A Ballistic Material Model For Cross-Plied Unidirectional Ultra-High Molecular-Weight Polyethylene Fiber-Reinforced Armor-Grade Composites PDF
A Ballistic Material Model For Cross-Plied Unidirectional Ultra-High Molecular-Weight Polyethylene Fiber-Reinforced Armor-Grade Composites PDF
A Ballistic Material Model For Cross-Plied Unidirectional Ultra-High Molecular-Weight Polyethylene Fiber-Reinforced Armor-Grade Composites PDF
a r t i c l e i n f o a b s t r a c t
Article history: The known fiber and polymeric-matrix material properties, unit-cell microstructural characteristics and
Received 17 May 2008 unit-cell level finite-element analyses are used to construct a new ballistic material model for 0◦ /90◦ cross-
Received in revised form 27 July 2008 plied oriented polyethylene fiber-based armor-grade composite laminates. The model is constructed in
Accepted 28 July 2008
such a way that it can be readily integrated into commercial finite-element programs like ANSYS/Autodyn
[ANSYS/Autodyn version 11.0, User Documentation, Century Dynamics Inc., a subsidiary of ANSYS Inc.,
Keywords:
2007] and ABAQUS/Explicit [ABAQUS version 6.7, User Documentation, Dessault Systems, 2007] as a user
Flexible armor
material subroutine. To validate the model, a series of transient non-linear dynamics simulations of the
Armor-grade composites
Material model
transverse impact of armor-grade composite laminates with two types of bullets/projectiles is carried out.
High-performance fibers The results obtained are next compared with their experimental counterparts. This comparison revealed
Ballistic performance that a relatively good agreement is obtained between the experimental and the computational analysis
relative to: (a) the success of the armor panels of different areal densities in defeating the bullets at
different initial bullet velocities; (b) post-mortem spatial distribution of damage within the panels; (c)
the temporal evolution of a bulge at the back-face of the armor; and (d) the existence of three distinct
armor-penetration stages (i.e. an initial filament shearing/cutting dominated stage, an intermediate stage
characterized by pronounced filament/matrix de-bonding/decohesion and the final stage associated with
the extensive bulging of the armor panel).
© 2008 Elsevier B.V. All rights reserved.
0921-5093/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2008.07.056
232 M. Grujicic et al. / Materials Science and Engineering A 498 (2008) 231–241
assessed in terms of two basic parameters: (a) a critical level of ina toward the rear of the laminate which induces shear cracks
the projectile’s velocity or the projectile’s kinetic energy (generally in the resin matrix parallel to the fibers and applies a trans-
referred to as the “ballistic limit”) below which no full perforation of verse load to the second lamina. This, in turn, causes separation
the armor takes place [14,15]; and (b) an extent to which material between the first two laminates, i.e. delamination. After the
ballistic-protection resistance is compromised in the armor sys- successive delamination had taken place through the entire
tems which are partially penetrated with projectile(s) or whose thickness of the laminate via the same mechanism and pen-
strike-face surface is damaged by the projectile(s). etration of the laminate has occurred, narrow strips of damage
Over the past decade, considerable efforts have been invested zone remain visible under the transmitted light and the strips
in carrying out various experimental investigations in order to are found to tend to follow the respective fiber orientation in
identify and elucidate various penetration-failure mechanisms of the panel. These strips typically contain numerous matrix/fiber
the armor-grade fiber-reinforced composites under transverse- interface cracks. In addition, a circular delamination zone is
impact loading and to compare and contrast these mechanisms generally seen around the perforation hole.
with those operating in the related resin-free fabrics and resin-rich (f) In contrast to the case of cross-plied fiber-reinforced composite
“structural-grade” composites. The main results obtained in these laminates, fabric-reinforced laminates exhibited much less lat-
investigations can be summarized as follows [16–23]: eral movements of reinforcing fibers during the penetration of
the projectile [3,4]. Even in thin panels, fibers apparently failed
(a) In sharp contrast to the penetration of resin-free fabrics which due to shearing/cutting in the laminae near the strike-face
is dominated by the successive fracture of individual yarns and in tension at the rear of the completely penetrated lami-
along the periphery of the penetrator head and by the side- nates. The presence of a narrow strip of the first lamina pushed
way/lateral movement of the yarns which enables them to slip forward by the penetrator is generally not observed. Instead,
off from the penetrator, the penetration of armor-grade com- the delamination zones are observed preferentially along the
posites is mainly governed by the failure of principal yarns (the two reinforcement directions of woven fabric. However, these
yarns which are in direct contact with the penetrator head). damage zones are closely integrated with the circular delami-
This observation is attributed to the effect of resin matrix on nation zone around the perforation hole. The occurrence of less
reducing yarn mobility which prevents them from slipping off anisotropic pattern of delamination was linked with the pres-
from the penetrator. In general, stiffer resin matrices (e.g. vinyl ence of resin-rich pockets between the reinforcing layers and
ester vs. polyurethane) constrained the yarn movement to a with a greater constraint to matrix crack propagation parallel
greater degree and force the penetrator to engage and frac- to the fibers/yarns.
