Journal of Physics: Conference Series

OPEN ACCESS Related content

- Blast response of curved carbon/epoxy
Ballistic impact behaviour of woven fabric composite panels: Experimental study and
finite-element analysis
composite: Finite element analysis and V A Phadnis, P Kumar, A Shukla et al.

- Analysis of nonlinear deformations and

experiments damage in CFRP textile laminates
H Ullah, A R Harland, T Lucas et al.

To cite this article: V A Phadnis et al 2013 J. Phys.: Conf. Ser. 451 012019 - Ultrasonically assisted drilling: A finite-
element model incorporating acoustic
softening effects
V A Phadnis, A Roy and V V Silberschmidt

View the article online for updates and enhancements.

Recent citations
- Characterisation of smart CFRP
composites with embedded PZT
transducers for nonlinear ultrasonic
applications
Christos Andreades et al

- Analytical and Experimental Investigation

of Ballistic Impact on Thin Laminated
Composite Plate
Reza Mohamadipoor et al

- Ballistic damage in hybrid composite

laminates
Vaibhav A Phadnis et al

This content was downloaded from IP address 203.199.213.6 on 16/11/2020 at 17:49

D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

Ballistic impact behaviour of woven fabric composite: Finite

element analysis and experiments

V A Phadnis 1,3, K S Pandya2, N K Naik2, A Roy1 and V V Silberschmidt1

1
Wolfson School of Mechanical and Manufacturing Engineering, Loughborough
University, Loughborough, LE11 3TU, Leics, UK
2
Aerospace Engineering Department, Indian Institute of Technology, Powai,
Mumbai - 400076, India

E-mail: v.a.phadnis@lboro.ac.uk

Abstract. A mechanical behaviour of plain-weave E-glass fabric/epoxy laminate composite

plate exposed to ballistic impact is studied using a finite-element (FE) code Abaqus/Explicit. A
ply-level FE model is developed, where a fabric-reinforced ply is modelled as a homogeneous
orthotropic elastic material with potential to sustain progressive stiffness degradation due to
fiber/matrix cracking, and plastic deformation under shear loading. The model is implemented
as a VUMAT user subroutine. Ballistic experiments were carried out to validate the FE model.
A parametric study for varying panel thickness is performed to compare impact resistance of
the studied composite.

1. Introduction
In recent years, woven-fabric polymer-matrix composites (PMCs) have been used increasingly in
defence-related applications due to their higher energy absorption, high through-thickness stiffness
and strength properties. Additionally, their shape and properties can be tailored to meet the needs of a
specific application. High-performance fibers such as carbon, boron and Kevlar are of interest for
military and aerospace applications of composites primarily due to their resilience at high
temperatures and corrosion resistance in comparison to traditional structural materials.
In ballistic impact, PMCs retard a projectile by absorbing its kinetic energy through different
mechanisms such as deformation, delamination, and shear between layers. The condition for
perforation, also called the ballistic limit velocity (V50) is the most important factor for design of a
suitable protective structure [1]. In this regard, significant research has been carried out on the
behavior of composite materials under ballistic impact loading. Few studies can be mentioned: Zhu et
al. [2] investigated the response of woven Kevlar/polyester laminates of varying thicknesses to quasi-
static and dynamic penetration by cylindroconical projectiles. Ballistic limits were also determined
and terminal velocities measured. It was reported that deliberately introduced delamination and
changes in the volume fraction of fibres did not result in significant changes in the impact resistance.
Some studies revealed that a damage pattern for dynamic loading was, significantly different from that
in the corresponding quasi-static penetration conditions. Cheng et al. [3] developed a model for high-
3
Author to whom any correspondence should be addressed.

