Vous êtes sur la page 1sur 11

ARTICLE IN PRESS

International Journal of Mechanical Sciences 48 (2006) 3343


www.elsevier.com/locate/ijmecsci

Finite element modeling of transverse impact on a ballistic fabric


Y. Duana,, M. Keefeb, T.A. Bogettic, B. Powersc
a

Center for Composite Materials, University of Delaware, Newark, DE 19716, USA


Department of Mechanical Engineering, University of Delaware, Newark, DE 19716, USA
c
US Army Research Laboratory, Aberdeen Proving Ground, MD 21005, USA

Received 5 August 2004; received in revised form 14 July 2005; accepted 17 September 2005
Available online 26 October 2005

Abstract
A 3D nite element analysis model is created using LS-DYNA to simulate the transverse impact of a rigid right circular cylinder onto a
square patch of plain-woven Kevlar fabric. The fabric is modeled to yarn level resolution and relative motion between yarns is allowed. A
frictional contact is dened between yarns and between the fabric and the projectile. Three different boundary conditions are applied on
the fabric: four edges left free; two opposite edges clamped; four edges clamped. Results from the modeling effort show that during initial
stage of the impact, the projectile velocity drops very quickly. There exists an abrupt momentum transfer from the projectile to the local
fabric at the impact zone. When the impact velocity is low, the fabric boundary condition plays an important role at later stages of the
impact. It signicantly affects the fabric deformation, stress distribution, energy absorption and failure modes. When the impact velocity
is high enough to cause the yarns to break instantaneously, the fabric fails along the periphery of the impact zone and the fabric
boundary condition does not take any effects.
r 2005 Elsevier Ltd. All rights reserved.
Keywords: Ballistic fabric; Transverse impact; Energy absorption; Finite element analysis

1. Introduction
Weight and exibility are two important design parameters for soft body armors that offer protection against
bullets and munitions fragments. Fabrics made from highstrength bers, also called ballistic fabrics, have the
attractive properties of low density, high exibility, high
strength-to-weight ratio, and outstanding ballistic resistant
property. Therefore, they have been widely used in making
soft body armors since their introduction to market. The
high-strength bers used in making ballistic fabrics include
aramid (Kevlar, Twaron, Technora), polyethylene (Spectra, Dyneema), and Polybenzoxazole (Zylon) [1,2]. These
bers are essentially elastic in tension and have very high
tensile modulus. They have relatively low tensile failure
strain and are generally not sensitive to strain rate.
Hundreds of the high-strength bers are grouped together
to make a yarn and yarns are woven to produce a single
layer ballistic fabric.
Corresponding author. Tel.: +1 302 831 0376; fax: +1 302 831 8525.

E-mail address: duan@ccm.udel.edu (Y. Duan).


0020-7403/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmecsci.2005.09.007

Typically, a soft body armor is constructed of a ballistic


panel, which is assembled from multiple layers of ballistic
fabric, and a carrier made from conventional garment
fabric. When a projectile impacts into such a fabric
structure, it is gradually slowed down and nally caught
by the ballistic panel. The impact energy is converted as
fabric kinetic energy, fabric strain energy, projectile
deformation energy and energy dissipated in frictional
sliding. The impact resistance of a ballistic panel depends
on its capability to absorb energy locally at the impact zone
and disperse energy quickly out of the impact zone. It is
affected by a number of factors, which include ber
density, ber tensile elastic modulus, ber tensile failure
strain, fabric weave style, fabric areal density, number of
fabric layers, fabric boundary condition, projectile shape,
mass and material property, impact velocity, and interfacial friction characteristics within impact system.
To optimize the design of a soft body armor, in other
words, increasing its ballistic resistance and at the same
time keeping or even reducing its weight and rigidity, one
must understand the impact behavior of the ballistic panel.
To understand the impact behavior of the ballistic panel,