ture more yarns. As a result, armor-grade composites reinforced (g) Up to the thickness of ∼3 mm, the dependence of the kinetic
with woven-yarn fabric are generally found to possess a higher energy for full perforation of armor-grade composites on
energy-absorption potential than their resin-free fabric coun- the laminate thickness is similar to that observed in ductile
terparts. However, as will be discussed in more detail later, monolithic materials such as poly-carbonate or aluminum but
excessive confinement of the yarns/filaments in collimated uni- displays some non-linearity which is attributed to their unique
directional armor-grade composites may have a deleterious mode of tensile failure for which the critical level of kinetic
effect on the ballistic-protection performance of this type of energy for full perforation is lowered by the fiber/yarn mobility.
armor.
(b) Since the energy absorbed by the armor-grade composite is The first use of fiber-based composites (primarily nylon (poly-
found to scale with the number of broken yarns in its fabric con- amide) fabric and E-glass fiber/ethyl cellulose composites) in
stituent, fiber tensile-straining and ultimate fracture is believed body-armor systems in place of the traditionally used metallic
to be the dominant mechanism for absorption of the projectile solutions can be traced back to the Korean war [24]. Although,
kinetic energy. primarily due to their low cost, nylon and E-glass fibers are still
(c) In addition to fiber fracture, both woven fabric-reinforced and being used today, high-performance polymeric fibers are now the
cross-plied fiber-reinforced composite laminates are generally standard in most fiber-reinforced body-armor applications. The
found to include additional complex failure processes such high-performance polymeric fibers used today are characterized
as: (i) delamination, (ii) a plug punch-out, (iii) resin matrix by substantially improved strength, stiffness and energy-absorbing
cracking, and (iv) fiber pull-out. These failure modes are also capacity. Among these high-performance fibers the most notable
typically observed in conventional structural-grade composites are: (a) poly-aramids (e.g. Kevlar® , Twaron® , Technora® ); (b) highly
reinforced with glass or carbon fibers. oriented ultra-high molecular-weight polyethylene, UHMWPE (e.g.
(d) In the case of multi-ply armor-grade composite laminates rein- Spectra® , Dyneema® ); (c) poly-benzobis-oxazole, PBO (e.g. Zylon® ),
forced with either cross-plied collimated Spectra fibers or with and (d) poly-pyridobisimi-dazole, PIPD (e.g. M5® ). When tested
woven Spectra fabrics, the following fracture modes are most in tension, all these materials differ significantly from the nylon
often observed: (i) sequential delamination, (ii) plug punch-out fibers, having very high absolute stiffness, extremely high spe-
induced by the through-the-thickness shear, and (iii) combined cific strength, and quite low (<4%) strains-to-failure. These fibers
fiber shearing/cutting and fiber tensile failure. In the cross-plied essentially behave, in tension, as rate-independent linear elastic
laminates, fibers in the top plies are typically found to fail by materials. When tested in transverse compression, however, these
shearing/cutting, primarily along the edges of the projectile. fibers are similar to nylon and can undergo large plastic defor-
Fibers located in the back layers of the laminates, on the other mation without a significant loss in their tensile load-carrying
hand, generally fail in tension although in thin laminates, the capacity. This behavior is quite different from that found in carbon
lateral motion of fibers and/or fiber pull-out rather than fiber or glass fibers, which tend to shatter under transverse compression
tensile-straining to fracture is sometimes observed. loading conditions.