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

velocity impact of thick composites based on a continuum orthotropic constitutive behavior with
stress-based failure criteria and a simplified degradation model of failure. The model was implemented
into a hydrodynamic finite element code. Punching, fiber breakage, and delamination were the major
energy-absorbing mechanisms of the penetration processes. Silva [4] carried out experiments to study
effect of ballistic impact on Kevlar-29 impacted with simulated fragments. Numerical modelling was
used to obtain an estimate for the limit perforation velocity (V50) and simulate failure modes and
damage. Good correlation between results of computational simulation and experimental data was
reported, both in terms of deformation and damage of the laminates. Naik and Doshi [5] presented the
ballistic impact behavior of typical woven fabric E-glass/epoxy thick composites analytically. It was
reported that shear plugging was the major energy-absorbing mechanism.
Numerical models for predicting the performance of woven fabric composites have been the
subject of interest for many years. However, due to the complexity involved in describing the response
of fabric panels to ballistic impact, most models attempt to provide the most acceptable trade-off in
performance analysis. To model a woven fabric down to a level of individual yarn crossovers would
certainly be preferred in order to study frictional and crimping effects, but such studies are
computationally impracticable. The complexity of the problem has forced researchers to accept certain
simplifying assumptions to make it computable.
The work presented here focuses on the finite-element analysis of a ballistic impact response of
plain-weave E-glass fabric/epoxy composite using the general-purpose finite-element software
ABAQUS/Explicit [6]. The experimental details are discussed first, followed by a physics-based finite
-element model that accounts for both delamination and in-plain failure of fabric-reinforced composite
plies. Next, the obtained numerical results are compared with the experimental data and followed by
conclusions.

2. Experimental studies
Experimental studies were carried out using a single-stage gas-gun-operated ballistic-impact test
facility shown in figure 1. The material studied was commercially available plain-weave E-
glass/epoxy fabric composite. The specimens were prepared in a way to obtain symmetric balanced
cross-ply laminates of dimension 125 mm × 125 mm with thicknesses of 2.5, 3, 4.5 and 5 mm. The
mass of the projectile of diameter 6.36 mm made of hardened steel was 6.42 g. Experimental studies
were carried out on at least 6 specimens for each impact condition. Further experimental details are
provided in [1].

Impact chamber

Gas gun
barrel

Specimen
fixture

Figure 1. Ballisic impact test facility [1]

2
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

3. Constitutive model for fabric-reinforced composites

A constitutive material model representing the mechanical response of the plain-weave glass
fabric/epoxy composite at ply level was developed. Each ply was modelled as a homogeneous
orthotropic elastic material with the potential to sustain progressive stiffness degradation due to
fiber/matrix cracking and plastic deformation under shear loading. The model was implemented as a
material subroutine, VUMAT in ABAQUS [6]. Its main details are discussed below.

3.1. Elastic continuum-damage-mechanics model

The fibre directions of woven fabric reinforcement are assumed to be orthogonal. The constitutive
stress-strain relations are formulated in a local co-ordinate system with its base vectors aligned with
fibre directions. The in-plane elastic stress-strain relations are given according to orthotropic elasticity
with damage [7].

 1 ν 12 
 − 0 
 ε11   (1 − d1 ) E1 E1  σ 
   ν 21 1   11 
ε 22=  − 0  σ 22  , (1)
ε el   E2 (1 − d 2 ) E2  σ 
 12     12 
1
 0 0 
 (1 − d12 )2G12 

( ) ( )
T T
where ε = ε11 , ε 22 , ε12el is the elastic stain vector; σ = σ 11 , σ 22 , σ 12el is the stress vector; E1 and
E 2 are the Young’s moduli in the principal orthotropic directions; G12 is the in-plane shear modulus;
ν 12 and ν 21 are the principal Poisson’s ratios; d1 and d2 are the damage parameters associated with
the fibre fracture along the principal orthotropic directions; and d12 is the damage parameter
associated with the matrix micro-cracking due to in-plane shear deformation. The damage parameters
vary between 0 and 1 and represent the stiffness degradation caused by micro-damage in the material.
The material properties were taken from [1] owing to the similarity of specimens.

