A NUMERICAL AND EXPERIMENTAL INVESTIGATION OF THE IMPACT BEHAVIOUR OF HYBRID AND MULTI-PLY FABRIC STRUCTURES by Wesley R. Novotny B.ENG. (Civil Engineering), McGi l l University 2000 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF APPLIED SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Civil Engineering) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A November 2002 © W.R.Novotny, 2002 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of ^-i ">} cr ^ i V i g g ^ 1 ' n ^ The University of British Columbia Vancouver, Canada Date Oe ^ci (oe r 2. )*~o 6 \ DE-6 (2/88) Abstract The focus of this work is to develop an understanding of the impact behaviour of single-and multi-ply fabric materials used in protective (armour) structures. Hybrid structures are also investigated to explore the effect of mixing different materials on armour system performance. Emphasis is placed on the behaviour of these systems during the early stages of the impact event prior to the return of boundary-reflected strain waves to the impact point. This allows one to separate the material response from the structural response and in so doing investigate the losses of material efficiency inherent in systems with greater areal densities. The latter can arise from increased ply count and/or greater mass per unit length of the yarns which make up the fabric. A series of ballistic impact experiments are carried out on the permutations of stacking sequences of 2- and 4-ply Kevlar Nylon hybrids. Instrumented impact tests as well as post-mortem examination are used to investigate the unique behaviour of these hybrid systems over the entire duration of the impact event. The early event behaviour of the hybrids is also characterised. The significant difference between the various hybrids tested is the increased transverse deformation of specific stacking sequences due to penetrated layers. A finite element code, T E X I M , is used to explore the response of single and multi-ply Kevlar fabric systems during the early stages of impact. The numerical results of single and multi-ply Kevlar are found to be in good agreement with the relevant experimental data. Parametric studies using the numerical model are then carried out to investigate the effect of various weave and stacking parameters. Single panels with lower areal densities (typical of panels made up of yarns with lower mass per unit length ) are found to have superior early event performance due to increased strains in the yarns. Multi-ply systems are shown to perform better in terms of the rate of energy absorption early in the impact event as the inter-ply spacing is reduced. Minimizing the spacing is shown to result in increased fibre strains and greater material involvement in absorbing the impact energy. ii Table of Contents Abstract ii Table of Contents . i i i List of Tables vi List of Figures vii Acknowledgments xiv 1 Introduction 1 1.1 Background 1 1.2 Scope of Work 4 2 Literature Review 6 2.1 Background 6 2.2 Experimental 7 2.2.1 Multi-Ply & Hybrids Targets 7 2.2.2 Local Target Damage 12 2.2.3 Hybrid Patents 13 2.3 Multi-ply Numerical Models 15 3 Experimental Procedure & Results: Entire Impact Event 25 3.1 Objectives 25 3.2 Experimental Set-up 25 3.2.1 Powder Gun 25 3.2.2 Projectiles 26 3.2.3 Test Fixture 26 3.2.4 Boundary Conditions 27 3.2.5 ELVS System... 27 3.2.5.1 Hardware 27 3.2.5.2 Data Acquisition and Processing 28 3.3 Material Systems 29 3.3.1 Kevlar 29 3.3.2 Nylon 29 3.4 Single Material Panels 30 3.4.1 Non-Perforating 30 3.4.2 Vs-Vr 31 3.5 Hybrid Panels 32 3.5.1 2-ply Combinations of Kevlar and Nylon 32 3.5.1.1 Non-perforating 32 3.5.1.2 Perforating 34 iii 3.5.1.3 V s -V r Curves 34 3.5.2 Comparison between 2-ply Hybrids and Single Material Systems 35 3.5.2.1 Non-perforating 35 3.5.3 4-ply Combinations of Kevlar and Nylon 36 3.5.3.1 Non-perforating 36 3.5.3.2 Perforating -. 39 3.5.3.3 V s -V r Curves 39 3.5.4 Comparison between 4-ply Hybrids and Single Material Systems 40 3.5.4.1 Non-perforating 40 3.6 Chapter Summary 41 4 Experimental Results: Early Impact Behaviour 68 4.1 Introduction 68 4.1.1 Theoretical Background 68 4.1.1.1 Relation Between Strain, Strain Wave Velocity and Projectile Velocity 69 4.1.1.2 Panel Strain Wave Velocity 70 4.1.2 Energy Absorption Mechanisms 71 4.1.2.1 Strain Energy 71 4.1.2.2 In-plane Kinetic Energy 72 4.1.2.3 Out-of-Plane Kinetic Energy 73 4.1.2.4 Rate of Energy Absorption 73 4.1.3 Early Event Ballistic Efficiency, B 74 4.1.3.1 Rate of Energy Absorption 74 4.1.3.2 Calculation of B 75 4.1.3.3 Calculation of B Based on Rule of Mixtures Analysis 75 4.1.3.4 The significance of the Constant B 77 4.2 Chapter Summary 77 5 Numerical Model and Results 83 5.1 Introduction 83 5.2 Numerical Codes 83 5.2.1 T E X I M 83 5.2.1.1 Mass-String Model 84 5.2.1.2 Constitutive Model 84 5.2.1.3 Projectile 84 5.2.1.4 Boundary Conditions 85 5.2.1.5 Multi-Ply Targets & Interlayer Contact 85 5.2.1.6 Failure 85 5.2.1.7 Time Discretisation 87 5.2.2 L S - D Y N A 88 5.3 Model Verification and Validation 89 5.3.1 Numerical Verification 90 5.3.2 Experimental Validation 90 5.4 Analysis of the Entire Impact Event 91 5.5 Modelling of the Early Impact Event 92 5.5.1 Background 92 5.5.2 Calculation of B 93 5.6 Parametric Studies In Loss of Ballistic Efficiency B 95 5.6.1 Areal Density Effect 95 5.6.1.1 Background 95 iv 5.6.1.2 Results 96 5.6.2 Multi-ply Effects 97 5.6.2.1 Background 97 5.6.2.2 Effect of Spacing 98 5.6.2.3 Effect of Contact 99 5.6.3 Validation of Numerical B Values 100 5.7 Practical Design Insight 101 5.8 Chapter Summary 102 6 Conclusions & Future Work 128 6.1 Summary 128 6.2 Conclusions 129 6.2.1 Experimental 129 6.2.2 Numerical 130 6.2.2.1 Single Plies 130 6.2.2.2 Multi-Plies 130 6.3 Future Work 131 References 133 V List of Tables Table 3-1 Material Properties for Kevlar 129 (Shahkarami, 1999) 42 Table 3-2 Properties of Kevlar 840d panel tested 42 Table 3-3:Material Properties for Nylon (Smith et al., 1956) 42 Table 3-4:Properties of Nylon 1050d panel tested 42 Table 3-5: A l l shots taken for 1-ply Nylon, 2-ply and 4-ply Kevlar-Nylon hybrids 43 Table 3-6Comparison of results for 1-ply Nylon and 1,2,4-ply Kevlar. Note: There is no data available for 1-ply Kevlar for the 28 Joule shot because the panel is perforated at this velocity 45 Table 3-7: Comparison of results for Kevlar-Nylon and Nylon-Kevlar comparison 45 Table 3-8: Comparison of results for 4-ply hybrids and 4-ply Kevlar 45 Table 4-1: Comparison of hybrid B values calculated from Rule of Mixtures (Equation 4.25) and experimental results 78 Table 5-1 Properties of Kevlar panel tested 103 Table 5-2:List of simulations of the early impact event 103 Table 5-3:Exponent values from power regression analysis performed on numerical normalised dE/dt versus V s t r ike data for different systems 107 Table 5-4:Simulations carried out on 1-ply Kevlar panels impacted at 18 Joules to study the effect of areal density 107 Table 5-5 Simulations on Kevlar 840d impacted at 18 Joules to study effect of multi-plies 108 Table 5-6:Effect of layer interaction on the energy absorption characteristics of multi-ply fabric armours. These are values predicted for an 18 Joule strike energy on a 4-ply Kevlar 840d system at 34 microseconds after impact, i.e. just before the arrival of the reflected strain wave in the initially impacted layer 108 VI List of Figures Figure 2-1: Absorbed energy as a function of areal density for a 1500 denier Kevlar 29 fabric. The line indicates the predicted absorbed fabric energy and the dots represent the measured values; the difference is attributed to system effects. The line is not a linearly increasing line because a 1-ply Vs-Vr curve is used to calculate this line: therefore the curvature inherent in the Vs-Vr curve is present here.