You are on page 1of 9

ARTICLE IN PRESS

International Journal of Impact Engineering 35 (2008) 10001008 www.elsevier.com/locate/ijimpeng

Ballistic impact response of laminated composite panels


H.L. Gower1, D.S. Cronin, A. Plumtree
Department of Mechanical Engineering, University of Waterloo, 200 University Ave., W, Waterloo, ON, Canada N2L 3G1 Received 16 October 2006; received in revised form 3 April 2007; accepted 23 July 2007 Available online 8 August 2007

Abstract Laminated ballistic composite panels are an important part of hard-plate protective body armour and may be subjected to a wide variety of impact conditions depending on the projectile, impact velocity and armour construction, to name a few. The ballistic response of laminated composite panels has been investigated through direct impacts of two non-deforming projectiles (7.5 mm diameter hardened steel 1201 cylindricalconical, and 9 mm hemispherical nosed) selected to enhance different failure mechanisms including penetration and delamination. Experimental and numerical studies were carried out to determine the ballistic response of laminated Kevlars 29 and 129 composite panels, commonly used in protective body armour. These panels were impacted at velocities between 130 and 250 m/s, which were below the penetration limit of the panels. A numerical parametric study was initially undertaken to determine those material properties which reduce back face signature (BFS; maximum dynamic displacement), one of the important performance indicators for assessing personal protection. Experimental material characterization then allowed mechanical property data to be determined for numerical simulations, which showed good agreement with the experimental data, particularly for the conical projectile impacts on both types of Kevlars panels. Numerical simulations of the impact tests accurately predicted the BFS and dynamic response for the conical projectile impacts, while the BFS for the hemispherical projectiles was slightly low. This can be attributed to the dominant delamination failure mechanisms, which may not be completely captured by the numerical model. Importantly, the numerical analysis accurately predicted the initial velocity of the panel back face for the hemispherical projectiles and the time to reach maximum BFS for the conical projectiles. r 2007 Elsevier Ltd. All rights reserved.
Keywords: Back face signature; Ballistic impact; Explicit numerical modelling; Fabric material properties woven; Laminated composites

1. Introduction Laminated ballistic composite materials may be used in protective helmets, or with ceramics and other materials for protective body armour. Standard hard plate protective body armour is made up of multiple layers, commonly including a ceramic plate to blunt and fracture projectiles and a laminated composite panel to stop the projectile while containing the ceramic particles. Armour may include an anti-trauma layer to reduce potential injury caused by dynamic deformation of the armour into the wearer. The maximum displacement of the armour during
Corresponding author.

E-mail address: dscronin@uwaterloo.ca (D.S. Cronin). Now at: Netherlands Institute for Metals Research, Mekelweg 2, P.O. Box 5008, Delft, The Netherlands.
1

impact is described as the back face signature (BFS) [1] and is a key measure of ballistic performance. Two dominant types of composite fabric may be used in protective body armour, including aramid, specically Kevlars, and polyethylene-based materials such as Spectras or Dyneemas. Layers of woven or unidirectional bres are bonded using a thermoplastic or thermosetting polymer matrix. The present study has focused on the ballistic response of K29 and K129 woven Kevlars (DuPont, USA) fabric (2 2 basket weave, 1500 denier) laminated using a polyvinyl butyral (PVB)-phenolic matrix (18% volume fraction). The design, performance and evaluation of composite panels undergoing ballistic impact require an understanding of material properties under high impact conditions. There are several models that describe the impact on woven composites. However, many deal with low-velocity impacts

0734-743X/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijimpeng.2007.07.007

