You are on page 1of 20

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320736489

Impact Modelling of Kevlar Fabric Composite Panels

Conference Paper · May 2017

CITATION READS
1 826

1 author:

Gang Li
National Research Council Canada
70 PUBLICATIONS   652 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Riveted lap joints View project

composite joints View project

All content following this page was uploaded by Gang Li on 31 October 2017.

The user has requested enhancement of the downloaded file.


Impact Modelling of Kevlar Fabric Composite Panels
Gang Li
Structures, Materials, and Manufacturing Laboratory, Aerospace
National Research Council Canada, 1200 Montreal Road, Ottawa, ON, Canada K1A 0R6
Gang.Li@nrc-cnrc.gc.ca, (613) 990-4989

Abstract
A three-dimensional explicit finite element low-velocity impact model of Kevlar fabric composite panels
was developed. This model was validated using test results from drop-weight impact on a 4.7 mm thick
panel. Good agreement was obtained in both deflection and impact force responses. The model was then
extended to simulate the behaviour of bonded layered Kevlar panels under low-velocity impact. Further,
perforation simulations were conducted for single and layered Kevlar panels. In the perforation
simulations, three-dimensional Hashin damage criteria were used for the Kevlar panel. The simulation
results obtained on deflection, impact force, local failure, and critical impact velocity were used to assess
the panel ballistic impact performance. The developed explicit model provides a cost-effective approach
for impact damage analysis of both aircraft composite structures and personal protective armour
equipment made of composite materials.
Keywords: Composites; Damage; Finite element; Impact; Perforation

1. Introduction
Impact damage is one of the major threats for the new generation of aircraft made with a considerable
amount of composite materials. This damage can lead to premature failure due to stiffness or strength
degradation and accelerated ageing from reduced protection from the environment. Impact damage can
be induced by ground service equipment, hail, tool drop, etc. Impact simulation with limited test data can
provide insightful information to support test task screening, assess the structural response to impact and
predict the effect of the resulting damage on structural integrity.
The mechanical performance of laminated composites and Kevlar fabrics is considerably affected by
impact velocities. Low-velocity impact can create damage like delamination, matrix cracking, and/or
fibre breakage [1-7]. An explicit model with the Hashin damage criteria and the cohesive zone modelling
technique can be used to simulate the formation and predict the extent of the induced damage. However,
the computational cost associated with these techniques is very high. In [1], a three-dimensional (3D)
finite element (FE) model was developed to simulate delamination initiation and propagation in a 24-ply
laminate [±45/90/0/45/04/45/02]s with cohesive elements inserted at each interface. This model required
4 to 5 days to complete a simulation using 32 CPUs. For another simulation work reported in [7], it took
over 90 hours to complete an analysis for a model with 112000 solid elements and only 8 surface-based
cohesive contact interfaces inserted into a 32 ply laminate [45/0/45/90]4s.

1
Results show that the woven fabric laminates are more resistant to impact damage [7]. For woven Kevlar
KM2 fabric panels, limited damage has been observed from low-velocity impact experiments [8] and
limited information on the modelling of woven Kevlar fabrics is reported in the open literature.
High-velocity impact generally creates perforation in the targeted panel. The material mechanical
behaviour can be considerably different under high strain rates. Zhu et al. [9] experimentally investigated
the strain rate effect on the failure behaviour of Kevlar 49 fabric. Their results show that the dynamic
material properties in terms of Young’s modulus, tensile strength, maximum strain, and toughness
increase as the strain rate increases. For Kevlar 49, they found that these values were approximately 120
GPa, 1500 MPa, 0.028, and 24.5 MPa, respectively, at a strain rate of 22 s-1. They increased to
approximately 145 GPa, 2375 MPa, 0.036, and 41 MPa, respectively, at a strain rate of 160 s-1. Shim et
al. [10] found that for the Twaron fabric the fibres become stiffer at high strain rates, whereas the strain to
failure is reduced. Tan et al. [11] tested the perforation behaviour of the Twaron woven fabric using
projectiles of different shapes. The energy absorption capability of the fabric was explained by
considering how much impact energy was converted to strain energy and kinetic energy. Each projectile
shape was also found to perforate the fabric through different mechanisms: yarn rupture, fibrillation,
failure by friction, and bowing. Moreover, Twaron, like most polymers, is viscoelastic and this property
becomes significant in impact problems. Similar studies about strain rate effects and perforation
behaviour were not found in the open literature for Kevlar fabrics.
This report presents impact simulations for single and bonded layered Kevlar fabric panels. Three-
dimensional (3D) FE analyses were carried out using a commercial software package, Abaqus/CAE,
version 6.14r2 [12]. The objective of this work was to develop and demonstrate a numerical methodology
capable of simulating the impact response of materials like Kevlar fabrics and laminated composite
panels. Low-velocity impact and high-velocity ballistic perforation simulations were conducted for
mpnolithic Kevlar fabric panels and bonded layered Kevlar-bond-Kevlar panels. The effects of the Kevlar
material parameters, adhesive thickness, and adhesive material parameters on the panel impact response
were investigated. For the low-velocity impact condition, the model did not consider the formation of
damage. For the perforation simulations, failure in the Kevlar fabric panel and the adhesive layer was
considered. The 3D Hashin damage criteria were employed for the Kevlar fabric panel [13]. Future work
is suggested for accurately modelling the impact response and damage of composite structures used in
either aircraft or personal protective armour equipment.