(Cunniff, 1992) 20 Figure 2-2: For a given areal density the lower denier panels absorb a greater amount of energy. These curves are derived from experimental data. (Cunniff, 1992) 20 Figure 2-3: The Vso of a 23-ply pack is found by finding the strike velocity that yields a residual velocity equal to the V5o for a 12-ply pack. This residual velocity is then the V50 of the 23-ply pack. (23-ply data is used due to the lack of 24-ply data) (Cunniff, 1999) 21 Figure 2-4: Comparison of V50 performance for different armour systems, of similar areal densities. Superior performance is seen in the lower denier armour systems. (Tejani, 2002) 21 Figure 2-5: Ratio of maximum energy absorbed in 1 and 2-ply systems for various projectile shapes. The projectiles are not necessarily the same mass, implying also that the ballistic limit velocities are different. The fabric samples are clamped on opposite edges, and free on the other two opposite edges. (Lim et al, 2002) 22 Figure 2-6: 22-calibre fragment simulating projectile impacting a Nylon fabric panel. Note the cutting mechanism at the sharp projectile edges. (Prosser, 1988b) 22 Figure 2-7: Idealization of a point impacted fabric panel made up of pin-jointed tension members. (Roylance and Wang, 1980a) 23 Figure 2-8:Plot of out-of-plane positions along the main load bearing fibre for a 9-panel Kevlar target. (Ting et al, 1993) 23 Figure 2-9:Projectile impacting multi-ply fabric panel modelled using plate theory, in this case the plate has 100% compressibility. (Taylor and Vinson, 1998) 24 Figure 2-10:Projectile impacting multi-ply fabric panel modelled using plate theory, in this case the plate has 0% compressibility. (Taylor and Vinson, 1998) 24 Figure 2-ll:Sketch of model illustrating the breakdown and geometries. (Ting et al, 1998) 24 Figure 3-l:The powder gun set-up 46 Figure 3-2:Projectile used in experimental testing. Left standing projectile shows the front face and the right standing projectile shows the tail end 46 Figure 3-3: Schematic view of testing jig used to hold fabric panels during testing.(Starratt, 1998) 47 vii Figure 3-4:Clamping mechanism used to grip the fabric panel in the testing jig. (Starratt, 1998) 47 Figure 3-5: Photo displaying how the boundary pull out was monitored and quantified, in this case for a 3-ply Kevlar shot 48 Figure 3-6:Simple schematic of the ELVS system. 1-Laser diode 2-Horizontal collimating Lens 3- Aperture device 4-Collector lens 5-Photo-voltaic Detector (Sanders, 1997) 48 Figure 3-7:A Voltage-Time output from the E L V S system. (Starratt, 1998) 49 Figure 3-8:Schematic displaying the position of the projectile relative to the voltage output shown in Figure 3-7. (Starratt, 1998) 49 Figure 3-9:Projectile velocity-displacement graph of 1-ply Nylon and 1,2,4 ply Kevlar for a 19 Joule strike energy, non-perforating event. Note the number following the material name refers to the test number. Kevlar data taken from Cepus (2002).50 Figure 3-10:Projectile velocity-time graph comparing 1-ply Nylon and 1,2,4-ply Kevlar for a 19 Joule strike energy, non-perforating event. Kevlar data taken from Cepus (2002) 50 Figure 3-1 LProjectile velocity-displacement graph of 1-ply Nylon and 2 and 4-ply Kevlar for a 28 Joule strike energy, non-perforating event. Kevlar data taken from Cepus (2002) 51 Figure 3-12:Projectile velocity-time graph of 1-ply Nylon and 1,2,4-ply Kevlar for a 28 Joule strike energy, non-perforating event. Kevlar data taken from Cepus (2002) 51 Figure 3-13: Vs-Vr curve for 1-ply Nylon and 1-ply Kevlar. Kevlar data taken from Cepus (2002) 52 Figure 3-14:Projectile velocity-displacement graph of both 2-ply hybrids for a 27 Joule strike energy, non-perforating event. Note: Kevlar text underlined indicates a penetrated ply 52 Figure 3-15:Projectile velocity-time graph of both 2-ply hybrids for a 27 Joule strike energy, non-perforating event. Note: Kevlar text underlined indicates a penetrated ply 53 Figure 3-16: Projectile velocity-time graph of Kevlar-Nylon hybrid for a 15 Joule strike energy, non-perforating event. Note the front Kevlar panel is not penetrated. .53 Figure 3-17:Picture of Kevlar panel from Kevlar-Nylon 03. This Kevlar panel has been penetrated, but the adjacent Nylon panel has not. This is the front panel from the 27 Joule event examined in Figure 3-14 and Figure 3-15 54 Figure 3-18: Nylon panel from Kevlar-Nylon 03 hybrid pack where the Kevlar panel has been penetrated. Note the wispy Kevlar fibres stamped on the panel at the point of projectile impact. This is the rear panel from the 27 Joule event examined in Figure 3-14 and Figure 3-15 54 viii Figure 3-19:Projectile velocity-displacement graph of both 2-ply hybrids for a 42 Joule strike energy, perforating event 55 Figure 3-20:Projectile velocity-time graph of both 2-ply hybrids for a 42 Joule strike energy, perforating event 55 Figure 3-21 Comparison of the Vs-Vr curves for both 2-ply hybrid systems 56 Figure 3-22Projectile velocity-displacement graph of Kevlar-Nylon hybrid, 2-ply Kevlar and 1-ply Nylon for a. 27 Joule strike energy, non-perforating event. 2-ply Kevlar data taken from Cepus (2002) 56 Figure 3-23: Projectile velocity-time graph of Kevlar-Nylon hybrid, 2-ply Kevlar and 1-ply Nylon for a 27 Joule strike energy, non-perforating event. 2-ply Kevlar data taken from Cepus (2002) 57 Figure 3-24:Projectile velocity-displacement graph of Nylon-Kevlar hybrid, 2-ply Kevlar and 1-ply Nylon for a 27 Joule strike energy, non- perforating event. 2-ply Kevlar data taken from Cepus (2002) 57 Figure 3-25Projectile velocity-time graph of Nylon-Kevlar hybrid, 2-ply Kevlar and 1-ply Nylon for a 27 Joule strike energy non-perforating event. 