ARTICLE IN PRESS
34

Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

one should rst understand the impact behavior of its


construction units such as single layer fabric, single yarn,
and single ber. During the past several decades, a lot of
experiments and theoretical work have been conducted to
understand the transverse impact behavior of single yarns
and single layer fabrics [311]. Smith et al. [3], Roylance [4],
Morrison [5], and Field and Sun [6] studied the response of
yarns to high-speed transverse impact while Wilde et al.
[7,8], Briscoe and Motamedi [9], Shim et al. [10], and
Shockey et al. [11] investigated the transverse impact
behavior of single layer fabrics.
In this paper, a nite element analysis (FEA) model is
created using LS-DYNA to simulate the transverse impact
of a right circular cylinder (RCC) on a single layer plainwoven Kevlar fabric. The model allows for denition of
contact between yarns and therefore takes into account the
physical interaction between yarns. It more realistically
describes the woven fabric structure than those computational models in which yarn crossovers are described as
links, joints or bulk continuum.
2. Transverse impact on yarns
2.1. Existing theory on transverse impact on a long straight
yarn
As stated previously, the yarn in a ballistic fabric is
composed of hundreds of high-strength bers. It is a very
complex structure. For simplicity in analysis, the interaction between bers in a yarn is generally ignored and the
yarn is assumed to be an elastic continuum. Fig. 1 shows a
heavy wedge-tipped projectile transversely impacting on a
long straight yarn. The yarn tensile elastic modulus is E
and volumetric density is r. The impact velocity is v and it
is not high enough to cause the yarn to break. According to
Refs. [3,4], two mechanical waves are generated by the
impact. One is a longitudinal wave, which propagates away
from the impact point at the sound speed of the yarn
material. The longitudinal wave speed c is given by
s
E
c
.
(1)
r
v

time = 0

time = t
Fig. 1. A wedge-tipped projectile transversely impacts on a long straight
yarn at a constant velocity of v.

Ahead of the longitudinal wave front, yarn material strain


is zero; behind, a constant tensile strain is developed. The
tensile strain, denoted as , is determined by the yarn tensile
elastic modulus E, volumetric density r, and the impact
velocity v. It is implicitly given by
p
rv2
.
(2)
1   2
E
The other mechanical wave generated by the impact is a
transverse one, which propagates away from the impact
point at a relatively lower speed. The transverse wave speed
u is given by
r

.
(3)
uc
1

2

Across the transverse wave front, the strain of yarn


material does not change; however, the motion of yarn
material experiences an abrupt change. Ahead of the
transverse wave front but behind the longitudinal wave
front, yarn material moves longitudinally toward the
impact point. Behind the transverse wave front, yarn
material moves transversely in the impact direction. It can
be seen from Eq. (3) that the transverse wave speed
positively correlates with the yarn tensile strain. It is larger
with a larger tensile strain, and vice versa. When the yarn
tensile strain is zero, the transverse wave speed is also zero.
The yarn kinetic energy E k and the yarn strain energy E s
during the impact can be obtained from Eqs. (4) and (5),
respectively.
s
r
3

2 E
E k At
,
(4)
32
1
r
s
E3
E s At2
,
r

(5)

where, A is the yarn cross-section area and t is the time


after impact. The two formulas may be obtained from the
above analysis of wave propagations in yarn material.
2.2. Comparison of FEA modeling results and predictions
from the existing theory
Consider an impact case: a heavy wedge-tipped projectile
transversely impacts on a long straight Kevlar yarn that
has a tensile elastic modulus E of 74 GPa, a volumetric
density r of 1,440 kg/m3 , and a cross-section area A of
5:83  108 m2 . The impact velocity v is 200 m/s. The yarn
tensile strain generated by the impact, , can be obtained
from Eq. (2) by using iteration method. Substitute the
values of E; r; A, and  into Eqs. (4) and (5), the yarn
kinetic energy and the yarn strain energy at any time t can
be obtained.
A 3D FEA model is created using LS-DYNA to simulate
the impact. Fig. 2 shows the FEA model for the straight
Kevlar yarn. The yarn is modeled as a continuum and the
yarn cross-section is dened by a pair of symmetric arcs.

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

Gasser et al. [12] have shown that an orthotropic elastic


continuum has yarn behavior if its Poissons ratios are zero
and the shear moduli and transverse elastic moduli are very
small with respect to the longitudinal elastic modulus. In
the FEA model, the yarn has locally orthotropic elastic
material property. Table 1 lists the nine orthotropic elastic
material data. Each element in the model denes a
referential coordinate whose three axes are determined by
the nodes of the element. The orthotropic elastic material
data is dened in the local referential coordinates.
The yarn kinetic energy and the yarn strain energy as a
function of time are obtained from the FEA modeling.
Fig. 3 shows a comparison of the modeling results and the
analytical results obtained from Eqs. (4) and (5). It can be
seen from this gure that the FEA modeling results agree
well with the analytical results. The good agreement
indicates that the FEA modeling approach and the
orthotropic elastic material data listed in Table 1 describe
well the transverse impact behavior of the Kevlar yarn.

Fig. 2. The 3D FEA model for the straight Kevlar yarn.