2. Experimental Information and Material Parameters

A 203x203x4.7 mm laminated flat Kevlar panel made of 20 plies of plain weave Kevlar KM2-600 fabric
reinforced polyamide (PA12) was tested using the drop-weight method. Fig. 1 shows the test setup [8].
The support was made of vinyl nitrile foam approximately 20 mm thick, with a density of 108 kg/m3, and
was on a 12 mm thick steel plate. A “Kevlar/foam/steel” piled structural configuration was used for the
test. The distance between any two adjacent clampers on the Kevlar panel was approximately 180 mm.
Fig. 2 shows the geometry of the steel striker. The mass of the striker was 0.764 kg and the total mass of

2
the drop-weight assembly was 6.786 kg. The impact energy could be adjusted and in-situ results such as
the impact force and panel deflection were captured [8, 14].

Fig. 1. Drop-weight impact setup.

(a) Striker drawing (dimensions: inch) (b) Photo of the striker

Fig. 2. Steel striker used in the drop-weight impact.

3
Limited material parameters, mostly only the Young’s modulus and/or possible tensile strength, could be
obtained via a literature survey. Values were assumed for the missing parameters to meet the modelling
requirements. Table 1 summarizes material parameters obtained from the open literature and highlights
the parameters that were used for the current modelling work. Considerable scatter in Kevlar parameters
was noticed. A total of four sets of Kevlar material parameters were deduced from available data and
were implemented in the simulations.

Table 1. Material properties for the materials used in the drop-weight impact test.

Mechanical parameters obtained from


Material Reference Parameters for current FE simulations
literature

Density 𝜌 = 7800 kg/m3, 𝐸 = 210 GPa, Density 𝜌 = 7800 kg/m3, 𝐸 = 210 GPa,
Steel ----
𝑣 =0.3 𝑣 =0.3

𝜌 = 108 kg/m3 [14, 15]


For 𝜌 = 16 to 35 kg/m3 foam: 𝐸 = 0.3 𝜌 = 108 kg/m3, 𝑣 =0.12,
Foam to 1 MPa. Trials on 𝐸 = 4 MPa and then 0.3 MPa.
[16]
For 𝜌 = 70 to 115 kg/m3 foam: 𝐸 = 4
to 12 MPa.
𝜌 = 1220 kg/m3 Set 1: KMat1:
0 0 0 0 0
𝐸11 = 𝐸22 = 22.4 GPa, 𝐸33 = 1.5 GPa, [15] 𝜌 = 1220 kg/m3, 𝐸11 = 𝐸22 = 22.4 GPa,
0 0 0 0 0 0 0
𝐺12 = 1.8 GPa, 𝐺13 =𝐺23 = 0.5 GPa, 𝐸33 =1.5 GPa, 𝐺12 = 1.8 GPa, 𝐺13 =𝐺23 =
𝑣12 =0.03, 𝑣13 =𝑣23 =0.41 0.5 GPa, 𝑣12 = 𝑣13 =𝑣23 =0.03
𝜌 = 1220 kg/m3 Set 2: KMat2:
0 0 0 0 0
𝐸11 = 𝐸33 = 5.82 GPa, 𝐸22 = 1.5 GPa, [15] 𝜌 = 1220 kg/m3, 𝐸11 = 𝐸22 = 5.82 GPa,
0 0 0 0 0 0 0
𝐺12 = 1.8 GPa, 𝐺13 =𝐺23 = 0.5 GPa, 𝐸33 =1.5 GPa, 𝐺12 = 1.8 GPa, 𝐺13 =𝐺23 =
𝑣12 =0.03, 𝑣13 =𝑣23 =0.41 0.5 GPa, 𝑣12 = 𝑣13 =𝑣23 =0.03
0
𝐸11 =51 GPa for Kevlar 29 fabric; Set 3: KMat3:
0 0
0 𝜌 = 1220 kg/m3, 𝐸11 = 𝐸22 = 50 GPa,
𝐸11 =78 GPa for Kevlar Ht; [17] 0 0 0 0
Kevlar 𝐸33 =2.5 GPa, 𝐺12 = 1.8 GPa, 𝐺13 =𝐺23 =
0
𝐸11 =84 GPa for Kevlar 49 fabric 0.5 GPa, 𝑣12 = 𝑣13 =𝑣23 =0.03
Kevlar 29 fabric: 𝜌 = 1440 kg/m3
0
𝐸11 = 62 GPa, 𝑋1𝑡 = 2600 MPa (tensile [18, 19]
strength).
0
Set 4: KMat4:
Kevlar KM2: 𝐸11 = 62 GPa, 𝑋1𝑡 = 3.4 0 0
𝜌 = 1220 kg/m3, 𝐸11 = 𝐸22 = 62 GPa,
GPa; 0 0 0 0
0 0 0
[20] 𝐸33 =3.6 GPa, 𝐺12 = 4.5 GPa, 𝐺13 =𝐺23 =
𝐺12 , 𝐺13 , and 𝐺23 should be 2 or 3 1.5 GPa, 𝑣12 = 𝑣13 =𝑣23 =0.03
0
orders of magnitude smaller than 𝐸11 .
0
𝐸11 =64 GPa for neat Kevlar [21]
0
𝐸11 =71 GPa for Kevlar made of PPTA [22]

