Beruflich Dokumente
Kultur Dokumente
461471, 1998
( 1998 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0734743X(98)000062 0734743X/98 $19.00#0.00
(Received 11 June 1997; in revised form 5 September 1997: in final form 8 January 1998)
SummaryA very simple one-dimensional and fully analytical model of ballistic impact against
ceramic/composite armors is presented in this paper. The analytical model has been checked both
with ballistic tests and numerical simulations giving predictions in good agreement with them. The
model allows the calculation of residual velocity, residual mass, and the projectile velocity and the
deflection or the strain histories of the backup material. These variables are important in describing
the phenomenological process of penetration. Described are modifications to previous work of
impact into ceramics combined with a new composites model. The development of this composite
model is based on studies of the impact in yarns, fabrics and finally composites. ( 1998 Elsevier
Science Ltd. All rights reserved
NOTATION
A percentage of volume which has been delaminated
A cross-sectional area of the projectile
1
APDS armor piercing discarding sabot
c ceramic longitudinal sound speed
c longitudinal speed of sound in the fabric yarns
:
d areal density of the composite
e strain on the fabric under the projectile
E Young modulus of the yarns
F force exerted by each yarn on the ceramic cone
FSP fragment simulating projectile
G energy necessary to delaminate 1 m2 of composite
h ceramic plate thickness
characteristic dimension of the composite panel
M mass of the ceramic cone
#
M mass of the projectile
1
M mass of the projectile at the end of the first phase
11
n number of layers of fabric
-
n number of yarns under the projectile
:
h angle between the yarn and the direction of impact
R failure constant
o ceramic density
#
o projectile density
1
S section of the yarn
t time in seconds, zero is the instant of contact between armor and projectile
t time when the first phase has finished
1
t time of failure of the fabric
&
u5 (t) velocity of the ceramic cone
0
velocity of the projectile at any time
ballistic limit
50
residual velocity
3
striking velocity
4
mass of the projectile in grams.
x5 (t) velocity of the projectileceramic interface
*Corresponding author. Current address: Southwest Research Institute, Engineering Dynamics Department,
PO Box 28510, San Antonio, TX 78228-0510, U.S.A.
Tel.: (210)-522-2657; fax: (210) 522 3042; email: schocron@swri.edu.
461
462 I. S. C. Benloulo and V. Sanchez-Galvez
1. INTRODUCTION
Ceramic backed by composite armors are becoming the subject of many investigations
because their performance against small and medium caliber projectiles is outstanding
when the weight is a design condition, for instance in light weight vehicles, airplane and
helicopter protection or body armors. The main role of the ceramic is the erosion and
rupture of the projectile. The composite absorbs the kinetic energy of the fragments
stopping them. The design of these armors is really complex and need many sophisticated
tools. Empirical methods are the most widely used ones because they offer reliability, but
they are extremely expensive and the results did not give enough information: the history of
the projectile, the trends when changing the configurations or the phenomenological
process cannot be obtained in detail with the experimental approach. Another way to face
the problem is to use hydrocodes to simulate numerically the physical process. Either finite
elements or finite difference schemes need many parameters for material description, which
very often are not even known, for a correct calculation. For instance, the impact of
a tungsten projectile against a ceramic/composite armor may need as many as 50 different
mechanical parameters, including the elastic constants for each material (in all directions),
dynamic yield stress and others without clear physical meaning such as erosion strain, etc.
Another problem is that the calculations are very long and usually need many hours or even
days of powerful computers like workstations. The numerical approach provides a lot of
information but again cannot give trends unless multiple configurations are calculated and
consequently making the design of an armor long and hard. The third approach is the
analytical one. Provided the appropriate assumptions it could be a simple and fast way
allowing to obtain phenomenological information of the penetration mechanics without
losing much accuracy with respect to the numerical models. To the knowledge of the
authors only Florences model [1] appears in the literature as a simple analytical model
being used to analyze impact onto ceramic/composite armor. Interesting tests may be found
in Refs. [24].
The analytical model presented in this paper begins with the problem of the impact of
a projectile into the ceramic already extensively treated in the literature. Then the composite
model is faced with the logic reductionism: yarnfabriccomposite. Combining the two
models a ceramic/composite model is derived and checked against experimental data and
numerical simulations.
2. ANALYTICAL MODEL
The extremely complex problem of a projectile impacting a ceramic backed by a metal
plate has already been studied by Florence [1] in 1969 and more recently by Woodward
[5], den Reijer [6], Hetherington [7] and Zaera [8]. It is not the objective of this paper to
improve their ceramic impact models but simply to use them, together with the model of
impact onto fabrics developed by the authors [9], to build a ceramic/composite one. In the
following the reader can find the most important results obtained by the authors and how
they have been applied to this problem. Two phases have been assumed during the impact.