(e) The delamination in the cross-plied Spectra fiber-reinforced The ballistic performance of high-performance polymeric fibers
composite laminates is typically found to resemble the “gener- is, in general, quantified with respect to their ability to: (a) absorb
ator strip” phenomenon [22] seen in conventional composites the projectile’s kinetic energy locally; and (b) spread out the
reinforced with glass fiber-reinforced epoxy resin. That is, under absorbed energy fast before local conditions for the failure are met.
transverse impact, the projectile pushes a strip of the first lam- In simple terms, the ability of high-performance fibers to absorb
M. Grujicic et al. / Materials Science and Engineering A 498 (2008) 231–241 233
Table 1
Typical mechanical properties of high-performance fibers
Fiber type Failure strength (GPa) Failure strain Axial modulus (GPa) Density (kg/m3 )
as a result of the repetition of this block in three orthogonal direc- 2.1. Unit-cell level finite-element analyses
tions.
Coupling between the continuum-material formulation and the The salient feature of the proposed computational approach is
unit-cell geometry and mechanical response is done in the follow- that the mechanical response of a continuum-level material point
ing way: (a) the deformation state of a continuum-material point (corresponding to a unit-cell in the armor-grade composite) and the
(as quantified by the corresponding deformation gradient) is used accompanying changes in constituent materials (primarily those
to update the unit-cell geometry; (b) the updated unit-cell geom- associated with the filament/matrix interfacial de-bonding) can be
etry and the state of the continuum material at the end of the inferred by carrying out a series of finite-element analyses pertain-
previous time increment are used to update the extent of struc- ing to relatively simple mechanical tests of the unit-cell. In these
tural damage in the unit-cell; and (c) the updated material state analyses, a detailed representation of the unit-cell microstructure
obtained in point (b) is then used to compute the stress state at the is considered. In this section, details are presented regarding the
end of the current time increment. geometrical models used in the construction of the unit-cell, mate-
It must be noted that in order for the aforementioned approach rial models assigned to the filament segments, matrix and the
to be valid (i.e. in order for homogenization of the armor-grade com- filament/matrix interfacial bonding, and the finite-element anal-
posite unit-cell response to be justified), the characteristic lengths yses used to determine the mechanical response and the material
in the numerical analysis in which the model is used (e.g. the projec- evolution under different loading conditions.
tile and the armor dimensions in a projectile/armor impact problem An example of the finite-element meshes used in the unit-cell
analyzed in the present work and the associated stress/strain gra- computational analyses is displayed in Fig. 3(a) and (b). 20,847 first-
dient ranges) must be large in comparison to the unit-cell edge order tetrahedron elements, C3D4, are used to discretize each of the
lengths. In most practical situations, this appears to be the case two filament segments, Fig. 3(a), while 21,606 elements of the same
since the unit-cell edge length is in the range between 10 and type are used to discretize the matrix, Fig. 3(b). Bonding between
30 m. the matrix and the filaments is represented using 7056 “cohesive”
elements, COH3D6.
The polymeric filaments (assumed to be based on the UHMWPE)
are modeled as orthotropic (more precisely as planar isotropic)
linear elastic materials (up to the point of failure under axial ten-
sion or transverse shear) with the unique material direction being
aligned with the filament axis. A summary of the elastic and failure
properties of the filament material is provided in Table 2. The poly-
meric matrix (assumed to be based on styrene–isoprene–styrene
tri-block copolymer [32]) is modeled, due to attendant high-
deformation rate conditions, as a linear isotropic material with
a Young’s modulus of 3 GPa and a Poisson’s ratio of 0.4. Bond-
ing between the filaments and the matrix is modeled using
traction vs. interfacial-displacement discontinuity relations (one
for the normal and one for the tangential displacements), these
two relations are characterized by a linear traction vs. displace-
ment/discontinuity relation unto the point of damage initiation and
with a linear “down-hill” post-damage relationship. Consequently
the two modes “normal and shear” of interfacial-bonding dam-
age are each characterized by three parameters: (a) critical normal
or shear interfacial-displacement discontinuities at which damage
Table 2
The orthotropic linear elastic material data for UHMWPE filaments [31]
where ε33,init and 12,init are pure normal and shear strains at which
damage initiation is first observed. In a ε33 /ε33,init vs. 12 / 12,init
plot, Eq. (3) defines a unit (failure-initiation) circle. Within the same
plane, the condition at which complete damage-induced degrada-
tion takes place is defined by an ellipse in the form:
2 2
ε33 12
+ =1 (4)
ε33,fail 12,fail
Fig. 3. Typical finite-element meshes used in the unit-cell computational analyses
to discretize: (a) the two filaments and (b) the matrix. where ε33,fail and 12,fail are pure normal and shear strains at which
complete degradation (D = 1) takes place.
