Int. J. Solids Structures Vol. 29, No.4, pp. 39<1-420, 1992 0020-7683/92 $5.00+.

00
Printed in Great Britain. Pergamon Press pic

PENETRATION OF LAMINATED KEVLAR BY

PROJECTILES-I. EXPERIMENTAL INVESTIGATION

GUOQI ZHU,t WERNER GOLDSMITH and C. K. H. DHARAN

Department of Mechanical Engineering, University of California, Berkeley, CA 94720, U.S.A.

(Received 18 January 1991; in revisedform II June 1991)

Abstract-The response of woven Kevlar/polyester laminates of varying thicknesses to quasi-static

and dynamic penetration by cylindro-conical projectiles with an apex of primarily 60° has been
investigated. The static properties of the material were obtained by means of standard testing
procedures, including in-plane tension, compression and shear, through-thickness compression,
fracture toughness and cone indentation. Compressive dynamic results were obtained using a
Hopkinson bar at rates up to 2000 S-I . Quasi-static force-indentation behavior for conical indenters
was determined for a specimen thickness ranging from three to 10 layers to provide constitutive
relations for the corresponding analytical model. These data exhibit a linear rise whose slope and
peak values are proportional to specimen stiffness, followed by a plateau that terminates in a rapid
reduction to a small, but finite force level when the shank of the penetrator exits. The curves were
found to be highly sensitive to the size of the projectiles and the tip angle.
Static investigations were also conducted to assess the effect on penetration of specimens
subject to special artificial initial conditions. This included a restraint on the global plate defor-
mation; the use of several samples of the same total thickness, but comprised of several assemblies
of stacked laminates; and embedment of an artificial delamination flaw located centrally at the mid-
plane of the plate over approximately half of its diameter.
The dynamic tests were executed using both pneumatic and powder guns, mostly with a
12.7 mm barrel diameter. Ballistic limits were determined for these plates and terminal velocities
were measured when perforation occurred. Deliberately introduced delaminations and changes in
the volume fraction did not result in significant changes in the impact resistance. The damage zone
created was square in shape for a 0/90 lay-up and circular for quasi-isotropic specimens. The damage
pattern for dynamic loading was, however, quite different from that for the corresponding quasi-
static penetration case. Kevlar laminates were found to be very resistant in arresting blunt-nosed
strikers in comparison with metals on a specific weight basis.

INTRODUCTION

Laminated composites have many applications; one ofthe most important and widely used
of these materials employs Kevlar which frequently serves as a shield in a variety of
environments, such as helicopters, tanks, personnel carriers and body armor. Its primary
function in these installations is the prevention of perforation of the protected surface by
fragments travelling at less than ballistic speed. In consequence, an understanding of the
behavior of Kevlar plates under impact conditions and, in particular, the conditions for
perforation (ballistic limit) are of critical importance in the design of suitable armor.
While some information has been developed for high-speed impact, the majority of
previous investigations involving impact of Kevlar and other laminated composites has
focused on the low velocity regime, up to lO m s- I (ASTM, 1975; Greszczuk, 1982;
Shivakumar et aI., 1985a, b). In this domain, it has been shown that among a variety of
composites, those containing Kevlar 29 had the highest energy absorption capability
(Wardle and Tokarsky, 1983; Wardle, 1982; Wardle and Zahr, 1987). Further, when a
sphere is used at these speeds, the failure mechanisms of matrix cracking and delamination-
with the maximum delamination shown to be linearly related to impact energy (Hong and
Liu, 1989)-are of primary concern, without any consideration of fiber failure. The latter
becomes critical at higher velocities when the possibility of perforation exists; few investi-
gators have been interested in this regime. Cristescu et al. (1975) were concerned with the
failure mechanisms of shear plugging, fiber debonding, stretching and breaking, and matrix
deformation in cracking produced by the impact of a blunt projectile. Experimental investi-
gations of matrix cracking (Takeda et al., 1982a, 1987), delamination extension (Takeda

t Current address: CPC Group, General Motors Corporation, Warren, MI, U.S.A.

SAS 29:4-A 399

400 G. ZHU et al.

et al., 1982b) and wave propagation due to ballistic impact by blunt-faced cylinders (Takeda
et al., 1981), have been conducted. Matrix cracking in laminates struck by a sphere was
studied by Joshi and Sun (1985) and by Liu and Malvern (1987).
Ballistic performance data for Kevlar have been reported by Du Pont (a, b). This
information is based on standardized circular cylindrical fragments with a diameter and
height of 8.75 mm and a weight of 64 grains. No data have been obtained for projectiles
with conical tips. Long-rod and jet penetration of Kevlar has been studied at speeds of 1.5
km S-1 and the deformation mechanisms believed to have occurred in the process have
been described (Scott, 1990).
The objective of the present paper is the experimental investigation of the static
and dynamic penetration and perforation response due to cylindro-conical projectiles of
laminated Kevlar plates. Impact speeds up to 204 m S-1 were employed. The quasi-static
properties of the substance were determined for in-plane tension, compression and shear,
through-thickness compression, fracture toughness and cone indentation. Compression
Hopkinson-bar tests with strain rates up to 2000 S-1 were also conducted. The study
required the fabrication of samples of various thicknesses, fiber orientations and volume
fractions, and with various deliberately embedded flaws. The effect of multiple adjacent
layers, compared to the response of a simple homogeneous target of the same total thickness,
was also investigated. Material damage, resisting force, ballistic limits and terminal velocities
in the penetration tests were carefully examined and, where applicable, compared to the
predictions of a phenomenological model described in a companion paper (Zhu et al.,
1991).