Table 1
Orthotropic elastic material data (GPa) for the Kevlar yarn
E11

E22

E33

G12

G13

G23

n12

n13

n23

74

0.74

0.74

0.148

0.148

0.148

0.05
Yarn kinetic energy; FEA
Yarn kinetic energy; analytical
Yarn strain energy; FEA
Yarn strain energy; analytical

Energy (J)

0.04
0.03
0.02
0.01
0

10

12

14

Time (s)
Fig. 3. A comparison of the FEA modeling results and the predictions
from theory.

35

2.3. Effect of yarn ends boundary condition


In the above analysis and modeling, the Kevlar yarn is
assumed to be innitely long. Therefore, the effect of yarn
ends boundary condition is not taken into account. For a
real impact situation, the stress/strain wave generated by
the impact soon arrives at the yarn ends. The yarn ends
boundary condition inevitably plays a role in the yarn
impact behavior.
To explore the effect of yarn ends boundary condition,
two cases are modeled where a heavy wedge-tipped
projectile transversely impacts at 200 m/s onto the center
of a Kevlar yarn that has a length of 49 mm. In the rst
case, both the yarn ends are clamped, while in the second
case both the yarn ends are left free. It can be seen from
Table 1 that the yarn longitudinal elastic modulus is much
larger than its shear moduli and transverse elastic moduli.
In this situation, the value of the maximum principal stress
is very close to that of the tensile stress along ber
direction. For convenient implementation in LS-DYNA, a
maximum principal stress failure criterion is used in the
modeling. When the maximum principal stress at a
material point exceeds 2.3 GPa, the material fails and the
corresponding element is deleted automatically from the
mesh. The release wave resulted from deleting the element
is taken into account in subsequent deformation process.
The maximum principal stress failure criterion is equivalent
to a maximum tensile strain failure criterion with a failure
strain of 3.1%.
Fig. 4 shows the yarn deformation when both of its ends
are clamped, while Fig. 5 shows the yarn deformation when
both of its ends are left free. It can be seen from the two
gures that the yarn ends boundary condition signicantly
affects the yarn deformation. The yarn is broken at the
impact point when its two ends are clamped while it is not
broken when its two ends are left free. The different yarn
behavior is a result of the different yarn ends boundary
condition. As stated previously, two mechanical waves are
generated by the impact: a longitudinal one and a
transverse one. The longitudinal wave propagates away
from the impact point at a very high speed while the
transverse wave propagates at a relatively lower speed.
Behind the transverse wave front, yarn material moves
transversely in the impact direction. Ahead of the
transverse wave front but behind the longitudinal wave
front, yarn material moves longitudinally toward the
impact point. When the longitudinal wave reaches the
clamped ends, it is reected back and propagates toward
the impact point. Behind the reected wave front, yarn
material stops moving longitudinally and the tensile stress
is doubled. The reected longitudinal waves meet at the
impact point where the tensile stress is superimposed on
each other. After three reections of the longitudinal wave,
the stress at the impact point reaches the failure criterion
and the yarn, as shown in Fig. 4, is broken. Similarly, when
the longitudinal wave reaches the free ends, it is also
reected back and propagates toward the impact point.

ARTICLE IN PRESS
36

Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

Fig. 4. Deformation of the straight Kevlar yarn at various instants of time


when both of its ends are clamped; the arrows indicate the transverse wave
fronts at 20 ms.

However, behind the reected wave front, the yarn tensile


stress becomes zero and the velocity of the yarn material
moving longitudinally toward the impact point is doubled.
As can be seen from Fig. 5, with reection of the
longitudinal wave the two ends of the yarn move longitudinally toward the impact point. The transverse wave
stops propagating when the reected longitudinal wave
meets the transverse one.

Fig. 5. Deformation of the straight Kevlar yarn at various instants of time


when both of its ends are left free; the arrows indicate the transverse wave
fronts at 20 ms.

y
RCC projectile
x

3. Modeling transverse impact on a ballistic fabric


Fig. 6 shows the initial geometry of an impact event: a
rigid RCC projectile transversely impacts onto the center
of a patch of plain-woven Kevlar fabric. The fabric is at
and aligned with the xz plane. It is composed of 39 yarns
in each of the warp direction (along the x-axis) and the weft
direction (along the z-axis). The fabric maximum thickness
is 0.23 mm, and both of its side length is 32.7 mm. The
fabric four edges are left free. The projectile diameter is
8 mm, its mass is 2 g, and its impact velocity v is 200 m/s.
During the impact, the rigid projectile can only move along
the y-direction and the other ve degrees of freedom are
constrained.
A 3D FEA model is created using LS-DYNA to simulate
the aforementioned impact. The impact system has
symmetry with respect to both the xy plane and the yz
plane, therefore only a quarter of the entire system needs to

v
f
f
f

Kevlar fabric

Fig. 6. A rigid RCC projectile transversely impacts onto the center of a


square patch of plain-woven Kevlar fabric; the impact velocity is v.