initiation begins; (b) the corresponding normal or shear interfacial When the (ε33 , 12 ) strain state of a material point (e.g. point
strengths; and (c) normal or shear interfacial-displacement dis- B in Fig. 4) lies between the damage initiation circle and the fail-
continuities at which complete filament/matrix decohesion takes ure ellipse, the corresponding extent of material damage is defined
place. A summary of the interfacial cohesion parameters used in the as the ratio of line segments AB and AC indicated in Fig. 4. The
present work is given in Table 3. These parameters were determined four continuum-level damage parameters ε33,init , ε33,fail , 12,init and
in a separate molecular-statics-based investigation of atomic-level 12,fail are determined using the unit-cell finite-element analyses
mechanical properties of the composite materials consisting of the described in this section. The values of these four parameters are:
unidirectional UHMWPE filaments and an amorphous polymeric 0.01, 0.04, 0.01 and 0.04, respectively.
matrix. The results of this investigation will be reported elsewhere
and the procedure used closely follows that presented in our recent 2.2. Determination of the current unit-cell geometry and
work [36] in which atomic-level properties of composite materials architecture
Once the extent of material damage is updated then E33 , G12 , G23
and G31 are degraded by multiplying their initial values by a factor
(1.0cD), where 0.0 ≤ c ≤ 1.0 is an elastic-modulus dependent param-
eter. Since the continuum material is modeled as a linear elastic
orthotropic material with degradable stiffness moduli, the stan-
dard relationships are used to compute the stress components from
the updated strain components and the updated material stiffness
matrix.
Once the stresses are updated, the occurrence of filament fail-
ure is investigated. Filaments are allowed to fail in one of the two
following modes: (a) in tension, when the tensile-strain/tensile-
stress reaches a critical value or (b) due to transverse shear, when
the corresponding transverse shear stress reaches a critical value.
When either of these two filament failure modes takes place, the
corresponding in-plane normal stress(es) and the corresponding
transverse shear stress are set to a small residual value associ-
ated with the remaining matrix ligaments. Once the stresses are
updated to include the effect of filament failure they are ready to
Fig. 4. A schematic of the computational procedure used to determine the onset and be returned to the finite-element solver for the computation of the
track the progress of material damage (interfacial de-bonding) within the unit-cell. global equilibrium.
The present material model is, thus, constructed in such a way
these parameters are related to the continuum-level deformation so that it can account for the competition between the follow-
state of the material point corresponding to the unit-cell in ques- ing two processes: (a) transverse shear loading which is promoted
tion. by good filament/matrix bonding and higher matrix stiffness. If
At the continuum level, the state of deformation at a given sufficiently high transverse shear stresses are developed they can
material point is described by the deformation gradient, F, whose cause shear/type failure of the filament(s). In this case, the energy
components in a Cartesian coordinate system are defined as: absorbed by the filaments is relatively small and, consequently,
ballistic/protection performance of the armor-grade composite
dxj (t) laminate is inferior; and (b) stretching of the filaments till the point
Fjk (t) = (5)
dXk of failure. This process is promoted by filament/matrix de-bonding
which enables the filaments to deform independently of the matrix.
where xj (t) is the jth component of a material point at time t, and
In this case, the energy absorbed by the filaments is maximum
Xk the kth component of the same point in the initial/un-deformed
and the ballistic/protection of the armor is greatly enhanced. It
configuration.
should be noted that some critical level of filament-matrix bond-
At the continuum level, the 0◦ and 90◦ filaments can be
ing is needed to ensure that the filaments are not simply pushed
described using vectors ai (i = 1 and 2) aligned with the axis of these
laterally by advancing projectile which can lead to the defeat of the
filaments and the length of these vectors can be set equal to the cor-
armor by the projectile via the so-called “wedge through effect”
responding current in-plane unit-cell edge lengths, ai (i = 1 and 2).
[35].
