Finite-Element-Based Fiber-Length-Scale Modeling of The Ballistic-Impact Response of KEVLAR® KM2 Fabric

Textiles and Light Industrial Science and Technology (TLIST) Volume 5, 2016
doi: 10.14355/tlist.2016.05.001
www.tlist-journal.org
Finite-element-based Fiber-length-scale
Modeling of the Ballistic-Impact Response of
KEVLAR KM2 Fabric
M. Grujicic1, S. Ramaswami, J. S. Snipes, R. Galgalikar
Department of Mechanical Engineering, Clemson University, Clemson SC 29634
gmica@clemson.edu
Abstract
In the present work, finite-element-based sub-yarn-level modeling of the Kevlar KM2 ballistic-fabric is carried out in order to
help establish the relationships between the fabric penetration resistance and various fiber-level phenomena such as fiber-fiber
friction, fiber twist, transverse properties of the fibers and the stochastic nature of fiber strength. In particular, a series of
transient non-linear dynamics finite-element analyses was carried out of the transverse impact of a test patch of Kevlar KM2
fabric by a solid spherical projectile, in order to investigate the role of the following two important sub-yarn phenomena on the
penetration resistance of this fabric: (a) fiber-transverse properties including non-linear elastic and plastic response; and (b)
stochastic nature of the fiber strength.
Keywords
Finite-Element Modeling; Kevlar KM2 Fibers and Fabric; Flexible Armor; Penetration Resistance; Ballistic Performance
Introduction
The present work deals with the high-performance polymeric-fiber-based fabrics and with the computational
modeling of the protection offered by such fabrics against ballistic threats. Therefore, the topics most relevant to
the present study include: (a) the basics of high-performance polymeric-fiber-based fabrics; and (b) an overview of
prior ballistic-fabric modeling efforts. In the remainder of this section, a brief description is provided of these topics.
The Basics of High-Performance Polymeric-Fiber-Based Fabrics
Military systems, in particular those supporting the U.S. ground forces, are being continuously transformed to
become faster, more agile, and more mobile, so that these systems can be quickly transported to operations
conducted throughout the world in order to respond to the new enemy threats and warfare tactics. Consequently,
an increased emphasis is being placed on the development of improved lightweight body-armor and lightweight
vehicle-armor systems as well as on the development of new high-performance armor materials. Highperformance fiber-based materials have been exploited for both body-armor (e.g. as soft, flexible fiber mats for
personal-armor vests) and for the vehicle-armor systems (e.g. as reinforcements in rigid polymer matrix composites,
PMCs, for lightweight vehicle-armor systems).
Flexible lightweight materials have been used, throughout history, in body-armor systems to provide protection
against specified threats, at reduced weight and without compromising a persons mobility. Early materials used
included leather, silk, metal chainmail and metal plates. Replacement of metal with a nylon (poly-amide) fabric and
an E-glass fiber/ethyl cellulose composite in body armor systems can be linked to the Korean War [1]. Although,
primarily due to their low cost, nylon and E-glass fibers are still being used today, other non-nylon highperformance polymeric fibers (typically used in the form of woven fabrics) is now the standard in most fiberreinforced body-armor applications. These high-performance polymeric fibers enabled the development of novel
flexible-armor systems with exceptional protection against bullet penetration [2].
The defining characteristics of high-performance polymeric fibers used today are their superior strength, stiffness
and ballistic performance. Among these high-performance fibers the most notable are: (a) poly-aramids (e.g. Kevlar ,
Twaron , Technora ); (b) highly-oriented poly-ethylene (e.g. Spectra , Dyneema ); (c) poly-benzobis-oxazole, PBO (e.g.
Zylon ), and (d) poly-pyrido-bis-imidazole, PIPD (e.g. M5 ). The present work is concerned with a particular grade of
Kevlar fibers, the Kevlar KM2 fibers.
When tested in tension, the high-performance fibers mentioned above differ significantly from the nylon fibers,
having very high absolute stiffness, extremely high specific strength (i.e. strength normalized by the mass-density),
and quite low (<4%) strains-to-failure. These fibers essentially behave, in tension, as rate-independent linear elastic
materials. When tested in transverse compression, however, these fibers are similar to nylon and can undergo
large non-linear elastic/plastic deformation without a significant loss in their tensile load-carrying capacity. On the
other hand, this behavior is quite different from that found in carbon or glass fibers, which tend to shatter under
transverse compression loading conditions [3].
The ballistic behavior of high-performance polymeric fibers is, in general, characterized by their ability to: (a)
absorb the projectiles kinetic energy locally; and (b) spread out the absorbed energy fast before local conditions for
the failure are met. In simple terms, the ability of high-performance fibers to absorb energy per their unit mass, Esp ,
is related to the fiber tenacity/failure-strength, fail , the fiber strain-to-failure,
fail ,
and the fiber density, , as
Esp 0.5 fail fail . The ability of fibers to spread out energy is governed by their speed of sound, vsound , which is
defined in terms of their axial modulus of elasticity, E , and their density as vsound E .
A summary of the key properties of the most-commonly used high-performance fibers is provided in Table 1. In
Fig. 1, the two ballistic-performance parameters are displayed for the same set of high-performance fibers.
TABLE 1. TYPICAL MECHANICAL PROPERTIES AND MASS DENSITY OF HIGH-PERFORMANCE FIBERS
Fiber
Type
Failure Strength
(GPa)
Failure
Strain
Axial Modulus
(GPa)
Density
(kg/m3)
Aramid
2.8-3.2
0.015-0.045
60-115
1,390-1,440
HMWPE
2.8-4.0
0.029-0.038
90-140
970-980
LCP
2.7-2.9
0.033-0.035
64-66
1,400-1,420
PBO
5.4-5.6
0.024-0.026
270-290
1,540-1,560
PIPD
3.9-4.1
0.011-0.013
320-340
1,690-1,710
Nylon
0.06-0.08
1.5-2.5
1.0-1.5
1,070-1,170
S-glass
4.64-4.66
0.053-0.055
82-92
2,470-2,490
FIGURE 1. SOUND SPEED VS. MASS-BASED ENERGY ABSORPTION CAPACITY FOR A NUMBER OF HIGH-PERFORMANCE FIBERS
While the attention above was focused on fibers and the associated materials, it should be recognized that flexiblearmor structures are made of fabric/textiles containing these fibers. The fabric itself possesses a hierarchical
structure, i.e. fibers are arranged into bundles called yarns and yarns are assembled into fabric in specific patterns
through weaving or braiding processes. It is important to note that fabric ballistic-penetration resistance is
influenced not only by fiber physical properties, such as strength, ductility, elastic modulus, mass-density,
viscosity and thermal properties, but also by fabric architecture in its as-fabricated state [46].
An Overview of Prior Ballistic-Fabric Modeling Efforts
Attempts were made as early as the 1960s to guide the development of new flexible-armor systems using
computer-aided-engineering and modeling/simulation approaches. Nevertheless, the design and development of
textile armor systems remains primarily governed by experiments and experience. Despite the fact that numerical
models typically have not been used to guide the design/development of the new flexible-armor systems, these
models have often been employed to quantify the effect of specific fiber/yarn/fabric parameters on the ballisticpenetration resistance of the armor [e.g. 715]. It should also be noted that, while it is well established that the
fabric performance is highly affected by the fabric-manufacturing process and the as-fabricated architecture, these
effects are generally not incorporated into the models. Furthermore, the effect of fabric micro-structural parameters
such as yarn denier, end count, yarn structures, filament spatial paths and fiber-to-fiber interactions are generally
either not included, or included incompletely-incorrectly by most of the flexible-armor ballistic-penetration
numerical models. Consequently, a quantitative relation among material properties, manufacturing process, fabric
micro-geometry, and ballistic performance are yet to be established by these models.
It is generally recognized that the level of detail of the fabric-architecture used in the numerical model influences
the resolution and the accuracy/fidelity of the numerical results obtained [1618]. For instance, if a fabric is
considered as a continuum (rather than a discrete-yarn woven structure), details of yarn-to-yarn interaction and
projectile-to-yarn interaction cannot be modeled. As a result, the effect of fabric weave pattern on ballistic
resistance cannot be inferred. Likewise, if the fabric is represented as an assembly of inter-woven yarns but the
yarns are represented as a continuum (rather than an assembly of engaged discrete fibers), details of fiber-to-fiber
interaction and projectile-to-fiber interaction cannot be investigated. As a result, the effect of yarn microstructure
on the ballistic resistance cannot be quantified.
The numerical models of the flexible armor systems reported in the open literature can be divided into two general
groups: (a) those in which the fabric is modeled as a continuum; and (b) those in which the fabric structure is
accounted for explicitly but the yarns are treated as a continuum. Since a detailed overview of these two groups of
material models was provided in our prior work [9], the same overview will not be given here. Instead, only a few
general points regarding the two groups of models will be highlighted.
Continuum-Fabric Material Models: There are a large number of models of this type, and they differ in the way the
internal (not explicitly represented) architecture of the fabric and the yarn is accounted for. While the models
which account for more details of the fabric architecture generally yield better results, the whole group of models
does not generally give satisfactory results. This is attributed to the fact that these models do not account for yarnto-yarn and projectile-to-yarn interactions, the interactions which strongly influence the flexible-armor ballistic
performance [19].
Discrete-Fabric/Continuum-Yarn Material Models: Within this type of models, an explicit account is taken of the
phenomena such as yarn crimp, slip at yarn crossover points, yarn slippage at yarn/projectile contact surface, yarn
weaving pattern, multiple-ply fabric architecture, etc. Since these important phenomena could not be accounted for
within the continuum-fabric material models, the discrete-fabric/continuum-yarn material models yield
substantially better results. However, the phenomena taking place at the fiber-length-scale are not accounted for by
the discrete-fabric/continuum-yarn material models. This shortcoming is addressed in the present work [20].
Main Objective
The main objective of the present work is to develop a finite-element-based model for ballistic fabric in which
fabric, yarns and fibers are all explicitly accounted for and to investigate the structural (including failure) response
of the fabric to ballistic impact. To validate the model, the key model predictions will be compared with their
experimental counterparts [18].
Formulation of the Computational Model

