Theoretical and Applied Fracture Mechanics 33 (2000) 185±190

www.elsevier.com/locate/tafmec

Optimization of two component ceramic armor for a given

impact velocity
G. Ben-Dor a, A. Dubinsky a, T. Elperin a,*, N. Frage b
a
Department of Mechanical Engineering, The Pearlstone Center for Aeronautical Engineering Studies,
Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel
b
Department of Materials Engineering, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel

Abstract
Using FlorenceÕs model a problem of two-component ceramic-faced lightweight armors design against ballistic
impact is solved. Approximate analytical formulas are derived for areal density and thicknesses of the plates of the
optimal armor as functions of parameters determining the properties of the materials of the armor components, cross-
section and mass of an impactor, and of the expected impact velocity. Ó 2000 Elsevier Science Ltd. All rights reserved.

1. Introduction shield [10]. Similar calculations for ceramic/GFRP

Ballistic perforation of multi-layered plates has
been a subject of intensive research during recent
years since non monolithic con®gurations are
considered feasible for designing shields or ele-
ments of the shields. Simpli®ed analytical models
were derived and used for the analysis and opti-
mization of the shields consisting of the layers
manufactured from di€erent materials, e.g., ductile
multi-layered shields [1±7], aluminum/Lexan com-
binations [8], ceramic-faced armors [9±19]. Since
the model [10] provides a relatively simple expres-
sion for the ballistic limit velocity, it is widely used
for the investigation of the ballistic properties of
two-component ceramic armors [11±13,15,16]. The
model [10] was used to determine the armor with
the minimum areal density with an example for
the numerical calculation of a ceramic/aluminum
*
Corresponding author. Fax: +972-7-6472813990.
E-mail address: elperin@menix.bgu.ac.il (T. Elperin).
(glass®bre reinforced plastic) shield can be found
also in [11]. The problem of designing an armor
with the maximum ballistic limit velocity was in-
vestigated analytically in the case when the areal
density is given in [14] and in the case when the
total thickness of the armor is given in [17].
In this study the problem of designing an armor
with the minimum areal density which was for-
mulated in [10], is completely investigated for an
arbitrary two-component armor, and the approx-
imate analytical solution is found.
2. Model description
Consider a normal impact by a rigid projectile
on a two-layer composite armor consisting of a
ceramic front plate and a ductile back plate. In this
study, the following model is employed:
ae2 r2 h2 z‰…q1 h1 ‡ q2 h2 †z ‡ mŠ
v2 ˆ ;
0:91m2 …1†
z ˆ p…R ‡ 2h1 †2 ;

0167-8442/00/$ - see front matter Ó 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 6 7 - 8 4 4 2 ( 0 0 ) 0 0 0 1 3 - 6
186 G. Ben-Dor et al. / Theoretical and Applied Fracture Mechanics 33 (2000) 185±190

where v is the ballistic limit velocity, m the pro- It is important to emphasize that the dimen-
jectile's mass, R the projectile's radius, h1 , h2 are sionless areal density A is a function of one vari-
the plate's thicknesses, r the ultimate tensile able h1 and depends on only two parameters, q1
strength, e is the breaking strain, q is the density,  If
and w.
subscripts 1 and 2 refer to a ceramic plate and a
hopt
1 ˆ u1 … 
q1 ; w† …8†
back plate, respectively.
For a ˆ 1 Eq. (1) describes the model suggested provides the minimum A,  then the dimensionless
in [10] as it is re-worked in [11]. We generalized minimum areal density Aopt and the optimal ratio
slightly this model introducing a coecient a of the areal density of the second plate to the areal
which can be determined using the available ex- density of the ®rst plate are also functions of q1
perimental data in order to increase the accuracy and w:
of the predictions. h i
Expressions for the areal density A of the armor Aopt ˆ A u1 …  q1 ; w
q1 ; w†;  ˆ uA … 
q1 ; w†; …9†
reads
A ˆ q 1 h 1 ‡ q2 h 2 : …2† q2 hopt q hopt Aopt
2
ˆ 2 2opt ˆ ÿ1
The objective of the present study is to ®nd the q1 h1
opt
q 1 h1 q1 hopt
1

thicknesses of the plates h1 , h2 which provide the uA … 

q1 ; w†
ˆ ÿ 1 ˆ u0 … 
q1 ; w† …10†
minimum areal density of the armor A for a given q1 u1 … 
q1 ; w†
ballistic limit velocity v .
Introduce the following dimensionless variables and