These vectors and the vector a3 which is aligned with the out-of-
plane unit-cell edge can be related to their initial counterparts, ai,0
2.4. Material model implementation in a user material subroutine
(i = 1–3) as:
ai = Fai,0 (i = 1–3) (6) The unit-cell-based material model described in the previ-
ous section is next implemented in the material user subroutine,
The length of each ai (i = 1–3) can be defined as: VUMAT, of the commercial finite-element program ABAQUS/
Explicit [2]. This subroutine is compiled and linked with the
ai = ai · ai = (Fai,0 ) · (Fai,0 ) = ai,0 (FT F)ai,0 (7) finite-element solver and enables ABAQUS/Explicit to obtain the
needed information regarding the state of the material and
where the inter-filament included angle can be computed from: the material mechanical response during each time step, for
a1 · a2 = (Fa1,0 ) · (Fa2,0 ) = a1 a2 cos (8) each integration point of each element. In the present work,
first-order six-node general-purpose reduced-integration solid ele-
An equation analogous to Eq. (8) can be used to define the out-of- ments (ABAQUS/Explicit designation C3D6R) are used.
plane shear angles and . Eqs. (5)–(8) thus show that the current The essential features of the coupling between the
geometry of the unit-cell can be determined using the original unit- ABAQUS/Explicit finite-element solver and the VUMAT mate-
cell edge lengths and the current value of the deformation gradient. rial user subroutine at each time increment at each integration
M. Grujicic et al. / Materials Science and Engineering A 498 (2008) 231–241 237
Table 4 thick lamina. To reduce the computational burden, the size of the
Experimental [33] and the corresponding computational (present work) results per-
C3D6 elements is chosen to match that of the C3D4 elements only
taining to the success of armor-grade composite test panel to stop a M855 bullet at
different initial bullet velocities in the region of the armor panel impacted and greatly affected by
the bullet. A courser mesh was used in the section of the armor less
Test-panel thickness (mm) Areal density (kg/m2 ) Bullet velocity (m/s)
affected by the bullet impact.
600 700 800 900 The three metallic materials (steel, lead and copper) present in
4.2 4 – – – G/G the two types of bullets are modeled using (a linear equation of
11 10.5 G/G G/G G/G G/G state, the Johnson–Cook strength model, the Johnson–Cook failure
15 13.7 – O/O O/O G/G model and an erosion algorithm based on the maximum allow-
22 21 U/U U/U U/U U/U
able instantaneous geometrical strain) [1]. Considering the fact
32 31 U/U U/U U/U U/U
that these material models were reviewed in our recent work [34],
Nomenclature: U, undermatched; O, overmatched; G, grossly overmatched; and they will not be discussed any further here. The armor panel was
experiment/computation.
assigned the material model developed in the present work.
To define the initial-conditions, zero initial velocities were
point of each element can be summarized as follows: assigned to all the nodes of the armor panel while a constant veloc-
ity in the negative Z direction was assigned to all the nodes of the
(a) The corresponding previous time-increment stresses and mate- bullet. Four initial bullet velocities were considered: 600, 700, 800
rial state variables as well as the current time-step deformation and 900 m/s. To mimic the effect of clamping along the armor edges,
gradient are provided by the ABAQUS/Explicit finite-element fixed boundary conditions are applied to all the peripheral nodes
solver to the material subroutine. In the present work, the strain of the armor panel.
components, two variables pertaining to the not-failed/failed The same hard interaction algorithm as that used in Section 2.1
status of the filaments and one variable pertaining to the dele- was also employed to model interactions between the bullet and
tion status of the finite element are used as the state variables. the armor. To account for the sliding-friction resistance between the
(b) Using the information provided in (a), and the unit-cell- bullet and the armor, a simple Coulomb friction model was used.
based material model presented in the previous section, the Computational analyses were run on a machine with two
material stress state as well as values of the material state 2.33 GHz Quad-core Intel Xeon processors with 16 GB of RAM. A typ-
variable(s) at the end of the time increment are determined ical run involving the 11-mm thick armor panel took ∼12 min while
within the VUMAT and returned to the ABAQUS/Explicit finite- in the case of a 32-mm thick panel the wall-clock computational
element solver. In addition, the changes in the total internal time was ∼45 min.
and the inelastic energies (where appropriate) are computed
and returned to the solver.