SPECIMEN FABRICATION

The composites fabricated for this study consisted of woven Kevlar 29 fiber plies in a
thermosetting polyester matrix. Although its modulus is lower than Kevlar 49 (used in
structural composites), Kevlar 29 has superior energy absorption qualities.
Laminated plates of Kevlar 29 woven fiber (style 735) in a room temperature curing
polyester matrix were manufactured by wet lay-up techniques. This is a square-weave fabric,
i.e. it consists of an equal number of fibers along the two orthogonal directions woven in
a 2-harness satin style, where each bundle of fibers loops alternately over and under two
bundles in the orthogonal direction. The thickness of the fabric was approximately
0.61 mm, with an areal density of 491 g cm- 2 •
The resin matrix employed was a low viscosity thermosetting polyester resin commonly
used for hand lay-up at room temperature. This resin is cured at 90°C and is designed to
wet easily the reinforcement fabrics employed in hand lay-up construction. The catalyst
used was methyl ketone polymer (MEKP). Table 1 lists the material properties of Kevlar
fiber and polyester resin.
Frabrication was conducted by first cutting 28 cm x 28 cm sections of the Kevlar sheets.
Laminates were constructed with fibers in each ply oriented in the same direction (0/90)
and with alternating woven plies oriented at 0° and 45° (0/90/±45, quasi-isotropic). The
fabric plies were placed in a mold consisting of steel plates separated by a window frame
that set the required final thickness of the laminate. The bottom steel plate, which was first
coated with a release agent, was wet thoroughly with the catalyzed polyester resin before
the first ply was placed on it. More catalyzed resin was applied to this first ply with a brush
until it was thoroughly wet. Following this, the remaining plies were placed in the mold
following the same sequence.

Table I. Material properties of Kevlar fiber and polyester resin

Kevlar 29 Kevlar 49 Polyester

Density (g em - 3) 1.44 1.44 1.12-1.46

Tensile modulus (GPa) 62 125 1.03-4.48
Tensile strength (MPa) 2758 2800-3600 28-69
Elongation to break (%) 4.0 2.4 \.7-2.6
 Penetration of laminated Kevlar by projectiles-I 401

Since the mold contained the window frame, which determined the final thickness of
the laminate (and thus the final fiber volume content), there was no need to control the
amount of resin placed on the fabric plies. Once all the plies were placed, the upper steel
plate (also treated with release agent) was placed on top of the stack. The assembly was
then inserted between the platens of a microprocessor-controlled heated platen press, a
Pasadena Hydraulics model with 45 cm x 45 cm heated platens capable of a maximum
compression force of 2 x 10 5 N. The platens were heated to 90°C and sufficient pressure
was applied until the steel plates bottomed out on the window frame. The cure time was
two hours to ensure complete cross-linking. The fiber volume fraction was determined to
be 55% (based on weighing the fabric before impregnation, and the laminate after curing,
and on using the weight density of the fiber and the resin).
The specimens with embedded delaminations were manufactured in a similar fashion
except for the introduction of a circular insert of a 0.1 mm thick fluoropolymer film at the
mid-plane of the laminate. All other parameters including cure temperature and fiber
volume fraction were unchanged. The neat resin sample was made by mixing the resin with
the catalyst and pouring the mixture into a 3.2 mm thick mold, which was then cured in an
oven at 200°e.
An abrasive water jet was used to cut Kevlar 29/polyester matrix samples with clean
edges, required for mechanical property experiments. For the static and dynamic penetra-
tion tests, where a clean edge was not essential, the laminated plates were cut into 14 cm
diameter circular plates using a diamond band saw and then finished by abrasive grinding
of the edges.

MATERIAL CHARACTERIZAnON TESTS

(a) In-plane tension and compression moduli, and Poisson's ratio measurements
In-plane tension and compression moduli were determined by using strain gages and
loading the laminate in an Instron testing machine at rates ranging from 1O~ 3-10~ 1.
Specimen dimensions were chosen for convenience as 6.3 mm (10 plies) thick and 25.4 mm
wide with gage lengths of 76.2 mm and 25.4 mm for tension and compression, respectively.
Poisson's ratio was found by using a strain gage at 90° to the loading direction in addition
to a longitudinal strain gage. An electro-optical displacement transducer (OPTFOLLOW),
described in more detail subsequently, was also used to measure the specimen elongation.
The results of the tension tests are shown in Fig. 1. The initial slope of the 0/90
laminate, about 19 GPa, decreases with further straining. This is a common observation in
cross-plied laminates and is attributed to the onset of resin microcracking. Further loading
results in a small increase in the modulus, possibly due to straightening of the woven fibers
in the laminate. The average tangent Young's modulus over the range ofloading employed
in these tests was found to be 7 GPa.

300

250

~ 200
0..
~
III 150
III
~
Ci5 100
0/90 with OPTFOLLOW
50 0/90 with strain gage

2 3 4 5 6
Strain (%)
Fig. I. Quasi-static in-plane tension test results for 0/90 laminates measured with strain gages or
the OPTFOLLOW system. Specimens have a gage length of 76.2 mill and a thickness of 6.35 mm.
402 G. ZHU et al.

The rule of mixtures, which is an upper bound of the longitudinal modulus, EI> has
been found to show good agreement with measured values in continuous fiber-reinforced
laminates. It is given by the expression,

(1)

where Er is the longitudinal tensile modulus of the fiber, Em is the tensile modulus of the
matrix and Vr is the volume fraction of the fiber in the composite. For VI = 0.55, the value
of Er = 62 GPa was obtained from Saghizadeh and Dharan (1986) and the value of Em = 1.8
GPa was an average measurement. The longitudinal modulus E1 is 3.49 GPa. With this
value for a 0/90 weave, the laminate modulus is 17.7 GPa which is in good agreement with
the measured initial slope modulus of 19 GPa. The Poisson ratio, VLT of the laminate, was
measured as 0.25, using longitudinal and transverse strain gages.
The shear modulus GLT can be calculated from the data resulting from tensile tests
conducted on 0/90 and ±45° laminates using the relation (Jones, 1975),

(2)

The Young's modulus in the 45° direction, E 45 , was measured to be 2.8 GPa. Since EL = ET
is taken as 18.5 GPa (approximately the average of the two measured values), the shear
modulus GLT of the laminates was determined to be 0.77 GPa. In-plane tension test data
at 0/90 and at ±45 are shown in Fig. 2.
The stress-strain curve for in-plane compression of 0/90 25.4 mm long samples using
the Instron tester is shown in Fig. 3. The response is typical of a composite material in which
the reinforcement phase has low compressive strength resulting in non-linear behavior.
Observation of the failed specimens indicated that failure was caused by local buckling of
the fibers. The poor compressive behavior and fiber buckling induced failure modes of
Kevlar laminates are well known.