hi pR3 qi q2 hopt opt ÿ q1 hopt

2 ˆ A 1
hi ˆ ; qi ˆ ; i ˆ 1; 2;
R s m ˆ uA …  ÿ q1 u1 …
q1 ; w† 
q1 ; w†
…3†
0:91q2 pR2 A ˆ u2 … 
q1 ; w†: …11†
w ˆ v ; A ˆ :
ae2 r2 m
The above property of the model, i.e., the ex-
Then Eqs. (1) and (2) can be rewritten, respec- istence of only two independent dimensionless
tively, as follows: parameters, allows us to investigate the problem
h  i completely and for a general case, namely, for
 2 ˆ q2 h2z q1 h1 ‡ q2 h2 z ‡ 1 ;
w …4† arbitrary combination of materials of the plates.
The results for a given materials can be determined
A ˆ q1 h1 ‡ q2 h2 ; …5† using the solution obtained in dimensionless form:
where hopt 
1 ˆ Ru1 …q1 ; w†;
z  2
m
z ˆ 2 ˆ 1 ‡ 2h1 : …6† hopt
2 ˆ u … 
q ; w†;
pR pR2 q2 2 1 …12†
Eq. (4) is a quadratic equation with respect to m
Aopt ˆ 2 uA … 
q1 ; w†;
q2 h2 Solving this equation and substituting the pR q2
obtained expression for q2 h2 into Eq. (5) yields where superscript opt indicates the corresponding
r

2 optimal parameters, q and w  are determined by
 
 q1 h1z ÿ 1 ‡ q1 h1z ‡ 1 ‡ 4w 2 Eq. (3).
 
A h1 ; q1 ; w
 ˆ :
2z In order to elucidate the analysis based on the
…7† dimensionless variables we will refer (where it is
possible) to the special kind of the armor which
Thus the problem is reduced to ®nding a positive will be referred to as a ``basic armor'' (BA). To this
 The value h2 can
h1 that provides the minimum A. end, select the ceramic/GFRP armor, and use the
be then found from Eq. (5). experimental data in [11] for perforation of the
 G. Ben-Dor et al. / Theoretical and Applied Fracture Mechanics 33 (2000) 185±190 187

armors with di€erent thicknesses of the plates by a 3. Optimal armor

0.50 inch projectile. For BA a change from the
dimensionless to the dimensional parameters can Now we will analyze the dependence on A on h1 .
2
be performed for areal density A…kg=m †, widths  ˆ …q ÿ 1†=2 where
Eqs. (6) and (7) imply that A…0†
of the plates hi (mm) and the ballistic limit velocity p
v (m/s) follows: A ˆ 370A;  hi ˆ 6:35hi ; v ˆ 133w
 q ˆ 4w  2 ‡ 1 > 0: …14†
q1 ˆ 0:060 corresponds to q1 ˆ 3499 kg=m3 ).
…  is independent on h1 When h1 ! 1
The value A…0†
The approach suggested in [11±13] is based on function A…  h1 † attains the asymptotic value
the assumption that the momentum transferred  
A ˆ q1 h1 . The derivative
from projectile to the armor is independent on the
impact velocity, while the model in [10] is used for dA
A0 ˆ 
calculating the ballistic limit velocity. This ap- dh1
  p 
proach implies the following correlation between q 1z ‡ 1 q1z z ÿ 4 ÿ 16w
h 2
the impact velocity vimp , the residual velocity vres q 2 1
ˆ 1 ‡ p ‡ r
2 
and the ballistic limit velocity v 2 z z p
2 z q1 h1z ‡ 1 ‡ 4w 2
vimp vres
ÿ ˆ 1: …13† …15†
v v
is too involved for the complete analytical analy-
If the experimental data on vimp and vres are sis. Consider its behavior for h1 ˆ 0
available, Eq. (13) can be used for ®nding the best
q1 ÿ 4 ÿ 16w 2
value of the parameter a in Eq. (1) and for deter- 2A0 …0† ˆ q1 ‡ 4 ‡p
mining the corresponding ballistic limit velocities. 1 ‡ 4w 2
Certainly, it is assumed that a is constant for a 2
q1 ‡ 4†q ‡ q1
ÿ4q ‡ …
given combination of the materials of the plates. ˆ : …16†
q
For the data presented in [11], it is found that
a ˆ 0:90. Comparison of Eq. (13) with the exper- The positive solution of the equation
imental data is showed in Fig. 1 where solid circles ÿ4q2 ‡ …q1 ‡ 4†q ‡ q1 ˆ 0 reads
correspond to the experimental vimp and vres [11]  q
together with the calculating v using Eq. (1). q0 ˆ 0:125 q1 ‡ 4 ‡ q21 ‡ 24 q1 ‡ 16 : …17†