be modeled. Fig. 7 shows a part of the 3D FEA model for


the plain-woven fabric. The fabric is modeled to yarn level
resolution and the yarns are modeled as continuum with
locally orthotropic elastic material property. The material
properties of the Kevlar yarn have been given in the

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

37

Projectile velocity (m/s)

200
197
194
191
188
185

10

20

30

40

50

Time (s)

Fig. 8. Time history of the projectile velocity for the case with four fabric
edges left free and v 200 m/s. The arrow on the projectile velocity versus
time curve indicates the stage of deformation shown in Fig. 9.
Fig. 7. The 3D FEA model for the plain-woven Kevlar fabric.

previous section. The cross-section of the crimped yarn is


the same as that of the straight yarn shown in Fig. 2. The
crimped proles of the warp and the weft yarns are
identical and are dened by a series of connected arcs. The
3D FEA model for the plain-woven fabric denes
yarnyarn contact and allows for relative motion between
yarns. Simple Coulomb friction is introduced between
yarns and between the projectile and the fabric. A friction
coefcient of 0.3, which is obtained from experiments on
Kevlar fabrics, is used for both the types of friction [13].
The inuence of interfacial friction on fabric impact
behavior will not be discussed in this paper. However, it
is worth noting that the friction between projectile and
fabric, between yarn and yarn, and between bers
themselves might have signicant effects on the impact
behavior of ballistic fabrics, especially when the projectile
is in spherical shape [9,14,15].
In order to comparatively investigate the effect of fabric
boundary condition, two additional cases are modeled
where all the conditions described previously are maintained except that different boundary conditions are
applied on the fabric. In one case, two opposite edges of
the fabric are clamped and the other two edges are left free;
in the other case, all the four edges of the fabric are
clamped. Finally, to explore the effect of impact velocity on
the fabric ballistic performance, three cases are modeled
where the impact velocity v is 400 m/s instead of 200 m/s
and the three different types of boundary conditions
described previously are applied on the fabric, respectively.
4. Results and discussion
4.1. Projectile-fabric interaction and energy transfer
Fig. 8 shows time history of the projectile velocity for the
case with four fabric edges left free and impact velocity
v 200 m/s. It can be seen from this gure that within a
very short period of time 0:3 ms, the projectile velocity
drops from 200 to 198.3 m/s. Afterwards, the projectile

gradually slows down and at 50 ms its velocity is 188 m/s.


Fig. 9 depicts contour maps of the fabric transverse
velocity and transverse displacement at 0:3 ms when the
projectile velocity is 198.3 m/s. It can be seen from this
gure that the local fabric that directly contacts the RCC
projectile abruptly moves with the transverse impact.
The projectile momentum is transferred to the local fabric.
The momentum transfer occurs so quickly that the
fabric located out of the impact zone is not affected at
all. At this moment, the fabric transverse displacement
coincides with the projectile-fabric contact zone and is in a
pie shape. The initial momentum transfer is responsible for
the abrupt drop of the projectile velocity from 200 to
198.3 m/s.
After the initial momentum transfer, the local fabric in
the impact zone moves together with the projectile. Due to
the sudden transverse motion of the local fabric, a
longitudinal wave and a transverse wave are generated in
the principal yarns (those yarns that directly contact the
projectile). The longitudinal wave propagates away from
the impact zone at a very high speed; behind the wave
front, yarn material is strained and moves longitudinally
toward the impact zone. The transverse wave propagates
away from the impact zone at a relatively lower speed;
behind the wave front, yarn material moves transversely in
the impact direction. Fig. 10 shows contour maps of the
fabric resultant displacement at initial stages of the impact.
It can be seen from this gure that at 0:5 ms, mainly the
principal yarns are affected. With propagation of the
mechanical waves, yarnyarn interactions cause the secondary yarns (those yarns that do not directly contact the
projectile) to move. At 1:0 ms, the impact-affected zone
takes a square form, and with time going on it gradually
expands outward. During the process, the fabric absorbs
energy from the projectile and the projectile gradually
slows down. At 3:0 ms when most of the fabric has been
affected by the impact, the projectile velocity is 197.9 m/s.
The projectile velocity is reduced by 1.7 m/s during the
initial 0:3 ms while it is only reduced by 0.4 m/s during the
period from 0.3 to 3 ms.