In this section, a detailed description is provided of the procedure used to construct the ballistic-fabric
computational model and of its application to the analysis of a simple ballistic-impact event.
Problem Formulation
The basic problem analyzed in the present work involves the impact of an 8-mm-diameter steel spherical projectile,
mass = 2.1 g, with a 33 mm 33 mm square patch of a Kevlar KM2 single-ply fabric clamped along all its four
edges. Only the normal (i.e. zero obliquity angle) impact case is considered, with the projectile incident velocity
varied in a 35210m/s range.
Finite-Element Model and Analysis
Modeling of the ballistic fabric and of its normal impact by a spherical steel projectile is carried out using the
conventional finite-element method/analysis. Within such an analysis, a complete problem formulation requires
specification of the following: (a) geometrical model; (b) meshed model; (c) computational algorithm; (d) initial
conditions; (e) boundary/loading conditions; (f) contact interactions; (g) material models; (h) computational tool;
and (i) computational accuracy, stability and cost. These items are briefly overviewed in the remainder of this subsection.
1) Geometrical Model
The initial geometrical model for the projectile/fabric assembly is shown in Fig. 2(a). A close-up of the fabric is
displayed in Fig. 2(b) in order to reveal the plain-weave fabric architecture and the sub-yarn microstructure
(involving seven fibers, in the present case).
The fabric architecture and its areal density will be discussed in the subsection dealing with the material models.
The fabric test patch consists of 42 warp yarns and 42 weft yarns. Each yarn is discretized into discrete circularcross-section fibers bundled into a regular (identical) hexagonal close-packed configuration. Various yarn
discretizations (containing 1, 7, 14, 19, 24 and 30 fibers) were considered in the initial portion of the work in
order to identify the one (containing 7 fibers) which provides the best trade-off between the physical accuracy
and computational cost. For each of the configurations considered, fiber diameter was calculated in such a way
that the correct fabric areal density is obtained. Fibers are treated as discrete circular-section solid entities. Yarns
are treated as bundles of collinear, transversely-interacting fibers while the fabric is treated as a plain-woven
structure consisting of effectively orthogonal warp and weft yarns.
FIGURE 2. (A) A GEOMETRICAL MODEL FOR THE PROJECTILE/FABRIC INTERACTION PROBLEM ANALYZED IN THE PRESENT
WORK; AND (B) A CLOSE-UP OF THE FABRIC REGION SHOWING SUB-YARN MICROSTRUCTURE AND THE YARN-WEAVE
PATTERN.
2) Meshed Model
Within the finite-element model each fiber is discretized into a number of three-dimensional beam elements
with the beam length chosen to match the fiber diameter. Since each yarn contained seven fibers, the fabric
contained approximately 400,000 beam elements.
The spherical projectile is discretized using continuum first-order tetrahedron elements with an edge length
comparable to the fiber beam-element length. Consequently, the projectile contained approximately 100,000
finite elements.
3) Computational Algorithm
To ensure robustness of the computational analysis, all the calculations carried out in the present work utilized
a transient, displacement-based, purely-Lagrangian, conditionally-stable, dynamic explicit finite-element
algorithm.
4) Initial Conditions
At the beginning of the analysis, the following initial conditions are specified: (a) the projectile is assigned a preselected negative value of the velocity z-component (the mid-plane of the fabric is placed on the xy-plane
associated with z=0, while the projectile center of gravity is placed at x=y=0 and z>projectile radius); and (b) the
fabric is assumed to be stress-free, deformation-free and at rest.
5) Boundary/Loading Conditions
As mentioned earlier, during the simulation the fabric is fully clamped along all its four edges. As far as the
projectile is concerned, its motion was left fully unconstrained.
6) Material Model for Projectile
Due to the fact that the steel projectile used in the experimental studies relevant to the present work [18], did
not suffer any permanent deformation during impact testing, the projectile material (steel) is treated as an
isotropic linear elastic material with a mass-density of 7850 kg/m3, a Youngs modulus of 210 GPa, and a
Poissons ratio of 0.3.
7) Material Model for Fabric
Fiber Properties: Experimental investigation of Chen et al. [18], clearly established that, in the longitudinal
direction, Kevlar KM2 fiber behaves as a rate-independent inelastic material with an average longitudinal
Youngs Modulus of 84.62 4.18 GPa with 95% confidence. As far as the longitudinal strength, , is concerned,
it is found to be a stochastic property (due to inevitable presence of various microstructural
defects/imperfections within the fibers) and to be implicitly given by a bi-modal Weibull-type failuredistribution function, F ( ) , in the form [22]:
L
F ( ) 1 exp
L0
mi