Fig. 1. Comparison of theoretical predictions with experimen-
tal data from [11] combined with the model given by Eq. (1).
Thus
A0 …0† 6 0 if q P q0 …18†
and
A0 …0† P 0 if q 6 q0 : …19†
The condition given by Eq. (19) is satis®ed only for
very small ballistic limit velocities (e.g., for BA for
v < 17 m=s). Therefore, consider the case when a
condition expressed by Eq. (18) is satis®ed.
The typical behavior of function A…  h1 † is
showed in Fig. 2 for q1 ˆ 0:06 and di€erent w 
where solid circles are plotted using experiments
[11]. The minimum of function A…  h1 † provides the
solution of the problem. However, the variation is
quite small in the neighborhood of the minimum
(see also [11]). This means that the thickness of the
188 G. Ben-Dor et al. / Theoretical and Applied Fracture Mechanics 33 (2000) 185±190

Fig. 2. Areal density of armor vs. thickness of ceramic plate;

Fig. 4. Optimal thickness of ceramic plate vs. ballistic limit
solid circles ± experimental data from [11] combined with model
velocity.
given by Eq. (1).

ceramic plate may be changed in the vicinity of the areal densities of the plates in the optimal armor
optimal value without considerable loss in areal A2 =A1 is plotted. Inspection of this ®gure shows
density.  the ratio A2 =A1 is
that even for a relatively small w,
The solution of the optimization problem in a close to a constant value 0.30 (for BA it corre-
graphical form is showed in Figs. 3 and 4. Useful sponds to v ˆ 400 m=s). Thus, for w  P 4 the ratio
conclusions about the properties of optimal armor of the thicknesses of the plates in the optimal ar-
can be made using Fig. 5 where the ratio of the mor is inversely proportional to the ratio of their
densities:
hopt
2 q1
opt  0:3 : …20†
h1 q2

Fig. 5. Optimal ratio of the areal densities of the plates vs.

Fig. 3. Areal density of optimal armor vs. ballistic limit velocity. ballistic limit velocity.
 G. Ben-Dor et al. / Theoretical and Applied Fracture Mechanics 33 (2000) 185±190 189

The families of curves plotted in Figs. 3 and 5 dure must be adjusted for testing the optimality of
can be approximated with the average accuracy the obtained solution.
of 3% in the range 0:04 6 q1 6 0:1, 1 6 w
 6 10 as Investigations of Functionally Graded Com-
follows: posites (FGC) have received considerable atten-
tion during recent years (see, e.g., [20]). The results
Aopt ˆ uA …  ˆ …0:04 ‡ 1:12
q1 ; w†  0:425 ;
q1 †w …21†
of our research in this ®eld [21, 22] show that
analysis of two-component materials can be useful
q2 hopt
2 for optimization of FGC.
ˆ u0 …
q1 ; w†  ÿ1:47 :
 ˆ 0:29 ‡ …0:1 ‡ q1 †w …22†
q1 hopt
1

Using Eqs. (9)±(12) hopt

1 and hopt
2 can be expressed Acknowledgements
through functions uA and u0 :
u … 
q ; w† We are grateful to Prof. A.L. Florence (Inter-
hopt
1 ˆ u1 …  ˆ
q1 ; w† h A 1 i national Stanford Research Institute) and Prof. R.
q1 u0 …  ‡1
q1 ; w† Zaera (Carlos III University, Madrid) for providing
…0:04 ‡ 1:12
q 1 †w 1:895 us with their research reports and reprints of their
ˆ  ; …23† papers. This study was partially supported by the
q1 q1 ‡ 1:29w
 1:47 ‡ 0:1 Israel Ministry of Science (Grant No. 8450-1-98).