(b) Through-thickness compressive strength and modulus measurements

Through-thickness compression tests of neat polyester resin and Kevlar/polyester
composites were conducted on an MTS testing system using 3.175 mm diameter, 6.35 mm
thick disks, at strain rates ranging from 2.7 x 10- 3 S-1 to 1.3 x 10 S-I. The polyester disks
were made with a core drill, while the composite disk was cut with an abrasive waterjet.
The specimens were compressed by a 6.35 mm diameter punch lubricated with a Teflon
coating.

300

iii
n. 200
~
III
III
------ 0/90
±.45

GI
.:
(/)
100

4 6 8 10
Strain (%)
Fig. 2. Quasi-static in-plane tension data of laminates with 0/90 lay-up. Specimens have a gage
length of 76.2 mm and a thickness of 6.35 mm.
 Penetration of laminated Kevlar by projectiles-I 403
60,---------------~~--------------~

50

40

30

20

10

2 4 6 8
Strain ("II.)
Fig. 3. Quasi-static in-plane compression results for 0/90 laminates. The specimen is 25.4 mm long
and 6.35 mm thick.

The polyester resin exhibited strong rate dependence, as shown in Fig. 4. The yield stress
increased 52.7% over the strain rate from 0.002 to 0.3 S-l. However, the Kevlar/polyester
laminates exhibited a much weaker strain rate dependence, as shown in Fig. 5, for through-
thickness specimens, with only a 5.5% increase when the strain rate was changed over the

---
--
300
strain rate-o.131s
strain rate-0.01/s
strain rate-O.002ls
200 strain rate-1500/s
iii" strain rate-2000/s
Q.
~
<II
<II
!
en 100

0.1 0.2 0.3 0.4

Strain
Fig. 4. Quasi-static and dynamic compression test results for polyester resin.

600

500

iii" 400
Q.
~

--
<II 300

--
<II

--
!
en 200 strain rate-O.131s
strain rate-o.OO27/s
100 strain rate-1500/s
strain rate-2000/s
0
0.0 0.1 0.2 0.3 0.4 0.5
Strain
Fig. 5. Comparison of quasi-static and dynamic through-thickness compression test results for
Kevlar/polyester laminates.
404 G. ZHU et al.

same range. Corresponding dynamic data from Hopkinson-bar tests at strain rates up to
2000 s- 1 are also shown in these figures. The viscoelastic nature of both materials is
evidenced by the substantially higher stresses in the dynamic response.
Data for dynamic loading of in-plane specimens is expected to exhibit a much lower
rate effect as fiber behavior, which is essentially elastic, dominates specimen responses.
Visual inspection of the crushed specimens of the polyester resin revealed that the material
failure was due to through-thickness inward propagating radial cracks which were formed
by circumferential tension. In composite specimens, the Kevlar fiber reinforcement inhibited
the formation and propagation of the through-thickness cracks. Double cantilever beam
specimens 25 mm wide, 3.2 mm thick, 215 mm long were used to determine the Mode I
delamination using a Teflon tape generated initial mid-plane delamination 50.8 mm long.
The average fracture toughness was found to be 455 ± 50 J m - 2; this can be compared to
values of 330 J m- 2 for Kevlar 49/epoxy and 525-1020 J m- 2 for glass-epoxy (Saghizadeh
and Dharan, 1986).

EXPERIMENTAL ARRANGEMENT AND PROCEDURE

(a) Quasi-static tests

Quasi-static tests on Kevlar/polyester laminates were conducted at loading rates from
10- 3 to 10- 1 s- 1 to ascertain the effect of thickness, separation layers and projectile
diameter on penetration characteristics and to compare observed failure phenomena with
those obtained under dynamic conditions. An MTS closed-loop servo-hydraulic testing
system using a load cell with a capacity of 400 kN was employed for force application. The
moving crosshead could be stopped manually or by setting a voltage limit in the controller
software which is run on an IBM-AT personal computer that also samples the data.
All penetrators consisted of hard-steel cylindrical projectiles with diameters of 12.7
(the standard size), 9.525 or 6.35 mm and a conical or blunt tip; cone angles employed
were 60°, 90° or 120°. The various circular Kevlar specimens of 139 mm diameter were
clamped in a fixture on a 114 mm diameter frame and the unit was placed on the base of
the MTS machine. Plates consisted of laminates ranging from 5 to 24 layers (3.1 mm to 15
mm thick) arranged in 0/90 and 0/45 lay-ups. These included a combination of multiple
plates, as well as specially prepared samples, some artificially delaminated over a portion
of the surface, while others were sandwiched or backed by steel for a particular investigation.
The testing machine head was lowered until the penetrator could be positioned so that its
tip was just touching the center of the sample, while its flat rear surface was in contact with
the head of the machine; the load was then applied. The tests were displacement controlled
so that the punch could be stopped at different positions for an inspection of material
damage. The load cell provided a readout of the applied force history, while displacement
data were obtained from the cross head speed.