Nevertheless, the composite model has been checked independently, as a previous step to
the ceramic/composite model, with very different projectiles, usually FSPs, ranging from
0.2 g to tens of grams showing good agreement with experiments, see for instance, Refs.
[910]. In this latter case the steel projectiles were assumed to be rigid.
2.2.1.1. First case: xR (t)((t), rigid plastic projectile. There is a plastic zone where the
projectile is being eroded. Tate and Aleksevskii equation is applicable
#1 o (!x5 )2"@ #1 o (x5 !u5 )2 , (3)
1 2 1 # 2 # 0
Fig. 1. Configuration at the end of the first phase. Fig. 2. Phenomenological description of the second
phase.
464 I. S. C. Benloulo and V. Sanchez-Galvez
@ is the dynamic strength of the broken ceramic. For a careful study of the terms like
#
and @ the reader should consult Ref. [12].
1 #
Deceleration of the projectile is found from Tates equation:
d
M (t) "! A , (4)
1 dt 1 1
where at t"t is , the initial velocity in this phase which coincides with the final
1 1
velocity of the first phase. The last equation to be written for the projectile in the first case is
the following geometrical condition:
dM
1"!o A (!x5 ), (5)
dt 1 1
where M at t"t is defined as M , the projectile mass at the end of Phase 1.
1 1 11
2.2.1.2. Second case: xR (t)"(t), rigid projectile. When solving Eqns (3)(5) it could
happen that the velocity of the projectileceramic interface equals or even becomes larger
than the projectiles one. This has obviously no physical sense: the reality is that the
projectile is no longer flowing, so Eqns (35) are no longer applicable. Now, the interface
and the projectile move at the same speed
x5 (t)"(t) (6)
and the force slowing the projectile is exerted by the ceramic. Two subcases should be
considered.
Subcase 1: The velocity of the cone is smaller than that of the projectile (u5 (t)((t)),
0
hence the projectile is penetrating into the ceramic cone until the ceramic is completely
eroded. The formula governing this subcase is Newtons equation
d
M "!@ A , (7)
12 dt # 1
where now the force on the projectile is exerted by the pulverized ceramic and the mass of
the projectile is constant and equal to the mass at the end of the first case.
Subcase 2: It could happen that the ceramic cone, not yet being totally eroded (cone mass
more than 0.1 g), reaches projectiles velocity. Then the two bodies, as a single projectile, will
penetrate the composite backup. To simplify the mathematical aspect, the kinetic energies
of the ceramic cone and projectile are added and a new homogenous projectile with the
total mass and the same total kinetic energy is assumed to impact the composite. The
discontinuity obtained in the velocity is very small because projectile mass is much higher
than the ceramic cone mass.
2.2.2. Ceramic equations. The ceramic cone begins to move at the end of the first phase. It is
pushed at one end by the projectile while, at the back, it is being retained by the composite
backup. Figure 3 shows the forces acting on the cone during the impact process. The
composite material is a textile fabric, for instance, aramid or polyethylene yarns, embedded
2.2.3. Composite equations. In this section a new model, based on the previous works
published by Roylance [13], Cunniff [14], Beaumont [15] and Navarro [16] will be
presented. A simple analytical model of impact onto a fabric has already been published by
Chocron [9] including the basis of the model which is now expanded to ceramic backed by
composite materials. The model has been checked with alumina/aramid and alumina/
polyethylene giving good results when compared with analytical and numerical ones. It
basically states that the kinetic energy of the projectile is transmitted to the backup plate
through two mechanisms: straining and breaking of the yarns, and delamination. The next
two paragraphs focus on these events.
When a point projectile impacts a linear elastic yarn the velocity u5 of the projectile and
0
the strain e are related by
J2eJe(1#e)!e2
sin h" . (10)
Je(1#e)
The force the yarns exert on the projectile can easily be calculated with
F"EeSn n , (11)
- :
where E is the Youngs Modulus, S the section of the yarn, n the number of layers and h is
-
shown in Fig. 2. n is the number of yarns directly in contact with the impacting body. All
:
the yarns are assumed to have the same strain through the textile thickness to keep the
analytical model simple enough, weave and crimp are not considered. Now that the first
term in the right-hand side of Eqn (8) has become clear the second shall be explained.