2-ply Kevlar data taken from Cepus (2002) 58 Figure 3-26Projectile velocity-displacement graph of all four 4-ply hybrids for a nominal 57 Joule strike energy, non-perforating event. Note the underlined text indicates a penetrated layer 58 Figure 3-27Projectile velocity-time graph of all four 4-ply hybrids for a nominal 57 Joule strike energy, non-perforating event. Note the underlined text indicates a penetrated layer 59 Figure 3-28:First Nylon panel in K K N N 08 impacted at 59 Joules. Note the wispy nature of the stamped on Kevlar fibres 59 Figure 3-29:Projectile velocity-displacement graph of K N K N 02 and K N K N 06 for strike energies of 67 Joule and 57 Joule, respectively. Note the underlined text indicates a penetrated layer 60 Figure 3-30: Projectile velocity-time graph of K N K N 02 and K N K N 06 for strike energies of 67 Joule and 57 Joule, respectively. Note the underlined text indicates a penetrated layer. 60 Figure 3-31:Front layer of Nylon in N N K K hybrid impacted at 57 Joule strike energy. Note the melted fibre ends at the projectile impact point 61 Figure 3-32:Second layer of Nylon in N N K K hybrid impacted at 57 Joule strike energy. Note the weave of the fibres from initial layer of Nylon have been cut from the front panel and stamped onto this layer of Nylon at the projectile impact point.61 Figure 3-33:First Kevlar layer in N N K K set-up impacted at 57 Joule strike energy. The projectiles impact point is quite evident. The arrows indicate the strained orthogonal yarns 62 ix Figure 3-34: Projectile velocity-displacement of all four 4-ply Hybrids for a 90 Joule strike energy, perforating event. They are not labelled because there is no discernible difference 62 Figure 3-35: Projectile velocity-time of all four 4-ply Hybrids for 90 Joule strike energy perforating event. They are not labelled because there is no discernible difference 63 Figure 3-36:Vs-Vr curve for all four 4-ply hybrids. Note both hybrids that have initial layers of Kevlar have slightly greater x-axis intercepts 63 Figure 3-37:Projectile velocity-displacement graph of 4-ply Kevlar and K K N N hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002) 64 Figure 3-38:Projectile velocity-time graph of 4-ply Kevlar and K K N N hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002) 64 Figure 3-39: Projectile velocity-displacement graph of 4-ply Kevlar and N N K K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002) 65 Figure 3-40:Projectile velocity-time graph of 4-ply Kevlar and N N K K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002) 65 Figure 3-41: Projectile velocity-displacement graph of 4-ply Kevlar and N K N K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002) 66 Figure 3-42: Projectile velocity-time graph of 4-ply Kevlar and N K N K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002) 66 Figure 3-43: Projectile velocity-displacement graph of 4-ply Kevlar and K N K N hybrid for a nominally 60 Joule strike energy, non-perforating event. Because of the differences in behaviour, the results of both K N K N 06 and K N K N 02 events are displayed. 4-ply Kevlar data taken from Cepus (2002) 67 Figure 3-44: Projectile velocity-time graph of 4-ply Kevlar and K N K N hybrid for a nominally 60 Joule strike energy, non-perforating event. Because of the differences in behaviour, the results of both K N K N 06 and K N K N 02 events are displayed. 4-ply Kevlar data taken from Cepus (2002) 67 Figure 4-1 Calculating the rate of energy absorption over the first 35 microseconds of impact for a Kevlar-Nylon hybrid system struck at 27 Joules of energy (non-perforating event). The graph on the left displays the whole event, whereas the graph on the right displays the first 35 microseconds 79 Figure 4-2: Log-Log normalised rate of energy absorption versus strike velocity graph for the 2-ply hybrid systems, along with the fitted regression line 79 Figure 4-3: Log-Log normalised rate of energy absorption versus strike velocity graph for the four 4-ply hybrid systems, along with the fitted regression line 80 Figure 4-4: Experimentally calculated values of B for the material systems relevant to this thesis. The single and multi-ply Kevlar results are taken from Cepus (2002)...81 Figure 4-5 Comparing normalised dE/dt curves for 2-ply hybrids 81 Figure 4-6:Comparing normalised dE/dt curves for 4-ply hybrids 82 Figure 5-1 :A net-like mass string system (Zhang et al, 1998) 109 Figure 5-2:Projectile patch configuration for T E X I M with mass of projectile lumped on the indicated nodes. The element used for strain-time evaluations is also indicated.(Shahkarami, 1999) 109 Figure 5-3: Symmetry of the fabric panel (Shahkarami, 1999) 110 Figure 5-4:Projectile velocity-time graph predicted by L S - D Y N A (patch and deformable projectile) and T E X I M for 1-ply Kevlar fabric, impacted with a strike energy of 18 Joules; non-perforating event. Note all three lines are on top of each other with the L S - D Y N A deformable projectile displaying the small oscillations. ..110 Figure 5-5: Energy-time graph predicted by L S - D Y N A and T E X I M for 1-ply Kevlar panel, impacted with a strike energy of 18 Joules Energy; non-perforating event. 111 Figure 5-6:Projectile velocity-time graph predicted by L S - D Y N A and T E X I M for a 2-ply Kevlar panel, impacted with a strike energy of 18 Joules; non-perforating event. Note the lines are on top of each other I l l Figure 5-7:Projectile velocity-time graph predicted by L S - D Y N A and T E X I M for 1-ply Nylon, impacted with a strike energy of 15 Joules, non-perforating event. Note the lines are on top of each other 112 Figure 5-8Projectile velocity-time graph predicted by T E X I M as well as experimental results for 1-ply Kevlar, impacted with a strike energy of 17 Joule; non-perforating event, experimental data taken from Cepus (2002) 112 Figure 5-9:Energy-time graph predicted by T E X I M as well as experimental results for 1-ply Kevlar, impacted with a strike energy of 17 Joules; non-perforating event, experimental data taken from Cepus (2002) 113 Figure 5-10:Divergence of behaviour between experimental and predicted T E X I M results after the first strain wave reflection for 1-ply Kevlar, impacted with a strike energy of 17 Joules; non-perforating event, experimental data taken from Cepus (2002) 113 Figure 5-ll:Energy-time graph predicted by T E X I M as well as experimental results for 1-ply Nylon, impacted with a strike energy of 15 Joules; non-perforating event. Note the difference in slope between the linear regression lines of the experimental and T E X I M results 114 Figure 5-12: Projectile velocity-time graph predicted by T E X I M for a 2-ply Kevlar-Nylon hybrid impacted with a strike energy of 27 Joules; non-perforating event. The front Kevlar panel is penetrated 114 xi Figure 5-13: Experimentally measured projectile velocity-time graph for a 2-ply Kevlar-Nylon hybrid impacted with a strike energy of 27 Joules; non-perforating event. The front Kevlar panel is penetrated 115 Figure 5-14:TEXIM predicted energy and strain-time graph (strain curve computed for element shown in Figure 5-2). The portion of the energy-time curve used to calculate the rate of energy absorption for calculation of B is indicated. This is for 1-ply Kevlar, impacted with a strike energy of 18 