(b) Dynamic investigation

The dynamic penetration tests utilized 12.7 mm and 9.5 mm diameter cylindro-conical
oil-hardened steel projectiles (62 RJ with a 60° tip angle, lengths of 38.1 and 34 mm and
masses of 28.9 and 15.4 g, respectively, to determine the ballistic limits of various Kevlar
composite combinations or the terminal velocity when complete perforation occurred. A
single test utilizing a flat-nosed striker was conducted to portray the difference in target
response between a blunt and a sharp-nosed projectile.
The propulsion equipment consisted of a pneumatic and two powder guns, all mounted
on a heavy steel table located in an isolation chamber requiring remote firing. The pneumatic
gun featured a barrel with an inside diameter of 12.7 mm and a length of 1.3 m mounted
on its support by three brackets. A solenoid valve separated the breech from a reservoir
containing nitrogen stored at the desired pressure by means of a control valve. A maximum
pressure of 102 atm can be achieved, transmitted from the supply cylinder which is charged
to 153 atm. Solenoid activation provides access of the back pressure to the rear of the
projectile and initiates the launch.
 Penetration of laminated Kevlar by projectiles-I 405

Two separate powder guns with barrel diameters of 12.7 and 9.5 mm, respectively,
were connected to the table at the breech by a massive brace. Projectiles for these impact-
activated devices were coated with copper to minimize barrel wear. The breech of the former
can accommodate 20 mm shells, but a 50 caliber shell was used for most of the tests fired
at velocities of about 300 m S-1 and for the single shot conducted with this diameter that
was propelled at 800 m S-I. The smallest-sized gun was adapted from a commercial
Magnum; its breech houses shells that are specifically designed for its bullet size. Rounds
were assembled individually; a steel shell with a volume of 6.48 cc filled with Du Pont IMR
3031 black powder, reusable upon replacement of the primer, was inserted into the breech
just behind the 12.7 mm diameter projectile placed in the barrel. The maximum speed
obtainable with the steel shell is 1000 m S-1 ; substantially higher velocities can be achieved
by use of a brass shell which can accommodate more powder. The 9.52 mm diameter striker
was inserted into the tip of the corresponding brass shell that was then loaded into the
Magnum breech; these shells were used repeatedly upon replacement of the primer.
Most of the targets were made of Kevlar laminates with a 45% matrix volume fraction;
a few consisted of smaller matrix volume fractions of 35% and 26%. The majority of the
targets were fully clamped on a 114.3 mm diameter target holder which was also mounted
on the steel table. A few samples were constrained only over a portion of the periphery to
permit photographic observation of the penetration process. All specimens were located
normal to the projectile trajectory which intersected the target at its center.
The initial velocity of the pneumatically-propelled projectile is measured from the
interruption upon bullet passage of two lasers focused onto two photodiodes, 152 mm
apart, passing through two slots in the barrel on opposite sides of a diameter. These slots
also serve to vent the back pressure and thus prevent further acceleration of the striker.
The change in voltage due to beam extinction is recorded on a Nicolet digital oscilloscope.
In the case of the powder gun, the laser beams are located just ahead of the gun muzzle.
A Beckman-Whitley model WB-2 high speed camera featuring a motor driven rotating
prism, with framing rates ranging from 20,000 to 10 6 frames S-I, was employed to observe
the motion of the projectile and target during penetration. There are 79 separate lens stages
and corresponding light paths focused onto the stationary Kodak Tri-X (400 ASA) 35 mm
film. The motor speed is controlled by a rheostat and continuously displayed on a time-
interval meter from a signal emitted upon passage of a protrusion on the rotor by a
stationary magnet. The light source was a stroboscopic unit with a single flash capability
of sufficient duration to illuminate the process for the duration of a quarter-circle sweep of
the prism, about 1.8 ms. It was activated by a signal obtained from the interruption of the
second laser beam delayed by a device so that the flash duration corresponded to the impact
event.
The final projectile velocity is obtained by means of two aluminum screens spaced
235 mm apart, located just behind the target. Each screen is composed of two aluminum
foils taped on the surface of a 10 mm thick wooden frame. One foil is directly connected
to an oscilloscope, while the second foil of each set is connected serially to a 6 V battery
and then to the oscilloscope. This unit is not subjected to a voltage initially, since the foils
are separated. When the projectile perforates each of the two screens, two foils are connected
by the conductive projectile, and two signals are thus imposed on the recorder, permitting
the calculation of the terminal striker speed.
An OPT FOLLOW 7000C electro-optical displacement systemt was used to track the
motion of the projectile. This unit consists of a camera and a microprocessor control and
monitors the movement of the interface between a black and white region with a digital
output transmitted to an oscilloscope. The device is calibrated quasi-statically by moving
such a boundary in the plane orthogonal to the camera at the same distance as used in the
dynamic test to achieve the proper field of view. In the dynamic application, the camera
follows the tail of the projectile on which the boundary is marked, its path being uniformly
illuminated by two DC lights. A background with a black and white boundary is placed on
the opposite side of the projectile path from the camera. When focused on the path of the

tYaman Ltd., San Jose, CA, U.S.A.