1 u …  0 …
q ; w†u 
q1 ; w†
hopt
2 ˆ u …  ˆ Ah 1
q ; w† i: …24† References
q2 2 1 q u …  ‡1
q ; w†
2 0 1
[1] V.N. Aptukov, A.V. Murzakaev, A.V. Fonarev, Applied
Taking into account Eq. (12) the expressions for Theory of Penetration [Prikladnaja Teorija Pronikanija],
the solution given by Eqs. (21)±(24) provides Moscow, Nauka, 1992 (in Russian).
characteristics of the optimal armor in term of the [2] G. Ben-Dor, A. Dubinsky, T. Elperin, Optimal 3D
impactors penetrating into layered targets, Theoret. and
parameters determining the properties of the ma- Appl. Fract. Mech. 27 (3) (1997) 161±166.
terials of the armor components, its cross-section [3] G. Ben-Dor, A. Dubinsky, T. Elperin, Some ballistic
area, mass of the impactor and the expected im- properties of non-homogeneous shields, Composites A 30
pact velocity. (6) (1999) 733±736.
[4] G. Ben-Dor, A. Dubinsky, T. Elperin, On the order of
plates providing the maximum ballistic limit velocity of a
layered armor, Int. J. Impact Eng. 22 (1999) 741±755.
4. Concluding remarks [5] G. Ben-Dor, A. Dubinsky, T. Elperin, E€ect of air gap
and of order of plates on ballistic resistance of two-
It is showed that the solution of the optimiza- layered armor, Theoret. Appl. Fract. Mech. 31 (1999)
233±241.
tion problem for a two-layer armor can be pre- [6] G. Ben-Dor, A. Dubinsky, T. Elperin, Optimization of
sented in terms of dimensionless variables whereby layered shield with a given areal density, Int. J. Fract. 91
all the characteristics of the impactor and the ar- (1998) L9±L14.
mor are expressed as a function of two indepen- [7] G. Ben-Dor, A. Dubinsky, T. Elperin, The optimum
dent dimensionless parameters. The latter allows arrangement of the plates in a multi-layered shields, Int. J.
Solids and Struct. 37 (2000) 687±696.
to present the solution of the optimization prob- [8] J. Radin, W. Goldsmith, Normal projectile penetration
lem for an arbitrary two-component composite and perforation of layered targets, Int. J. Impact Eng. 7
armor in analytical form which is convenient for (1988) 229±259.
practical applications. [9] I.S. Chocron Benloulo, V. Sanchez-Galvez, A new analyt-
ical model to simulate impact onto ceramic/composite
The validity of the result obtained using the
armors, Int. J. Impact Eng. 21 (6) (1998) 461±471.
above simpli®ed model must be investigated ex- [10] A.L. Florence, Interaction of projectiles and composite
perimentally for di€erent materials applied for armor. Part 2. Stanford Research Institute, Menlo Park,
armor component. Clearly, experimental proce- AMMRC-CR-69-15, August 1969.
190 G. Ben-Dor et al. / Theoretical and Applied Fracture Mechanics 33 (2000) 185±190
[11] J.G. Hetherington, B.P. Rajagopalan, An investigation
into the energy absorbed during ballistic perforation of
composite armors, Int. J. Impact Eng. 11 (1) (1991) 33±40.
[12] J.G. Hetherington, B.P. Rajagopalan, Correspondence on
an investigation into the energy absorbed during ballistic
perforation of composite armors, Int. J. Impact Eng. 12 (2)
(1992) 325±327.
[13] J.G. Hetherington, B.P. Rajagopalan, Energy and momen-
tum changes during ballistic perforation, Int. J. Impact
Eng. 18 (3) (1996) 319±337.
[14] J.G. Hetherington, The optimization of two-component
composite armors, Int. J. Impact Eng. 12 (1992) 229±259.
[15] C. Navarro, M.A. Martinez, R. Cortes, V. Sanchez-
Galvez, Some observations on the normal impact on
ceramic faced armors backed by composite plates, Int. J.
Impact Eng. 13 (1) (1993) 145±156.
[16] B.P. Rajagopalan, J.G. Hetherington, Characterization of
ceramic/steel and ceramic/aluminum armors subjected to
oblique impact, Int. J. Impact Eng. 19 (9-10) (1997) 811±819.
[17] B. Wang, G. Lu, On the optimization of two-component
plates against ballistic impact, J. Mater. Process. Technol.
57 (1996) 141±145.
[18] R.L. Woodward, A simple one-dimensional approach to
modeling ceramic composite armor defeat, Int. J. Impact
Eng. 9 (4) (1990) 455±474.
[19] R. Zaera, V. Sanchez-Galvez, Analytical modeling of
normal and oblique ballistic impact on ceramic/metal
lightweight armors, Int. J. Impact Eng. 21 (3) (1998)
133±148.
[20] M. J. Pindera, J. Aboudi, S. M. Arnold, Recent advances
in the mechanics of functionally graded composites, in:
E.A. Thornton (Ed.), Aerospace Thermal Structures and
Materials for a New Era, Progress in Astronautics and
Aeronautics, vol. 168, 1995, pp. 181±203.
[21] Y. Seidman, N. Frage, M.P. Dariel, Graded TiB2 /TiC
composites in®ltrated with liquid aluminum, 101st Annual
Meeting of the American Ceramic Society, 25±28 April
1999, Indianapolis, Indiana. Abstracts, p. 229.
[22] S. Sabatello, N. Frage, M.P. Dariel, Graded cermets based
on the non-stoichiometric TiC in®ltrated with a Fe±C
alloy, 101st Annual Meeting of the American Ceramic
Society, 25±28 April 1999, Indianapolis, Indiana. Ab-
stracts, p. 254.