ARTICLE IN PRESS
38

Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

Fig. 9. The fabric deformation at 0:3 ms. (a) Contour map of the fabric transverse velocity 103 m/s). (b) Contour map of the fabric transverse
displacement (mm).

Fig. 10. Contour maps of the fabric resultant displacement (mm) at the initial stage of the impact (v 200 m/s, four fabric edges left free).

Deformed congurations of the fabric at various instants


of time are illustrated in Fig. 11 for the case with four
fabric edges left free and impact velocity v 200 m/s. It can
be seen from this gure that during the impact, the fabric at
the impact zone conforms to the at round nose of the
RCC projectile. Lim et al [16] have observed by using highspeed camera the conformation of fabric to the nose shape
of a at-nosed projectile during transverse impact. It is
found from the deformed congurations of the fabric that
two warp yarns and one weft yarn are broken during the
impact. Except for the three yarns, no other yarns in
the fabric are broken. It can be seen from Fig. 11 that the
fabric transverse displacement is in the form of a conical
frustum at 10 ms. At that instant of time, the four fabric
edges slightly bow toward the impact zone but most of the
fabric is not affected by the transverse wave yet. Bowings
of the fabric edges indicate that the longitudinal wave has
been reected back from the fabric edges. With time going
on, the four fabric edges gradually bow toward the impact

zone and the fabric transverse wave gradually propagates


outward. It is noted that the fabric transverse wave front
evolves during the impact. It is in the form of a circle at
10 ms while it is in the form of a round-lleted square at
30 ms. The corresponding projectile velocity for each of
the deformed congurations of the fabric can be found in
Fig. 8.
There is no external force acting on the system during the
impact. The energy in the system is therefore conserved.
The lost projectile kinetic energy is completely absorbed by
the fabric. Fig. 12 shows time history of the energy transfer
between the projectile and the fabric. It can be seen that at
the initial momentum transfer, the projectile loses 0.68 J of
kinetic energy, of which 80% is absorbed by the fabric in
the form of yarn kinetic energy, 19% is absorbed by the
fabric in the form of yarn strain energy, and the remaining
1% is dissipated as heat through friction between the
projectile and the fabric and between yarns themselves. It is
evident that yarn kinetic energy is the dominant energy

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

39

Fig. 11. Top and side view of the fabric deformation at various instants of time (v 200 m/s, four fabric edges left free).

4.2. Effect of fabric boundary condition


Two additional cases with different fabric boundary
conditions (two opposite edges clamped; four edges
clamped) are modeled to comparatively study the effect
of fabric boundary condition. Fig. 13 shows the projectile
velocity as a function of time for the three cases that have
the same impact velocity v of 200 m/s but different fabric
boundary conditions. It can be seen from this gure that
within the initial 5 ms, the projectile velocity is the same for
all the three cases. This result indicates that the fabric
boundary conditions do not play a role in decelerating the
projectile during that period of time. After 5 ms, the fabric
boundary conditions take effect and the projectile velocity
becomes different for the three cases; the projectile velocity
drops most slowly when all the four fabric edges are left
free while it drops most quickly when all of the four fabric
edges are clamped. It can be seen that at 10 ms, the
projectile velocity is 197.0 m/s for the case with four fabric
edges left free while it is 196.1 m/s for the case with two
opposite fabric edges clamped and 195.1 m/s for the case
with four fabric edges clamped. Though the fabric most
quickly decelerates the projectile when its four edges are

6
Loss of projectile kinetic energy
Yarn kinetic energy
Yarn strain energy
Friction dissipated energy

5
Energy (J)

absorption mechanism during the impact. At 50 ms, around


67% of the lost projectile kinetic energy is absorbed by the
fabric in the form of yarn kinetic energy, while 25% in the
form of yarn strain energy and 8% in the form of friction
dissipated energy.

4
3
2
1
0

10

20

30
Time (s)

40

50

Fig. 12. Time history of energy transfer between the projectile and the
fabric (v 200 m/s, four fabric edges left free).

clamped, it loses function at the earliest time; the projectile


punches through the fabric at 12 ms and moves away at a
constant velocity of 194.4 m/s. When two opposite edges of
the fabric are clamped, the projectile gradually slows down
until 45 ms when the fabric is punched through and loses its
capability to decelerate the projectile. As can be seen, the
fabric most effectively slows down the projectile when all of
its four edges are left free; at 50 ms, the projectile velocity is
188.0 m/s when the four fabric edges are left free while it is
193.1 m/s when two opposite edges of the fabric are
clamped and 194.4 m/s when all of the four fabric edges
are clamped.