4
3. Drop-Weight Impact Modelling

3.1 Model for simulating the drop-weight impact test


Damage prediction was not included in the explicit model developed for the Kevlar KM2 panel under the
low-velocity impact, due to little damage observed from the test using 55 J of impact energy [8]. For the
high-velocity or perforation simulations, damage in Kevlar fabric panels was modelled using the 3D
Hashin damage criteria.
Fig. 3, (a) and (b), shows the FE model for a 4.7 mm square thick Kevlar panel with the foam/steel
backup used in the drop-weight impact test. In the model, the striker, with a mass of 0.787 kg, was
created and just contacted the Kevlar front surface at its centre. The remaining 6 kg mass of the striker
assembly was applied at a reference point at the mass centre of the impact assembly, 40 mm away from
the striker rear surface. This reference point was attached to the striker and hence the total mass of the
impact assembly would be the same as in the test. The impact velocity was assigned to the striker in the z
direction and the displacement were constrained to be zero in the 𝑥 and 𝑦 directions. The panel sides
were 180 mm long and its four vertices were centred on rigid circular surfaces 20 mm in diameter. The
four rigid surfaces were motionless and parallel to the panel surface, located within 𝑥𝑦 plane, and were
used only to restrict panel bounce-back during the impact via contact elements. In addition, contact pairs
between: “striker and Kevlar”, “Kevlar and foam”, “foam and steel”, and “all and self” were defined in
the “Contact Domain” within the “Interaction” module.
For the displacement boundary conditions (BCs) the foam was assumed to be tied with the steel plate and
the steel plate was fully constrained with zero displacement in the 𝑥, 𝑦, and 𝑧 directions. The Kevlar four
back surface corners were tied to their corresponding foam surface corners. These simplifications should
not affect the major response characteristics at the centre of the panel.
A total of 48066 elements were generated in the model. Linear hexahedron elements, C3D8R, were used
to mesh the panel and rear impact parts; linear tetrahedron elements, C3D4, were used to mesh the
impactor head part; linear quadrilateral surface elements, R3D4, were used to mesh the four rigid
surfaces. The Kevlar panel thickness was meshed uniformly using 8 elements and its four edges were
meshed with 40 elements with a double bias ratio of 5 to generate a fine mesh within its central area. As
a result, the largest elements, with a size of 8.96x8.96x0.59 mm, were located at the four edge corners,
and the smallest element, with a size of 1.57x1.57x0.59 mm, was at the panel centre, at the impact
location. Six uniform elements spanned the 20 mm foam thickness and 2 uniform elements spanned the
12 mm steel plate thickness. The foam edges were meshed with 45 elements with a double bias ratio of 5
to generate a fine mesh within its central area. Uniform 8 mm long elements were set to the four steel
plate edges. Fine mesh was also applied to the striker head part and rigid surfaces.
In Abaqus, the “Hyperfoam” material was used for the foam layer by inputting several parameters without
test data. Through numerical trials, the foam Young’s modulus was found to be lower than 4 MPa to
generate impact force comparable with the experimental force results. The foam was then assumed as
flexible polymer foams (VLD) with Young’s moduli from 0.3 MPa to 1 MPa [16].

5
Rigid surface

Striker
Kevlar/foam/steel

(a) A 3D view of the FE model

69.8 mm
25.4 mm

25.4 mm

15 mm

15.08 mm
Kevlar: 4.7 mm
Foam: 20 mm thick
Steel: 12 mm thick

180 mm

(b) A side view of the FE model

Fig. 3 Three-dimensional FE model, viewed from a 3D (a) and a side (b) angle conditions, for simulating
drop-weight impact testing on a 4.7 mm thick Kevlar panel. The Kevlar panel was placed on a 20 mm
thick foam that was on a 12 mm thick steel plate (i.e.: a foam/steel backup).

6
Numerical trials were conducted due to the paucity of material properties, it was found that the peak
impact force, obtained from the explicit model using set 1 Kevlar material properties and the 4 MPa
Young’s modulus foam (Table 1), was more than twice as high as the peak force obtained from the
experimental impact test. Therefore, the small 0.3 MPa foam Young’s modulus was used for the model.