ARTICLE IN PRESS
H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008 1001

Nomenclature CS through-thickness compression strength E1 elastic modulus in the warp direction E2 elastic modulus in the weft direction E3 through-thickness compressive modulus G12 in-plane shear modulus G23, G13 through-thickness shear moduli ISS interlaminar shear strength S1 tensile strength in the warp direction

tensile strength in the weft direction S2 S12 in-plane shear strength S23, S13 through-thickness shear strength SN interlaminar normal strength STF1 strain to failure in the warp direction STF2 strain to failure in the weft direction u12 in-plane Poissons ratio u23, u13 through-thickness Poissons ratio r density

such as drop weight tests to simulate tool drop [2], where the failure mechanisms differ signicantly from those under ballistic impacts. Among the current high-speed impact models, many are analytical and have difculty accounting for deformable projectiles, or the coupled response of the projectile and composite [3,4]. Very few numerical models simulating high-speed impact of projectiles involve detailed composite models as well as the associated response and failure mechanisms [3]. A recent model was developed to simulate the impact of a rigid cylinder on woven Kevlars fabric [5]. The fabric tows were modelled independently and the properties used were those for Kevlars yarn. The failure mode of a woven fabric cannot be directly applied to a composite panel, unfortunately, since the presence of the resin and multiple layers was not considered [5]. Silva et al. [6] developed a model for predicting the perforation limit velocity (V50) of an fragment simulating projectile (FSP) on woven Kevlars 29/Vinylester panels. Accurate predictions of the failure modes and stress waves were made, despite the extent of penetration being determined by visual examination. Although this study provides some important information regarding modelling of ballistic impacts, the analysis focused on projectile penetration and not on the transient response of the panel during the impact. The current study considers high-speed impact tests and applies an existing numerical model for laminated ballistic composites [7] to evaluate the dominant material properties affecting ballistic performance. This model provides some guidance to improve the design and enhance the development of new composite panels. In addition, the study has been undertaken to evaluate the effect of the composite panel properties on such factors as the number of plies penetrated and the extent of BFS. This information leads to an increased understanding of the composite panel behaviour under ballistic conditions, while improving the predictive capabilities of the numerical model. A small change in one particular material property can result in a large change in BFS as well as the number of plies sheared, whereas another property may have more limited effects, as shown by Van Hoof [7]. Knowledge of the signicant properties that contribute to reducing BFS would enhance the selection of new and appropriate materials, leading to

improved protection from blunt trauma and possible weight reduction. 2. Initial ballistic composite numerical model The ballistic numerical model initially developed by van Hoof [7] to simulate the impact and post-failure response of a FSP on a laminated woven Kevlars 29 composite formed the basis for this parametric study. The model was developed from that proposed by Matzenmiller, Lubliner and Taylor (MLT) [8], which assumed that each lamina was represented as a homogenized continuum. This approach has been widely used to simulate impact and failure of composite materials [7,9,10]. To verify the numerical model, van Hoof conducted a series of experimental tests with a variety of projectiles using a range of velocities. Four different types of projectiles were used: 1.1 g FSP; 2.8 g FSP; 40 mm long 1201 conical hardened AISI 4340 steel and 40 mm long 371 hardened AISI 4340 steel. The conical projectiles had a diameter of 7.56 mm. All the experiments used the same targets, 10.16 15.24 cm Kevlars 29 plates. Each plate was 19 plies or 9.5 mm thick. The Kevlars 29 (Mil-C-44050 Type II, Class I) used was 1500 denier in a 2 2 basket weave. The resin used contained 50% phenol formaldehyde and 50% PVB, with the low resin volume fraction of 18% being typical for armour grade composites [12]. Three types of measurement systems were used to obtain velocitytime data for the projectile. These included a highspeed X-ray IMAcon (I MAX), a Velocity Interferometer System for Any Reector (VISAR) and Enhanced Laser Velocity Sensor (ELVS)[11]. BFS was determined from projectile displacement using the ELVS and VISAR data. The IMAX was capable of measuring the BFS in addition to the projectile displacement. As expected, BFS increased with impact velocity. For the initial parametric study by Van Hoof [7], an FSP impact at a velocity of 461 m/s was considered. This was representative of previous ballistic tests from which the model was developed. Hence, since the impact and failure mode characteristics were known for the FSP, the results could be validated. The model used was implemented in LS-Dyna [13], a commercial explicit nite element code. Van Hoof