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

40

Fig. 14 shows the maximum principal stress distribution


in the fabric at 8 ms for the three impact cases. It can be
seen that the maximum principal stress mainly distributes
in the clamped principal yarns and it is very small in the
unclamped principal yarns or in the secondary yarns. The
fabric boundary condition signicantly affects the stress
distribution pattern in the fabric. Fig. 15 shows contour
maps of the fabric transverse displacement at 10 ms for the
three cases. As can be seen, the transverse wave front is in
the form of a circle when all the four fabric edges are left
free, while it is in the form of an ellipse with the long axis

Projectile velocity (m/s)

200

196

192
Four fabric edges left free
Two opposite fabric edges clamped
Four fabric edges clamped

188

184
0

10

20

30

40

50

Time (s)
Fig. 13. The projectile velocity as a function of time for the three cases
that have the same impact velocity of 200 m/s but different boundary
conditions.

along the clamped yarns when two opposite fabric edges


are clamped and a lleted square when all the four fabric
edges are clamped. The different stress distributions and
the different transverse displacements are due to the
different fabric boundary conditions. As stated previously,
with the transverse impact, two mechanical waves originate
in the impact zone. The longitudinal wave propagates away
from the impact zone at a very high speed while the
transverse wave propagates at a lower speed. Behind the
longitudinal wave front, yarn material is tensioned and
moves longitudinally toward the impact zone, while behind
the transverse wave front, yarn material moves transversely. When the longitudinal wave arrives at the free fabric
edges, it is reected back and propagates toward the impact
zone. Behind the reected wave front, the yarn tensile stress
fades away and the velocity of the yarn material moving
toward the impact zone is doubled; the interactions
between yarns gradually produce bowings along the free
fabric edges. When the longitudinal wave arrives at the
clamped edges, it is also reected back and propagates
toward the impact zone. However, behind the reected
wave front, the yarn tensile stress is doubled and the yarn
material stops moving longitudinally toward the impact
zone. At 8 ms, the longitudinal wave has been reected by
the fabric edges and propagated back to the impact zone.
Therefore, the stress in the clamped principal yarns is much
larger than the stress in the unclamped principal yarns and
the secondary yarns. As shown by Eq. (3), the transverse
wave speed in a yarn is determined by the sound speed of

Fig. 14. Distribution of maximum principal stress (103 GPa) in the fabric at 8 ms for the three cases that have the same impact velocity of 200 m/s but
different boundary conditions. (a) Four fabric edges left free. (b) Two opposite fabric edges clamped. (c) Four fabric edges clamped.

Fig. 15. Contour maps of the fabric transverse displacement (mm) at 10 ms for the three cases that have the same impact velocity of 200 m/s but different
boundary conditions. (a) Four fabric edges left free. (b) Two opposite fabric edges clamped. (c) Four fabric edges clamped.

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

41

Fig. 16. The fabric deformation at 40 ms for the three cases that have the same impact velocity of 200 m/s but different boundary conditions. (a) Four
fabric edges left free. (b) Two opposite fabric edges clamped. (c) Four fabric edges clamped.

4.3. Effect of impact velocity


Three cases with an impact velocity of 400 m/s and
different boundary conditions (four fabric edges left free;
two opposite fabric edges clamped; four fabric edges
clamped) are modeled to comparatively study the effect
of impact velocity. Fig. 17 shows time history of the
projectile velocity while Fig. 18 shows the fabric deformed
congurations at various instants of time for the case with
four fabric edges left free. It can be seen that the projectile
is decelerated very quickly at initial stage of the impact.
Within 0:3 ms, the projectile velocity drops from 400 to
396.7 m/s. The abrupt drop of the projectile velocity is due
to the initial momentum transfer from the projectile to the