3.2. Elastic-plastic shear model

The in-plane shear response is dominated by a non-linear behavior of the matrix that exhibits both
stiffness degradation due to matrix micro-cracking and plasticity [8]. The elastic part of the shear
response was calculated using (1). Here, the plastic shear response of the material is considered. The
matrix response may be inelastic due to extensive cracking or plasticity. This leads to permanent
deformations in the ply upon unloading. To account for these effects, a classical plasticity model with
an elastic domain function and a hardening law, which is applied to the effective stresses in the
damaged material, is used. The elastic domain function F is given by

σ12 − σ 0 (ε pl ) ≤ 0 .
F= (2)

The hardening law is of the form:

σ 0 (ε =
pl
) σ y 0 + C (ε pl ) p , (3)

where σ y 0 is the initial effective shear yield stress; C and p are coefficients; and ε pl
is the equivalent

3
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

plastic stain due to shear deformation. The vales of σ y 0 , C and p are listed in Table 2.

4. Description of the finite element model

4.1. Intralaminar response of the plate

The in-plane response of the fabric plies was modelled using the constitutive model described. The
elasticity constants include the Young’s moduli in the fiber directions, the principal Poisson’s ratio,
and the in-plane shear modulus. The damage initiation coefficients account for tensile and
compressive strengths along the fiber directions and shear strength at the onset of shear damage. The
values of the user material constants and damage initiation coefficients are given in Table 1. The
damage-evolution coefficients are represented by tensile and compressive fracture energies per unit
area along the fiber directions and by parameters α12 and d12max .We assume α12 = 0.15 and d12max =
1, in our simulations. The fracture energies can be determined through the testing procedure described
in [9].

Table 1. Material properties of plain-weave E-glass fabric/epoxy composite [1, 9, 11]

Parameter Variable Value
Longitudinal modulus (GPa) E11 12.5

Transverse moduli (GPa) E 22 = E 33 11.8

Principal Poisson’s ratio ν 12 0.169

Shear moduli (GPa) G=

12
G=
23
G 31 3.5

Longitudinal tensile strength (MPa) X 1+ 322

Longitudinal compressive strength (MPa) X 1− 204

Transverse tensile strength (MPa) Y1+ 290

Transverse compressive strength (MPa) Y1− 200

In-plane shear strength (MPa) S 39

Mode-I fracture toughness (J/m2) GIC 504

Mode-II fracture toughness (J/m2) GIIC 1566

Ultimate failure strain ε max 4.2%

The critical values of the fracture energies along the fiber direction 2 were calculated so that the ratios
G 2f + / G1f+ and G 2f − / G1f− were approximately equal to the ratios Y2+ / X 1+ and Y2− / X 1− ,
respectively.

4
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

Table 2. Shear plasticity coefficients [11]

Parameter Variable Value

Initial effective shear yield stress (MPa) σ y 0 150

Hardening coefficient in equation (3) C 1125

Exponent in equation (3) p 1.05

4.2. Interlaminar response of plate

In order to account for adhesive bonds between adjacent plies and model the delamination
phenomenon, a generalized traction-separation law was used. Damage initiation at the interfaces was
defined using a quadratic nominal-stress criterion, according to which damage was assumed to initiate
when the quadratic interaction function involving the stress ratios reaches a value of one [10]:
2 2 2
 t n   t s   tt 
 0 + 0 + 0 = 1, (4)
 t n   t s   tt 

where t = {tn , ts , tt} is the nominal traction vector. The interface damage-initiation properties used in
the present analysis are listed in Table 3. These properties are usually difficult to determine
experimentally with good accuracy; therefore, they can be used as calibration parameters, if required.
In the present work, the variation of the damage initiation parameters within 15% from the typical
values given in Table 3 did not influence significantly the overall analysis results. It should be noted
that experimental determination of material properties typically yields a variation of ~15%.
Progression of damage at the interfaces was modelled using a critical mixed-mode energy behaviour
based on the Benzeggagh–Kenane (BK) law [7]:
η
G 
G + (G − G )  S  =
c
n
c
s
c
nGc , (5)
 GT 

where GS =
Gs + Gt and GT =
Gn + GS ; Gn , Gs and Gt refer to the work done by the normal and the
first and the second shear forces acting in the interface, respectively; Gnc and Gsc are critical fracture
energies required to cause failure in normal and shear directions, respectively. G c is the total critical
mixed-mode fracture energy and η is the power coefficient taken to be 2.284. After damage
initiation, its evolution until failure was modelled using exponential softening law [7].