Joules; non-perforating event 115 Figure 5-15:Predicted Normalised dE/dt versus Vstrike for 1-ply Kevlar 116 Figure 5-16:B values calculated from T E X I M for various single material and hybrid multi-layer fabrics 116 Figure 5-17:Comparison of numerical and experimental normalised dE/dt versus Vstrike curves for 1, 2, 3, 4-ply Kevlar 840d. Note the good agreement of the numerical results with the experimental results, as well as the V8'3 relationship, experimental data taken from Cepus (2002) 117 Figure 5-18:Comparison of numerical and experimental normalised dE/dt versus Vstrike curves for 8 and 16-ply Kevlar 840d. Note the good agreement of the Vs/S relationship for the 8 and 16-ply cases, and the good agreement of the numerical and experimental results in absolute terms for the 8-ply case, experimental data taken from Cepus (2002) 118 Figure 5-19Comparison of numerical and experimental normalised dE/dt versus Vstrike curves for 1-ply Kevlar 1500d and 1-ply Nylon 1050d. Note the good agreement of the numerical and experimental results for Kevlar 1500d, but not for the Nylon. However the Vs relationship agrees well.for the numerical results of both Kevlar 1500d and Nylon. Kevlar experimental data taken from Cepus (2002) 118 Figure 5-20:Comparison of numerical and experimental normalised dE/dt versus Vstrike curves for 2-ply Kevlar-Nylon hybrids. Note the good agreement of the V813 relationship, but the poor agreement between the predictions and measurements in absolute terms 119 Figure 5-21 Comparison of numerical and experimental dE/dt versus Vstrike curves for 4-ply Kevlar-Nylon hybrids. Note the good agreement of the V8'3 relationship, but the poor agreement between the predictions and measurements in absolute terms 120 Figure 5-22:Normalised energy-time graphs for 1-ply Kevlar fabric with a constant yarn count of 12 yarns per cm, for an 18 Joule Strike Energy event. The lower areal density panel absorbs normalised energy more quickly 121 Figure 5-23: Normalised energy-time graphs for 1-ply Kevlar fabric with a constant linear yarn density of 800d, for an 18 Joule Strike Energy event. The lower areal density panel absorbs normalised energy more quickly 121 Figure 5-24: Normalised energy-time graphs for 1-ply Kevlar fabric with a constant areal density of 213g/m2, for an 18 Joule Strike Energy event. A l l three lines of xii outputs are directly on top of each other, indicating energy absorption at exactly the same rate 122 Figure 5-25: Strain-time graphs for 1-ply Kevlar fabric with a constant yarn count of 12 yarns per cm, for an 18 Joule Strike Energy event. The lower areal density panel has greater strains 122 Figure 5-26: Strain-time graphs for 1-ply Kevlar fabric with a constant linear yarn density of 800d, for an 18 Joule Strike Energy event. The lower areal density panel has greater strains 123 Figure 5-27: Strain-time graphs for 1-ply Kevlar fabric with a constant areal density of 213 g/m2, for an 18 Joule Strike Energy event. A l l three outputs are directly on top of each other, indicating exactly the same strains 123 Figure 5-28:Predicted normalised dE/dt versus Vslrike for 840d Kevlar fabric displaying the loss in efficiency as the ply count is increased 124 Figure 5-29: Normalised energy-time graph of multi-ply Kevlar for an 18 Joule Strike Energy event. The smaller the ply count, the greater the rate of normalised energy absorption 124 Figure 5-30: Normalised energy-time graph for 4-ply Kevlar fabric with a varying gap values, for an 18 Joule Strike Energy event. Note the smaller the gap, the greater the rate of normalised energy absorption 125 Figure 5-31 Predicted normalised dE/dt versus Vstrike for 50-ply Kevlar 840d displaying the loss in efficiency as the gap is increased .125 Figure 5-32:Strain-Time curves for the element on the 1st ply (impact face) in 1,2,4,8-ply Kevlar systems for an 18 Joule Strike Energy event. Note reduced strains in the front ply due to engagement of subsequent plies 126 Figure 5-33:Strain-time curves of 1,2,3,4,5,6,7,8th plies in an 8-ply Kevlar system, for an 18 Joule Strike Energy event. Note lower strains in plies towards the rear of the pack 126 Figure 5-34:Cross-plot of experimental and numerically obtained B values for various materials and ply counts 127 Figure 5-35:Plot of B versus number of plies. Note the effect of gap on the calculated B for increased ply count 127 xiii Acknowledgments I would like to thank my two supervisors: Professors Reza Vaziri and Anoush Poursartip. Their enthusiasm and clever means of tackling problems made the conducting of this research both exciting and challenging. Their different styles of operation and wise advice gave me a well-rounded and fruitful atmosphere towards the accomplishment of this thesis, and I thank them. I also have to acknowledge the members of the U B C Composites Group, the camaraderie at the lab made my life most enjoyable. I should especially thank Elvis and A l i , fellow members of the Impact Group for the technical discussions and help. Karim, Jason and Mike must be thanked for not having me removed from FF105. Roger Bennett cannot go unmentioned for help with the practical aspects of the experiments and all the knowledge of the BC wilderness. Thanks also to Serge Milaire for the electronics advice. NSERC, Pacific Safety Products, DuPont and the Canadian Department of National Defence are recognised for supplying materials and financial support. I would most importantly like to thank my parents and my sister as well as my relatives for the quiet confidence and caring that always buoys my spirits and gives me hope. Lastly I wish to thank Mrs.Tracy Hobson for being a patient, honest and supportive friend. A . M . D . G . xiv Introduction 1 Introduction 1.1 Background Man has constructed body armour to protect himself from injury since the dawn of time. From the leather belts first used by Egyptian warriors, the full metal suits in use during the high middle ages by European knights to the first appearance of Nylon flak jackets for aviators in World War II, armour systems have been used to reduce combat casualties (Laible, 1980). At times these armours have been used together in order to achieve a certain performance which one system could not achieve alone: for example in the middle ages chain mail was used with a leather under-cloth; the chain mail to stop sharp edges and the leather to stop sharp points. Armours have progressed as threats have evolved and the materials available have advanced. In recent times the advent of incredibly strong and tough new fibres such as Kevlar has led to the development of highly effective soft, flexible and light personnel body armours. These bulletproof vests are primarily in use by soldiers and security personal around the world as an inexpensive method of reducing casualties from ballistic threats. These fabrics are also beginning to find uses in other areas where there is a need for impact resistant materials. These include high performance sporting good applications such as bike helmets and protective hockey equipment, as well as aerospace applications such as engine nacelle and fuselage linings to limit damage from engine rotor failure (Shockey et al, 2000). The use of high performance fibres as a protection alternative began with the use of ballistic Nylon in aviator flak jackets during the Second World War. Nylon is a material with a large strain to failure, but a relatively low specific strength and Young's modulus. 