3.1.1 Effects of foam and striker stiffness on the impact force variation
Fig. 4 shows the comparison of the impact force variations obtained from the experimental and numerical
results. Information on the material properties used in the simulation is provided in Table 2. Two
different foam materials with the same 0.3 MPa Young’s modulus and 0.12 Poisson’s ratio were assessed:
hyperfoam [12] and linear elastic foam. The effect of the striker stiffness, rigid and elastic, was also
studied. It can be seen from Fig. 4 that (i) almost identical predictions were obtained between the two
simulations, and (ii) the peak impact force was overestimated by approximately 15% by the numerical
models, as compared to the testing. Results show that the effects of the foam behaviour (hyperfoam and
linear elasticity) and striker stiffness on the impact force variation were very limited, and that the model
worked very well and gave reasonable predictions. The only difference was the computational cost. The
model with the rigid striker took less than half of the total time ( 3 hours in rigid striker model vs  6
hours in deformable striker model) that was used in the model with the deformable striker. The
discrepancy between the experimental and numerical results could also be caused by the assumptions
made concerning the material behaviours, for both Kevlar and foam.

Fig. 4. Variation of impact force versus time obtained from experimental and numerical results of a 4.7
mm thick Kevlar panel with the foam/steel backup under a 55 J energy impact.

7
Table 2. Additional information about the modelling results presented in Fig. 8 for the Kevlar with the
foam/steel backup under a 55 J energy impact and the experimental results [2].

Label Note Striker Kevlar fabric panel Foam


Exp: 55 J impact of the Vinyl nitrile foam
Deformable
Kevlar fabric coupon in Test Kevlar KM2-600/PA12 with density 𝜌 =108
steel
B3462_xx [1] kg/m3

FE-def-striker: Fz- Set 1 material behaviour: Hyperfoam (with N=1,


Deformable
KMat1-hyperfoam1- FEM 𝜌 = 1220 kg/m3, 𝜇1 =0.137 N/mm2,
steel 0 0
steel 𝐸11 = 𝐸22 = 22.4 GPa, 𝛼1 =2, 𝑣1 =0.12 [12])
0 0
𝐸33 =1.5 GPa, 𝐺12 = 1.8
0 0
FE: Fz-KMat1-foam1- Assumed GPa, 𝐺13 =𝐺23 = 0.5 GPa, Linear elasticity: E =
FEM
steel rigid 𝑣12 = 𝑣13 =𝑣23 =0.03 0.3 MPa, v = 0.12

3.1.2 Effect of Kevlar material mechanical parameters on the impact response


Five numerical cases using the linear elastic foam model were assessed in this section. Comparisons of
the deflections and impact forces between the experimental and numerical results are shown in Figs. 5
and 6. The numerical labels, “foam1a” and “foam1” refer to the foam with 𝑣 = 0.12 Poisson’s ratio, 𝐸 =
0.2 MPa and 0.3 MPa Young’s modulus. “KMat 𝑖” (𝑖 =1 to 4) refers to the Kevlar material in property
set 𝑖 (Table 1). Only two simulations, “KMat1-foam1-steel” and “KMat3-foam1-steel”, were completed
successfully. The remaining three simulations, “KMat1-foam1a-steel”, “KMat2-foam1-steel”, and
“KMat4-foam1-steel”, were terminated at the peak deflection or force phase due to iteration divergence
because the ratio of deformation speed to wave speed exceeded 1.0 in at least one foam element.
Excessive element deformation was generated and led to convergence difficulty that terminated the job.
The effect of the material properties on the impact response is clearly visible in the two Figs. 5 and 6.
Stiffer Kevlar fabrics generated smaller deflection and higher impact forces during the impact period. For
instance, the impact peak force predicted by the model “KMat1-foam1-steel” was higher than that of the
model “KMat1-foam1a-steel”. The model “KMat2-foam1-steel” gave very good predictions for both
peak deflection and force using the set 2 Kevlar material properties suggested in [15]. However, the very
0 0
low Kevlar stiffnesses, 𝐸11 = 𝐸22 =5.82 GPa (see Table 1) made in [15] through artificially tuning the
parameters for fitting the test results, are not in the realistic range, as compared to the published data from
open literature [16-22]. The mechanical parameters usually increase as the strain rate increases [7-11].
Therefore, the results obtained from the “KMat2-foam1-steel” model may not be practical. The Kevlar
material property sets 3 and 4, “KMat3” and “KMat4”, were deduced from the open literature [17-22]. A
summary of the peak values is given in Table 3. Based on the current experimental data, the best
predictions were made using the set 1 Kevlar material. Overall, good agreement (2.6 to 25% difference
depending on the Kevlar material parameters) in the peak force variation trends was obtained between the
experimental and numerical results. The impact model gave reasonable predictions, as validated by
testing.