i 1 0 i
(1)
where n 2 (due to the bi-modal nature of the distribution function), L0 0.025 m (the fiber reference gauge
length), L L0 (the fiber actual gauge length), 01 3.323 GPa and 02 3.7004 GPa (the two Weibull scale
parameters) and m1 18.0023 and m2 24.8803 (the two Weibull shape parameters). The bi-modal nature of
the Kevlar KM2 fiber strength is related to the distinct defect/flaw potency distributions in the core and in the
skin regions of this class of fibers.
While in the transverse direction, Kevlar KM2 fibers show non-linear and inelastic behavior, following the
work of Wang and co-workers [21], the constitutive response of the fibers in the transverse direction is modeled
at three different levels of physical accuracy/fidelity:
(a) at the highest level of approximation, the response is treated as being linear elastic with the associated
transverse modulus of 1.34 0.35 GPa (with 95% confidence based on the normal distribution of this quantity
[10]);
(b) The response is considered to be non-linear elastic with the loading/unloading nominal stress vs. nominal
strain relation given as:
0.006e9 0.5296e9 1.2208e9 2 6.9417e9 3
(2)
where the nominal stress is defined as a ratio of the contact force and the product of the fiber diameter and the
fiber length, while the nominal strain was defined as the ratio of the fiber diameter change and the fiber
diameter; and
(c) At the highest level of fidelity, the transverse response is considered as being elastic/plastic with Eq. (2)
defining the strain hardening yielding-behavior of the fiber during loading (involving plastic deformation)
while the stress during elastic unloading/reloading is defined as:
2
K 0.0646e9 7.0632e9 max 18.666e9 max
(3)
where max is the maximum transverse normal strain experienced by the local material point of the fiber.
It should be noted that since fiber transverse stiffness provides little contribution to either fabric or yarn
transverse stiffness, one cannot generally determine fiber transverse properties by testing yarns or the fabric.
Instead, testing must be done using individual fibers [21].
It should be noted, as will be discussed in more detail below, that due to the beam character of the fiber finite
elements, the longitudinal properties of the fibers are specified within the material definition while the
transverse properties are specified as a part of the fiber/fiber contact definition.
As established above, fiber fracture strength is a stochastic quantity. To assess the effect of the stochastic nature
of the fiber fracture strength on the fabric ballistic-impact response, analyses are carried out in which fiber
fracture strength is treated as a deterministic quantity, and in other analyses the same is treated as a stochastic
quantity. In the latter case, a Monte Carlo procedure is employed within which a random number in a range (0,
1) is drawn (for each fiber element) from the uniform distribution and assigned to the failure probability of the
element in question. Then, the Eq. (1) is solved for the unknown fracture strength of the material residing
within this element. Within the ABAQUS/Explicit [23] computational environment, the most efficient way to
implement the stochastically distributed fiber fracture strength is through the use of a user material subroutine
VUMAT. Within this subroutine, material strength associated with a given beam-element/integration-point is
treated as a material-state variable. This variable retains its initially-assigned value until the onset of fiber failure
(i.e. until the axial stress becomes equal to the material strength) and then decreases monotonically to zero (in a
manner consistent with the pre-assigned value of the material fracture toughness).
Yarn Properties: It should be noted that yarns are represented in the present work as a bundle of collinear
discrete fibers. Hence, yarn properties are not defined as input parameters to the numerical model but rather
could be assessed through the appropriate post-processing schemes. Nevertheless, it should be recognized that
the yarn longitudinal properties are directly controlled by the fiber longitudinal properties. For example, the
longitudinal stiffness modulus of the yarns is defined as a product of the fiber longitudinal stiffness modulus
and a ( 1 F ( ) ) factor in order to account for the effect of potential fiber failure.
As far as the yarn transverse properties are concerned, they are mainly controlled by yarn microstructure,
fiber/fiber friction and yarn tension.
Fabric Properties: As in the case of the yarns, fabric properties are not defined as input parameters to the
numerical model but rather could be assessed through the appropriate post-processing schemes. In general,
both the longitudinal and transverse properties of the fabric depend on the fabric areal density, yarn count per
fabric unit length, longitudinal properties of the fibers and yarns, fabric architecture/weave pattern, fiber/fiber
friction and yarn tension. The Kevlar KM2 fabric analyzed in the present work has an areal density of 180 g/m2
and 13.4 yarns per centimeter. The fabric contains a plain-woven architecture which is obtained by snaking
through orthogonally-oriented warp and weft yarns.
8) Contact Interactions
The response of the fibers in the transverse directions, due to fiber/fiber and fiber/projectile normal interactions,
is modeled using a contact algorithm implemented within the VUINTER user subroutine of the commercial
finite-element package ABAQUS/Explicit [23]. Within this subroutine, the fiber transverse constitutive relations
defined in Eqs. (2) and (3) are implemented. Within the same subroutine, tangential fiber/fiber and
fiber/projectile interactions are modeled using a simple Coulomb-type friction law within which the
transmission of shear stresses across the contact interfaces is defined in terms of static and kinetic friction
coefficients and an upper-bound shear stress limit (a maximum value of shear stress which can be transmitted
before the contacting surfaces begin to slide). Based on the results obtained in our prior work [24], the friction
coefficient was set to a value of 0.1 (unless otherwise stated). This value is very consistent with its experimental
counterpart as reported in Ref. [21].
9) Computational Tool
All the calculations carried out in the present work were done using ABAQUS/Explicit [23] within which the
problem at hand, defined in terms of a set of mass, momentum and energy conservation differential equations
along with the material constitutive relations and contact, initial and boundary conditions, is solved using a
finite-element approach.
10) Computational Accuracy, Stability and Cost
A standard mesh sensitivity analysis was carried out (the results not shown for brevity) in order to ensure that
the results obtained are accurate, i.e. insensitive to further refinements in the finite-element size used.
Due to the conditionally-stable nature of the explicit finite element analysis used, the maximum time increment
during each computational step had to be lower than the attendant stable time increment. To keep the
computational cost reasonable while ensuring accuracy and stability of the computational procedure, a mass
scaling algorithm is used. This algorithm adaptively adjusts material density in the critical stable-timeincrement-controlling finite elements without significantly affecting the computational analysis results.
A typical 210
clock) time on a 12-core, 3.0 GHz machine with 16 GB of memory.
7-fiber yarns required 3.5 hours of (wall-
Results and Discussion