4. Results and discussion
3. Validation of the material model
As discussed earlier, the unit-cell continuum-damage material
model for 0◦ /90◦ cross-ply UHMWPE filament-based armor-grade
As stated earlier, the main objective of the present work was
composites is validated by comparing the computational results
to develop a microstructure-dependent physically based mate-
obtained in a series of transient non-linear dynamics finite-element
rial model for the armor-grade composites based on 0◦ /90◦
analyses discussed in Section 3 with their experimental coun-
cross-plied unidirectional UHMWPE filaments and a low-content
terparts obtained in the work of Iremonger [33]. Iremonger [33]
(<20 mass%) polymeric matrix. In this section, a simple projec-
investigated five armor panels with a thickness from 4.2 to 32 mm,
tile/armor impact problem is described. The problem is used to
used two types of bullets (M855 and M193) and four initial
carry out preliminary testing and validation of the proposed unit-
velocities. Thus the full-factorial test matrix involves 5 × 2 × 4 = 40
cell continuum-damage-based material model.
experiments. Not all 40 experiments were conducted in the work
Two types of full-metal jacketed bullets both of the 5.56-mm
of Iremonger [33] and only selected results were reported for the
caliber were considered. The first type of bullet (M855) has a hard-
experiments that were carried out. Three types of results were
ened steel tip attached to the lead core and a 0.5-mm thick copper
reported: (a) the success of armor in stopping the bullet. Iremonger
jacket. The weight of this bullet is ∼4.0 g and due to the presence
[33] used the following nomenclature: under matched (to denote
of the hardened steel tip the bullet behaves as an armor-piercing
the cases when the armor was successful in stopping the bullet),
(AP) projectile. The second bullet type (M193) does not contain the
overmatched (to denote the cases when the armor was fully pen-
a hardened steel tip but its otherwise geometrically quite similar
etrated by the bullet but was able to absorb a substantial amount
to the M855 bullet. The weight of the M193 bullet is ∼3.5 g.
of the bullets kinetic energy), and grossly overmatched (to denote
Since the computational results for the projectile/armor impact
the cases when the bullet was able to fully penetrate the armor
obtained in the present work were compared with their experi-
by punching out a circular-disk shaped plug of the armor material
mental counterparts obtained in the work of Iremonger [33], five
without a significant loss in the bullets kinetic energy; (b) post-
armor panels with a thickness of 4.2, 11, 15, 22 and 32 mm were
mortem micrographs of the vertical cut sections of the armor panel
investigated in the present work. The corresponding armor areal
passing through the axis of the penetration hole and; (c) temporal-
densities can be found in Table 4. In each case the armor panel had
evolution plots for the armor back-face bulge height and the bulge
a circular/disk shape with a radius of 90 mm.
diameter. In the remainder of this section, a comparison is made
An example of the initial configuration of the projectile/armor
between the computational and the experimental results for each
finite-element system analyzed here is shown in Fig. 5. Due to the
of these three types of results.
inherent symmetry of the problem, only one quarter of the model is
analyzed and the appropriate symmetry boundary conditions are
applied along the planes of symmetry. Typically (one quarter of) the 4.1. Outcome of the bullet impact onto the armor test panel
projectile is discretized in terms of 1955 first-order tetrahedron ele-
ments C3D4, while (one quarter of) the armor panel is discretized The success of different armor panels in stopping the M855 bul-
using 382 first-order six-node brick elements, C3D6, per 1.1-mm let at different bullet velocities as determined experimentally by
238 M. Grujicic et al. / Materials Science and Engineering A 498 (2008) 231–241
Fig. 5. Typical final element meshes used in the transient non-linear finite-element analyses of the impact of a full-jacketed metal bullet impact onto an armor-grade
composite test panel. Due to inherit symmetry of the problem only one-quarter of the model is analyzed.
Iremonger [33] and computationally in the present work as dis- in the armor panel obtained in the present work are displayed in
played in Table 4. It is seen that the overall agreement between the Figs. 6(b), 7(b) and 8(b), respectively.
two sets of results is quite reasonable suggesting that the proposed A comparison of the results displayed in Figs. 6(a), 7(a) and 8(a)
unit-cell continuum-damage-based material model for cross-plied with the ones displayed in Figs. 6(b), 7(b) and 8(b), respectively,
UHMWPE filament-based armor-grade composites is capable of reveals that the overall agreement between the two sets of results
capturing the essential features of the mechanical behavior of this is reasonable. More specifically:
material under ballistic loading conditions. It is unfortunate that in
Ref. [33], the results analogous to those displayed in Table 4 were (a) In the case of 11-mm thick armor panel impacted by an M855
not reported for the non-armor-piercing bullet M193. Hence a sim- projectile at an initial velocity of 600 m/s, both the experiment,
ilar experiment/computation comparison cannot be made for this Fig. 6(a), and the numerical results, Fig. 6(b), show a “punch
type of bullet. through” mode of penetration which is dominated by transverse
shearing/cutting of the filaments and associated with relatively
4.2. Type, extent and spatial distribution of damage low absorption of the projectiles kinetic energy. In addition, the
size of penetration hole and its changes through the armor-
In Ref. [33], three micrographs of the vertical cut sections of panel thickness are reasonably well reproduced by the present
the armor panel (each passing through the axis of the penetration material model, Fig. 6(a) and (b).