400
Projectile velocity (m/s)

the yarn material and the tensile strain in the yarn. It is


higher with a larger strain and lower with a smaller strain.
Due to reection of the longitudinal wave from the fabric
edges, the transverse wave propagates much quicker along
the clamped yarns than along the free yarns. Therefore, for
the case with two opposite fabric edges clamped, the
transverse wave front is in the form of an ellipse, with its
long axis along the clamped yarns and its short axis along
the free yarns.
Fig. 16 illustrates the deformed congurations of the
fabric at 40 ms for the three cases. It can be seen from this
gure that the fabric boundary condition signicantly
affects the fabric deformation at later stage of the impact.
For the case with four fabric edges left free, large bowings
are produced along the four edges; the integrity of the
fabric is maintained well and only a few yarns are broken
along the periphery of the impact zone. For the case with
two opposite fabric edges clamped, bowings are produced
along the two free edges; most of the clamped yarns are
broken. For the case with four fabric edges clamped, all the
principal yarns are broken at the impact zone. As Fig. 13
shows, the fabric most effectively slows down the projectile
when all its four edges are left free. The reason for the high
performance of the fabric is that only few yarns are broken
during the impact when all the four fabric edges are left
free. Due to the local failure at the impact zone, the fabric
loses capability to decelerate the projectile at later stages of
the impact when two or four of its edges are clamped.

399
398
397
396
395
0

10

Time (s)
Fig. 17. Time history of the projectile velocity for the case with four fabric
edges left free and v 400 m/s.

local fabric at the impact zone. The projectile velocity


drops by 0.5 m/s during the period from 0.3 to 4 ms when
the fabric completely loses capability to decelerate the
projectile. The fabric fails along the periphery of the impact
zone. The local fabric at the impact zone is punched out
while most of the fabric does not move transversely. The
fabric deformation is very different from that with an
impact velocity of 200 m/s (see Fig. 11). The energy transfer
between the projectile and the fabric is illustrated in Fig. 19
for the case with v 400 m/s and four fabric edges left free.
It can be seen that yarn kinetic energy is the dominant
energy absorption mechanism; it accounts for around 63%
of the total absorbed energy.
Fig. 20 shows the projectile velocity as a function of time
for the three cases with the same impact velocity of 400 m/s
but different boundary conditions, while Fig. 21 depicts
contour maps of the fabric transverse displacement at
10 ms. It can be seen that the time history of the projectile
velocity is the same for all the three cases and the fabric
deformed congurations show little difference for the three
different boundary conditions. The fabric boundary condition does not take any effect when the impact velocity is
400 m/s. The phenomena are very different from those with
an impact velocity of 200 m/s (see Figs. 13 and 16). The
fabric energy absorption as a function of time is shown in
Fig. 22 for the six cases that have different impact velocities

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

42

Fig. 18. Top and side view of the fabric deformation at various instants of time (v 400 m/s, four fabric edges left free).

400

Energy (J)

5
4

Projectile velocity (m/s)

Loss of projectile kinetic energy


Yarn kinetic energy
Yarn strain energy
Friction dissipated energy

3
2

Four fabric edges left free


Two opposite fabric edges clamped
Four fabric edges clamped

399
398
397
396

1
395

0
0

10

Time (s)
Fig. 19. Time history of energy transfer between the projectile and the
fabric (v 400 m/s, four fabric edges left free).

and different fabric boundary conditions. It can be seen


from this gure that the fabric responses under 200 and
400 m/s are very different. When the impact velocity is
200 m/s, the fabric boundary condition signicantly affects
the fabric energy absorption. However, when the impact
velocity is 400 m/s, the fabric boundary condition does not
have any effect on the fabric energy absorption. Further
modeling work shows that the transition takes place at
around 300 m/s. The results indicate that fabric boundary
condition plays an important role only when the impact
velocity is low. When the impact velocity is high enough to

6
Time (s)

10

Fig. 20. The projectile velocity as a function of time for the three cases
that have the same impact velocity of 400 m/s but different boundary
conditions.

cause yarns to break instantaneously, the fabric deformation is localized at the impact region and the fabric far eld
boundary condition does not take any effects.
5. Conclusions
A 3D FEA model is created using LS-DYNA to simulate
the transverse impact of a rigid RCC projectile on a single
layer plain-woven Kevlar fabric. The fabric is modeled to
yarn level resolution and relative motion between yarns is
allowed. A frictional contact is dened between yarns and

ARTICLE IN PRESS
Y. Duan et al. / International Journal of Mechanical Sciences 48 (2006) 3343

43

Fig. 21. Contour maps of the fabric transverse displacement (mm) at 10 ms for the three cases that have the same impact velocity of 400 m/s but different
boundary conditions. (a) Four fabric edges left free. (b) Two opposite fabric edges clamped. (c) Four fabric edges clamped.