406 G. ZHU et al.

235mm
background
~ ~
~~I
I
II ~
I
I
II
control
device

~~
I I gun
I. I
•f U)
N I

e>~
:

----.
I - I

Pd.
g, / '>\
I •

!j
screens E
Ln\)j
..: DC light ·51
DC light .E
~

.----~=;_-1 oscillo-
scope

Fig. 6. Schematic of experimental arrangement using the OPTFOLLOW system.

projectile, the OPTFOLLOW recognizes the blurry boundary on the background before
the projectile appears in the field of view. The clear boundary on the projectile overlaps the
blurry boundary when the projectile enters the view of the OPTFOLLOW. The unit now
switches to track the motion of the boundary on the projectile, resulting in an electrical
signal. The framing camera and the OPTFOLLOW unit could not be used concurrently.
All operations were performed from outside the chamber. The signal produced by the
interruption of the first laser beam was fed into a Tektronix AM502 differential amplifier
and sequentially to a delay unit whose output was employed to trigger the stroboscope.
The proper delay time was calculated from the anticipated projectile speed and the distance
from the first laser to the target.
Operation of the OPTFOLLOW system required proper setting of the control unit,
the establishment of a suitable field of view and a subsequent initial calibration; two
methods were used for the latter procedure. In the first, a dummy projectile was slowly
moved through the viewing window so that the output could be correlated with the position
of the projectile. The second method involved the firing of the striker through this region
at constant velocity, whose value was obtained from laser measurements. This provided a
calibration factor for the concurrent OPTFOLLOW output. A schematic of the exper-
imental arrangement for the OPTFOLLOW system is shown in Fig. 6.

RESULTS AND DISCUSSION

(a) Static tests

The load-displacement curves of Kevlar/polyester laminates ranging from 5 to 10 plies
perforated quasi-statically by a 60° conically-nosed steel penetrator are presented in Fig.
7; these combine the local indentation and the global deflection of the plate. Both the initial
slope, which is a measure of the global deflection, and the maximum resistance of the
laminates increased with thickness. The first peak on the curves corresponds to the initiation
of fiber failure which occurred on the distal side and is due to local bending and bulging of
the material. An increase in the laminate thickness reduces global deflection for a given
load and delays fiber failure. The thinner plates exhibit a load plateau with small oscillations
occasioned by continuing fiber failure; for the thicker specimens, the load in this plateau
domain relaxes at an increasing rate as the number of layers increases. This is attributed to
the breakage of a larger number of fibers after initial penetration than occurs for laminates
containing a smaller number of plies.
 Penetration of laminated Kevlar by projectiles-I 407

15~ ~---------------------------------,

o 10 20 30
Displacement ofProjec:tilc (mm)
Fig. 7. Load-displacement curves of the quasi-static perforation of Kevlar/polyester laminates with
thicknesses ranging from 5 to 10 plies by a 12.7 mm diameter hard steel cylinder with a 60° tip
angle. The loading speed is 12 mm min - I •

Figure 8 presents four sectioned samples at successive stages of quasi-static loading to

a 1O-ply (6.35 mm thick) laminate by a 12.7 mm diameter cylindro-conical steel projectile
with a 60° tip angle, corresponding to an advance of the projectile in increments of 6.35
mm. These photographs depict three of the four most important responses of the plate
when it is subjected to increasing load by a sharp-tipped penetrator; this behavior is
characteristic of specimens of moderate thickness. Initially, Fig. 8a, only global plate
deflection is observed. In the next stage, Fig. 8b, some fiber failure and bulging at the distal
side is manifested. The next two phases show an increasing pattern of fiber failure until
complete perforation occurs; this process is accompanied by a successively reduced global
deformation. In addition, delamination is present which cannot be distinguished in the
photographs. The dominance of the global deformation process does not have a counterpart
in the dynamic case because the energy of deformation is transformed into that of traveling
waves and the system is not in a state of equilibrium.
The effect of a restraint on global deformation is shown in Fig. 9 where two lO-ply
laminates, clamped around the edges, but unrestrained in back or supported by an adjacent
steel plate with a 25.4 mm central hole, respectively, were perforated by a 12.7 mm cylindro-
conical steel projectile with a 60° tip. The peak load for the two cases is about the same,
but the restrained system exhibits a much lower deflection both at the maximum load and
at complete perforation. Thus, the energy consumed in global deformation of the backed
plate is about 60% of that of the unrestrained case where it represents a significant fraction
of the total consumed during the perforation process.
The quasi-static perforation of several laminates with a total of 24 plies was also
investigated to assess the performance of a homogeneous target relative to that composed
of several plies of the same total thickness. The samples consisted of a single 24-ply laminate
as well as adjacent blocks of 2 x 12, 3 x 8 and 4 x 6-ply laminates; the test results are shown
in Fig. 10. The bending stiffness of the targets decreased with the increasing number of
adjacent, but unconnected layers. At the initiation of perforation, up to which point global
deformation constitutes the major energy absorption mechanism, the four-ply sample
absorbed approximately 68% of the energy of the homogeneous laminate and exhibited
58% of its peak load. These figures provide an idea of the effect of delamination; however,
these values differ substantially from those for the case of pure bending. The reasons for
this discrepancy are largely due to the presence of a sharp tip and the transfer of loading
in the tests from pure bending to in-plane tension.
408 G. ZHU et at.
The quasi-static loading curve for a 10-ply laminate with an embedded delamination
of 85 mm diameter and that of a fully laminated plate when perforated by a 12.7 mm
diameter, 60° cylindro-conical steel projectile are presented in Fig. 11. The loading pattern
was transformed from bending to in-plane tension. The delaminated specimen has a lower
initial slope and a slightly low\!r peak load, corresponding to the initiation of effective fiber
failure, that occurs at a significantly greater displacement. However, the energy absorbed
in view of the more rapid unloading for this case is approximately lO% lower than for the
intact specimen. Both of these results are qualitatively confirmed by the data shown in
Fig. 10.
The size of the penetrator also plays a key role in the response of the composite to
quasi-static loading. Experiments were conducted using 12.7, 6.35 and 3.175 mm steel
cylinders with 60° conical tips to perforate both 20- and lO-ply laminates; the results are
shown in Fig. 12. The initial slope for these three cases is about the same, but the peak
force is roughly proportional to the diameter of the projectile. A comparison of the response
of the lO-ply laminate with that of the 20-ply sample upon loading by the same (12.7 mm)
penetrator indicates the divergence in the slope, previously noted, and a peak force ratio
of one-half, with unloading occurring sooner for the thinner plate.
The effect of the cone angle of a 12.7 mm penetrator on the response of a lO-ply sample
is presented in Fig. 13. The initial slopes for the 60°,90° and 120° tips are almost the same
because global deflections dominate at this stage. The peak force increased with tip angle,
as expected, amounting to a 40% difference over the range investigated with a corresponding
increase in the energy absorbed by the plate. Figure 14 shows the distal side of the sample
upon perforation by the 120° tip projectile, together with the plug produced. This was the
only instance of plugging found in the present investigation. This type of phenomenon is
similar to that observed in metallic targets and some other composites when a blunt
projectile strikes at a speed above the ballistic limit. In contrast, a blunt projectile
did not perforate a lO-ply laminate under quasi-static conditions; the large forces applied
resulted in the specimen being pulled out of the clamp with severe distortion and
delamination.

