This simultaneous loading is considered to be transverse bulkhead of the rst-layer cabins to reduce the pressure
synergistic in the sense that the simultaneous damage is greater buildup in a partially conned space and the resulting structural
than the sum of the impact and non-impact loadings. Considering damage [9e12]. The cabins of the second layer are usually lled
with liquid such as water, and denoted by G in Fig. 1(b). These liquid
cabins are designed to stop fragments and to reduce the blast
* Corresponding author. Tel./fax: 86 27 86551193. pressure associated with an explosive event. When the high-speed
E-mail address: kongxs@whut.edu.cn (X.-s. Kong). fragments penetrate a liquid cabin, high pressure shock wave is

0734-743X/$ e see front matter 2013 Elsevier Ltd. All rights reserved.

Downloaded from http://www.elearnica.ir

X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 147

plates with idealized pre-formed holes to investigate the combined

effect of pressure and fragment loading on steel plates. In their
work, a simplication was proposed that fragments struck and
perforated the exible target before the pressure load arrived. Their
results were valid for a combined blast and fragment loading case
in condition that the stand-off distance was large enough for
fragments to reach the plate before the blast wave and that frag-
ments must perforate the structure. However, an explosion of a
cased charge that occurs in a conned or partially conned cabin of
a multi-layer protective structure corresponds to a much more
complex case. The complexities mainly lie in: () the fragments
produced by the metal casing will perforate plates of the cabins, ()
the blast wave will aggravate the damage of plates and () the in-
tensity of blast wave will decrease with the increase of the venting
area due to the damage of plates.
To understand the behavior of structures under the syner-
gistic effects of blast and fragment loadings, full scale blast tests
are required. However, these tests are limited due to security
restrictions and lack of considerable resources required. Dynamic
tests are usually conducted on small scale models to determine
the response characteristics of a geometrically similar full-size
structure. Because tests with cased charges will induce more
uncertainties and complexities, no scaling law adequately suit-
able for the problem is available. Recent advances in FEM make it
possible to investigate the uid structure interaction effects due
to large deformations in the plated structure [30,31]. However,
further efforts are still needed to improve the numerical proce-
dure for general applications of response prediction of structure
under the synergistic effect of blast wave and fragment impact
The motivation of this paper was to investigate the synergistic
Fig. 1. A sketch and arrangement of cabins of a multi-layer protective structure. (a) A effect of blast wave and fragment impact loadings and the subse-
sketch of multi-layer protective structure; (b) Arrangement of cabins. quent damages on the multi-layer protective structure. Another
purpose of the model experiment was to develop an experimental
data set which would validate the numerical method. This would
generated in the liquid and kinetic energy of the fragments transfer enable the validated model to be used with condence to predict
to the surrounding walls of the cabin, which is known as hydro- the response of a full scale structure under the loadings of anti-ship
dynamic ram [13]. The high pressure combined with the fragments weapons. The outline of the paper is as follows:
can result in the deformation and failure of the uid-contained
structure, the velocities of fragments decrease signicantly in the (1) In Section 2, the experimental setup was presented,
process of penetrating the liquid cabin according to the previous including the details of the multi-layer protective structure
researches [14e22]. The third layer is composed of an empty cabin designed and manufactured by the present authors, the
denoted by H in Fig. 1(b). It includes a defensive bulkhead, which is MCTC, the measuring point arrangement and the experi-
usually armored and used to prevent the fragments that have mental method. Besides, the distribution of fragments and
penetrated the liquid cabin. All in all, for a multi-layer protective the equivalent bare charge of the MCTC were obtained by
structure, the waterproof bulkhead must be protected from being using a numerical method.
destroyed by anti-ship weapons. (2) Experimental results and discussions were presented in
The damage effects from the cased charge of anti-ship weapons Section 3. The failure pattern of the multi-later protective
mainly include the blast wave and fragments, and these two factors structure under the synergistic effect of blast wave and
inuence each other [23e26]. The combined effect of blast wave fragment impact loadings, the releasing effect of the venting
and fragment impact loadings on the protective structure is still not hole in the transverse bulkhead, the function of liquid cabins
completely understood. Knowledge of how the blast wave and the of the multi-layer protective structure and the shock re-
fragment impact inuence the structural behaviors is quite limited. sponses of inner cabins were investigated. The synergistic
Nevertheless, the standard design methodology is that these effect of blast wave and fragment impact loadings for a
threats are treated independently of each other [27]. A stiffened stiffened plate was also presented by comparing the defor-
steel plate is a basic structural member in ship and offshore mation and rupture of the air-backed and water-backed
structures. Most previous works on synergistic effect of blast wave plates in the experiment.
and fragment impact loadings on steel plates assumed that the (3) The computational procedure was presented in Section 4. The
fragments have already hit and perforated the target plate, and description of the FE models and the material models utilized
these perforations were idealized as pre-manufactured holes or was listed in detail. The comparisons between experimental
generalized shapes. Schleyer et al. [28] presented a study on square results and numerical simulations were also presented,
plates with a central hole, and the results produced by a simplied including the deformation and rupture of the multi-layer
analytical approach combined with Lagrangian nite element structure, the shock response and strain-time history.
simulation indicated no signicant reduction in the resistance. (4) In Section 5, the synergistic effect of blast and fragment
Rakvg et al. [29] conducted a series of tests on thin ductile steel loadings was discussed. Some conclusions were drawn and
148 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

suggestions for the design of multi-layer protective structure

were put forward.

2. Experimental setup

2.1. The experimental model of the multi-layer protective structure

The experimental model of the multi-layer protective structure

was made of stiffened plates. The manufacture of the experimental
model was completed in Ship Structure Laboratory at Wuhan Uni-
versity of Technology. The design of experimental model referred to
the arrangement of cabins presented in Fig. 1. Considering that no
sophisticated scale law for such problem was available, the experi-
mental model was not used to predict the response of a real multi-
layer protective structure. The present paper aims at experimental
and numerical investigation of a multi-layer structure subjected to a
cased interior explosion. The overriding consideration governing
our design of experimental model was the manufacturability to
satisfy the function of multi-layer protective structure.
The experimental structure was made of mild steel 235A, which
was commonly used in model experiments. Table 1 lists the material
parameters used in this study. A sketch of the experimental model is
shown in Fig. 2. The arrangement of decks and cabins is shown in
Fig. 2(a). The longitudinal plates of protective structure except the
broadside plate Fig. 2(b). The cabins, transverse bulkheads and a
venting hole in the rst layer of protective structure are shown in
Fig. 2(c). The total length of the model is 5 m with three cabins in the
longitudinal direction. The lengths of a middle cabin and two side
cabins are 2 m and 1.5 m, respectively. The experimental model was
also divided into three layers of cabins by four decks in the vertical
direction. The height of the total model is 1.8 m, and the height of
each cabin is 0.6 m. The model was divided into four layers in the
transverse direction. The total width is 2.2 m. In the transverse di-
rection, the experimental model of this multi-layer protective
structure consists of empty cabins (0.6 m width) of the rst layer,
liquid cabins (0.6 m width) of the second layer, empty cabins (0.5 m
width) of the third layer and the personal passageway (empty cabin
with 0.5 m width) of the fourth layer. The transverse bulkheads were
only arranged in the rst and second layers. There was a circular hole
with diameter of 200 mm in the center of the broadside plate to
simulate the perforation caused by an anti-ship missile. As discussed
in the introduction section, when the explosion occurs, the venting
hole in the transverse bulkhead can lead the blast wave to propagate
in the longitudinal direction. Empty cabins of the rst layer supply
large space for the attenuation of the blast wave in the longitudinal
direction. In order to investigate the inuence of the venting hole on
the propagation of the blast wave, a hole with diameter of 200 mm
was set in a transverse bulkhead of Cabin 3 in which the explosion
occurs, as shown in Fig. 2(d).
There were two types of stiffeners welded on the plates of
experimental model, i.e., T type and at type. T type stiffeners were
arranged with equidistant distance of 500 mm in the transverse
direction. The at type stiffeners were arranged equidistantly on
the cabin walls in the longitudinal direction. The detailed thickness