hole) were provided. Due to copyright restrictions, only schemat- (b) In the case of a 22-mm thick armor panel impacted by an
ics of these micrographs are included in the present work. The M855 projectile at an initial velocity of 800 m/s, both the
three micrographs correspond to the following armor/bullet/test experiment, and the numerical results show only a partial
conditions: (a) 11-mm thick test panel/M855/600 m/s, Fig. 6(a); (b)
22-mm thick test panel/M855/800 m/s, Fig. 7(a); (c) 22-mm thick
test panel/M193/800 m/s, Fig. 8(a). The corresponding computa-
tional counterparts revealing the spatial distribution of damage
Fig. 6. A comparison between (a schematic of) the experimental results obtained Fig. 7. A comparison between (a schematic of) the experimental results obtained
in Ref. [33] (a), and their computational counterparts obtained in the present work in Ref. [33] (a), and their computational counterparts obtained in the present work
(b), pertaining to the spatial distribution of damage in case of a M855 bullet initially (b), pertaining to the spatial distribution of damage in case of a M855 bullet initially
propelled at a velocity of 600 m/s impacting an 11-mm thick test panel. propelled at a velocity of 800 m/s impacting a 22-mm thick test panel.
M. Grujicic et al. / Materials Science and Engineering A 498 (2008) 231–241 239
Acknowledgements
References
Fig. 9. A comparison between the experimental results obtained in Ref. [33] and [1] ANSYS/Autodyn version 11.0, User Documentation, Century Dynamics Inc., a
their computational counterparts obtained in the present work pertaining to the subsidiary of ANSYS Inc., 2007.
temporal evolution of: (a) the height and (b) the diameter of a delamination-induced [2] ABAQUS version 6.7, User Documentation, Dessault Systems, 2007.
bulge at the armor back-face for the case of a M855 bullet initially propelled at a [3] B.L. Lee, J.W. Song, J.E. Ward, J. Compos. Mater. 28 (13) (1994) 1202–1226.
[4] B.L. Lee, T.F. Walsh, S.T. Won, H.M. Patts, J.W. Song, A.H. Mayer, J. Compos. Mater.
velocity of 900 m/s impacting a 32-mm thick test panel.
35 (2001) 1605–1635.
[5] D.C. Prevorsek, Y.D. Kwon, H.B. Chin, Polym. Eng. Sci. 34 (1994) 141–152.
[6] G. Zhu, W. Goldsmith, C.K.H. Dharan, Int. J. Solids Struct. 29 (4) (1992) 399–420.
[7] G. Zhu, W. Goldsmith, C.K.H. Dharan, Int. J. Solids Struct. 29 (4) (1992) 421–424.
[8] T.L. Schuman, Proceedings of the 24th SAMPE International Technical Confer-
of Iremonger [33]. The key experimental results obtained in Ref. ence, 1992, pp. 280–290.
[33] were then compared with their computational counterparts [9] C.L. Segal, Proceedings of the 23rd SAMPE International Technical Conference,
1991, pp. 651–660.
obtained in the present work to judge the validity of the present [10] P.G. Riewald, F.H.H. FolgarYang, W.F. Shaughnessy, Proceedings of the 23rd
model. The obtained level of qualitative and quantitative agreement SAMPE International Technical Conference, 1991, pp. 684–695.
between the two sets of results suggests that the proposed material [11] T.S. Thomas, Proceedings of the 22nd SAMPE International Technical Confer-
ence, 1990, pp. 304–318.
model is capable of capturing the essential behavior of a proto-
[12] D.C. Prevorsek, H.B. Chin, Development of a Light Weight Spectra Helmet, Phase
typical 0◦ /90◦ cross-plied UHMWPE-based armor-grade composite I Interim Technical Report from AlliedSignal Inc. to U.S. Army Natick RD&E
material. While the present work was focusing on the initial devel- Center, Natick, MA, DAAK60-87-C-0089/D, 1988.