ARTICLE IN PRESS
1002 H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008

[7] showed that those variables having the most effect on BFS were through-thickness elastic modulus, in-plane elastic modulus, through-thickness compressive strength and in-plane strain to failure. Of these, a low throughthickness elastic modulus and a high value for each of the other parameters gave a low BFS. To reduce the number of plies penetrated, however, a low through-thickness modulus, low in-plane moduli, high compressive strength, low through-thickness shear strength and high in-plane strength would be benecial. It is worthwhile noting that no ranking was provided for these properties in terms of the effect on BFS. 3. Ballistic composite numerical model development Previously [7], small 10.16 15.24 cm panels were used for impact testing and simulation, which was somewhat unrealistic since delaminations were seen to extend to the edge of the panels in some cases, resulting in reduced impact resistance. To address this limitation, typical armour-sized panels, 25.4 cm 30.48 cm (1000 1200 ), were used for the current study. Consequently, any edge effects in these panels would be present in actual body armour. These larger panels were made with Kevlars 129. Six of these panels were 19 plies thick and three were 37 plies thick. In addition, eight 19-ply Kevlars 29 composite panels were included in the test program to allow comparison with the earlier work of van Hoof [7]. Two different types of projectiles were used. To ensure continuity and comparison with earlier work [7], a 1201 conical 7.5 mm diameter projectile was chosen. In addition, a 9 mm diameter hemispherical nosed projectile was included since it was considered to more accurately simulate the geometry of a projectile after passing through a ceramic facing. Both projectiles were made of heattreated AISI 4340 steel and were 40 mm in length. The current version of the numerical model and material constitutive model was implemented into a recent version of LS-DYNA (LS-DYNA 960) from VEC-DYNA [13].

This included improving the nite element meshes and contact denitions. The nite element mesh is shown in Fig. 1, where each ply in the laminated panel was modelled as a separate entity with a tiebreak interface used to model inter-ply delamination and contact. This model has several advantages over that developed earlier [7]. The elements have minimal distortion and the increase is more gradual from the small element size in the impact zone to the large size at the edge of the panel. A similar mesh for the smaller 10.16 cm 15.24 cm panel was shown to accurately reproduce the results obtained by van Hoof [7]. The new meshes were developed using the MSC/Patran pre-processor (MSC/Patran, MSC Software Corporation, Sanat Ana, CA), which dened the element sizes according to their location on the panel. The fewest number of elements were distorted by using the same element size beside the impact zone, as that dened inside the impact zone (0.26 mm), and by using an element size of 8 mm at the outer edges. This technique maintained a reasonable transition in element size and an acceptable processor time, approximately 1420 h for a single simulation. Quarter models were developed for the three congurations tested experimentally, including a 19-ply composite impacted by a the 1201 conical projectile, as well as a 19-ply composite impacted by the 9 mm hemispherical projectile and a 37-ply panel impacted by the 9 mm hemispherical projectile. Since a quarter model was used, appropriate boundary conditions were applied along the planes of symmetry. The nodes along the exterior edge of the shorter sides of the panel were xed in the z direction, as movement was also restricted along these edges during the experiments. No restriction was placed along the longer sides. 4. Experimental panel impact testing Experimental testing was carried out to validate the improved numerical model. Seventeen experimental tests were conducted, including three types of composite panels and two types of projectiles. The panels were

Fig. 1. Finite element mesh of composite panel.