Fig. 2. A sketch of the experimental model (a) Arrangement of decks and cabins; (b)
Longitudinal plates of protective structure except broadside plate; (c) Cabins, trans-
Table.1 verse bulkheads and venting hole in the rst layer of protective structure; (d) The
Material properties and JohnsoneCook material constants used in this study. venting hole in the transverse bulkhead of Cabin 3.
Density r Yield strength Tensile strength A
kg/m3 s0 (MPa) sb (MPa) of plates and stiffeners are listed in Table 2. It should be noted that
7800 249.2(>235) 530 249.2 there were no stiffeners on the longitudinal defensive plate. The
B n C m stitch welding technique was adopted in the manufacture of stiff-
889 0.746 0.058 0.94
ened plates in order to minimize the distortion of the plates due to
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 149

Table 2
Dimension of experimental structure.

The parts of structurea L-D bulkhead Deck 01,B-plate, Deck 02-04, Waterproof Transverse
inner plate outer plate bulkhead bulkhead

Thickness of plate 5 4 3 2 3
Longitudinal stiffenerat type thickness  height e 3  30 3  30 3  30 3  30
Frame or beam T type web thickness  height e 3  80 3  80 3  80
panel thickness  height e 4  30 4  30 4  30
L-D bulkhead, B-plat, inner plate and outer plate are short for broadside plate, inner and outer plate of water cabin and waterproof bulkhead, respectively.

the heat generated in the welding process. Schleyer et al. [32] with a length-to-diameter ratio of 1.46, and the thickness is 6 mm.
conducted a series of three-point bend tests on welded stiffened It should be noted that two sections of the cylindrical casing with
plate specimens to determine the performance of the stitch welds a length of 10 mm adjacent to both ends are thickened to 10 mm.
under large deformations. In their tests, the specimens failed due to These two parts are hereinafter referred to as thickened parts. The
the stiffener buckling rather than the failure of the weld. Therefore, ends near to and far from the initiation point of the charge are
it was reasonable to assume that the stitch welds will perform named as the near end and far end, respectively. The mass of
similarly in the stiffened plate tests. metal casing is 4.10 kg, with 1.9 kg TNT charge in it. The explosive
was initiated at one end of the centerline.
2.2. Characteristics of fragments and equivalent bare charge of In order to determine the characteristics of fragments and
MCTC equivalent bare charge of MCTC, numerical simulations were
performed. The fragmentation process of metal casing in the
The blast wave and the fragment impact loadings are two main numerical simulation is shown Fig. 4 [33]. In the numerical
threats to the multi-layer protective structure, and these two simulation, 2356 fragments were produced, mainly resulting
threats originate from the equivalent bare charge and the frac- from the cylindrical casing and both ends. The maximum frag-
tured metal casing, respectively. These loadings should be deter- ment with mass of 36.1 g, and average speed of 1389.1 m/s, was
mined approximately before the experiment was conducted. A produced by the cylindrical casing. It was found that all frag-
numerical method was setup by the present authors to investigate ments with mass more than 20 g were produced by the two ends
the explosive fragmentation of metal casing, which was described and the part of cylinder approximately 110 mm away from the
in detail and validated in another study of authors [33]. This initiation point (the total length of the cylinder is 160 mm) in the
numerical method was employed here to design the MCTC used in axial direction. The fragments with the maximum kinetic energy
the present experiment. Several different sizes of metal casings were from the far end, with velocity 1.3 times higher than that of
were simulated numerically and nally a MCTC was determined to the fragments produced by the near end. The mass of equivalent
be used in the experiment, as shown in Fig. 3. The metal casing bare charge was 0.905 kg, which was also obtained from the
adopted here was fabricated from mild steel 235A. The inside numerical simulation by comparing the energy change of TNT
diameter of the cylindrical casing is 110 mm, the length is 160 mm charge.

2.3. Arrangement of measuring points and experimental method

2.3.1. Arrangement of measuring points

Sensors at various positions, as indicated in Fig. 5, enable the
responses of the multi-layer protective structure to be measured.
As shown in Fig. 5(a), the dynamic strain gauges were arranged on
the outer and inner plates of liquid cabins and used to measure the
dynamic strain responses of the different plates of liquid cabins
under the blast wave and fragment impact loadings. The strain
gauge 1 was mounted on the non-shock side of the plate of Cabin 3,
in order to minimize the thermal effect caused by the temperature
uctuations within the ow on the strain gauge. The strain gauges
2e5 were mounted on the surface of inner plate of the liquid cabin.
The accelerometer 1 was placed on Deck 02 of the personnel
passageway, and accelerometers 2e4 were placed on the longitu-
dinal defensive bulkhead, as shown in Fig. 5(b).

2.3.2. Experimental method

According to the characteristics of fragment distribution of the
MCTC presented in Section 2.2, the MCTC model was placed
longitudinally in Cabin 3 of the multi-layer protective structure. As
a result, more fragments strike on the second-layer cabins (liquid
cabins) of the experimental model. Besides, a velocity measuring
device of fragments was located near the circular hole in the
broadside plate to obtain the velocity of fragments from the cy-
Fig. 3. The MCTC model. lindrical metal casing, as shown in Fig. 6.
150 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

Fig. 4. The fragmentation process of the metal casing [33]. (a) t 1.39  102 ms; (b) t 2.75  102 ms; (c) t 5.30  102 ms; (d) t 6.53  102 ms; (e) t 9.00  102 ms; (f)
t 1.52  101 ms; (g) t 2.14  101 ms; (h) t 2.27  101 ms; (i) t 4.87  101 ms.