Table 3. Interface damage-initiation properties

V
Parameter Variable
alue
Maximum nominal stress in normal mode (MPa) tn0 35

Maximum nominal stress in shear directions (MPa) tso = tto 65

4.3. Element deletion

In simulations, failed continuum shell elements were removed from the model to prevent excessive
distortion and premature termination of the analysis. In the present work, damage-based element

5
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

deletion is combined with a deformation-based deletion criterion. The damage-based element deletion
was activated when any damage variable along the fiber directions or the equivalent plastic strain due
to shear deformation reached a maximum specified value. The deformation-based deletion criterion
was used when either the tensile or compressive principal logarithmic strain reached its maximum or
minimum specified value, respectively. Additionally, detached composite fragments were deleted
when they moved far away from the impact zone to prevent non-physical numerical distortions.

4.4 Loading and boundary conditions

A schematic of the developed FE model is shown in figure 2. Both woven fabric plate and bullet were
modelled as 3D continuum deformable solids. The dynamic explicit solver was used in simulations to
account for the time-dependent loading and the complex interaction between the bullet and the fabric
composite plate. A 3 mm thick symmetric cross-ply laminate in this FE model consisted of 12 plies
with an individual ply thickness of 0.25 mm. The local co-ordinate systems were defied to account for
orientations of individual plies and to model the laminate and material behaviour precisely. The
cylindrical bullet of mass 6.42 g and length 25.3 mm impacts the centre of the workpiece in the axial
direction using a velocity boundary condition. The details are listed in Table 4.

Cylindrical bullet
V = 98 m/s Boundary
condition – plate
is fixed at all four
edges
2

Plain-weave E-
glass fabric /epoxy
3 1 composite plate

Figure 2. FE model: bullet impact on plain weave E-glass fabric/epoxy composite plate

A parametric study was carried out to analyse the effect of plate thickness on the ballistic limit
velocity. The mechanical properties and geometry of the bullet used earlier simulations were kept
constant. The plate thicknesses studied are listed in Table 4.

4.5 Finite elements and mesh sensitivity

In the rectangular composite plate, each ply was discretised with eight-node linear brick elements with
reduced integration (C3D8R) through its thickness in the vicinity of the impact zone. Mesh-sensitivity
analysis is important in simulations involving high deformations with a non-linear material response.
Thus, a rigorous mesh-sensitivity study was carried out to obtain a computationally accurate finite-
element mesh. In the current study all results are presented based on simulations performed with an
optimised mesh. A computing clock time was reduced by introducing different mesh sizes in distinct
regions of the FE model. A planar mesh size of 0.25 mm × 0.25 mm (in 1–2 plane) in the vicinity of

6
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

the impacted area was used; while a coarser mesh of 1.25 mm × 1.25 mm was employed in the area
away from the zone of interest.

Table 4. Loading and boundary conditions

Material studied Plain-weave E-glass fabric composite (symmetric
balanced cross-ply), thickness : 3mm
Parametric study Plate thickness : 2.5 mm, 4.5 mm and 5 mm
Impactor Cylindrical bullet (mass : 6.42 g, length : 25.3
mm)
Impactor velocity, V50 (m/s) 98 m/s
Boundary condition All four edges of plate fixed