1 Introduction Nylon was in turn, in large part, replaced by the appearance of aramid fibres, the original commercial example being Kevlar. The aramid fibre's highly crystalline and oriented structure gives rise to a material with a high specific strength and Young's modulus. Although it has a lower strain to failure than Nylon its vastly increased modulus gives it better dynamic energy absorption characteristics, which makes Kevlar a superior ballistic resistant material. There have been further developments (Bajaj and Srirami, 1997) with the introduction of U H M W P E (ultra-high molecular weight polyethylene) and PBO (para-phenylene benzobizoxale). These materials take the advancements made in Kevlar fibres a degree further, with greater moduli and specific strength parameters, but they have yet to gain widespread use as bulletproof vest materials. As the use of bulletproof vests has become more common, so has the push to increase their performance and reduce their cost. Some of the problems found are that increased ply counts of fabric and increased yarn deniers used in armour system construction lead to reduced ballistic efficiencies (Cunniff, 1992). In order to design more efficiently with these materials for assorted applications we have to understand how they behave and why. In this vein a significant amount of research has been carried out to understand these fabric materials. This research has proceeded from developing an understanding of individual yarn behaviour, founded on fundamental physics, to understanding the assembly of yarns into usable fabric panels. A standard engineering approach has been undertaken by the scientific world with regards to this research: development of closed form analytical solutions, followed by the development of progressively more sophisticated computer models with the use of experimental results as validation for both models. For the cases of interest in this thesis the experiments are ballistic in nature and 2 Introduction therefore happen over very short time periods. This makes the acquisition of useful data difficult. The most common method of capturing data has been to measure the strike and residual velocities of a projectile impacting a fabric panel. The measurement of these velocities has been achieved using high-speed photography or a counter-timer method (Cunniff, 1996; Wilde et al, 1973; Prosser, 1988a). With the use of very expensive high-speed photography equipment discrete snapshots of the fabric panel can be obtained throughout the event, but this provides limited data density at best. Although a great deal of fundamental insight has been achieved using this method of experimentation, the limitation of this method is that little data is available during the period of interest: when the projectile is interacting with the fabric target. The U B C Composites Group has developed a novel method of measuring the position of the projectile continuously through a ballistic event (Starratt et al, 2000). This instrumentation technique called the Enhanced Laser Velocity Sensor (ELVS) provides a continuous stream of data that can be analysed to yield the displacement, velocity and acceleration of the projectile throughout the event. Since the projectiles in use are essentially non-deforming at the velocities of interest, the projectile information allows us to track the fabric panel's behaviour. In this way we can validate our analytical and numerical models for the whole ballistic event, versus only the beginning and end points and hypothesizing what is happening in the middle, as is traditionally done. With this insight we have found that a significant problem with numerical modelling of fabric panels is accurately modelling the boundary conditions, which have a significant effect on the overall behaviour of the fabric panel. Concurrent with this study Cepus (2002) has 3 Introduction developed an analysis technique to understand the early experimental behaviour of a fabric panel in a ballistic event, which is free of the influence of boundary conditions. A n important design consideration, as our understanding of high performance fabric panels develops, is the mixing of different materials together to achieve a superior performing, or cheaper system. Some extrapolations have been made (Cunniff, 1992) to give guidance on what effect mixing these materials will have, but much work remains to be done towards understanding these systems. 1.2 Scope of Work 1. In Chapter 2 a literature review is conducted. The focus of this review is to bring the reader up to date on the work carried out so far in terms of multi-ply and hybrid, fabric armour systems both in terms of experiments and numerical models. 2. Chapter 3 deals with the experimental method and results. This chapter is focused primarily on hybrid systems. Two and four ply combinations of Kevlar 129 and ballistic Nylon are tested using the E L V S and the results are discussed. There is also a discussion of the damage characteristics of each panel. 3. Chapter 4 introduces the early impact event behaviour analysis developed by Cepus (2002). This analysis method is then applied to the different hybrid systems. 4. Chapter 5 presents a previously developed numerical code T E X I M (Shahkarami, 1999) (Zhang et al, 1998) and the commercial code L S - D Y N A (Shahkarami et al, 4 Introduction 2000, 2001, 2002a, 2002b). It introduces some background information and a description of the different numerical codes as well as the benchmarking necessary to develop confidence in the model outputs. Parametric studies are then carried out to understand the early impact event behaviour of Kevlar 129 targets of different deniers and ply counts. 5. Chapter 6 draws conclusions and makes armour system recommendations based on the results obtained in this study. It also discusses some of the shortcomings of the current work and what can be done in the future to overcome these problems. 