Beaumont in his Ph.D. Thesis [15] studied low-velocity impact, around 100 m/s, into
brittle matrix composites and found a law to predict the evolution of delamination during
the impact process: it is assumed that the delamination travels together with the transversal
wave in the fabric, then
A BG C A A B D H
dA 2@3 ~1 1@3
AQ , "4 c 2 1!5 JA (12)
dt : 4c
:
466 I. S. C. Benloulo and V. Sanchez-Galvez
2.2.4. Failure of the composite backup. A failure criteria of the composite should be used in
order to stop the model calculations when penetration occurs. As it is assumed that the
failure happens just under the projectile a maximum strain or strength criteria would not
work: according to Smiths equation (9) the higher the impact velocity, the greater the strain
implying that both maxima coincide at the beginning of the impact process (when the
velocity of the projectile is the highest). If a maximum strain criterion is applied it will give
failure at the very first microsecond, which has no physical meaning.
A failure model based on the energy absorbed until failure by the backup is another way
to calculate the instant of failure. In this work one part of the energy is supposed to be
absorbed by delamination and the other one by the fabric, as elastic energy. The latter one is
supposed to be a constant amount of energy when failure occurs, independently of the
impact velocity. A similar hypothesis has already been used by Prosser [18]. Then if the
mass of the projectile is constant, as for instance in the case of the impact of a projectile into
a textile fabric
2!2"2 (13)
4 3 50
where is the striking velocity of the projectile, the residual velocity and the ballistic
4 3 50
limit.
The equation of energy for the projectile may be written as
P
1 1 1 5
M 2! M 2(t)" n n SEc e2 dt, (14)
2 1 4 2 1 2 : - :
0
which means that the kinetic energy lost by the projectile at any time t has been dissipated
through the yarns under it (e is the strain just under the projectile). Friction between yarns
or between the projectile and the fabric has been neglected in order to gain simplicity. If t is
selected as the failure instant and, as the energy until failure is considered constant
independently of the striking velocity, then
P
1 5&
n n SEc e2 dt"const., (15)
2 : - :
0
where t is the time of failure. Another way to write Eqn (15) is
&
P
5& M 2
e2 dt" 1 50 ,R, (16)
n n SEc
0 : - :
where R is called the failure constant. Thus, knowing (or, in fact, any residual velocity
50
from one single firing test), it is easy to calculate the time of failure for any impact velocity
because R will not change if the configuration of the armor does not change. It is suggested
not deriving R directly from the material properties because of the lack of accuracy in
measuring the section, the modulus (that could depend on the strain rate, see Ref. [19]), or
. With one single test the designer can calculate R, by forcing the analytical and
50
experimental residual velocity to be equal, and then use it to analyze all impact velocities.
Eqns (13)(16) are a summary of the failure model for impact into textile fabrics already
published by Chocron [9]. A more general failure model is easily derived: as R depends on
the configuration, let R be the failure constant for the target configuration i and R be the
i 0
one for a reference target configuration, from Eqn (16) it follows:
(M ) n (n ) (2 )
R "R 1i :0 - 0 50 i . (17)
i 0 (M ) n (n ) (2 )
10 :* - 0 50 0
Analytical model for ceramic/composite armors 467
This formula can be very useful if some empirical value for is assumed or found
50
experimentally. For instance, Van Gorp [20] found that Dyneema (or Spectra) armors
verify for fragment simulating projectiles:
"232 d0.5~1@6 , (18)
50
where d is the areal density of the panel and the mass of the projectile in g.
Then Eqn (17) may be rewritten as
(M )2@3 n
R "R 1* :0 , (19)
* 0 (M )2@3 n
10 :*
which only depends on the mass of the projectile and the number of yarns under it (say the
number of yarns per meter in the fabric). Equation (19) allows the calculation of the failure
constant for any configuration (the projectile, the thickness of the backup panel, the striking
velocity can be changed) if one firing test is provided in order to calculate firstly R .
0
The fabric failure model presented above has been, in the same way, applied for the
analysis of impact onto ceramic/composite armors taking into account that it is only a first
approximation to the actual process. Now the reference case for calculating R should be
0
a ceramic/composite one and the mass considered is not the projectiles one but the mass of
the impacting body.
Fig. 4. Comparison of the residual velocities ob- Fig. 5. Comparison of the residual lengths obtained
tained analytically and experimentally for a 20 mm analytically and experimentally for a 20 mm APDS
APDS projectile at 1250 m/s. Circles represent ex- projectile at 1250 m/s. Circles represent experimental
perimental data and crosses the result of the analyti- data and crosses the result of the analytical model.
cal model. Error bars represent the accuracy of the
Doppler radar.