References

Energy absorbed by fabric (J)

7
v=200 m/s, four fabric edges left free
v=200 m/s, two fabric edges clamped
v=200 m/s, four fabric edges clamped
v=400 m/s, all boundary conditions

6
5
4
3
2
1
0
0

10

20

30

40

50

Time (s)
Fig. 22. Time history of the fabric energy absorption for the six cases with
different impact velocities and boundary conditions.

between the fabric and the projectile. Three boundary


conditions are applied on the fabric: four edges left free;
two opposite edges clamped; four edges clamped. Modeling
results show that during initial stage of the impact, the
projectile velocity drops very quickly. There exists an
abrupt momentum transfer from the projectile to the local
fabric at the impact zone. When the impact velocity is low,
the fabric boundary condition plays an important role. It
signicantly affects the fabric deformation, stress distribution, energy absorption, and failure modes. The fabric
most effectively slows down the projectile when all its four
edges are left free. The reason for the high performance of
the fabric is that only few yarns are broken during the
impact when all the four fabric edges are left free. When the
impact velocity is high and causes yarns to break
instantaneously, the fabric deformation is localized at the
impact region and the fabric far eld boundary condition
does not take any effects on the fabric ballistic performance.
Acknowledgements
The support of the US Army Research Laboratory at
Aberdeen Proving Ground and the Center for Composite
Materials at University of Delaware (UD-CCM) during
this research is gratefully acknowledged.

[1] Phoenix SL, Porwal PK. A new membrane model for the
ballistic impact response and v50 performance of multi-ply brous
system. International Journal of Solids Structure 2003;40(24):
672365.
[2] Jacobs MJN, Van Dingenen JLJ. Ballistic protection mechanisms
in personal armour. Journal of Materials Science 2001;36(13):
313742.
[3] Smith JC, Blandford JM, Schiefer HF. Stress-strain relationships in
yarns subjected to rapid impact loading. Part VI: velocities of strain
waves resulting from impact. Text Research Journal 1960;30:75260.
[4] Roylance D. Ballistics of transversely impacted bers. Text Research
Journal 1977;47(10):67984.
[5] Morrison C. The mechanical response of an aramid textile yarn to
ballistic impact. PhD thesis, University of Surrey, UK, 1984.
[6] Field JE, Sun Q. A high speed photographic study of impact on bers
and woven fabrics. In: Gareld B, Rendell J, editors. Proceedings of
the 19th international congress on high-speed photography and
photonics; 1990. p. 70312.
[7] Wilde AF, Roylance DK, Rogers JM. Photographic investigation of
high-speed missile impact upon nylon fabric. Part I: energy
absorption and cone radial velocity in fabric. Text Research Journal
1973;43(12):75361.
[8] Wilde AF. Photographic investigation of high-speed missile impact
upon nylon fabric. Part II: retarding force on missile and transverse
critical velocity. Text Research Journal 1974;44(10):7728.
[9] Briscoe BJ, Motamedi F. The ballistic impact characteristics of
aramid fabrics: the inuence of interfacial friction. Wear 1992;
158(12):22947.
[10] Shim VPW, Tan VBC, Tay TE. Modeling deformation and damage
characteristics of woven fabric under small projectile impact.
International Journal of Impact Engineering 1995;16(4):585605.
[11] Shockey DA, Erlich DC, Simons JW. Improved barriers to turbine
engine fragments. US Department of Transportation, Federal
Aviation Administration, DOT/FAA/AR-99/8 III, 2001.
[12] Gasser A, Boisse P, Hanklar S. Mechanical behavior of dry fabric
reinforcement. 3D simulations versus biaxial tests. Composite
Materials Science 2000;17(1):720.
[13] Martinez MA, Navarro C, Cortes R, Rodriguez J, Sanchez-Galvez V.
Friction and wear behaviour of Kevlar fabrics. Journal of Materials
Science 1993;28(5):130511.
[14] Duan Y, Keefe M, Bogetti TA, Cheeseman BA. Modeling friction
effects on the ballistic impact behavior of a single-ply high-strength
fabric. International Journal of Impact Engineering 2005;31(8):
9961012.
[15] Duan Y, Keefe M, Bogetti TA, Cheeseman CA. Modeling the role of
friction during ballistic impact of a high-strength plain-weave fabric.
Composite Structure 2005;68(3):3317.
[16] Lim CT, Tan VBC, Cheong CH. Perforation of high-strength doubleply fabric system by varying shaped projectiles. International Journal
of Impact Engineering 2002;27(6):57791.

Vous aimerez peut-être aussi