(c) Dynamic investigations

The ballistic limits of Kevlar laminates with random fiber orientations as a function
of their thickness, a structural property of the system, were determined for both 9.525 and
12.7 mm diameter 60° cylindro-conical steel projectiles with masses of 12.5 and 28.9 g,
respectively. This limit was considered here to have been attained when the projectile was
either stuck in the target as shown in Fig. 15, or else exited with negligible velocity. The
difference in these conditions is considered to be a variation in initial speed of less than 5
m s- I. The results are shown in Fig. 16; within this range, the ballistic limit varies linearly
with laminate thickness and is higher for the smaller striker due to its lower mass. A
similar conclusion was reached by Bless et al. (1985) for glass composites penetrated by a
hemispherically-nosed projectile. The smaller projectile exhibits a slightly steeper slope. The
ballistic limit of 3.175 mm thick 2024-0 aluminum plates struck by the 12.7 mm diameter
cylindro-conical projectile was about 90 m S-I (Bless and Hartman, 1989); on the basis of
areal density, the Kevlar has an equivalent ballistic limit of 110 m S-I, 20% greater than
that for aluminum.
Laminates with various lay-ups as well as with an embedded delamination between
two layers of 40 mm diameter were examined to assess the effects of these factors on
dynamic perforation. The results, indicated in Fig. 17, show the final velocity of the
projectile as a function of the initial speed and indicate that neither lay-up nor prior partial
delamination played a significant role in the perforation phenomenon.
On the other hand, Fig. 11 indicates that the effect of a delamination was a reduction
of lO% in the energy absorption during quasi-static loading. The ballistic limit for a 5-ply
Kevlar plate with a single-level 40 mm diameter delamination was found to be 13% less
than that of a full-strength sample. As the impact speed increases above the ballistic limit,
this difference rapidly disappears, and partly delaminated specimens exhibit the same
resistance as intact samples. As in similar cases of comparison of theory and experiment
 Penetration of laminated Kevlar by projectiles-I 409

Fig. 8. Photographs of the damage in four 6.35 mm Kevlar/polyester laminates due to quasi-static
penetration by a 12.7 mm diameter hard steel cylinder with a 60° tip angle to various end positions
from a fixed reference: (a) 6.35 mm, (b) 12.7 mm, (c) 19.05 mm and (d) 25.4 mm. The loading
speed is 12 mm min-I.
410 G. ZHU et al.

Fig. 14. Damage due to the 120° penetrator whose action is shown in Fig. 12. A plug was generated.

Fig. 15. Photograph of a 20-ply target perforated by a 12.7 mm diameter, 60° cylindro-conical
projectile of 32.2 g mass at the ballistic limit (initial velocity = 170 m S-I).
 Penetration of laminated Kevlar by projectiles- I 411

Fig. 19. Exit side damage in a 10-ply laminate due to perforation by a 60°, 12.7 mm diameter
projectile at an initial velocity of 783 m s - I .

Fig. 20. Cross-sections of a 20-ply laminate perforated by a 60°, 12.7 mm diameter cylindro-conical
projectile at an initial velocity of 236.5 m s-'.
412 G. ZHU et al.

Fig. 21. High-speed photographs of a lO-ply laminate perforated by a 60°, 12.7 mm diameter
cylindro-conical projectile at an initial velocity of 188.6 m S-I. The framing rate is 20.2 Ils/frame.
 Penetration of laminated Kevlar by projectiles-I 413

jlj)ftft ....
. ." . .,.,.,
ft . . . . . . . .
WWW~~-....

..... .............. . ........ . . '

..~ --. ..

Fig. 22. High-speed photographs of a IO-ply laminate struck by a 12.7 mm diameter blunt hard-
steel projectile at an initial velocity of 170 m s -I . The framing rate is 20.2 J1.s/frame.
414 G. ZHU et al.

Fig. 23. Damage to the specimen shown in Fig. 22 : (a) Impact side; (b) Cross-section ; (c) Distal
side.
 Penetration of laminated Kevlar by projectiles-I 415

15000 ~--------------------------------~

10000

€
""oS.

5000

o 10 20 30
Displacement of Projectile (mm)

Fig. 9. Load-displacement curves of the quasi-static perforation by a 12.7 mm diameter cylindro-

conical penetrator of a 6.35 mm thick laminate with and without a back-up of a steel plate with a
5.4 mm diameter central hole.