All the four cases can be calculated in 5 s in a regular workstation or about 2 min in
a personal computer.
In Fig. 5, residual length of the projectile from the experiments and the model are
compared. Initial length of the projectile was 35.2 mm. The model provides always the best
performance attained by the ceramic element. Some of the ceramic targets may have some
defects that make them fail before expected, eroding the projectile less than desired and
giving the data dispersion observed.
As it is understood that nine tests, four different configurations, are not enough to
validate the analytical model, an additional series of eleven numerical tests, now changing
the projectile caliber, impact velocities and armor thickness, have been performed with
Autodyn-2D hydrocode, its results being compared with those of the analytical model.
Table 2. Configurations calculated and comparison of the results obtained with numerical and analytical methods
for a Tungsten projectile, 10 mm in diameter and 41.7 g
Table 3. Configurations calculated and comparison of the results obtained with numerical and analytical methods
for a Tungsten projectile, 14 mm in diameter and 114.3 g
Fig. 6. Graphic comparison of the residual velocities Fig. 7. Graphic comparison of the residual lengths
obtained analytically and numerically for the config- obtained analytically and numerically for the config-
urations of Tables 2 and 3. urations of Tables 2 and 3.
Fig. 8. Residual Velocity versus Striking velocity for the 10 mm projectile against ceramic/dyneema
armor, calculated analytically and numerically. The thicknesses were 20 mm/20 mm.
the summary of the results for the larger projectile. The results are also shown graphically in
Figs 6 and 7 to assess better the correlation of the analytical and numerical models.
Although not experimentally determined it is estimated that cases 1, 4 and 6 are close to
the ballistic limit of the armor. The residual velocity versus impact velocity for the analytical
model is shown in Fig. 8 for the 41.7 g projectile and 20 mm thick ceramic, the same
thickness as that of the Dyneema composite. The numerical calculations are also shown in
this figure. It is observed that the results are very similar except where the ballistic limit is
approached. Near the ballistic limit, residual velocity versus striking velocity response is
very sensitive to initial conditions, e.g. a small change in the striking velocity results in
a large change in the residual velocity. The slope of the versus curve at the ballistic
3 4
limit is, in fact, theoretically infinite. This is also clearly observed in the experiments
reported by Anderson et al. [21]. Near the ballistic limit the details of the failure process
470 I. S. C. Benloulo and V. Sanchez-Galvez
Fig. 9. Velocity history of the rear of the projectile, calculated numerically and analytically.
become increasingly important. Hence, because the numerical and analytical models have
different failure models it is expected that near the ballistic limit the results will not match so
well. But apart from the cases close to the ballistic limit, the analytical predictions are quite
good considering the number of simplifications. This could permit an easy design of the
light-weight armor as hundreds of cases may be analyzed in a short period of time.
The numerical simulation also allows the analysis of the history of stresses, displace-
ments, velocity, etc. For instance, Fig. 9 plots the velocitytime history of the projectile for
the numerical and analytical models. It is seen that for the first 11 or 12 ls, while the ceramic
is not moving, the velocities match perfectly. This result confirms that the Tate model is very
good for impact of a rod onto a rigid surface. The second phase is when 11 ls(t(20 ls;
the projectile pushes the ceramic cone against the composite. Numerical and analytical
solutions are quite different here. In fact, this is expected because of the derivation of the
analytical composite model: The composite is assumed to fail abruptly at around 20 ls, and
no erosion process has been defined in order to keep the simplicity of the model. On the
other hand, the numerical code evolves more smoothly and naturally due to the (assumed)
erosive process during penetration. Perhaps the next step to take in the analytical model is
to develop a simple erosive model, like the failure of each ply in the composite.
5. CONCLUSIONS
A simple analytical model of high-velocity impact onto ceramic/composite has been
developed. The model is easily encoded, for instance, in FORTRAN, and a complete
impact/penetration problem takes a few seconds to run on a personal computer. The model
is divided into three phases of penetration: intact ceramic, fractured ceramic and initial
response of the composite substrate, and fabric response and failure. The assumptions of
each phase of the model were checked separately, giving reasonable engineering results.
Predictions from the combined model are in relatively good agreement with a limited set of
experimental data. The model was also compared with results from numerical simulations.
There was relatively good agreement on the residual velocity and residual length of the
projectile for most of the cases compared. However, near the ballistic limit velocitywhere
the details of the failure are importantdiscrepancies existed. This is to be expected since
failure is modeled completely differently in the analytical model and finite element ap-
proaches. The approach presented in this paper is a first step in developing an analytical
model of the impact against ceramic/composite armors.