In the experiment, all liquid cabins of the second layer of multi- 3. Experimental results and discussions
layer protection structure model were lled with water, and Cabin
5 of the rst layer of experimental model was also fully lled with 3.1. The perforations caused by fragments and fragment velocity
water while the corresponding Cabin 1 opposite to Cabin 5
remained empty. The water was poured into the second-layer The perforations caused by fragments from the cylindrical metal
cabins and Cabin 5 through a hole located in Deck 01. casing of the MCTC are shown in Fig 11(a). Since the distribution

Fig. 5. Placement of measuring points. (a) Measuring points of dynamic strain on outer and inner plate of liquid cabin; (b) Measuring points of acceleration on deck 02 and
longitudinal defensive bulkhead.
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 151

The velocities of the three parts of the metal casing measured

from the experiment were Vc 1207 m/s (velocity of fragment from
the cylindrical casing), Vn 1210 m/s (velocity of fragment from the
near end), Vf 1613 m/s (velocity of fragment from the far end).
Fragment velocities of the three parts of the metal casing deter-
mined by numerical simulations were 1440.2 m/s,1323 m/s and
1745.1 m/s, respectively. The experimental measured velocities
were slightly smaller than the corresponding numerical results.
However, the ratio of velocities of the far end and near end was Vf/
Vn 1.333, which was close to the numerical result 1.319.
The initial velocity of fragment predicted by Gurney is,

p C=M
V0 2E (1)
1 0:5C=M

Fig. 6. The experimental setup.

where 2E 520 0:28De, De 6860 m/s is the detonation ve-
locity of TNT. C and M are masses of TNT and metal casing,
range of the fragments from the cylindrical metal casing was respectively.
limited in scope, as presented in Fig. 7. The perforations in one wall The initial velocity predicted by Eq. (1) is 1383.8 m/s. In the
of Cabin 3 are concentrated in a narrow band, marked by a red numerical simulation, the range of expansion velocities of the cy-
rectangle in Fig. 7(a). The perforations caused by the near end and lindrical metal casing was from 1187.3 m/s to 1440.2 m/s. However,
far end of the MCTC model are shown in Fig. 7(b) and (c), respec- the fragment with maximum velocity was from the far end, with
tively. The plate in Fig. 7(b) refers to transverse bulkhead 1 (TB1), expansion velocity of 1745.1 m/s. This special feature of fragment
and the perforated plate in Fig. 7(c) represents transverse bulkhead velocity distribution was also observed in the experimental test.
4 (TB4). By checking the perforations on the other plates of the However, it cannot be predicted by Gurney formula.
structure, which were out of the penetration range of the fragments
from the MCTC, it can be found that there were rare perforations.
Besides, the perforations caused by the fragments from the ends 3.2. The destruction of the multi-layer protective structure
were almost concentrated on a square plate with the side length of
600 mm, although the travel distance of these fragments in the air When the MCTC explodes inside Cabin 3, the blast wave and
nearly reached 2.5 m. high-speed fragments caused a serious damage to the structure, as

Fig. 7. Perforations produced by fragments. (a) Perforations produced by fragments from cylindrical metal casing; (b) Perforations caused by fragments of near end; (c) Perforations
caused by fragments of far end. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)
152 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

seam on the broadside was longitudinal, so the failure of the

broadside plate was not due to the weakness of weld seam but as a
result of blast wave and fragment impact loadings. As stated above,
the conned explosion can cause more severe damage than a
similar external free-air explosion [7,8]. Besides, the duration of the
blast pressure from a bare charge detonated within a vented or
unvented structure is much longer than that of a similar external
free-air explosion [34]. The maximum velocity of fragments from
cylindrical metal casing was 1389.1 m/s, which was obtained
numerically in Section 2.2. The fastest fragment reached the closer
plate of Cabin 3 in 0.2 m s. Thus, the loadings of blast wave and
fragment impact exerted on plates of Cabin 3 simultaneously. In
condition that the MCTC exploded inside Cabin 3, the blast wave
rstly releases from the venting holes in the broadside plate and
transverse bulkhead 2 (TB2), and the plates of Cabin 3 were sub-
jected to the blast pressure. The fragments from the metal casing
played a crucial role in the damage of stiffened plates of the

Fig. 8. Destroy and deformation of the broadside plate. (a) Destroy and deformation of
the broadside plate; (b) Arrangement of deformation measuring points; (c) Detailed
deformation obtained by measuring data.

shown in Fig. 8(a). The broadside plate, including the stiffeners on

it, was ripped from the top to the bottom. The ripped line was
approximately located at the middle of the broadside plate, and it
was also located in the impact band of fragments from the cylin-
drical metal casing. It was noteworthy that the direction of weld

Fig. 10. Deformation and rupture of transverse bulkheads. (a) Destroy of TB 2; (b)
Destroy of TB 3; (c) Deformation and rupture of TB 1; (d) Deformation and perforation
Fig. 9. Destroy of deck 01 and deck 02. of TB 4.
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 153