ARTICLE IN PRESS
H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008 1003

30.48 cm 30.48 cm, 19-ply (9.5 mm) Kevlars 29, 25.4 cm 30.48 cm, 19-ply (9.5 mm) Kevlars 129 and 25.4 cm 30.48 cm, 37-ply (18.5 mm) Kevlars 129. All were similar in size to those used in personal body armour. The projectiles used were the 1201 conical projectile used in previous research [7] and a 9 mm diameter hemispherical ended cylinder. Both were made of heat-treated AISI 4340 steel, 40 mm long. The impact velocities ranged from 100 to 250 m/s, selected to provide a range of impact energy while not penetrating the panels. A gas gun actuated with a high-speed valve was used for the ballistic tests. A vacuum chamber was incorporated into the gas gun to achieve higher projectile velocities and more consistent impact velocities, while absorbing the pressure wave and reducing noise during ring. A laser measurement system was used to detect the speed of the projectile at the barrel exit. Two ELVS systems were located on the vacuum chamber. The rst tracked the displacement of the composite backplane, from which the BFS was determined, and the second that of the projectile. Fig. 2 shows both progressive displacements for the 9 mm diameter hemispherical-ended projectile impacting a 19-ply Kevlars 29 composite panel at 146 m/s. This gure is representative of all the ballistic tests. On impact, some backplane oscillation was apparent, as seen in Fig. 2. The projectile trace shows the displacement of its tail starting at 42 mm from the composite panel (Point A0 ), which was the maximum width of the ELVS beam. The projectile continued to impact the panel to Point A corresponding to a total displacement of 65 mm (25 mm from A-A0 , plus the projectile length of 40 mm).

The projectile then rebounded or fell away at Point B. The small oscillations in the curve were due to debris from the projectile sabot crossing the ELVS beam, and have no inuence on the panels displacement. Since no increase in amplitude of oscillation for the projectile took place, as seen for the backplane of the composite panel, it can be assumed that the backplane oscillation was the result of the delaminated plies vibrating, not the entire panel. This was subsequently veried after viewing high-speed videos of these ballistic impacts [14]. Fig. 3a is a summary plot of the BFS and corresponding projectile velocity for all the tests. In general, the thicker panels displayed a greater ballistic resistance evidenced by a lower BFS for all projectile velocities. It is interesting to note that Kevlars 29 composite panels deformed less under impact from conical projectiles than the Kevlars 129 composite panels at low-velocity impacts. However, they deformed in a similar manner for higher-velocity impacts. It is important to note that, for the velocities considered in this study, all of the projectiles were stopped by the panels before complete penetration occurred. From Fig. 3a it can be seen that the 9 mm hemispherical projectile produced a higher BFS than the conical projectile for the same velocity. This is to be expected since the 9 mm projectile is heavier and has higher kinetic energy for the same velocity. However, when comparing the results based on kinetic energy before impact, as shown in Fig. 3b, it can be seen that the shape of the projectile was not signicant for the Kevlars 129 panels. For the Kevlars 29 panels, however, the hemispherical-nosed projectile produced a higher BFS than the conical projectile for the same energy.

Fig. 2. A hemispherical-ended projectile, 9 mm diameter, impacting a 19-ply Kevlars 29 composite panel at 146 m/s.

ARTICLE IN PRESS
1004 H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008 Table 1 Comparison of numerical with experimental values for Kevlars 29 composite properties Property Values reported in literature [7] Initial numerical model values [7] Experimental value (static testing) 17.86 8.3

E1 (GPa) E2 (GPa) E3 (GPa) u12 u31,32 G12 (GPa) G13,23 (GPa) r (g/cm3) S1 (MPa) S2 (MPa) CS (MPa) S12 (MPa) S13,23 (MPa) Sn (MPa) ISS (MPa) STF1 STF2

18.5 18.5 6.0 0.25 0.33 0.77 5.43 1.23 1850.0 1850.0 1200.0 77.0 543.0 34.5 9.0 0.1 0.1

18.5 18.5 6.0 0.25 0.33 0.77 2.71 1.23 555.0 555.0 1200.0 77.0 898.0 34.5 9.0 0.03 0.03

0.85

444.9 225.5 $900 56.7

0.045 0.050

Fig. 3. (a) BFS versus impact velocity, (b) BFS versus projectile energy.