5 Literature Review 2 Literature Review 2.1 Background The analysis of fabric armour starts with understanding the mechanics of a single yarn under impact (of particular interest to us is the transverse impact event). As with any scientific field there are some classic papers from which other work draws its roots. In the case of single yarn mechanics this starts with Von-Karman and Duwez (1950) and Taylor's (Taylor, 1958) study of stress wave propagation in solid wires in the 1940s and 50s. In these papers the wave differential equation is applied to understand the propagation of a stress wave in a material. The most important work in taking Von Karman and Duwez and Taylor's developments a step further and developing the fundamental equations for a transversely impacted single yarn was done by Smith et al (1958). Most importantly Smith et al. developed equations relating the velocity of the projectile to strains in the fibre and the material properties. It should be noted that this theory assumes that the fibre tension versus strain curve is not rate sensitive as well as short-time creep and relaxation effects are ignored. Ringleb (1957) a researcher at the Naval Air Engineering Facility in the mid 1950s developed equations to understand the development of the stress waves in transversely impacted cables (single yarns). He found that the strain in the cable in relation to the velocity of the impactor is defined by the following relationship: 6 Literature Review 1 4 rn 3 (\ strike v 4 , \ c ) (2.1) where C is the longitudinal strain wave speed in the material of the cable. It should be noted that a fundamental assumption of this equation is that the velocity vector is constant. 2.2 Experimental 2.2.1 Multi-Ply & Hybrids Targets Some of the most important work in recent memory, due to the insights and the sheer quantity of data is the work of Cunniff at U.S. Army Natick Research Development and Engineering Center (Cunniff, 1992, 1996, 1999). Cunniff s landmark paper (Cunniff, 1992) ' A n Analysis of the System Effects In Woven Fabric Under Ballistic Impact' Cunniff discusses, among other things, the losses of efficiency seen with increased ply count and increased panel denier, which are themes of interest in this thesis. In order to ascertain the ballistic efficiency loss due to the presence of multiple plies of material the multi-ply packs are compared with a single ply spaced armour system of the same ply count. This is carried out in the following fashion. First the strike velocity versus the residual velocity (Vs-Vr) data for the single plies is compiled. Then a semi-empirical expression is derived from the experimental data to express projectile residual velocity as a function of impact conditions and system characteristics. This equation is then used to determine the maximum strike velocity that would yield a residual velocity equal to the ballistic limit of a single ply fabric system. This velocity is then assumed to be the ballistic limit of a two-ply system, without system effects (multi-ply interactions). Ballistic limit is Literature Review defined as the maximum strike velocity without causing perforation of the fabric panel. This iterative procedure is then used to find the ballistic limit of a pack with any ply count. Plots comparing the expected fabric energy absorption (calculated using the ballistic limit from the iterative procedure described above), and the observed fabric energy absorption are shown in Figure 2-1, clearly displaying the existence of system effects. Cunniff also investigated the increased specific energy absorption in lower denier plies of fabric made from the same material. For example a 1000 denier1 plain weave fabric panel absorbs more energy compared to 1500 denier basket weave Kevlar 29 panel with equivalent areal densities (as shown in Figure 2-2). He also proposes that deleterious system effects are reduced with the use of lower denier panels. Cunniff puts forward some ideas about why system effects occur. He hypothesises about the existence of a strain gradient located in the area of the transverse tent deformation. This strain gradient leads to highly localised strain at the projectile's impact point, potentially leading to premature failure. As the ply count of a system is increased the strike velocities also tend to increase, leading to larger strain gradients therefore contributing to a reduced ballistic limit. He also suggests that in a multi-ply system the force resisting the projectile is due to the tensile force component in the contacting yarns acting opposite to the projectile. Since the point of contact of the projectile is where the force from each ply acts, each layer must be transferring its resisting force through the previous layers. This increases the transverse stress component of the first few layers of a multi-ply pack, leading to premature failure. 1 Denier is defined as the mass in grams of 9000 meters of the yarn. 8 Literature Review Cunniff also examines the effect of transverse deflection constraint. As additional plies of material are added to a system the transverse deflection of the initial layers is constrained. This has the consequence of amplifying the tensile stresses in the region of the deflection cone causing adverse system effects. Cunniff tests this hypothesis with two hybridising schemes. First a 375 denier Spectra 1000 ply and a 1000 denier Kevlar 29 panel are tested together. The ballistic limit of the Spectra/Kevlar set-up is 114 m/s whereas the Kevlar/Spectra set-up yields a ballistic limit of 269 m/s. The material properties dictate that the Spectra 1000 panels have a faster transverse wave speed than those of Kevlar 29. Therefore, the transverse deflection cone of a Spectra 1000 panel placed on the impact side is constrained by the presence of the slower expanding Kevlar 29's transverse deflection cone. Reversing the panel order removes the transverse deflection constraint on the front panel. The reduced ballistic limit of the Spectra/Kevlar set-up versus that of the Kevlar/Spectra suggests that transverse deformation constraint does indeed have a negative impact on performance. This was further tested with a 1000 denier Kevlar 29 and a 1040 denier Kevlar 49 system. Kevlar 49, due to its higher Young's modulus, is expected to have a faster transverse wave speed than Kevlar 29. In this case no difference in the ballistic limit of the systems was found, implying that transverse deflection cone constraint has no effect. Cunniff concludes this to be uncertain evidence requiring further work. The work done is this thesis in Chapter 3 also provides no clear conclusion as to the effect of transverse deflection interaction. Cunniff s next significant work (Cunniff, 1996) develops a semi empirical model which is a closed form algebraic equation of the residual velocity of a projectile in terms of projectile mass, presented area, striking velocity, striking obliquity and system areal density. The 9 Literature Review equation gives a first order approximation of the areal density requirements for a given threat. This model also gives some insight into the energy absorption characteristics of body armour by non-deforming projectiles. For impact velocities significantly greater than the critical velocities of the fabric panels the systems absorb energy primarily as kinetic