experimental structure. Stiffened plates were perforated by frag- completely torn apart. The TB 1, which was adjacent to TB 2 shoot
ments in the impact zone of the fragments, as shown in Fig. 7(a). In out from the structure, and the escaping distance was more than
the present experiment, these perforations caused by fragments 50 m. The main reason was that the high pressure blast wave blew
acted as the crack initiation locations, and through-section crack through the venting hole in TB 2 and directly imposed on the
appeared immediately. Under the pressure from the blast wave, adjacent cabin, resulting in the rupture of the weld seams of the TB
plates were torn and large deformation emerged along the 1. As a result, the TB 1 was seriously distorted, and there were
through-section crack. several punch holes, as shown in Fig. 10(c).
The detailed deformation of the broadside plate was obtained by The destruction of TB 3 without a venting hole is shown in
arranging the marks, measuring the spatial coordinates of the Fig. 10(b), and TB 3 was severely distorted. The weld seams between
marks and reproducing the measuring data of the plate, as shown TB 3 and the adjacent structures were almost torn. The destruction
in Fig. 8(b) and (c). The broadside plate was torn into two parts. The of TB 4 adjacent to TB 3 is shown in Fig. 10(d). The weld seams
white dotted lines corresponded to the locations of transverse between the transverse bulkhead and the adjacent structures
bulkheads 2 and 3. It was observed that the deformation of the left remained intact, for the intensity of blast wave acting on the TB 4
part was larger than that of the right part of the broadside plate, was not high enough to tear the weld seams. A large amount of
which was mainly due to that the area bearing the blast wave load punch holes appeared in TB 4, which was mainly caused by high-
of the left part was larger than that of the right part. speed fragments from the far end of the MCTC.
The damages and deformations of Deck 01 and 02 of the rst By comparing the destructions of the TB 1 and TB 4, it can be
layer are shown in Fig. 9. Many perforations were observed in the found that the venting hole in the transverse bulkhead played an
Deck 01 of Cabin 1, and these perforations were also concentrated obvious guiding role in the propagation of the blast wave. The
in the impact band of fragments. The Deck 01 of Cabin 1 projected shock wave can spread to the adjacent cabin through the hole,
upward, but it was not torn apart. The reasons included that (1) the which reduced the damage effect on the internal structure. In the
distribution density of the perforations on Deck 01 of Cabin 1, design of a multi-layer protective structure, venting holes should be
which was far from the MCTC, is smaller, (2) the intensity of the arranged appropriately in the transverse bulkhead.
blast wave was not high enough due to the releasing of blast wave
from the broken broadside plate. The damage of Deck 02 of Cabin 3, 3.4. The function of liquid cabins in the multi-layer protective
which was closer to the MCTC was more severe. It seemed that the structure
damage characteristic of the Deck 02 of Cabin 3 was similar to that
of the broadside plate. Under the combined blast and fragment The liquid cabins were arranged in the second layer of the multi-
loadings, the Deck 02 of Cabin 3 was torn apart totally at the impact layer protective structure to prevent high-speed fragments from
band of fragments and upturned. causing serious damage to inner structures. In the experiment, the
cabins of the second layer and Cabin 5 were fully lled with water.
3.3. The releasing effect of the venting hole on the transverse The rupture of the wall of liquid cabin is shown in Fig. 11(a). Frag-
bulkhead ments damaged the wall of liquid cabin close to the MCTC. A large
number of perforations appeared. These perforations were almost
A venting hole with diameter of 200 mm was arranged in the TB connected to each other and formed a large rupture zone. However,
2 of Cabin 3 in order to estimate the releasing effect of the blast this wall of liquid cabin did not deform inward under the blast wave
wave and the subsequent damage to the structure. As presented in load. The deformation of the undamaged zone of this wall was
the Introduction section, the blast wave can release from the outward and not large. Photos of the inner plates of the liquid cabin
venting hole and reduce the pressure buildup in a partially conned are shown in Fig. 11(b). No fracture or punch holes appeared at the
space. inner plate of liquid cabins.
When the MCTC model exploded inside Cabin 3, blast wave and The combined effect of the blast and fragment loadings
high-speed fragments, as shown in Fig. 10, also destroyed trans- exerted on liquid cabins can be analyzed in two steps. (1) When
verse bulkheads. The destruction of TB 2 with a venting hole is the high-speed fragments pierced into liquid cabin, the semi-
shown in Fig. 10(a). The TB 2 was distorted seriously. The weld spherical shock wave produced in the liquid. The pressure of
seams between the transverse bulkhead and the adjacent struc- the shock wave results in the outward pressure exerted on the
tures were almost torn, and only the connection to Deck 03 was not outer plate [35,36], and (2) the blast wave of the partially

Fig. 11. The damage of liquid cabins. (a) Perforations in the outer plate of liquid cabins; (b) The inner plate of the liquid cabin.
154 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

conned explosion from the charge in the MCTC exerted on the in the liquid, and the nal deformation of the cabin walls. The strain-
outer plate of the liquid cabins. Thus, the pressures from the blast time responses of the gauges on the outer and inner plates of liquid
wave and the shock wave produced by fragments moving in the cabins are shown in Fig. 12. The strain gauge 1 was located near the
liquid exerted on the outer plate simultaneously, but in the perforated zone of the outer plate, at which relatively large residual
opposite direction. As a result, the deformation of the walls of strain produced. The strain responses of gauges 2e5 are shown in
Cabin 5 and the walls of other liquid cabins close to the MCTC Fig. 12(b)e(e). Residual deformation was found at the locations of
will be smaller when compared to the Deck 02 of Cabin 3, which the inner plate where the gauges were arranged. This was mainly
was an air-backed plate. due to the pressure produced by the fragments moving in the liquid
The process from the moment of fragment impact to the nal cabin. It was noted that the time interval between the initial strain
response of the liquid cabin was generally divided into three stages, response and the peak strain increased with the increase of the
namely, the initial perforation, the subsequent shock wave produced distance between the gauges location and the fragment impact zone.

Fig. 12. Dynamic strain-time history curve of strain gauges on the outer and inner plate of liquid cabin. (a) Strain gauge 1; (b) Strain gauge 2; (c) Strain gauge 3; (d) Strain gauge 4;
(e) Strain gauge 5.
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 155

Fig. 13. Comparison of deformations of air-backed and water-backed stiffened plates. (a) air-backed stiffened plate; (b) water-backed stiffened plate.

The function of the liquid cabin was to convert the concentrated Stepka et al. [37] analyzed the velocity decay of the projectiles in
impact load of fragments to a distributed load as a form of shock the liquid with Newtons second law in the form,
pressure in the liquid and transfer the kinetic energy of fragments to
the walls of liquid cabin eventually. dvp 1
When high-speed fragments penetrate a liquid tank, the kinetic mp  Cx trl Ap v2p (2)
dt 2
energy of the fragments transferred to the surrounding walls of the
tank, and high pressure shock wave formed in the liquid, which was where the liquid is assumed to be incompressible, and cx is a con-
known as hydraulic ram [13]. In the studies of the hydraulic ram, stant, cx 0.491 [22]. mp, Ap, and vp are the mass, cross-sectional
researchers were particularly interested in the decay of the pro- area and velocity of the projectile, respectively. rl is the density of
jectile velocity, the cavity formation and growth in the liquid and the liquid. The projectile is assumed to be a spherical body. The
the damage mechanism of the liquid-lled tank. solution to Eq. (2) is

Fig. 14. Shock response curves of accelerometers. (a) A1; (b) A2; (c) A3; (d) A4.
156 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

vp x vp0 eCx rl Ap x=2mp (3)

The maximum mass of fragments obtained from numerical

simulation in Section 2.2 was 36.1 g, with average speed of
1389.1 m/s. When the fragment traveled through the liquid cabin
with width of 0.6 m, the nal velocity of the fragment can be
derived from Eq. (3), and vpf 355.1 m/s. The velocity of the frag-
ment dropped dramatically when the fragment moved in the
liquid. On the other hand, the kinetic energy of fragments, which
transfers to the surrounding walls, caused the deformation and
rupture of plates of the liquid cabin.