8
Fig. 5. Variation of deflection versus time curves obtained from experimental and numerical results of a
4.7 mm thick Kevlar panel with the foam/steel backup under a 55 J energy impact.

Fig. 6. Comparison of impact force versus time curves obtained from experimental and numerical results
of a 4.7 mm thick Kevlar panel with foam/steel backup under a 55 J energy impact.

9
Table 3. Comparison of peak values in deflection and force induced on the 4.7 mm thick Kevlar fabric
with foam/steel backup by the 55 J impact, where the experimental results were for coupon B3462_xx [8].
Relative difference in Relative difference in
Peak Peak impact
Model peak 𝑈𝑧 against the 17.04 peak force 𝐹𝑧 against
deflection 𝑈𝑧 force 𝐹𝑧
mm test result the 6222 N test result
KMat1-foam1a-steel 17.5 mm 2.6% 6062 N −2.6%
KMat1-foam1-steel 16.9 mm −1% 7155 N 15%
KMat2-foam1-steel 18.5 mm 8.5% 6378 N 2.5%
KMat3-foam1-steel 15.6 mm −8.7% 7596 N 22%
KMat4-foam1-steel 14.2 mm −16.5% 7796 N 25%

3.2 Simulation of layered Kevlar panels with foam/steel backup


After the above model validation, a layered Kevlar structure configuration was assessed numerically.
This layered configuration was made by adhesively bonding two 2.35 mm thick Kevlar panels to form a
“Kevlar-bond-Kevlar” structure that was used to replace the 4.7 mm thick single-layer Kevlar panel.
Fig. 7 shows the finite element model for the “Kevlar-bond-Kevlar” structure with the original foam/steel
backup. In this study, three different adhesive thicknesses, 0.5 mm, 1 mm, and 2 mm, were assessed for
the layered panel. The same loading condition and BCs were used, as described in Section 3.1. The
material properties used were: Kevlar in set 1 material property, adhesive in linear elasticity with 1 GPa
Young’s modulus and 0.34 Poisson’s ratio, and foam in linear elasticity with 0.3 GPa Young’s modulus
and 0.12 Poisson’s ratio and a density of ρ= 1200 kg/m3 of film adhesive AF163-2K [23] .
Fig. 8 shows deflection variations of the Kevlar panels with the foam/steel backup during the impact
period. Deflections of the layered Kevlar panels were smaller than the original monolithic 4.7 mm thick
Kevlar panel. Relative reductions at the peak deflection were approximately −2.7% for the 0.5 mm thick
adhesive, −5% for 1 mm thick adhesive, and −9.5% for the 2 mm thick adhesive, as compared to the
monolithic 4.7 mm thick Kevlar panel. Fig. 9 shows the impact force variations during the impact period.
The peak force increased and the deflection decreased as the adhesive layer thickness in the layered
panels increased. The increases of the peak force were approximately 0.8% for the 0.5 mm thick adhesive,
2.5% for the 1 mm thick adhesive, and 6.5% for the 2 mm thick adhesive in the layered Kevlar panels, as
compared to the 4.7 mm thick Kevlar panel. It can be noticed that the stiffness of the layered Kevlar
panels are higher than the original 4.7 mm thick Kevlar panel.

10
A 0.5 mm thick adhesive layer

A 1 mm thick adhesive layer

A 2 mm thick adhesive layer

Kevlar-bond-Kevlar
Foam: 20 mm thick
Steel: 12 mm thick

Fig. 7 Three-dimensional FE model to study effect of adhesive thickness on the impact response of a
layered Kevlar-bond-Kevlar structure, with the foam/steel backup, subjected to a drop-weight impact.

Fig. 8. Comparison of deflection versus time curves of Kevlar panels with the foam/steel backup under
55 J energy impact. The adhesive layer thicknesses were 0.5, 1, and 2 mm in the layered Kevlar panels.

11
Fig. 9. Comparison of impact force versus time curves of Kevlar panels with foam/steel backup under
55 J energy impact. The adhesive layer thicknesses were 0.5, 1, and 2 mm in the layered Kevlar panels.

3.3 Simulation of layered Kevlar panels without foam/steel backup


Fig. 10, (a) and (b), shows the finite element model for the Kevlar panels without the foam/steel backup.
A simple supporting condition, the displacement 𝑈𝑧 = 0 along one edge and 𝑈𝑥 = 𝑈𝑦 = 𝑈𝑧 =0 along the
other opposite edge of the bottom panel surface, was assumed to the layered Kevlar panel.

(a) A 3D view of the FE model (b) A side view of the FE model

Fig. 10 Three-dimensional FE model, viewed from a 3D (a) and a side (b) angle conditions, to study the
impact response of a layered Kevlar-bond-Kevlar structure under 55 J impact energy. The Kevlar panel
was simply supported at two opposite edges of the bottom panel surface.