Although the Kevlars 29 panels deformed less than Kevlars 129 panels under impact from the conical projectile, the backplane displacement of the Kevlars 129 panels was less than that of the Kevlars 29 panels for the hemispherical-nosed projectile. In this case, the penetration mode for the two projectiles was different. The conical projectile tended to cut through the plies due to its sharp edges, whereas the hemispherical projectile caused early delamination to occur. This study showed that, when considering blunt projectiles, the Kevlars 129 panels showed reduced BFS compared with the Kevlars 29 panels, but had less resistance to penetration from sharp projectiles. 5. Material properties and numerical model Experimental tensile tests were conducted to determine or verify the composite panel material properties. Original data used in the model were obtained from different sources in the literature. Table 1 compares the Kevlars 29 composite panel properties obtained from earlier data [7] with those properties used in the initial numerical model, as well as those obtained from the present experimental test program. The most signicant differences found in the literature were in the values for S13 and S23, which were used to

calibrate the model previously [7]. The through-thickness shear strength was determined for one impact condition and then veried with the other impact tests. The highest impact velocity was appropriately chosen so that the numerical model would correctly predict no penetration since this was the case. For calibration of both composite panel types, the higher velocity conical projectile was chosen since the computing time was 15 times shorter than that for the hemispherical-ended projectile. This allowed more simulations to be carried out, thereby giving more reliable results and a better match. The data for the 1201 conical projectile impacting the 30.48 cm 30.48 cm Kevlars 29 composite panel at 241 m/s was used to calibrate the current model and determine the shear strengths (S13, S23). Considering various values for the shear strength, a value of 898 MPa was found to produce the most consistent results based on BFS. These strength values were used to simulate all other impact conditions. The experimental and numerical results for tensile modulus in the warp (01) direction, E1, were very similar. Any difference could be accounted for by strain rate effects. The difference between the experimental and numerical values for weft (901) direction, E2, was quite signicant, as can be seen in Fig. 4. However, this difference is not considered in the current model, and the experimental results are only approximate for static deformation conditions. This result highlights the fact that the properties may vary signicantly, often depending on the processing conditions. Improved data can be obtained by using biaxial load conditions and testing at appropriately high strain rates. The difference in E1 and E2 found during the present tensile testing may be attributed in part to the type of weaving and tension control [15]. It was observed that the warp tows were nearly straight; hence, a

ARTICLE IN PRESS
H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008 1005

Fig. 4. Stressstrain curves for a 4-ply Kevlars 129 composite.

Table 2 Comparison of numerical and experimental values for Kevlars 129 material properties Property E1 (GPa) E2 (GPa) E3 (GPa) u12 u31,32 G12 (GPa) G23,31 (GPa) r (g/cm3) S1 (MPa) S2 (MPa) CS (MPa) S12 (MPa) S23,31 (MPa) Sn (MPa) ISS (MPa) STF1 STF2 Value used in numerical model 22 22 9.0 0.25 0.33 0.77 2.715 1.23 800.0 800.0 1200.0 77.0 1000.0 34.5 9.0 0.036 0.036 Experimental value 16.67 14.0