3.5. The synergistic effect of blast and fragment loadings

In order to investigate the synergistic effect of blast wave and Fig. 15. FE model of multi-layer structure.
fragment impact loadings, Cabin 5 of the rst layer of experimental
model was fully lled with water while the corresponding Cabin 1
were 9949.8 g, 3527.8 g, and 2212.2 g, respectively. It was found
remained empty. The relative positions between Cabin 1 and Cabin
that two sharp peaks appeared in the acceleration-time curve of
5 with the MCTC model were equal, so these two cabins bore equal
point A1 and one sharp peak in that of point A2 which lied on the
blast wave and fragment impact loadings from the explosion of the
longitudinal defensive bulkhead between Deck 01 and Deck 02.
MCTC. In the experiment, it was observed that the outer plates
The sharp peaks of A1 and A2 may result from the sudden
(close to the MCTC) of the cabins fully lled with water can bear
rupture of the decks in the rst layer of the experimental model.
blast wave and fragments impact without rupture and large
The red dotted lines in Fig. 20 marked the range of acceleration
deformation, but with perforations produced by fragments. The
value from 2000 g to 2000 g. The acceleration value of point A3
damage pattern of this plate was similar to that just caused by
was larger than those of other points at the beginning stage.
fragments. The plates of the Cabin 1 and the Cabin 5 close to the
Considering the locations and directions of the accelerometers,
MCTC model were represented by water-backed and air-backed
the decks of the experimental model turned out to be the main
stiffened plate, respectively. It was considered that the fragment
shock e transmission pathways.
impact loading caused the response of water-backed stiffened plate
and that the damage of air-backed stiffened plate was the result of
blast and fragment loadings. The synergistic effect of blast and
4. Numerical simulations and results
fragment loadings for the stiffened plate can be obtained by
comparing the deformation and rupture of the air-backed and
In the design of the multi-layer protective structure using non-
water-backed plates, which is shown clearly in Fig. 11(a). The air-
linear FE simulations, it is essential that numerical models have
backed and water-backed stiffened plates belong to thin wall
some special features, of which the material model should be
structure. They were usually designed to bear the in-plane force,
simplied, the computational time should be minimized and the
and they can be easily perforated by high-speed fragments.
numerical results are conservative in order to achieve a safe design
Although the impact area of fragments on stiffened plate was not
[39]. Numerical simulations in this study were carried out after the
large, the section of the stiffened plate can be weakened dramati-
experimental test was nished. In present study, numerical simula-
cally by a group of fragments. Besides, the perforations acted as the
tions were performed with the commercial code ANSYS AUTODYN
crack initiation locations due to the stress concentrations when
14.0. This software is particularly suitable for nonlinear dynamic
loaded by blast wave [38]. The air-backed stiffened plate was totally
problems, such as impact or explosions. It also allows the employ-
ruptured and upturned, as shown in Fig. 13(a). The perforations on
ment of different techniques such as ALE or SPH to solve uid-
it were connected to each other by cracks due to the following blast
structure and impact-penetration problems. In order to reproduce
wave, and the plate was torn and upturned quickly. However, only
the uid-structure (blast wave and structure) interaction and
local perforation appeared on the water-backed plate, as shown in
Fig. 13(b). From the comparison of the deformation and rupture of
these two plates, it can be seen that the synergistic effect of blast
and fragment loadings on stiffened plate were mainly reected by
the connection of perforations and the large deformation from the
weakened section. Therefore, it can be inferred that the spatial
distribution of fragments relate directly to the damage of the
structure when loaded by near led explosion of the MCTC.

3.6. Dynamic responses of inner cabins

The shock responses of the typical measurement points on the

Deck 02 (A1) and longitudinal defensive bulkhead (A2, A3 and A4)
are shown in Fig. 14.
At the moment of the MCTC explosion, strong shock occurred
in the personnel passageway and resulted in large shock accel-
eration value. The maximum acceleration value of point A1 on
the Deck 02 was 10,095.6 g, and the maximum accelerations of
points A2, A3 and A4 on the longitudinal defensive bulkhead Fig. 16. Sketch of initial position between multi-layer structure and Euler region.
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 157

Fig. 17. Propagation of blast wave and rapture of structure. (a) t 0.1 ms; (b) t 0.8 ms; (c) t 1.1 ms; (d) t 1.3 ms; (e) t 2.2 ms; (f) t 3.1 ms; (g) t 3.7 ms; (h) t 4.5 ms; (i)
t 5.4 ms; (j) t 6.3 ms; (k) 6.6 ms; (l) 7.2 ms.

fragments impact-penetration, the coupled EulerianeLagrangian AUTODYN is capable of predicting an unsteady, dynamic motion of a
and SPH-Lagrangian approaches were both adopted to model the material system by solving the appropriate mass, momentum and
experimental test in Section 3. The multi-solver coupling method was energy conservation equations, subjected to the associated initial and
used in numerical simulations. The principle of this method was boundary conditions [41]. Thus, the FE model of structure and uid
presented in detail by Yun et al. [40]. Hydrodynamic code such as should be correctly provided. Besides, parameters of equation of state
158 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

It is well-suited for numerical simulations, and only ve parameters

have to be determined by means of material tests. It has been
implemented in AUTODYN codes. The relation is dened through
the following multiplicative equation,

s A Bnp 1 C ln_* 1  T *m (4)

where A,B,C,n and m are JohnsoneCook material constants. p is the

effective plastic strain, _ * _ p =_0 the effective plastic strain rate at
a reference strain rate _ 0 1s1 and the homologous temperature
T* (TTr)/(TmTr) where T is the material temperature, Tr is the
room temperature, and Tm is the melting temperature of material.
The values for JohnsoneCook constitutive relation were ob-
tained from material tests conducted by authors, as shown in
Table 1. The ow stress parameters A, B and n are obtained from
quasi-static tension test on smooth specimens at a strain rate of
order 2  104 s1. The value of C was from the split Hopkinson
pressure bar tests at room temperature of 20  C with strain rates in
the range from 900 to 5500 s1. The value of parameter m is from
the Split Hopkinson Pressure Bar tests at three different tempera-
tures 200  C, 400  C and 600  C, with strain rates about 3500 s1.
The failure mode should be dened in the AUTODYN to provide a
suitable failure criterion of material. The rapid increase in compu-
tational power allows analysis of large and complex problems.
However, problems arise when fracture is included. Presently, there
is no practical formulation capable of simulating both fracture
initiation and propagation in large-scale structures using shell ele-
ments. One reason for this is that fracture is usually modeled by
removing over-strained elements. This is an engineering approach
which makes FE solutions very mesh sensitive. Is also makes
modeling of fracture propagation more difcult in the sense that
Fig. 18. Damage of decks. (a) Damage and perforations of Deck 02-04; (b) Damage and
crack tip effects are concentrated over very few elements [44,45].
perforations of Deck 01.
Alsos et al. [44] investigated the performance of two failure criteria,
which are referred to as the BWH instability criterion and the RTCL
(EOS), constitutive relations and failure criterion of materials should damage, and studied the inuence of the element size with respect
be dened properly. The following is the description of FE models and to onset of failure. They found that the element size effect varied
the material model utilized in the present study.