involving metals (Landkof and Goldsmith, 1985), a discrepancy near the ballistic limit is
more likely to occur because of the importance of certain mechanisms (neglected in an
analysis) whose effect becomes negligible when the initial striker speed is well above the
critical value. Sectioned specimens examined visually exhibited a delamination radius of
25-45 mm diameter, well below those used in the artificially separated targets; the exact
extent of the cracks was difficult to measure because the dye penetrant did not clearly
delineate the fracture region. This confirms the finding that the global deformation absorbs
proportionally a much lower amount of energy in the higher velocity regime, unlike the
situation under quasi-static loading.
Similarly, the effect of varying the volume fraction of the matrix in a 5-ply laminate
was investigated relative to the dynamic resistance. As shown in Fig. 18, the terminal
velocity and the ballistic limit vary only between 75 and 77.9 m s- 1, and are not affected,
within experimental error, by changes in composition over the range of Vm = 26% to 45%
examined here. This compares to the 10% difference in ballistic limits for Kevlar/polymer
combinations involving various volume fractions perforated by cylindrical bullets of
8.75 mm diameter and a mass of 4.15 g (Du Pont (b)).
The distal side view of a 10-ply laminate due to very high-speed perforation (initial
velocity Vo = 783 m S-I) by a 60°, 12.7 mm diameter bullet is presented in Fig. 19. The
principal damage is clearly fiber failure; a substantial bundle has been detached from the
matrix and pushed along the direction of the striker motion. The shape of the damage
region depends upon the weave and lay-up of the target. A 0/90 laminate showed a square-
shaped damage zone on the distal side with an extent 1.5-2 times the projectile diameter,
similar to the static case. A random lay-up resulted in a circular pattern. As only one type
of weave was utilized in the present tests, the effect of weave geometry could not be
determined. It was reported (Du Pont (b)), that while fabrics with smaller size yarns were
superior to constructions based on larger size yarns in resisting perforation of handgun
projectiles, neither yarn size nor weave pattern seemed to affect the penetrability of frag-
ments.
The delamination in perforated plates was studied by noting the damage in sections of
the target. This region was generally found to be 25-45 mm in length, as shown in Fig. 20.
Bulging and fiber failure are also depicted in this photograph.
The perforation process was captured by high-speed photography, as shown in
Fig. 21 for a 1O-ply laminate struck by a 60°, 12.7 mm diameter bullet at a velocity of
416 G. ZHU et al.

~ ~--------------------------------,

20000

10000

o 10 20 30 40 so
Displacement of Projectile (mm>
<a>

700 ~--------------------------------~

600

500

300

200

100

o 10 20 30 40 so
DisplacemCDt of Projectile <mm>
(b)

Fig. 10. Load-displacement (a) and energy-displacement curves (b) for the quasi-static perforation
by a 12.7 mm diameter penetrator with a 60° tip angle of various adjacent layered Kevlar/polyester
targets with a total thickness corresponding to 24 plies. Loading speed is 12 mm min - 1 •

188.6 m s - 1. The framing interval was 20.2 its. This sequence depicts the evolution of the
event, with initial bulging and subsequent dominant fiber failure, but relatively small global
deformation due to an initial speed approximately 50% above the ballistic limit.
A similar test using a blunt projectile striking a lO-ply target at a speed of 170 m s - 1
is depicted in Fig. 22. The sequence exhibits the enormously greater global displacement
of the plate and correspondingly large delaminations, but the absence of any significant
penetration; the deformed impact and distal sides of the specimen as well as a central
section are presented in Fig. 23. The large global deformations are clearly evident; the
laminate was found to be wrinkled with delaminations extending to the outer border. A
plug with the diameter of the projectile had been sheared out of only the first layer (0.635
mm thick). Clearly, the material exhibits a much better perforation resistance to blunt
projectiles compared to sharp-nosed strikers.
An analytical model of the dynamic perforation process has been developed which
takes into account global plate deformation, bulging on the distal side, penetration of the
conical striker, delamination and fiber failure. This work, as well as a comparison of its
predictions with experimental data, is presented in a companion paper (Zhu et al., 1991).
 Penetration of laminated Kevlar by projectiles-I 417

1~ ~---------------------------------.

8000

4000

o 10 20 30
Displacement of Projectllc <mm>
Fig. 11. Load-displacement curves of the quasi-static perforation of 10-ply laminates with and
without a built-in 85 mm radius central delamination by a 12.7 mm diameter penetrator with a 60°
conical tip. Loading speed is 12 mm min - I.

30000

20-ply, 12.7mm dla. .... 10-ply, 12.7mm dla.

25000

~
20000

15000
t~'·r··
~
I
1
10000

1
5000

0
0 10 20 30 40
Oisplacemem (mm)
Fig. 12. Load-displacement curves of the quasi-static perforation by 60° cylindro-conical hard steel
penetrators with diameters of 3.75,6.35 and 12.7 mm.

CONCLUSIONS

An experimental investigation of the static and dynamic penetration of various Kevlar/

polyester laminates primarily by sharp-pointed projectiles has been conducted for various
laminate constructions, target thicknesses and impact speeds. The hard-steel strikers were
generally cylindro-conical with a tip angle of 60° ; a 12.7 mm diameter projectile had a mass
of 28.5 g, while the 9.525 mm diameter striker had a mass of 12.5 g. It is concluded that:
(1) Global and shear stiffness played a very important role in resisting quasi-static
penetration, but were of considerably smaller importance in dynamic tests.
418 G. ZHU et al.

20000 . . . - - - - - - - - - - - - - - - - - - - - ,

cone angle ----:>~ 120

15000

,-.
Z
'-'
"0 10000
«I

oS
5000

o 10 20 30
Displacement (mm)
Fig. 13. Load-displacement curves of the quasi-static perforation of 10-ply laminates by 12.7 mm
diameter cylinders with cone angles of 60°, 90° and 120°.

300

• 12.7 mm dia. projectile

250
iil a 9.525 mm dia. projectile
~ 200
·e:::;
150
.2
~
iii 100
m

50

0
0 2 4 6 8 10 12 14
Thickness of Laminate (mm)

Fig. 16. Ballistic limits of Kevlar/polyester composites penetrated by 60° cylindro-conical projectiles
with two different diameters.