Localised stiffness reduction due to internal damage can cause excessive element distortion that
could lead to difficulties in numerical convergence. To resolve this numerical issue, ‘distortion
control’ was used in ABAQUS, and the damage variables were limited to a maximum value of 0.999.
At each ply interface, cohesive elements of type COH3D8 with a thickness of 10 µm were embedded
and used to model delamination initiation and growth with the failure criterion discussed in Section
4.2. The degradation parameters were set to 0.99, and failed cohesive elements were removed from the
FE model, once the failure criteria were satisfied.
It was observed in the wave-stability study that cohesive zone elements govern stability of the
solution (due to near-zero element size) with a very low stable time increment of the order of 10-9 s
which affected the overall solution run time and, hence, a selective mass-scaling technique was used
[10]. The density of cohesive-zone elements was artificially increased by a factor of 25 so that the total
mass of the laminate remained practically unaffected. The bullet was meshed with one-integration-
point, hexahedral elements of type C3D8R.
The model consisted of 575000 finite elements and analysis was completed in two and half hours
using eight parallel processors on Intel Core i7 CPU with 8 GB RAM.
5. Simulation results and discussion
The effect of increase in the thickness of composite plate on the ballistic limit velocity (V50) is shown
in figure 3 and the magnitudes of V50 are listed in Table 5.

Table 5. Magnitude of ballistic limit velocity (V50) at varying composite plate thickness
Thickness of composite plate Ballistic limit velocity (V50)
(mm) FE Analysis Experiment

2.5 73.5 72
3 98 98
4.5 118.5 118
5 130.5 130

It can be observed form Table 5 that, the increase in plate thickness by 66% (considering t = 3 mm
as the base plate thickness) resulted in a higher ballistic limit increase by 33%, while decrease in plate
thickness by 16% caused reduction in V50 by 25%.
It was found that the kinetic energy absorbed by the composite plate during impact did not increase
linearly with increase in the plate thickness, thus the primary reason of this increase in the ballistic
limit can be attributed to the prolonged contact. It can be safely assumed that at high loading rates, as
normally observed in ballistics, the glass fabric undergoes considerable hardening, before failure –
thus providing extended resistance.

7
D2FAM 2013 IOP Publishing
Journal of Physics: Conference Series 451 (2013) 012019 doi:10.1088/1742-6596/451/1/012019

Figure 3. Effect of composite plate thickness on ballistic limit velocity (V50)

6. Conclusions
The ballistic impact behaviour of plain-weave E-glass fabric/epoxy composite was studied
experimentally and numerically. A ply-level 3D finite element model was developed to analyse the
response of the plate of this composite impacted by a cylindrical bullet. This FE model was validated
with the experimental results and later used to predict the ballistic limit velocity for varying plate
thicknesses. A good correlation was obtained with the experimental data emphasising the effectiveness
of this model to predict the mechanical response of woven fabric composites under high-velocity
impact loading.

Acknowledgement
The authors are grateful to the Royal Society, UK and Department of Science and Technology,
Government of India for the partial support of this study.

References
[1] Pandya KS, Pothnis JR, Ravikumar G and Naik NK 2013 Mater. Design. 44 128-135
[2] Zhu G, Goldsmith W and Dharan CKH 1992 Int. J. Solids Struct. 29 399–420
[3] Cheng WL, Langlie S and Itoh S 2003 Int. J. Impact Eng. 29 167–84
[4] Silva M 2005 Int. J. Impact Eng. 31 289–306
[5] Naik N and Doshi A 2008 Compos. Struct. 82 447–64
[6] Abaqus 6.11 2011 User manual Dassault System Rhode Island US
[7] Maimí P, Camanho PP, Mayugo JA and Dávila CG 2006 A thermodynamically consistent
damage model for advanced composites. Technical Report NASA/TM-214282 Langley US
[8] Sokolinsky VS, Indermuehle KC and Hurtado JA 2011 Compos. Part A. 42(9) 1119-1126
[9] Pinho ST, Robinson P and Iannucci L 2006 Compos. Sci. Tech. 66 13
[10] Phadnis VA, Makhdum F, Roy A and Silberschmidt VV 2013 Compos. Part A. 47 41-51
[11] Ullah H, Harland AR and Silberschmidt VV 2012 Comput. Mater Sci. 64 130-135