1.35

470.1 420.4 $900 24.9

0.037 0.073

The material properties for the woven Kevlars 129 composite panel are given in Table 2. Since little data for the material properties of these composite panels are available in the literature, they were based on extending values for the bre to the composite panel. In this case, the failure strains STF1,2 were based on the bre failure strain of 0.03 [16]. The static elastic modulus of Kevlars 129 bres is 43% higher than that for Kevlars 29, and, allowing for the bre volume fraction, the modulus of the composite was increased by 20%, giving a value of 22 GPa. The quasi-static experimental value for the composite panel was 17 GPa. The higher value of 22 GPa is supported by Todo et al., who reported that the tensile strength of a woven aramid/carbon bre composite was 28% higher under high strain rates [18]. In the model, the in-plane failure stress was based on the product of the failure strain and the elastic modulus; hence, the strengths S1, S2 were recalculated applying the new values for modulus and failure strain [7]. The values of S13 and S23 were unchanged from those used for Kevlars 29 owing to lack of experimental data. Since the experimental data supported a change in elastic modulus, strength and thus failure strain were used to calibrate the model for Kevlars 129. In this case, the impact on the 25.3 cm 30.48 cm panel with the conical projectile at a velocity of 244 m/s was used. Sixteen combinations of in-plane elastic modulus, compression strength and through-thickness shear strength were considered. From this parametric study, the optimum values were found using the in-plane modulus of 22 GPa, an inplane strength of 800 MPa and a through-thickness shear strength of 1000 MPa. These values resulted in a BFS that was slightly high, but the BFS occurred at the correct time of 0.0003 s. It is recognized that other combinations of material properties may also produce satisfactory results, and this will be investigated in future studies. 6. Comparison of experimental and numerical analysis The experimental and numerical results were compared using BFS, time to reach BFS and displacementtime response of the composite panel. In all cases, the rst 0.0003 s of the impacts was simulated since this was the most signicant period during which maximum backplane deection (and BFS) occurred. The experimental and numerical BFStime relationships were compared for 1201 conical projectiles impacting Kevlars 129 and Kevlars 29 composite panels, and 9 mm hemispherical projectiles impacting the Kevlars 29 composite panels. Fig. 5a shows the displacementtime plot for the 1201 conical projectile impacting Kevlars 129 at 244 m/s. This gure represents typical behaviour and presents the data used to calibrate the through-thickness shear strength for Kevlars 129. The maximum dynamic backplane displacement obtained from the numerical model was within 5% of the experimental value. Fig. 5b presents similar data for the 9 mm hemispherical projectile impacting a Kevlars 29 composite panel at 146 m/s.

tensile test in this direction measured the force required to stretch the warp tows. The weft tows, however, undulated over and under the warp tows, resulting in a lower initial modulus due to bre straightening. Although the model has the exibility and capacity to assign separate values to the warp and weft directions, this was not used since the stressstrain curves become linear at high strain rates [17]. The experimental and numerical data for the in-plane strengths, corresponding to the warp and weft directions, were similar. The values for S1 were about the same, any differences being accounted for by strain rate effects. S2 was determined to be approximately half that of S1the same proportion as that for the measured tensile moduli E2 and E1. The experimental shear modulus, G12, was higher than in the numerical model, but should be regarded as an approximation since the lateral strain required to estimate u12 was determined from measurements of the specimen after failure, not from strain gauges.

ARTICLE IN PRESS
1006 H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008

parameter for determining the accuracy of the model is the time required to attain maximum BFS. Although the initial slope was low for the predicted conical projectile impact response curves, the model performed well by accurately estimating the time to reach maximum displacement. The experimental and numerical results for maximum BFS for both Kevlars panels impacted at various velocities for the three simulated congurations mentioned above are given in Figs. 6ac. The results for the conical

Fig. 5. (a) A 1201 conical projectile impacting 19-ply Kevlars 129 at 244 m/s. (b) A 9 mm hemispherical projectile impacting Kevlars 29 at 146 m/s.

The predicted time to BFS is in good agreement with the experimental result, although the magnitude of BFS is slightly lower. The initial slope of the numerical displacementtime curves can be used to assess the accuracy of the numerical analysis. There was good agreement for the initial slopes of the experimental and numerical curves for the hemispherical projectile impacts, as shown in Fig. 5b. For all conical projectile impacts the initial slope was lower than measured experimentally, as shown in Fig. 5a. Changing material properties such as compressive strength, in-plane modulus, in-plane strength and through-thickness shear strength had little effect on the initial slope. This indicates that either a material property not investigated or a missing parameter such as strain rate sensitivity may have had an inuence. It was observed that the 1201 conical projectile penetrated more plies than the 9 mm hemispherical projectile due to the sharp edges. This will be investigated in future studies. In addition to the maximum BFS and initial slope, another

Fig. 6. (a) Experimental and numerical maximum backplane displacementvelocity relationships for 19-ply Kevlars 29 composite panels. (b) Experimental and numerical maximum backplane displacementvelocity relationships for 19-ply Kevlars 29 composite panels. (c) Experimental and numerical maximum backplane displacementvelocity relationships for 19-ply Kevlars 129.