energy (not strain energy), due to failure occurring instantaneously with little strain wave propagation. Critical velocity is defined in this paper as the greatest impact velocity at which the no penetration occurs (this is very similar to the ballistic limit definition defined earlier). In the final Cunniff paper to be reviewed, (Cunniff 1999) extends and explores some of the observations made in his 1992 paper. He states that at strike velocities well above the V5o of the armour system the materials near the strike face fail before they can absorb a significant amount of strain energy. V$o for NIJ standards Level I, II, IIA and IIIA is defined as: the arithmetic mean of 10 shots, where 5 shots completely penetrate the pack and 5 partially penetrate the pack with the difference between the maximum velocity and minimum velocity fired being less than 45 m/s (U.S.Department of Justice, 2001). At these velocities, well above the V50 for the system, the impact is deemed to be "inelastic" according to Cunniff (exactly what is meant by inelastic is not clear), and the energy absorption is governed by the armour system's areal density.arid the amount of material involved in the impact event. This leads to what Cunniff refers to as the decoupled response of the armour system: namely that the armour system performance can be approximated by assuming the response of the first portion of the armour system has little effect on the behaviour of the remaining portion. This implies that the first portion of an armour system can be replaced with a cheaper material, as its high performance characteristics are not harnessed. Cunniff states that the response of the first plies of an armour system behave as if they are free standing and not backed by any 10 Literature Review material. Therefore the V50 performance of a system can be approximated from the performance of an armour system with a fraction of the ply count. Cunniff proves this for 12 and 23-plies of KM2™ (an improved Kevlar product from DuPont), see Figure 2-3. As an example the V50 of a 23-ply pack (a 23-ply pack is used due to the unavailability of 24-ply pack data) is found by taking the V50 of a 12-ply pack, which is the smallest pack Cunniff mentions, and finding the strike velocity that yields a residual velocity equal to the V5Q value. This strike velocity is then taken as the V50 of the 23-ply pack. Cunniff found that this approximates the V50 within the experimental error. Cunniff mentions that if the difference between the V50 velocity of the entire armour system and the critical velocity (ballistic limit) of a single ply is sufficiently small then all panels in the system will absorb an appreciable amount of strain energy. Therefore, replacement of any material panels in this type of system with a lesser performing material will result in a reduction in the performance of the system. Different projectile geometries can also sufficiently reduce the V50 of a system, such that replacement of the initial layers of the system with cheaper, lesser performing materials can have a significant effect on the performance of the overall pack. Cunniff concludes that if the tradeoffs in performance are understood, it is possible to hybridise an armour system to reduce the costs without hurting its overall performance. A n internal report from the DuPont Company (Tejani, 2002) contains an indication of the affect of different deniers on ballistic vest performance. The lower the denier of the yarn used to construct the fabric panels, the greater the performance of the armour system, at a given areal density. This is seen in Figure 2-4, which displays how different deniers have different Fjo's against the same threats. 11 Literature Review Lim et al. (2002) investigates the energy absorption characteristics of two plies of Twaron stacked together as a system. They state that for hemispherical, ogival and conical 60-degree projectiles the 2-ply system absorbs more than twice the amount of energy than a single ply, see Figure 2-5. For a flat nosed projectile the additional ply results in only about 1.25 times the energy absorption of a single ply panel. Lim et al. suggests this occurs due to the flat nose projectile head failing the panel via a transverse shearing mechanism, rather than stretching the fibres to breakage, which requires more energy. The ballistic performance of the thicker 2-ply system is less sensitive to projectile profile than the 1-ply system. With the exception of flat nosed projectiles, spaced armour systems are found to have lower ballistic limits than the comparable plied armour systems. The opposite is found for the flat nosed projectiles, with the spaced armours having a greater ballistic limit than the plied armour system. It should be remembered that this applies to a 2-ply system only. 2.2.2 Local Target Damage Prosser (1988a and 1988b) carried out work on the experimental behaviour of Nylon panels under impact loading at the U.S. Army Natick Research Center. In a series of papers he develops an understanding of Nylon panel behaviour and the most common failure mechanisms. He spends much effort in developing a simple model to predict the V50 of a fabric system based on the number of layers of Nylon in the pack. This is carried out for many different projectile calibres, allowing it to be developed as a predictive tool for Nylon armour systems. Prosser then develops the concept that Nylon fails via a cutting and shearing mechanism. . The projectile used by Prosser in these tests is a 22-calibre FSP. The photographic evidence (see Figure 2-6) directly suggests a cutting mechanism of failure. To further verify, a stiff 12 Literature Review screen was added as the last layer of a Nylon fabric armour system. This restricted the fabric panels from deforming (reduced transverse deformation), thus acting as a 'butcher's block' facilitating the cutting of the Nylon yarns and inhibiting their failure (especially in the final layers) due to elongation. The presence of the stiff screen did not change the V50 of the system significantly, suggesting elongation is not the primary means of failing a Nylon panel in a ballistic event. Prosser also found that projectiles with many 90-degree edges seem to penetrate Nylon and Kevlar ballistic panels with unexpected ease. These edges are more likely to cut the fabric thus explaining the lower ballistic limit. Lastly, Prosser also considers the behaviour of Kevlar. When Kevlar is exposed to ultraviolet light its tensile strength is significantly reduced, but its ballistic performance is not. This phenomenon suggests that impacted Kevlar fails via a mechanism other than elongation. 