The work carried out in the present study was aimed at helping answer the following two questions: (a) what role
the transverse properties of the fibers play in the ballistic-penetration resistance of the fabric; and (b) what effect
the stochastic nature of the fiber strength has on the ballistic-penetration resistance of the fabric. The results
obtained, which are most pertinent to helping answer these two questions are presented and discussed in this
section.
The Role of Fiber Transverse Properties
When a projectile strikes the flexible-armor/fabric, it applies transverse loading either directly through the
projectile/fiber interactions or indirectly through the interaction of the projectile-engaged fibers and other fibers in
contact with them. It should be noted that when a projectile strikes flexible armor, there are two general outcomes:
either (a) the projectile is defeated by the flexible armor and rebounds; or (b) the projectile penetrates the flexible
armor while retaining a portion of its initial velocity (typically referred to as the residual velocity). An important
question to be answered is to what extent fiber transverse properties affect the mechanical response of the fabric
[including the magnitude of the projectile residual velocity (the velocity of the projectile after it has defeated the
flexible armor or has been defeated by the armor), fabric transverse displacement profile, fabrics ability to deform
and absorb/dissipate the projectiles kinetic energy, the stress magnitude/distribution within the fibers and the
overall fabric-armor strength].
As mentioned earlier, three material models are used to describe the transverse behavior of the Kevlar KM2 fibers
in the present work. In increasing order of complexity and computational burden, these models are: (a) linear
elastic; (b) nonlinear elastic; and (c) rate-independent elastic-plastic.
The results obtained in the present work pertaining to the radial distribution of the fabric back face nodal
displacements in the projectile-propagation (z) direction for the case of projectile initial velocity of 63m/s and the
post-impact time of 0.174ms are displayed in Fig. 3. Three sets of the computational results are displayed in this
figure, one for each of the three (linear elastic, non-linear elastic and elastic-plastic) fiber transverse-response
conditions. For comparison, the experimental results reported in Ref. [21] are also depicted in this figure.
Examination of the results displayed in this figure clearly reveals that the computational analysis in which an
elastic-plastic material model was used to represent the transverse response of the fibers yields the best overall
agreement with the experimental data.
Frame 001 23 Jul 2015 | | |
200
Projectile Residual Velocity, m/s
Experimental [18]
Linear Elastic
150
Nonlinear Elastic
Elastic-Plastic
100
50
-50
50
100
150
Projectile Strike Velocity, m/s
FIGURE 3. VARIATION OF THE FABRIC BACK-SURFACE