ARTICLE IN PRESS
H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008 1007

projectile impacting 19-ply Kevlars 29 and Kevlars 129 composite panels (Figs. 6a and c) illustrate the predictive capability of the numerical model. This compares well with previous work [7], indicating that the model was accurate for a 1201 conical projectile impacting a smaller 10.16 cm 15.24 cm 19-ply Kevlars 29 panel. Simulations using the hemispherical projectile (Fig. 6b) consistently gave lower BFS values. This is attributed to the dominant delamination and in-plane tension failure modes, which may not be adequately captured by the model. Although the material properties were calibrated for a particular set of test conditions, these gures indicated that the properties used had the capacity of being applied to other congurations for the same material. This would be acceptable if all other factors were correct. The material properties were determined for a particular Kevlars composite panel using values from the literature and experiments, then calibrating the through-thickness shear strength using the highest impact velocity of the conical projectile. This procedure proved to be very satisfactory for impacts on Kevlars 29 composite panels. The selection of material properties was well balanced. The numerical results for the hemispherical projectile impacting Kevlars 129 composite panels resulted in a BFS that was slightly lower than that observed experimentally. This was also the

case for the 9 mm hemispherical projectile impacting Kevlars 29 composite panels. It is interesting to note that the model accurately simulated the failure modes occurring in the composite panels. On impact, penetration of the rst few plies occurred, followed by delamination of subsequent layers. Fig. 7 is representative and shows the progression of the hemispherical projectile on impact of the 19-ply Kevlars 29 composite panel, giving the relative positions of the projectile and panel for three different times during impact. Fig. 7a shows the projectile on immediate impact of the panel (t 0 s). Fig. 7b shows the panel slightly after impact (t 0.00005 s). Penetration of the rst plies has already taken place. Fig. 7c shows the projectile and panel (t 0.00015 s) where delamination has already taken place. Fig. 7d shows the panel at the end of the simulation run, (t 0.0003 s) and shows an increase in the number and area of delaminations. Fig. 7 conrms that the model simulates the expected failure modes in the correct sequence. To evaluate the failure modes and projectile penetration, the original test samples from van Hoof [7] were mounted in epoxy and sectioned for comparison with numerical model predictions. Fig. 8a gives a representative comparison with the experimental and numerical (Fig. 8b)

Fig. 7. Quarter model of 9 mm hemispherical projectile impacting Kevlars 29 at 203 m/s.

Fig. 8. (a, b) Cross-section comparing numerical [right] with experimental [left] results (2.8 g FSP projectile impact).

ARTICLE IN PRESS
1008 H.L. Gower et al. / International Journal of Impact Engineering 35 (2008) 10001008

cross-sections for a 2.8 g FSP impacting a 10.16 cm 15.24 cm panel at 490 m/s. This gure illustrates the anticipated penetration and delamination failure modes at the corresponding levels within the panel. It is important to note that the predicted penetration into the panel is in excellent agreement with that observed experimentally. 7. Conclusions An experimental and numerical ballistic study has been undertaken to investigate the dynamic response of Kevlars composite panels to impacts from non-deforming conical and hemispherical projectiles. Importantly, the sizes of the panels used in this study were chosen to be representative of those used in personal body armour. An initial numerical study of ballistic impacts on Kevlars composite panels indicated that an increased inplane modulus, in-plane strain to failure and throughthickness compression strength would reduce the BFS of woven Kevlars composites. In addition, reducing the through-thickness elastic modulus was predicted to be benecial in reducing backplane displacement. Experimental characterization of the stressstrain behaviour of the warp and weft exhibited two distinct slopes. The rst corresponded to straightening, and the second to the stretching of the tows. Mechanical testing data gave different property values than those used in the numerical model, especially for the in-plane strengths. The values used for the simulations were more than twice that found during experiments, attributed to strain rate effects in the material and consistent with data reported in the literature. The BFS predictions for the simulations of conical projectile impacts on Kevlars 29 were in good agreement with the experimental results, whereas the predicted BFS for the hemispherical projectile impacts on Kevlars 29 were consistently lower. This is attributed to increased delamination taking place within the panel. The experimental and numerical results for the conical projectile impacts on Kevlars 129 were in close agreement. Simulations correctly predicted the time to BFS for the conical projectiles. The numerical prediction of the velocitytime history for the hemispherical projectile impacting the Kevlars 129 panels was accurate, although the BFS prediction was low. Experimental testing showed that Kevlars 29 exhibited a lower BFS than the Kevlars 129 at low impact speeds; however, this changed for higher velocity impacts. In this study, when considering blunt projectiles, the Kevlars 129 panels showed reduced BFS compared with the Kevlars 29 panels, and displayed less resistance to penetration from sharp projectiles.