4.1. FE Model of multi-layer structure

Since the thicknesses of the plates are quite small comparing to

the geometry size of the multi-layer structure, the plates and
stiffeners of the whole structure were discretized by means of the
bilinear four-node quadrilateral shell elements with one quadra-
ture point and an hourglass control. According to the research done
by Guo et al. [42], when the perforation of a projectile through a
target is dominated by certain failure modes, shell elements are
able to model the material failure of the target with good accuracy.
In the present study, in order to consider the synergistic effect of
blast and fragment loadings, different sizes of shell elements were
adopted. The elements with size of 6.25 mm were used in the
impact-penetration zone of fragments. Shell elements with sizes of
12.5 mm, 25 mm and 50 mm were used in uid-structure inter-
action area adjacent to impact-penetration zone, transitional re-
gion of different sizes of elements and regions far away from the
explosion, respectively. Finally, the mesh of the structure consisted
of 212,997 elements. The FE model of multi-layer structure is
shown in Fig. 15.
The JohnsoneCook constitutive relation [43] was selected to
model the material behavior of the structure. There are other more
sophisticated hardening relations, but JohnsoneCooks is probably
the most widely used among those accounting for equivalent Fig. 19. Damage of broadside plate. (a) Damage result from numerical simulation; (b)
plastic strain, equivalent plastic strain rate and temperature effects. Damage results from experiment.
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 159

Fig. 20. Dynamic strain-time history curves of strain gauges in numerical simulations. (a) Strain gauge 1; (b) Strain gauge 2; (c) Strain gauge 3; (d) Strain gauge 4; (e) Strain gauge 5.
(For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

with structural geometry and load scenario. In this paper, from the dx  dy  t dx1 x  dy 1 y  tr (5)
experimental results presented in Section 3, the failure pattern of
the experimental structure was related to the synergistic effects of tr t1 t (6)
blast and fragment loadings. The perforations on stiffened plates
produced by fragments acted as local fracture. The tests of Alsos et al. Combining Eqs. (5) and (6),
[44] show that when the plate fractures, the crack path is restricted  
by the area conned by the stiffeners. So in the numerical simulation 1 x 1 y 1 t 1 (7)
of present study, for simplicity, the principal strain failure model is
It is assumed that the strains of plate near the crack are equal in
used to simulate the crack propagation on plates. The failure strain
x and y direction, i.e. x y. Thus,
was determined by measuring the residual thicknesses of plate
along the crack produced on broadside plate of experimental model. t tr
It is assumed that the size of a small element near the crack is tdxdy, q q (8)
of which t is the initial thickness of plate. The size of deformed small x y 1 1 t 1
1t tr
element is tr(1x)dx(1y)dy, of which tr is the residual thickness p
and x,y,t are strains in x,y,t direction. Based on the volume So the failure strain can be dened as f t=tr  1, and an
invariance hypothesis, equations are presented as follows, average value of 0.08 is selected as failure strain of ruptured plates.
160 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

The failure strain value of 0.08 seems too low for the mild steel Table 3
material. However, by measuring the thickness of the crack edge, JWL EOS parameters of TNT charge.

no obvious thinning or local necking was found. The authors Density r Detonation velocity CeJ pressure C1(Pa)
inferred that the perforations with irregular shape on plates had (Kg/m3) D (m/s) (Pa)
relatively high stress concentration factor. The plate would fracture 1630 6800 2.10 E10 3.74 E11
and the cracks propagate at low strain. C2 (Pa) r1 r2 u
3.75 E9 4.15 0.9 0.35

4.2. Model for the uid

The air and water were rstly modeled using the multi-material 4.3.1. Propagation of blast wave and rupture of structure
Euler formulation in AUTODYN-3D. The Euler-Shell and SPH-Shell Blast wave and fragments loadings almost simultaneously
interaction techniques were used to investigate the response of exerted on cabin walls shortly after the MCTC model exploded in
multi-layer structure under blast and fragment loadings. Different Cabin 3. The multi-layer structure was seriously damaged. The
types of processors were used in different regions providing an propagation process of blast wave and rapture of structure are
optimal solution technique for coupled problems. The air and shown in Fig. 17. For the sake of clarity, the broadside plate was
water were modeled in same Euler region by means of ll func- concealed in Fig. 17, and the propagation of blast wave was limit-
tion in AUTODYN. According to the experimental results, a proper edly shown in the middle section in longitudinal direction of the
Euler region of air was set in numerical simulation, of which the multi-layer structure. It can be seen that the blast wave was rstly
size is 6000 mm, 2400 mm, 2400 mm in length, width and height conned in Cabin 3, and exerted on stiffened plate of Deck 02 and
direction, respectively. This mesh size was considered to be Deck 03 simultaneously, as shown in Fig 17(a). When the blast wave
appropriate to couple with the shell elements of multi-layer reached the transverse bulkheads, TB 2 and TB 3 marked in
structure. Finally, the size of Euler elements is selected as experimental model of Cabin 3, it propagated through the venting
30 mm  30 mm  30 mm, resulting 1,280,000 elements in Euler hole in TB2. Meanwhile, TB 3, as shown in Fig. 17(b), reected the
region. It should be noted that Euler region was large enough to blast wave. According to fragment velocities of metal casing from
cover the deformed structure. The sketch of initial position of numerical simulation in Section 2.2, fragments from metal casing
multi-layer structure and Euler region is shown in Fig. 16. Besides, struck the plates of decks, TB 3 and TB 2 one after another, after the
the ow-out boundary is applied to surfaces of Euler region. The MCTC was initiated. These fragments caused perforations and
simulation was carried out in a workstation with eight dual core deformation of corresponding zone of the plates. Furthermore,
processors, and the parallel processing environment was provided catastrophic failure of plates was observed as time went on. The
to improve the calculation efciency. blast wave, which was rstly conned in Cabin 3 herein propagated
The air was modeled with an ideal gas form of equation of state, through the damage plates and exerted on walls of adjacent cabins,
dened as as shown in Fig. 17(c)e(e). It was clearly shown that damages of
plates were closely related to the distribution of fragments, and
P g  1re P0 (9) that the boundary of the damaged structure had inuence on the
propagation path of blast wave, as shown in Fig. 17(f)e(l).
where g is dened as the ratio between the specic heat at constant The longer duration of blast wave in conned space would
pressure and volume, respectively, and g1.4. P0 is the initial aggravate structure failure. Plate of Deck 02 was upturned seriously.
pressure, r is the density while e is the internal energy per unit The damage situation of Deck 02 was similar to that of the exper-
reference volume. This form of the equation has much to commend imental test, as shown in Fig. 18(a). There were perforations and
for its simplicity and ease of computation. The standard constants deformation occurred in the area of Deck 01 corresponding to Cabin
of air from the AUTODYN material library were used [41]. 1, which was the result of the combined effect of fragments and
A standard JWL (JoneseWilkinseLee) equation of state was blast wave, as shown in Fig. 18(b).
employed to describe the adiabatic expansion of the detonation The damage situations of broadside plate from experiments and
products. The pressure is expressed as a function of the volume and numerical simulations were compared with observation in Fig. 19.
energy, The broadside plate was torn from top to bottom, and the observed
    appearances were consistent with each other. The deformation of
u r1 v u r2 v ue the area corresponding to Cabin 1 was larger than that of Cabin 5.
pT C1 1  e C2 1  e (10)
r1 v r2 v v As discussed in experimental result, it was mainly caused by high-
pressure load in water cabin. Beyond that, the deformation of
where C1,C2,r1,r2 and u are constants. pT,v and e are the pressure, broadside plate in experimental test was larger than that of nu-
relative volume and specic internal energy, respectively. merical result. It can be inferred that defect of weld seam between
The material properties of TNT charge and parameters used for decks and broadside plate was the main cause of the disparity,
the JWL equation are listed in Table 3. which was not considered adequately in numerical simulations.