DISPLACEMENT WITH THE RADIAL DISTANCE FROM THE
FABRIC CENTER FOR THE CASE OF 2.1g STEEL SPHERICAL
PROJECTILE IMPACTING THE KEVLAR KM2 FABRIC AT AN
INITIAL VELOCITY OF 63m/s. THE EXPERIMENTAL RESULTS
FROM REF. [21] ARE SHOWN.
FIGURE 4. VARIATION OF THE PROJECTILE

RESIDUAL/REBOUND VELOCITY WITH THE PROJECTILE
INITIAL VELOCITY FOR THE CASE OF 2.1g STEEL SPHERICAL
PROJECTILE IMPACTING THE KEVLAR KM2 FABRIC. THE
EXPERIMENTAL RESULTS FROM REF. [21] ARE ALSO SHOWN.
REBOUND VELOCITIES ARE ASSIGNED NEGATIVE VALUES.
Other results generated in the present work but not shown for brevity confirm that the elastic-plastic material
model for fibers in the transverse direction yields the best results relative to the magnitude of the projectile residual
velocity (the velocity of the projectile after it has defeated the flexible armor or has been defeated by the armor),
fabric transverse displacement profile, fabrics ability to deform and absorb/dissipate the projectiles kinetic energy,
the stress magnitude/distribution within the fibers and the overall armor strength. Consequently, this type of
material model was adopted and used in the remainder of the present work.
The Role of the Stochastic Nature of Fiber Strength
Depending on the quality of the fabrication process, fibers may contain different populations of flaws and, thus
fiber strength becomes a stochastic quantity. This effect is not generally taken into account and high-performance
fibers are often assigned a constant stress value (set equal to the fiber mean strength). The results presented in this
section should help address the question of the role of the stochastic nature of fiber strength in the ballisticpenetration resistance of the fabric. Since the extent of this role may be affected by the magnitude of the fiber/fiber
friction and the fiber/projectile friction coefficients, the results will be discussed in the context of the attendant
values of the friction coefficients.
Typical results pertaining to the spatial distribution of the fabric damage during impact by a 2.1g steel spherical
projectile traveling at an initial velocity of 63m/s for the case of a deterministic fiber fracture strength (=3.4GPa) at
four different post-impact times (75
. 5(a)-(d). It should be noted
that for improved clarity, the results in this figure are displayed using the undeformed fabric mesh and the
projectile is not shown. The undamaged elements are shown in blue while the damaged elements are shown in a
color signifying the extent of the damage (red color denotes completely damaged and deleted elements).
Examination of the results displayed in Fig. 5(a)-(d) shows that the fabric damage, in the deterministic fiberstrength case, is mainly localized in the region underneath the projectile. This finding is not consistent with the
experimental results reported in Ref. [21], which indicate that, for the most part, fiber damage is a non-local
phenomenon.
FIGURE 5. SPATIAL DISTRIBUTION OF THE FABRIC DAMAGE DURING IMPACT BY A 2.1g STEEL SPHERICAL PROJECTILE
TRAVELING AT AN INITIAL VELOCITY OF 63 m/s FOR THE CASE OF A DETERMINISTIC FIBER FRACTURE STRENGTH (=3.4GPa) AT
POSTELEMENTS (OTHER COLORS) ARE SHOWN USING THE UNDEFORMED FABRIC MESH, AND THE PROJECTILE IS NOT SHOWN. THE
FABRIC IS VIEWED FROM THE SIDE NOT DIRECTLY IMPACTED.
Typical results pertaining to the spatial distribution of fiber axial stress under the impact conditions identical to the
ones stated in conjunction with Fig. 5(a)-(d) are shown in Fig. 6(a)-(d). In this case, the deformed mesh of the fabric
and the projectile are shown. Examination of the results displayed in Fig. 6(a)-(d) confirms the general
experimental findings in Ref. [21] that fabric penetration is a local process. In other words, penetration of the fabric
is mainly controlled by the creation and evolution of damage in the fabric region which is beneath and, hence,
directly impacted by the projectile.
FIGURE 6. SPATIAL DISTRIBUTION OF FIBER AXIAL STRESS DURING IMPACT BY A 2.1 g STEEL SPHERICAL PROJECTILE TRAVELING
AT AN INITIAL VELOCITY OF 63 m/s FOR THE CASE OF A DETERMINISTIC FIBER FRACTURE STRENGTH (=3.4 GPa) AT POST-IMPACT
TIMES OF: (A)75
Typical results pertaining to the spatial distribution of the fabric damage during impact by a 2.1g steel spherical
projectile traveling at an initial velocity of 63m/s for the case of stochastic fiber fracture strength at four different
post-(d). It should be noted that for
improved clarity, the results in this figure are displayed using the undeformed fabric mesh and the projectile is not
shown. The undamaged elements are shown in blue while the damaged elements are shown in a color signifying
the extent of the damage (red color denotes completely damaged and deleted elements). Examination of the results
displayed in Fig. 7(a)-(d) shows that the fabric damage, in the stochastic fiber failure-strength case, is not entirely
localized within the fabric region directly impacted by the projectile. Rather, some of the fibers and yarns outside
this region have also suffered damage and failed. This finding is fully consistent with the experimental results
reported in Ref. [21].
FIGURE 7. SPATIAL DISTRIBUTION OF THE FABRIC DAMAGE DURING IMPACT BY A 2.1 g STEEL SPHERICAL PROJECTILE
TRAVELING AT AN INITIAL VELOCITY OF 63 m/s FOR THE CASE OF STOCHASTIC FIBER FRACTURE STRENGTH AT POST-IMPACT
TIMES OF: (A) 140
COLORS) ARE SHOWN USING THE UNDEFORMED FABRIC MESH, AND THE PROJECTILE IS NOT SHOWN. THE FABRIC IS VIEWED
FROM THE SIDE NOT DIRECTLY IMPACTED.
10
Typical results pertaining to the spatial distribution of fiber axial stress under the impact conditions identical to the
ones stated in conjunction with Fig. 7(a)-(d) are shown in Fig. 8(a)-(d). In this case, the deformed mesh of the fabric
and the projectile are shown. Examination of the results displayed in Fig. 8(a)-(d) confirms the general
experimental findings reported in Ref. [21] that fabric penetration is a non-local process. In other words, fiber
damage is typically initiated in the fabric region which is not directly in contact with the projectile but contains
high-potency flaws/defects. The resulting redistribution of the stress within the fibers and the formation of high
stress regions leads to spreading of damage towards the region impacted by the projectile. Thus, fabric penetration
is no longer a local phenomenon and is highly affected by the stochastic and distributed nature of fabric failure
strength.
FIGURE 8. SPATIAL DISTRIBUTION OF FIBER AXIAL STRESS DURING IMPACT BY A 2.1 g STEEL SPHERICAL PROJECTILE TRAVELING
AT AN INITIAL VELOCITY OF 63 m/s FOR THE CASE OF STOCHASTIC FRACTURE STRENGTH AT POST-IMPACT TIMES OF:
It should be noted that the distribution of fiber axial stress are shown at different post-impact times in Figures 6(a)
(d) and 8(a)(d). The reason is that the times selected were chosen to best make the point that in the case of the
deterministic-fiber-strength, Figures 6(a)(d), failure is a local phenomenon, while in the case of the stochastic
fiber-strength, Figures 8(a)(d), failure is a non-local phenomenon.
FIGURE 9. VARIATION OF PENETRATION FREQUENCY OF THE FABRIC BY THE PROJECTILE WITH THE MAGNITUDE OF THE FIBERFIBER FRICTION COEFFICIENT FOR THE CASE OF: (A) DETERMINISTIC FIBER-FRACTURE STRENGTH; AND (B) STOCHASTIC FIBERFRACTURE STRENGTH.
11
The effect of the magnitude of the fiber-fiber friction coefficient on the penetration frequency of the fabric by the