Acknowledgements The authors gratefully acknowledge nancial support during the course of this research from Ontario Centres of Excellence (OCE), Defence R&D Canada - Valcartier (Valcartier, Quebec) and Barrday Inc. (Cambridge, Ontario). The authors also thank Professor Michael Worswick at the University of Waterloo for his support and assistance. References
[1] National Institute of Justice. Ballistic resistance of personal body armor, NIJ Standard 0101.04. [2] Roy T, Chakraborty D. Delamination in hybrid FRP laminates under low velocity impact. J Reinf Plast Compos 2006;25(18): 193956. [3] Naik NK, Shirirao P, Reddy BCK. Ballistic behaviour of woven fabric composites: formulation. Int J Impact Eng 2006;32(9):15212. [4] Naik NK, Shirirao P, Reddy BCK. Ballistic behaviour of woven fabric composites. Mater Sci Eng A 2005;412(12):10416. [5] Duan Y, Keefe M, Bogetti TA, Powers B. Finite element modeling of transverse impact on a ballistic fabric. Int J Mater Sci 2006;48(1): 3343. [6] Silva MAG, Cismasiu C, Chiorean CG. Numerical simulation of ballistic impact on composite laminates. Int J Impact Eng 2005;31(3): 289306. [7] van Hoof J. Modelling impact induced delamination in composite materials. Ottawa: Carleton University; 1999. [8] Matzenmiller A, Lubliner J, Taylor RL. A constitutive model for anisotropic damage in ber-composites. Mech Mater 1995;20(2): 12552. [9] Floyd A, Cronin D, Vaziri R, Poursartip A, Worswick M, Plumtree A. Three-dimensional numerical simulation of the ballistic response of laminated composite plates. Contract report for the Defence Research Establishment, Valcartier, June 2001. [10] Williams KV, Vaziri R. Application of a damage mechanics model for predicting the impact response of composite materials. Comput Struct 2001;79(1):9971011. [11] Starratt D, Sanders T, Cepus E, Poursartip A, Vaziri R. An efcient method for continuous measurement of projectile motion in ballistic impact experiments. Int J Impact Eng 2000;24(2):15570. [12] Lee B, Walsh TF, Won ST, Patts HM, Song JW, Mayer AH. Penetration failure mechanisms of armour-grade bre composites under impact. J Compos Mater 2001;35(18):160533. [13] Halquist JO. LS-Dyna theoretical manual LSTC. California, 1998. [14] Gower HL. Ballistic impact response of woven Kevlar composites. Waterloo: University of Waterloo; 2003. [15] Ito M, Chou T. An analytical and experimental study of strength and failure behavior of plain weave composites. J Compos Mater 1998; 32(1):230. [16] Dooraki FB, Nemes JA, Bolduc M. Study of parameters affecting the strength of yarns. J Phys IV France 2006;134:11838. [17] Chocron Benloulo IS, Rodr guez J, Mart nez MA, Sanchez Galvez V. Dynamic tensile testing of aramid and polyethylene ber composites. Int J Impact Eng 1997;19(2):13546. [18] Todo M, Takahashi K, Beguelin P, Kausch H. Strain-rate dependence of the tensile fracture behaviour of woven-cloth reinforced polyamide composites. Compos Sci Tech 2000;60(5): 76371.

You might also like