4.3. Numerical results 4.3.2. Strain and shock response of structure

Dynamic strain-time history curves of strain gauges in numer-
The hydrodynamic code simulations presented in this paper ical simulations are shown in Fig. 20. It indicated that plastic
provided an insight to the response of multi-layer protective model deformation occurred on the walls of liquid cabins. Among these
under the synergistic effects of blast and fragments loadings. The measuring points, the maximum strain and maximum residual
hydrodynamic code simulations were able to capture the propa- strain occurred on the outer plate of liquid cabin. With the increase
gation of blast wave with change of deformation and rapture of the of distance from the positions of strain gauges to the explosion site,
structure, the pressure contours in different cabins, the strain-time the maximum strain and residual strain decreased gradually. By
and acceleration-time history of certain positions of structure. Be- comparing the strain responses from experiment and numerical
sides, the numerical method employed in present study would be simulation, it was found that the values of strain gauge 1 and 2
validated by comparing numerical results and experimental data. were higher in numerical simulation than that obtained by
X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162 161

Fig. 21. Shock response curves of certain points in numerical simulations. (a) A1; (b) A2; (c) A3; (d) A4.

experiment, while the results of other strain gauges was inverse. stiffened plate act as the crack initiation locations due to the stress
That is to say, the deformation of walls of liquid cabins was more concentrations and connect to each other when loaded by blast
localized in numerical simulation. However, the same trends were wave and the stiffened plate is torn and upturned eventually. Soon
observed in both experimental test and numerical simulation. afterwards, blast wave in conned space would aggravate the
Numerical simulation results of shock response curves of points deformation of the weakened stiffened plates and large vent ap-
corresponding to the measuring point arrangement for experiment pears. Blast wave propagates through the vent and exerts on
are shown in Fig. 21. The shock response of every point was similar adjacent cabins.
to that obtained from experimental test. The variation range of Besides, based on experimental and numerical investigation,
acceleration in numerical simulation was larger than that obtained suggestions for the design of multi-layer protective structure are
from experimental test. The cause of this phenomenon was that the put forward as follows.
damping ratio of the structure was not employed in numerical The hole on the transverse bulkhead can effectively guide the
simulation. blast wave to spread along the longitudinal direction. In the design
of a multi-layer protective structure, the venting holes should be
arranged in the transverse bulkhead appropriately.
5. Discussions and conclusions The liquid cabin plays an important role in the enhancement of
the anti-explosion capacity of the multi-layer protective structure.
Experimental investigation and numerical simulations of a cased The liquid in the cabin can effectively absorb the energy of high-
charge exploded inside a multi-layer protective structure were speed fragments and prevent them from punching into internal
carried out. Special interest is given to study the synergistic effect of cabins. The function of the liquid cabin is to change the concen-
blast and fragment loadings. It has been demonstrated that the trated impact load of fragments to the distributed load as a form of
numerical method presented in this paper is capable of simulating shock pressure in the liquid and transfer the kinetic energy of
the experimental test. The experimental result shows an acceptable fragments to the walls of liquid cabin eventually.
agreement with the computation result. So far, knowledge of how It was found that the decks are the main paths of shock prop-
the blast wave and the fragment impact inuence the structural agation. If it is possible, connections between layers of protective
behaviors is quite limited. The numerical method used in present structure should be weakened.
study would be helpful to researches on synergistic effect of blast
and fragment loadings and the design of protective structure.
In the present study, the experimental structure is composed of Acknowledgments
stiffened plates, which are consisted of relative thin plates and
stiffeners on them. When subjected to the loading of blast wave and This work was done with the nancial support of the Defense
fragments, stiffened plates can be easily penetrated by fragments. It Industrial Technology Development Program reference
is different from the condition of thick structure under impact by A1420080184, and of the Fundamental Research Funds for the
fragments, such as concrete walls, in which fragments cause only Central Universities reference 2011YB08. The authors wish to ex-
localized damage. The section of the stiffened plate can be damaged press gratitude to Prof. Zhu xi and Dr. Hou Hai-liang for their
dramatically by a group of fragments. The perforations on the valuable help during experimental test.
162 X. Kong et al. / International Journal of Impact Engineering 65 (2014) 146e162