12
3.3.1 Effect of adhesive thickness
The same material properties as in section 3.1.2 were used. Fig. 11 (a) shows the deflection variations
and Fig. 11 (b) shows the force variations. As compared to the original monolithic 4.7 mm thick Kevlar
panel, the peak deflection of the layered Kevlar panels were decreased by approximately 12.8% for the
0.5 mm thick adhesive, 22.3% for the 1 mm thick adhesive, and 35.8% for the 2 mm thick adhesive; the
corresponding peak force were increased in the layered Kevlar panels by approximately 14.4%, 28.4%,
and 54.5%. High adhesive thickness considerably increased the panel bending stiffness, which shortened
the time required to generate the peak deflection and force, as well as the entire impact period.
Another observation was made on the effect of displacement boundary conditions on panel response using
the tested monolithic 4.7 mm thick Kevlar panel for the comparison. Without the foam/steel backup
condition its peak deflection and force were approximately 20.7 mm and 5416 N respectively (labelled as
“4.7mm_thick_kevlar only” in Fig. 11), while the corresponding values were 16.9 mm and 7155 N for the
Kevlar panel with the foam/steel backup, labelled as “KMat1-foam1-steel” in Figs. 8 and 9. The relative
difference was approximately 23% in peak deflection and – 24% in peak force for the Kevlar panel only,
as compared to the panel with the foam/steel backup.

(a) (b)

Fig. 11. Comparison of (a) deflection and (b) force curves of the Kevlar panels under 55 J energy impact.
The Kevlar panels were simply supported at two opposite bottom edges.

3.3.2 Effect of adhesive material property


The same displacement boundary and loading conditions as in section 3.3.1 were used here. The effect of
adhesive material property on the impact response was studied using the layered Kevlar panels bonded
with a 2 mm thick adhesive. Fig. 12 shows the deflection and force variations under the 55 J energy

13
impact. Three adhesive Young’s moduli were used: 1 GPa (labelled in the figure as “-Adhe-Mat1”), 0.1
GPa (labelled as “-Adhe-Mat2”), and 0.01 GPa (labelled as “-Adhe-Mat3”). Results suggest that the most
compliant adhesive material, with the 0.01 GPa Young’s modulus, did not result in a higher bending
stiffness in the layered Kevlar panel than the monolithic Kevlar panel. Consequently, higher deflection
and smaller forces were generated in this layered Kevlar panel than with the 4.7 mm single Kevlar panel.
The other two adhesives with 1 GPa and 0.1 GPa Young’s moduli increased the bending stiffness in the
layered Kevlar panels. The relative differences in the peak values were approximately: (i) −35.8% in
deflection 𝑈𝑧 and 54.4% in 𝐹𝑧 for the “K-2mm_bond-K-Adhe-Mat1”, (ii) −24.8% in 𝑈𝑧 and +36.4% in 𝐹𝑧
for the “K-2mm_bond-K-Adhe-Mat2”, and (iii) +17.2% in 𝑈𝑧 and −16.6% in force 𝐹𝑧 for the “K-
2mm_bond-K-Adhe-Mat3”, as compared to the monolithic 4.7 mm Kevlar panel. Moreover, the impact
period decreased with the increase of the panel stiffness. It can be seen that substantial effect of the
adhesive material on the layered panel responses was displayed under the low-velocity impact condition.
In some applications, high deflection must be avoided, because it increases the potential damage to the
structures. Also, for other applications, such as an aircraft composite structure, the generated impact
force, eventually equilibrated by the load carrying element, must be less than a prescribed limit to avoid
damage. Good structural design needs to consider two conflict properties: deflection and impact force
magnitudes. Results showed that the modelling approach presented can be used for such a design
assessment.

(a) (b)
Fig. 12. Effect of adhesive material on (a) deflection and (b) force curves of the Kevlar panels under 55 J
energy impact. The Kevlar panels were simply supported at two opposite bottom edges of the bottom
surface.

4. Perforation Modelling of the 4.7 mm Thick Kevlar Panel

Perforation modelling was conducted to simulate Kevlar damage induced by impact. Fig. 13, (a) and (b),
shows the model for the 4.7 mm thick Kevlar fabric panel and layered Kevlar panels impacted by a

14
20 mm diameter steel ball in rigid assumption. The four edge faces of this Kevlar fabric were clamped
via 𝑈𝑥 = 𝑈𝑦 =𝑈𝑧 = 0. The 3D Hashin damage criterion was employed for the Kevlar fabric panel.
Referring to the set 1 Kevlar material parameters in Table 1, the Kevlar material parameters used were:

𝜌 = 1220 kg/m3
0 0 0 0 0
𝐸11 = 𝐸22 = 22.4 GPa, 𝐸33 = 1.5 GPa, 𝑣12 = 𝑣13=𝑣23 =0.03, 𝐺12 = 1.8 GPa, 𝐺13 = 0.5 GPa
0
𝐺23 = 0.5 GPa, 𝛽= 1E09
𝑋1𝑡 = 1000 MPa, 𝑋1𝑐 = 200 MPa, 𝑋2𝑡 =1000 MPa, 𝑋2𝑐 = 𝑋3𝑡 = 𝑋3𝑐 = 200 MPa
𝑆12= 𝑆13 = 𝑆23 = 200 MPa
𝑤here: 𝜌 = density;
0 0 0 0 0 0
𝐸11 , 𝐸22 , 𝐸33 , 𝑣12 , 𝑣13 , 𝑣23 , 𝐺12 , 𝐺13 , and 𝐺23 are initial elastic constants;
𝛽= damping factor (units of time, here is in ms),
𝑋1𝑡 = tensile failure stress in fibre direction;
𝑋1𝑡 = compressive failure stress in fibre direction;
𝑋2𝑡 = tensile failure stress in direction 2 (transverse to fibre direction);
𝑋2𝑐 = compressive failure stress in direction 2;
𝑋3𝑡 = tensile failure stress in direction 3 (transverse to fibre direction);
𝑋3𝑐 = compressive failure stress in direction 3 (transverse to fibre direction);
𝑆12 = failure shear stress in 1-2 plane;
𝑆13 = failure shear stress in 1-3 plane; and
𝑆23 = failure shear stress in 2-3 plane.

(a) Front view of the model (b) Side view of the model

Fig. 13 Three-dimensional FE model, front view (a) and size view (b), for perforation analysis of the 4.7
mm thick Kevlar fabric panel with its four side edge faces clamped. The 20 mm diameter steel ball rigid
striker was initially placed on the panel front surface centre.

15
A critical velocity of 93.5 m/s was identified for this Kevlar fabric panel through several simulations.
Fig. 14(a) shows the full-field von Mises stress contours on the panel front surface. Fig. 14(b) shows the
deflection of the panel back surface induced by the critical impact velocity (95.5 m/s). The ball did not
penetrate through the panel, even though the induced damage was completely through the panel. The
residual peak deflection was approximately −22.6 mm in the failed area after impact. Local ruptures
were created at the impacted area and its two major cracks formed a cross path pattern. In addition to the
material anisotropic behaviour, this failure pattern might also be affected by the rectangular mesh
condition used.
Fig. 15 shows full-field contours of the Kevlar panel impacted by the ball with a velocity of 94 m/s,
slightly higher than the critical value, thus causing the ball to penetrate completely through the panel.
Fig. 15(a) shows the von Mises stress distribution viewed from the panel front surface. Fig. 15(b) shows
the residual deflection observed from the panel back surface. Similar failure patterns were observed as the
93.5 m/s velocity. The peak deflection was approximately −24.4 mm.

(a) Front surface (b) Back surface

Fig. 14. Full-field contours for the (a) von Mises stress (MPa) on the front surface and (b) residual
deflection of the back surface with the 4.7 mm thick Kevlar panel being impacted by a 20 mm diameter
steel ball with a 93.5 m/s velocity. The ball did not penetrate through the panel.

16
(a) Front surface (b) Back surface

Fig. 15. Full-field contours for the (a) von Mises stress (MPa) on the front surface and (b) residual
deflection of the back surface with the 4.7 mm thick Kevlar panel being impacted by a 20 mm diameter
steel ball with a 94 m/s velocity. The ball penetrated through the panel.

5. Summary and Future Work

Due to the paucity of material mechanical parameters, behaviours induced by strain rate and
viscoelasticity for the Kevlar fabric, foam, and adhesive were not considered in the current work.
Simplifications in the material mechanical behaviours were used. Using accurate material mechanical
parameters and failure behaviours are crucial for achieving practical numerical predictions. Knowledge in
key material parameters and associated fundamental theories must be identified for setting up the explicit
model. In order to choose correct materials with reasonable parameters, understanding the relevant
constitutive relations and theories on damage mechanics is crucial.
Three-dimensional explicit FE models were developed and validated using drop-weight impact test results
on a 4.7 mm thick Kevlar fabric panel with foam/steel backup. Good agreement with the experiment was
obtained for deflection versus time and for impact force versus time results. The model was extended to
study the response of layered bonded Kevlar panels under low velocity impact. Perforation simulations
were performed on modelling a Kevlar panel. In the perforation simulations, a 3D Hashin damage model
was used along with assumed strength parameters for the Kevlar fabric panel. The results obtained for
deflection, local failure, and critical impact velocity could be used to assess ballistic impact performance.
This effort provided a sound base for further developing practical explicit models to support impact
damage in composite aircraft structures.
To improve the explicit model accuracy and applicability, some future work is suggested:
(i) Data collection for actual composite structures and associated materials behaviours including:
 Damage tolerance requirements (single hit and/or multi-hit) on strength, stiffness, anti-
perforation performance, and major threats types;

17
 Test information on penetration vs. time profile, striker and its velocity, ballistic deformation
range, and failure/damage features in actual structures;
 Mechanical parameters and behaviour determination of relevant composite materials
 Damage characterisation and damage mechanics of relevant materials (ceramic, Kevlar,
CFRP);
(ii) User-subroutine writing on damage criteria for ceramics, Kevlar, and laminated composites;
(iii) Integration of user-subroutine with Abaqus/CAE for explicit simulations;
(iv) Parametric study of layered composite and armour structures with different layer thicknesses
setup and shapes; and
(v) Critical impact velocity determination for representative structures and armours.
The overall effort is to develop cost-effective computational tools/approaches to support specific aircraft
composite structures damage tolerance assessment and residual strength analysis.