projectile for the case of deterministic fiber-fracture strength is displayed in Fig. 9(a). Clearly, due to the
deterministic nature of the fiber strength, the results obtained under nominally identical conditions are fully
repeatable. Hence, the curves shown in Fig. 9(a) are of a step-function type. The results displayed in this figure
confirm the general experimental findings in Ref. [21] that: (a) a higher value of the fiber-fiber friction coefficient
increases the fabrics penetration resistance (as quantified by the projectile velocity at which the penetration
frequency changes from zero to one); and (b) the effect of friction coefficient on the fabric penetration resistance is
relatively weak.
The effect of the magnitude of the fiber-fiber friction coefficient on the penetration frequency of the fabric by the
projectile for the case of stochastically distributed fiber fracture strength is displayed in Fig. 9(b). Due to the
stochastic nature of the fiber strength, the penetration frequency vs. projectile strike velocity curves no longer
resembles a step function. Instead, there is a range of projectile velocities between the maximum velocity at which
no penetration occurs and a minimum velocity at which penetration always occurs. The results displayed in this
figure confirm the general experimental findings in Ref. [21] that:
(a) There is an optimal value of the fiber-fiber friction coefficient which is associated with the maximum value of
the fabric penetration resistance. The existence of an optimum friction coefficient is generally rationalized in the
following way. At lower values of the friction coefficient, the fibers can be readily pushed laterally by the projectile,
and the fabric defeated without, or with little, damage to the fibers. At too-high values of the friction coefficient,
fibers are over-constrained and get defeated by the projectile by transverse shear or transverse crushing before
experiencing sufficient axial scratching (the latter deformation mode is of key importance for the ballistic
performance of fabric); and (b) The effect of fiber-fiber friction coefficient on fabric penetration resistance is
substantially stronger in the case of the stochastic nature of the fiber strength than in the deterministic-case.
Additional results (not shown for brevity) were also generated in this portion of the work and their analysis and
comparison with the experimental results in Ref. [21] revealed:
(a) Failure of a (stretched) fiber is followed by a fairly complex stress-redistribution/unloading process which can
be divided into three well defined stages:
(i) the propagation of the release waves from the fractured ends of the two fiber fragments towards the
clamped end of these fragments. The swept region of the fiber fragments is fully relieved of the prior tensile
longitudinal stress;
(ii) the release waves become arrested at one of the yarn/yarn crossover points producing locally a high level of
tensile stress; and
(iii) transmitted release waves are emitted from the high-stress region towards the clamped end of the fiber
fragments, reducing the extent of stress concentration and increasing the average level of tension in the
loaded portion of the fiber fragments;
(b) the distance travelled by the release waves, before their arrest, scales inversely with the magnitude of the fiberfiber friction coefficient;
(c) if and when the enhanced average level of tension defined in (a),(iii) exceeds the fiber local failure strength, the
fiber fracture as well as the aforementioned stress-redistribution processes reoccur;
(d) in the case of the reoccurring fracture of the fibers, the average fragment length scales inversely with the
magnitude of the fiber-fiber friction coefficient;
(e) the deterministic vs. stochastic nature of the fiber strength affects both the magnitude of the fabric ballistic
penetration resistance, the morphology of the fabric-fracture process as well as the role of the fiber-fiber friction
coefficient;
(f) when fiber strength is treated as a deterministic quantity, fabric failure appears to be a local phenomenon and
occurs in the highest-stress region (the region which is in direct contact with the projectile tip). In this case, a lower
magnitude of the friction coefficient gives rise to a longer length of the tension-free region of the fractured fiber(s).
12
This enables the projectile to more readily push aside the fractured fibers and to defeat the fabric. However, due to
the local nature of the fiber failure process, the effect of the magnitude of fiber-fiber friction coefficient on the fabric
ballistic strength is relatively weak;
(g) when fiber strength is treated as a stochastic quantity, fabric failure appears to be a non-local phenomenon and
generally occurs not in the high stress region but rather in the regions associated with the most potent
microstructural imperfections/flaws in the fibers. Typically, these regions are located outside the impacted fabric
region and the associated release waves mentioned above can cause a spread of the fiber fracture into the fabricimpact region. In other words, the fabric fracture process becomes a non-local phenomenon in this case. Since the
average fiber fragment length is controlled by the fiber-fiber friction coefficient, the magnitude of the friction
coefficient tends to play a more significant role than in the deterministic case; and
(h) when the projectile incident velocity substantially exceeds V50 (the incident velocity of the projectile at which
the probability of the target defeat is 50%), the deterministic vs. stochastic nature of the fiber strength has little
effect on the projectile residual velocity or on the projectile/fabric interaction process in general. The fabric failure
in this case is a local phenomenon occurring in the fabric impacted region. That is, although in the stochastic case
the fabric initial failure can occur outside the impacted region, insufficient time is available for this fracture to
spread into the impact region before full penetration of the fabric takes place.
Summary and Conclusions
1. A finite-element-based fiber-length-scale model has been developed for Kevlar KM2 ballistic fabric.
2. The model is used to reveal the mechanical response of a fabric test patch subjected to a transverse impact by a
spherical projectile.
3. The model is also used to reveal the role of the transverse properties of the fibers, fiber/fiber and fiber/projectile
friction coefficient and the stochastic nature of the fiber strength on the penetration-resistance of the fabric.
4. The model is validated by comparing its key predictions with the available experimental counterparts.
ACKNOWLEDGEMENTS
The material presented in this paper is based on work supported by an Army Research Office (ARO) sponsored
grant entitled Friction Stir Welding Behavior of Selected 2000-series and 5000-series Aluminum Alloys (Contract
Number W911NF-11-1-0207). The authors are indebted to Dr. Asher Rubinstein of ARO for his continuing support
and interest in the present work.
REFERENCES
[1]
Wittman, R. E., and R. F. Rolsten, Armor of Men and Aircraft, 12th National SAMPE Symposium, SAMPE, 1967.
[2]
The Interceptor System, US Marine Corps, (http://www.marines.mil/marinelink/image1.nsf/lookup/200532317129?opendocument).
[3]
Grujicic, M., B. Pandurangan, D. C. Angstadt, K. L. Koudela and B. A. Cheeseman. Ballistic-Performance Optimization of