AcknowledgementsAuthors are indebted to Philip Cunniff, Res. Engineer at U.S. Army Natick RD and E Center,
MA, for his help, dedication and useful conversations.
Authors would also like to thank, Dr Charles E. Anderson, Dr James D. Walker, from Southwest Research
Institute, and Prof. Carlos Navarro, from Universidad Carlos III de Madrid, Spain, for their suggestions and
Comision Interministerial de Ciencia y Tecnolog a, Empresa Nacional Santa Barbara and Ministerio de
Educacion y Ciencia for the economical support.
Analytical model for ceramic/composite armors 471
REFERENCES
1. A. L. Florence, Interaction of projectiles and composite armor, Internal Report, US Army, August, 1969.
2. J. F. Mackiewicz, Advanced ceramic/composite armour for the defeat of small arms. Proc. ightweight
Armours Systems Symp. 95, 1995, Paper 24.
3. A. Davey, Ceramic faced bristol armour. Proc. ightweight Armor Systems Symp., Shrivenham 95, UK, June
1995, Paper 23.
4. C. Navarro, M. A. Mart nez, R. Cortes and V. Sanchez-Galvez, Some observations on the normal impact on
ceramic faced armours backed by composite plates. Int. J. Impact Engng, 13(1), 145156 (1993).
5. R. L. Woodward, A simple one-dimensional approach to modeling ceramic composite armour defeat. Int. J.
Impact Engng 9(4), 455474 (1990).
6. Paulus Cornelis den Reijer, Impact on Ceramic Faced Armour. Ph.D. Thesis, Delft University, 1991
7. J. G. Hetherington, The optimization of two component composite armours. Int. J. Impact Engng, 12(3), 1992,
409414
8. R. Zaera and V. Sanchez-Galvez, Analytical model of ballistic impact on ceramic/metal lightweight armours.
Proc. Ballistics 96 Conf., vol. T13 San Francisco, 2328 September 1996, 487495 (1996).
9. Isaias Sidney Chocron Benloulo, J. Rodr guez and V. Sanchez Galvez, A simple analytical model to simulate
textile fabric ballistic impact behavior. extile Res. J. Tentative date: July (1997).
10. I. S. Chocron-Benloulo and V. Sanchez-Galvez, Impact resistance of polymeric matrix composites. Proc.
ICCE/3 Symp., July 1996.
11. A. Tate, A theory for the deceleration of long rods after impact. J. Mech. Phys. Solids, 14 387399 (1967).
12. Z. Rosenberg and E. Dekel, A critical examination of the modified Bernoulli equation using two-dimensional
simulations of long rod penetrators. Int. J. Impact Engng, 15 (5), 711720 (1994).
13. D. Roylance, A. Wilde and G. Tocci, Ballistic impact of textile structures. extile Res. J. 3441 (1973).
14. Philip M. Cunniff, An analysis of the system effects in woven fabrics under ballistic impact. extile Res. J.,
62(9), 495509 (1992).
15. N. Beaumont, Contribution a` lEtude de lImpact dune Bille sur une Plaque en Materiau Composite, Ph.D.
Thesis, 1991
16. C. Navarro, J. Rodr guez and R. Cortes, Analytical modeling of composite panels subjected to impact loading.
J. de Physique I, Colloque C8, supplement au J. de Physique III, 4, 515520 (1994).
17. J. C. Smith, F. L. McCrackin and H. F. Schiefer, Stressstrain relationships in yarns subjected to rapid impact
loading. Part V: wave propagation in long textile yarns impacted transversely. extile Res. J., 288302 (1958).
18. R. A. Prosser, Penetration of nylon ballistic panels by fragment-simulating projectiles. Part I: a linear
approximation to the relationship between the square of the V50 or Vc striking velocity and the number of
layers of cloth in the ballistic panel. Part II: mechanism of penetration. extile Res. J., 6185 (1988).
19. I.S. Chocron Benloulo, J. Rodr guez, M.A. Mart nez and V. Sanchez Galvez, Dynamic tensile testing of aramid
and polyethylene fiber composites. Int. J. Impact Engng 19(2), 135146 (1997).
20 E. H. M. van Gorp, L. L. H. van der Loo and J. L. J. van Dingenen, A model for HPPE-Based lightweight
add-on armour. Ballistics 93 (1993).
21. Charles E. Anderson Jr., Volker Hohler, James D. Walker and Alois J. Stilp, Time-resolved penetration of long
rods into steel targets. Int. J. Impact Engng, 16(1), (1995) 118 (1995).