6. References
[1] Lopes C, Gurdal Z (2009), Simulation of low-velocity damage on composite laminates, in proceedings of
50th AIAA/ASME/ASCE/ASC Structures, Structural Dynamics, and Materials, 4-7 May 2009, Palm Springs,
California USA.
[2] Lopes CS, Camanho PP, Gürdal Z Maimí P, González EV (2009), Low-velocity impact damage on dispersed
stacking sequence laminates. Part I: Experiments, Composite Science and Technology, 69; 926-936.
[3] Lopes CS, Camanho PP, Gürdal Z Maimí P, González EV (2009), Low-velocity impact damage on dispersed
stacking sequence laminates. Part II: Numerical simulations, Composite Science and Technology, 69; 937-
947.
[4] Lopes CS, Sadaba S, Naya F, Gonzalez C (2014), Multiscale simulation strategy for low-velocity impact on
FRP, in proceedings of the American Society for Composites 29th Technical Conference, 8-11 September
2014, San Diego California, USA.
[5] Zhang J, Zhang X (2015), An efficient approach for predicting low-velocity impact force and damage in
composite laminates, Composite Structures, 130; 85-94.
[6] Long S, Yao X, Zhang X (2015), Delamination prediction in composite laminates under low-velocity impact,
Composite Structures, 132; 290-298.
[7] Yan B, Wang Z, Zhou L, Zhang J, Liang W (2015), Experimental and numerical investigation of interplay
hybrid composites based on woven fabric and PCBT resin subjected to low-velocity impact, Composite
Structures, 132; 464-476.
[8] Zhang Y, Benak T, Desnoyers R, Johnston A, MacMillan J (2009), Four-point bending and impact tests of
KM2-600/PA12 laminates with various matrix contents and skin-layer configurations, NRC Report, LM-
SMPL-2009-0157.
[9] Zhu D, Mobasher B, Rajan SD (2011), Dynamic tensile testing of Kevlar 49 fabrics, J Materials in Civil Eng,
23(3);1-10.
[10] Shim VPW, Lim CT, Foo KJ (2001), Dynamic mechanical properties of fabric armour. Int. J Impact Eng., 25;
1–15.
[11] Tan VBC, Lim CT, Cheong CH (2003), Perforation of high-strength fabric by projectiles of different
geometry, Int. J Impact Eng., 28; 207-222.
[12] Abaqus, User’s manual version 6.14r2.
[13] Abaqus, Section 24.3 Damage and failure for fiber-reinforced composites in Analysis user’s guide, volume
III: materials, version 6.14r2.
[14] Zhang Y, Desnoyers R, Johnston A (2009), Low-speed impact test for foam supported composite laminates,
NRC Report, LM-SMPL-2009-0089.
[15] Downes D (2010), Characterization of Kevlar and carbon laminates in LS-DYNA for applications involving
personal protection systems, NRC report, LTR-SMPL-2010-0120.
[16] Materials data book, 2003 edition, Cambridge University Engineering department.

18
[17] Martinez MA, Navarro C, Cortes R, Rodriguez J, Sanchez-Galvez V (1993), Friction and wear behaviour of
Kevlar fabric, J Materials Science, 28; 1305-1311.
[18] Yang HH (1993), Kevlar aramid fiber, John Wiley & Sons Inc, New York 1993.
[19] Sebastian MS, Unnikrishnan KC, Narayanan S (2008), Viscoelastic properties of Kevlar 29 fabric tape
strength member, Mechanics of Materials, 20; 949-960.
[20] Rao MP, Duan Y, Keefe M, Powers BM, Bogetti TA (2009), Modeling the effects of yarn material properties
and friction on the ballistic impact of a plain-weave fabric, Composite Structures, 89; 556-566.
[21] Bohannan A, Fahrenthold E (2008), Hypervelocity impact simulation using membrane particle-elements, Int.
J Impact Eng., 35; 1497-1502.
[22] Sha SX, Chen Y, Liu XY (2010), Finite-element simulation of Kevlar yarns and fabrics in assessing its
mechanical protective performance under projectile impact loading, J Fibre Bioengineering and Informatics,
3(2); 111-116.
[23] Li G. and Li C. (2013), “An analytical analysis of energy release rate in bonded composite joints in a mode I
condition”, Composites Part B, 44; 704-713.

19

View publication stats

You might also like