a Hybrid Carbon-Nanotube /E-glass Reinforced Poly-Vinyl-Ester-Epoxy-Matrix Composite Armor. Journal of Materials
Science 42(2007): 53475359.
[4]
LaBarre, E. D., X. Calderon-Colon, M. Morris, J. Tiffany, E. Wetzel, A. Merkle
and M. Trexler. Effect of a carbon
nanotube coating on friction and impact performance of Kevlar. Multidiscipline Modeling in Materials and Structures
50(2015): 54315442.
[5]
Park, Y., Y. Kim, A. H. Baluch, C. G. Kim. Numerical simulation and empirical comparison of the high velocity impact of
STF impregnated Kevlar fabric using friction effects. Multidiscipline Modeling in Materials and Structures 125(2015): 520
529.
13
[6]
Tapie, E., V. P. W. Shim and Y. B. Guo. Influence of weaving on the mechanical response of aramid yarns subjected to
high-speed loading. International Journal of Impact Engineering 80(2015): 112.
[7]
Grujicic, M., W. C. Bell, L. L. Thompson, K. L. Koudela and B. A. Cheeseman. Ballistic-Protection Performance of CarbonNanotube Doped Poly-Vinyl-Ester-Epoxy Composite Armor Reinforced with E-glass Fiber Mats. Material Science and
Engineering A 479(2008): 1022.
[8]
Grujicic, M., W. C. Bell, S. B. Biggers, K. L. Koudela and B. A. Cheeseman. Enhancement of the Ballistic-Protection
Performance of E-glass Reinforced Poly-Vinyl-Ester-Epoxy Composite Armor via the Use of a Carbon-Nanotube ForestMat Strike Face. Journal of Materials: Design and Applications 222(2008): 1528.
[9]
Grujicic, M., A. Hariharan, B. Pandurangan, C-.F. Yen, B. A. Cheeseman, Y. Wang, Y. Miao and J. Q. Zheng. Fiber-level
Modeling of Dynamic Strength of Kevlar KM2 Ballistic Fabric. Journal of Materials Engineering and Performance
21(2012): 11071119.
[10] Grujicic, M., W. C. Bell, P. S. Glomski, B. Pandurangan, C-F.Yen and B. A. Cheeseman. Filament-level Modeling of
Aramid-based High-Performance Structural Materials. Journal of Materials Engineering and Performance 20(2011): 1401
1413.
[11] Grujicic, M., W. C. Bell, P. S. Glomski, B. Pandurangan, C-F. Yen and B. A. Cheeseman. Multi-length Scale Computational
Derivation of Kevlar Yarn-level Material Model. Journal of Materials Science 46(2011): 47874802.
[12] Grujicic, M., S. Ramaswami, J. S. Snipes, R. Yavari, G. C. Lickfield, C.-F. Yen and B. A. Cheeseman. Molecular-Level
Computational Investigation of Mechanical Transverse Behavior of p-phenylene terephthalamide (PPTA) Fibers.
Multidiscipline Modeling in Materials and Structures 9(2013): 462498.
[13] Grujicic, M., R. Yavari, J. S. Snipes, S. Ramaswami, C.-F.Yen, and B. A. Cheeseman. Molecular-Level Study of the Effect of
Prior Axial Compression/Torsion on the Axial-Tensile Strength of PPTA Fibers. Journal of Materials Engineering and
Performance 22(2013): 32693287.
[14] Grujicic, M., S. Ramaswami, J. S. Snipes, R. Yavari, C.-F. Yen and B. A. Cheeseman. Axial-Compressive Behavior,
Including Kink-Band Formation and Propagation, of Single p-Phenylene Terephthalamide (PPTA) Fibers. Advances in
Materials Science and Engineering Volume 2013, Article ID 329549, 2013. doi: 10.1155/2013/329549
[15] Grujicic, M., R. Yavari, J. S. Snipes, S. Ramaswami, C.-F. Yen, and B.A. Cheeseman. The Effect of Plain-Weaving on the
Mechanical Properties of Warp and Weft P-Phenylene Terephthalamide (PPTA) Fibers/Yarns. Journal of Materials
Science 49(2014): 82728293.
[16] Grujicic, M., W. C. Bell, T. He and B. A. Cheeseman. Development and Verification of a Meso-scale based Dynamic
Material Model for Plain-woven Single-ply Ballistic Fabric. Journal of Material Science 43: 63016323, 2008.
[17] Grujicic, M., W. C. Bell, G. Arakere, T. He and B. A. Cheeseman. A Meso-Scale Unit-Cell Based Material Model for the
Single-ply Flexible-Fabric Armor. Materials and Design 30: 36903704, 2009.
[18] Chen, M. W., and T. Weerasooriya. Experimental Investigation of the Transverse Mechanical Properties of a Single
Kevlar KM2 Fiber. International Journal of Solids and Structures 41(2004): 62156232.
[19] Grujicic, M., G. Arakere, T. He, M. Gogulapati and B. A. Cheeseman. A Numerical Investigation of the Influence of YarnLevel Finite-Element Model on Energy Absorption by a Flexible-fabric Armor During Ballistic Impact. Journal of
Materials: Design and Applications 222(2008): 259276.
[20] Grujicic, M., W. C. Bell, G. Arakere, T. He, X. Xie, B. A. Cheeseman, Development of a Meso-scale Material Model for
Ballistic Fabric and its Use in Flexible-armor Protection Systems, Journal of Materials Engineering and Performance, 19(1):
2239, 2010.
[21] Miao, Y., Y. Wang, J. Yu, C-F. Yen, D. Swenson and B. A. Cheeseman, Effects of Fiber Transverse Properties on Fabric
Penetration Resistance, International Journal of Impact Engineering, submitted for publication, February 2011.
[22] Nilakantan, G., M. Keefe, J. W. Gillespie Jr. and T. A. Bogetti. Modeling the Material and Failure Response of Continuous
14
Filament Fabrics for use in Impact Applications. TexComp 9 - International Conference on Textile Composites, Newark,
DE: DEStech Publications, 205214, 2008.
[23] Dassault Systems, ABAQUS Version 6.14, User Documentation, 2014.
[24] Grujicic, M., J. S. Snipes, S. Ramaswami, R. Yavari, C-F.Yen and B. A. Cheeseman. Analysis of Steel-with-Composite
Material Substitution in Military-Vehicle Hull-Floors Subjected to Shallow-Buried Landmine-Detonation Loads.
Multidiscipline Modeling in Materials and Structures 10(2014): 416448.
15