References [24] Zhang Q, Miao C-Q, Lin D-C, Bai C-H. Relation of fragment with air shock wave
intensity for explosion in a shell. Int J Impact Eng 2003;28(10):1129e41.
[25] Jim D, Dennis F, James W. Blast algorithm development: denition of modied
[1] Kurki T. Contained explosion inside a naval vessel e evaluation of the struc-
blast algorithms for PBX based explosives. Bristol, UK: IMEMTS; 2006. p. 1e
ture response. Master Thesis. Finland: Helsinki University of Technology;
10. http://www.imemg.org/res/IMEMTS%202006_Dunnet_paper_post.pdf.
2007. p. 4e17.
[26] Crowley AB. The effect of munition casings on reducing blast overpressures.
[2] Joosef L. Experiments and numerical analyses of blast and fragment impacts
Bristol, UK: IMEMTS; 2006. p. 1e10. http://www.imemg.org/res/IMEMTS%
on concrete. Int J Impact Eng 2005;31(7):843e60.
[3] Marchand KA, Vargas MM, Nixon JD. The synergistic effects of combined blast
[27] U.S. Army Engineers Waterways Experiment Station. TM 5-855-1 e funda-
and fragment loadings. Florida: Southwest Research Institute Final Report No.
mentals of protective design for conventional weapons. Washington, D.C: U.S.
ESL-TR-91-18 to Air Force Engineering & Services Center, National Technical
Department of the Army; 1991. p. 6e11.
Information Service; 1992. p. 1e102.
[28] Schleyer GK, Underwood NJ, Do HM, Paik JK, Kim BJ. On the simplied analysis
[4] Nystrm U, Gylltoft K. Numerical studies of the combined effects of blast and
of square plates with holes. ASME Conference Proceedings 2011;2011(44342):
fragment loading. Int J Impact Eng 2009;36(8):995e1005.
[5] Friedman Norman, Baker AD. Aircraft carriers: an illustrated design history.
[29] Rakvg KG, Underwood NJ, Schleyer GK, Brvik T, Hopperstad OS. Transient
Annapolis: Naval Institute Press; 1983. p. 255e86.
pressure loading of clamped metallic plates with pre-formed holes. Int J
[6] Saunders CS. Janes ghting ship, 2004-2005. Janes Information Group
Impact Eng 2013;53:44e55.
Limited, Sentinel House, Surry; 2004. p. 828e55.
[30] Brvik T, Hanssen AG, Langseth M, Olovsson L. Response of structures to
[7] Hu Yi, Wu Chengqing, Lukaszewicz Matthew, Jonathan Jr Dragos, Haskett M.
planar blast loads e a nite element engineering approach. Comput Struct
Characteristics of conned blast loading in unvented structures. Int J Protect
Struct 2011;2(1):21e44.
[31] Brvik T, Olovsson L, Hanssen AG, Dharmasena KP, Hansson H, Wadley HNG.
[8] Edri I, Savir Z, Feldgun VR, Karinski YS, Yankelevsky DZ. On blast pressure
A discrete particle approach to simulate the combined effect of blast and sand
analysis due to a partially conned explosion: I experimental studies. Int J
impact loading of steel plates. J Mech Phys Solids 2011;59(5):940e58.
Protect Struct 2011;2(1):1e20.
[32] Schleyer GK, Hsu SS, White MD, Birch RS. Pulse pressure loading of clamped
[9] Ferrara G, Di Benedetto A, Salzano E, Russo G. CFD analysis of gas explosions
mild steel plates. Int J Impact Eng 2003;28(2):223e47.
vented through relief pipes. J Hazard Mater 2006;137(2):654e65.
[33] Kong X, Wu W, Li J, Liu F, Chen P, Li Y. A numerical investigation on explosive
[10] Razus DM, Krause U. Comparison of empirical and semi-empirical calculation
fragmentation of metal casing using smoothed particle hydrodynamic
methods for venting of gas explosions. Fire Saf J 2001;36(1):1e23.
method. Mater Des 2013;51:729e41.
[11] Feldgun VR, Karinski YS, Yankelevsky DZ. A simplied model with lumped pa-
[34] Anderson J, Charles E, Baker WE, Wauters Donna K, Morris Bruce L. Quasi-
rameters for explosion venting simulation. Int J Impact Eng 2011;38(12):964e75.
static pressure, duration, and impulse for explosions (e.g. HE) in structures.
[12] Janovsky B, Selesovsky P, Horkel J, Vejsa L. Vented conned explosions in
IJMS 1983;25(6):455e64.
stramberk experimental mine and AutoReaGas simulation. J Loss Prev Process
[35] Townsend D, Park N, Devall PM. Failure of uid lled structures due to high
Indust 2006;19(2e3):280e7.
velocity fragment impact. Int J Impact Eng 2003;29(1e10):723e33.
[13] Varas D, Lpez-Puente J, Zaera R. Experimental analysis of uid-lled
[36] Chou PC, Schaller R, Hoburg J. Analytical study of the fracture of liquid-lled
aluminium tubes subjected to high-velocity impact. Int J Impact Eng
tanks impacted by hypervelocity particles. DIT Report 160-9, NASA CR-72169.
Washington, D.C: NASA, Ofce of Scientic and Technical Information; 1967.
[14] Lecysyn N, Dandrieux A, Heymes F, Aprin L, Slangen P, Munier L, et al. Ballistic
p. 1e23.
impact on an industrial tank: study and modeling of consequences. J Hazard
[37] Stepka FS, Morse CR, Dengler RP. Investigation of characteristics of pressure
Mater 2009;172(2e3):587e94.
waves generated in water lled tanks impacted by high-velocity projectiles.
[15] Lee M, Longoria RG, Wilson DE. Ballistic waves in high-speed water entry. JFS
TN D-3143. Washington, D.C: NASA, Scientic and Technical Information Di-
vision; 1965. p. 1e24.
[16] Varas D, Zaera R, Lpez-Puente J. Experimental study of CFRP uid-lled tubes
[38] Veldman RL, Ari-Gur J, Clum C. Response of pre-pressurized reinforced plates
subjected to high-velocity impact. Compos Struct 2011;93(10):2598e609.
under blast loading. Int J Impact Eng 2008;35(4):240e50.
[17] Lecysyn N, Bony-Dandrieux A, Aprin L, Heymes F, Slangen P, Dusserre G, et al.
[39] Brvik T, Dey S, Clausen AH. Perforation resistance of ve different high-
Experimental study of hydraulic ram effects on a liquid storage tank: analysis
strength steel plates subjected to small-arms projectiles. Int J Impact Eng
of overpressure and cavitation induced by a high-speed projectile. J Hazard
Mater 2010;178(1e3):635e43.
[40] Yun S-H, Jeon H-K, Park T. Parallel blast simulation of nonlinear dynamics for
[18] Nishida M, Tanaka K. Experimental study of perforation and cracking of
concrete retrotted with steel plate using multi-solver coupling. Int J Impact
water-lled aluminum tubes impacted by steel spheres. Int J Impact Eng
Eng 2013;60:10e23.
[41] ANSYS AUTODYN 14.0. Theory manual. ANSYS, Inc., Pittsburgh, p. 147e148.
[19] Disimile PJ, Davis J, Toy N. Mitigation of shock waves within a liquid lled
[42] Guo J, Shi G, Wang Y, Lu C. Efcient modeling of panel-like structures in
tank. Int J Impact Eng 2011;38(2e3):61e72.
perforation simulations. Comput Struct 2003;81(1):1e8.
[20] Varas D, Zaera R, Lpez-Puente J. Numerical modelling of partially lled
[43] Johnson GR, Cook WH. A constitutive model and data for metals subjected to
aircraft fuel tanks submitted to hydrodynamic ram. Aerosp Sci Technol
large strain, high strain rates and high temperatures. In: Proceedings of the
7th international symposium on Ballistics 1983. p. 541e7. The Hague, The
[21] Varas D, Zaera R, Lpez-Puente J. Numerical modelling of the hydrodynamic
ram phenomenon. Int J Impact Eng 2009;36(3):363e74.
[44] Alsos HS, Amdahl J, Hopperstad OS. On the resistance to penetration of stiffened
[22] Lecysyn N, Dandrieux A, Heymes F, Slangen P, Munier L, Lapebie E, et al.
plates, part II: numerical analysis. Int J Impact Eng 2009;36(7):875e87.
Preliminary study of ballistic impact on an industrial tank: projectile velocity
[45] Simonsen BC, Trnqvist R. Experimental and numerical modelling of ductile
decay. J Loss Prev Process Indust 2008;21(6):627e34.
crack propagation in large-scale shell structures. Mar Struct 2004;17(1):1e27.
[23] Hutchinson MD. The escape of blast from fragmenting munitions casings. Int J
Impact Eng 2009;36(2):185e92.

Похожие интересы