[13] J.W. Song, G.T. Egglestone, Proceedings of the 19th SAMPE International Tech-
opment, parameterization and validation of the material model, in
nical Conference, 1987, pp. 108–119.
our future work, the model will be used to investigate in greater [14] J.G. Donovan, B. Kirkwood, F. Figucia, Development of Lower Cost Ballis-
details the competition and interplay between various deforma- tic Protection, U.S. Army Natick RD&E Center, Natick, MA, Technical Report
tion, fracture and energy-dissipation phenomena which control Natick/TR-85/019L, 1985.
[15] L.C. Lin, A. Bhatnagar, H.W. Chang, Proceedings of the 22nd SAMPE International
armors ability to defeat projectiles by absorbing their kinetic Technical Conference, 1990, pp. 1–13.
energy. [16] J.C. Smith, J.M. Blandford, H.F. Schiefer, Text Res. J. 30 (10) (1960) 752.
M. Grujicic et al. / Materials Science and Engineering A 498 (2008) 231–241 241
[17] J.C. Smith, J.M. Blandford, K.M. Towne, Text Res. J. 32 (1) (1962) 67. [27] A.M. Floyd, K. Williams, R. Varziri, K. Kanji, A. Pourtasip, Proceedings of 18th
[18] D. Roylance, A. Wilde, G. Tocci, Text Res. J. 43 (1973) 34–41. International Symposium on Ballistics, San Antonio, November 1999, p. 877.
[19] P.M. Cunniff, Text Res. J. 62 (9) (1992) 495–509. [28] P.M. Cunniff, J. Ting, Proceedings of 18th International Symposium on Ballistics,
[20] R.W. Dent, J.G. Donovan, Projectile Impact with Flexible Armor—An Improved San Antonio, November 1999, p. 822.
Model, U.S. Army Natick RD&E Center, Natick, MA, Technical Report Natick/TR- [29] G.R. Johnson, R. Beissel, P.M. Cunniff, Proceedings of 18th International Sympo-
86/044L, 1986. sium on Ballistics, San Antonio, November 1999, p. 862.
[21] C.Y. Hsieh, A. Mount, B.Z. Jang, R.H. Zee, Proceedings of the 22nd SAMPE Inter- [30] X. Guan, N. Birnbaum, Proceedings of 18th International Symposium on Ballis-
national Technical Conference, 1990, pp. 14–27. tics, San Antonio, November 1999, p. 1107.
[22] N. Critescu, L.E. Malvern, R.L. Sierakowski, Failure Mechanisms in Composite [31] M.J.N. Jacobs, J.L.J. Van Dingenen, J. Mater. Sci. 36 (2001) 3137–3142.
Plates Impacted by Blunt-Ended Penetrators, Foreign Object Impact Damage to [32] M.J.N. Jacobs, J.H.M. Beugels, M. Blaauw, Process for the Manufac-
Composites, ASTM, Philadelphia, PA, ASTM STP #568, 1975, pp. 159–172. ture of a Ballistic-Resistant Moulded Article, US Patent 7,288,314 B2
[23] D.S. Cairs, P.A. Lagace, AIAA J. 27 (11) (1989) 1590–1596. (October 2007).
[24] R. Frissen, T. Peijs, A. Verlinde, Proceedings of 16th International Symposium [33] M.J. Iremonger, International Symposium on Ballistics, San Antonio, November
on Ballistics, San Francisco, September 1996. 1999, pp. 946–953.
[25] P.M. Cunniff, Proceedings of 18th International Symposium on Ballistics, San [34] M. Grujicic, B. Pandurangan, U. Zecevic, K.L. Koudela, B.A. Cheeseman, Multi-
Antonio, November 1999, p. 1303. discip. Model. Mater. Struct. 3 (2007) 287–312.
[26] R. Clegg, C. Hayhurst, J. Leahy, M. Deutekom, Proceedings 18th International [35] B.A. Cheeseman, T.A. Bogetti, Compos. Struct. 61 (2003) 161–173.
Symposium on Ballistics, San Antonio, November 1999, p. 791. [36] M. Grujicic, Y.P. Sun, K.L. Koudela, Appl. Surf. Sci. 253 (2007) 3009.