Finite-Element-Based Fiber-Length-Scale Modeling of The Ballistic-Impact Response of KEVLAR® KM2 Fabric

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Finite-Element-Based Fiber-Length-Scale Modeling of The Ballistic-Impact Response of KEVLAR® KM2 Fabric

Diunggah oleh

Hak Cipta:

Format Tersedia

Textiles and Light Industrial Science and Technology (TLIST) Volume 5, 2016

and the fiber density, , as

Formulation of the Computational Model

strain relation given as:

0.006e9 0.5296e9 1.2208e9 2 6.9417e9 3

7-fiber yarns required 3.5 hours of (wall-

Results and Discussion

Projectile Residual Velocity, m/s

Projectile Strike Velocity, m/s

FIGURE 3. VARIATION OF THE FABRIC BACK-SURFACE

FIGURE 4. VARIATION OF THE PROJECTILE

The Interceptor System, US Marine Corps, (http://www.marines.mil/marinelink/image1.nsf/lookup/200532317129?opendocument).

Grujicic, M., B. Pandurangan, D. C. Angstadt, K. L. Koudela and B. A. Cheeseman. Ballistic-Performance Optimization of

LaBarre, E. D., X. Calderon-Colon, M. Morris, J. Tiffany, E. Wetzel, A. Merkle

and M. Trexler. Effect of a carbon

Anda mungkin juga menyukai