300

iil 250 0 0/45 lay-up

~ A 0/90 lay-up
with delamination
•
>. II
200
1:i0 •
,
random lay-up II
Qj
> 150
iii

.
c
'eiii 100
A
l-
SO
•
0
0 50 100 150 200 250 300
Initial Velocity (m/s)
Fig. 17. Terminal velocity as a function of initial striker speed for the perforation of 5-ply laminates
of various constructions by a 60°, 12.7 mm diameter cylindro-conical projectile.
 Penetration of laminated Kevlar by projectiles-I 419
300

Ii> 2S0 III Vm-4S%

§. • Vm-3S%
~
'u
0
200
• Vm-26%
~.
III

Qj lS0
>

III··
III
Iii
c: 100
'eQ;
SO
•
l-

0
0 50 100 150 200 250 300
Initial Velocity (m/s)

Fig. 18. Effect of matrix volume fraction on the ballistic performance of a 60°, 12.7 mm diameter
projectile striking a 5-ply target.

(2) Local deformation and fiber failure constituted the major energy absorption mech-
anisms in impact perforation. This is documented both by high speed photography and by
post mortem examination of the targets.
(3) The existence of deliberately introduced delaminations, changes in the fiber volume
fraction and variations in the lay-up did not significantly influence impact resistance, but
affected static penetration to a greater degree. Delamination, either in the static or dynamic
case, did not seem to dissipate a major amount of energy.
(4) Perforation with blunt 12.7 mm diameter steel projectiles could not be achieved
in lO-ply laminates in the velocity range up to 200 m S-1 compared to a ballistic limit of
75 m s- I for cylindro-conical strikers.
(5) Kevlar laminates exhibit better impact resistance to cylindro-conical projectiles on
a specific weight basis when compared to aluminum.

REFERENCES

ASTM STP 568 (1975). Foreign Object Impact Damage to Composites. ASTM, Philadelphia.
Bless, S. J., Hartman, D. R. and Hanchak, S. J. (1985). Ballistic Performance of S-2 Glass Laminates. Univ. of
Dayton Research Institute, UDR-TR-85-88A.
Bless, S. J. and Hartman, D. R. (1989). Ballistic penetration of S-2 glass laminates. Proc. 21st SAMPE Con!
(Edited by R. Wegman et al.), pp. 852-866.
Cristescu, N., Malvern, L. E. and Sierakowski, R. L. (1975). Failure mechanisms in composite plates impacted
by blunt-ended penetrators. In ASTM STP 568,159-172.
Du Pont Corporation (a). Kevlar lightweight protective armor. Wilmington, DE Publication E-60153.
Du Pont Corporation (b). Composite armor technical information. A guide to designing and preparing ballistic
protection of Kevlar aramid, Publication E-09164.
Greszczuk, L. B. (1982). Damage in composite materials due to low velocity impact. In Impact Dynamics (Edited
by L. A. Zukas et al.), pp. 55-94. John Wiley & Sons, New York.
Hong, S. and Liu, D. (1989). On the relation between impact energy and delamination area. Exp. Mech. 29, 115-
120.
Jones, R. M. (1975). Mechanics of Composite Materials. McGraw-Hill, New York.
Joshi, S. P. and Sun, C. T. (1985). Impact induced fracture in a laminated composite. J. Compo Mater. 19, 51-
66.
Landkof, B. and Goldsmith, W. (1985). Petaling of thin, metallic plates during penetration by cylindro-conical
projectiles. Int. J. Solids Structures 21, 245-266.
Liu, D. and Malvern, L. E. (1987). Matrix cracking in impacted glass/epoxy plates. J. Compo Mater. 21, 594--
609.
Saghizadeh, H. and Dharan, C. K. H. (1986). Delamination fracture toughness of graphite and aramid epoxy
composites. J. Engng Mater. Tech. 108, 125-130.
Scott, B. R. (1990). The penetration of Kevlar laminate by long rods and jets. Proc. 12th International Symposium
on Ballistics 3, 42--49.
Shivakumar, K. N., Elber, W. and Illg, W. (1985a). Prediction of impact force and duration due to low-velocity
impact on circular composite laminates. J. Appl. Mech. 52, 674--680.
Shivakumar, K. N., Elber, W. and Illg, W. (1985b). Prediction of low-velocity impact damage in thin circular
laminates. AlA A J123, 442--449.
Takeda, N., Sierakowski, R. L. and Malvern, L. E. (1981). Wave propagation experiments on ballistically
impacted composite laminates. J. Compo Mater. 15; 157-174.
420 G. ZHU et al.

Takeda, N., Sierakowski, R. L. and Malvern, L. E. (1982a). Microscopic observation of cross sections of impacted
composite laminates. Compo Tech. Rev. 4, 4-10.
Takeda, N., Sierakowski, R. L., Ross, C. A. and Malvern, L. E. (1982b). Delamination-crack propagation in
ballistically impacted glass/epoxy composite laminates. Experimental Mech. 22, 19-25.
Takeda, N., Sierakowski, R. L. and Malvern, L. E. (1987). Transverse cracks in glass/epoxy cross-ply laminates
impacted by projectiles. J. Materials 19, 51-66.
Wardle, M. W. (1982). Impact damage tolerance of composites reinforced with Kevlar aramid fibers. Prog. Sci.
Engng Camp. Proc. ICCM-IV in Tokyo.
Wardle, M. W. and Tokarsky, E. W. (1983). Drop weight impact testing of laminates reinforced with Kevlar
aramid fibers, E-glass and graphite. Camp. Tech. Rev. 5,4-10.
Wardle, M. W. and Zhar, G. E. (1987). Instrumented impact testing of aramid-reinforced composite materials.
ASTM STP 936, 219-235.
Zhu, G., Goldsmith, W. and Dharan, C. K. H. (1991). Penetration of laminated Kevlar by projectiles-II.
Analytical model. Int. J. Solids Structures 29, 421-436.