2.2.3 Hybrid Patents A plethora of patents exist for soft body armours, many of which deal with the hybridising of armour systems. The scientific claims made in these patents are somewhat dubious, but they do relate some of the successes in industry with regard to hybrids. A small number are surveyed here for completeness of the literature study. Pacific Safety Products Inc. (Field and Soar, 1995) found improved armour performance, both V50 and blunt deformation, by interleaving plies of Spectra Shield with aramid panels. Spectra Shield is an armour panel where Spectra™ (UHMPE) fibres are imbedded in a flexible resin matrix. The aramid, on the other hand is a woven fabric. Field and Soar claim there is a 20% reduction in areal density demand for the specified NIJ (National Institute of Literature Review Justice) threat . The ratios of interleaved layers, as well as the stacking sequence are discussed in the patent. The patented lay-up dictates that the interleave ratio is between 1:1 and 4:4 Spectra to aramid panels. The reasoning offered for the superior performance of the hybrid pack is due to increased reflection of the strain wave (called the reflection of energy in the patent), allowing greater in-plane energy absorption. The increased reflections occur at the material interfaces, due to different material properties. Zufle (1990) has also patented the idea of combining dissimilar materials to produce synergistic results. The patent refers to the use of different materials, in this case Kevlar and Nylon, in the warp and weft yarns of the fabric panel as a beneficial blending of the different material characteristics. This patent also suggests a sandwich placement of aramid fabric panels on the outside of interior Nylon panels. The patent claims the Kevlar panels are very hard (greater Young's modulus) and non-melting while the interior Nylon layers are yielding (large strain to failure) and melt readily from the heat of the incoming projectile as it attempts to pass through the initial aramid layers. This causes friction at the entry point of the aramid layer, thus dissipating a large portion of the projectile energy before it begins to traverse the interior Nylon layers. The rearmost layers of Kevlar support the Nylon from deforming and thereby trap the projectile. The use of Nylon in the set-up also reduces the cost of the body armour. Harpell et al. (1985) at Allied Signals examined the use of mixing plies of extended chain polyethylene (ECPE) with more conventional ballistic resistant materials Kevlar and Nylon. The patent recommends the material with the greatest resistance to displacement (the fabric 2 The NIJ standard mandates a minimum V50 and maximum allowable backface deformation performance for specified projectiles. The NIJ standard has different classifications for a variety of threats, from simple .22-calibre rifle bullets to armour piercing projectiles. 14 Literature Review panel whose constituent fibres require the greatest force to be displaced relative to an adjacent fibre in the plane of the layer) be placed at the back of the pack. When hybridising ECPE with Kevlar 29, better V$o performance for 22-calibre fragment threat is seen when the Kevlar panels are placed at the back of the pack. This is carried out for a 15-ply pack with 6 layers (areal density 2.3 kg/m ) of ECPE, and 9 layers (areal density 2.6 kg/m ) of Kevlar for a total system areal density of 4.9 kg/m2. The effect of stacking sequence can be seen in the V50 data provided with the V50 of ECPE/Kevlar 29 and Kevlar 29/ECPE being 1833 ft/sec and 1632 ft/sec respectively. When Nylon and ECPE are hybridised a similar result is observed, with the pack with Nylon at the back having superior performance. 2.3 Multi-ply Numerical Models The discussion of numerical models must start with Roylance and Wang and their development of the first numerical simulation of fabrics. The intractability of a closed form solution due to the occurrence and interaction of strain wave reflections at the boundaries and fibre crossover points necessitates the creation of a numerical model. The model is best described in 'Penetration Mechanics of Textile Structures' (Roylance and Wang, 1980a). The model incorporates previous work by Roylance (1973) that developed and implemented effects such as a visco-elastic material constitutive model. Roylance models a fibre panel as an assemblage of pin-jointed, flexible fibre elements each having a mass which makes the areal density of the mesh equal to that of a fabric panel (see Figure 2-7). The strike velocity of the projectile is imposed on the center point and as a result strain is induced in the adjacent elements. The constitutive material model is then used to relate the strain to tension in the element. This tension is then used to calculate the acceleration in the adjacent elements. The projectile velocity is then recalculated from the surrounding tensions using a momentum 1 5 Literature Review balance and this velocity is used in the calculations for the next time increment. Roylance also investigates numerical stability issues and its implication on the size of the time step used for numerical integrations. This model is the basis or origin of many fabric panel numerical models since developed, including the U B C Composites Group developed code, T E X I M , employed in this thesis. Roylance (1977) uses this model to make suggestions as to what type of materials would make effective ballistic armours, and what are the important factors in armour design. Roylance and Ting have further developed the model (Ting et al, 1993) to include multiple layers of fabric. Initially multi-ply targets were modelled by scaling the effective modulus and mass of a single ply fabric armour system to correspond to that of a multi-ply target. Further developments allowed the multiple plies to be modelled as discrete layers. These layers are separated by an adjustable spacing, which in this case is set to 1.0 mm. The nodes of the fabric panels are allowed to move in and out of plane. When the out of plane nodal displacements exceed the initial separation, contact is initiated (see Figure 2-8). When contact occurs the code calculates the slowdown due to the inelastic collision and then computes the nodal displacements by summing all the resisting forces due to the fibres that are in contact. Taylor and Vinson (1998) developed a multiple layer fabric target model. This work builds on the model developed by Vinson and Zukas (1975). The fabric panel is considered to be a homogeneous, isotropic, elastic plate, which deforms into a conical shell. Based on the fabric density and areal density of the plate as well as the projectile mass and striking velocity the geometry of the deformation cone is calculated. From this deformed shape a strain distribution is attainable using the structural theory for conical shells. Using the 16 Literature Review instantaneous modulus, available from wave theory, one can then calculate the stress distribution in the plate. The force on the projectile is then calculated (from the tension field acting on it) followed by the deceleration of the projectile, thereby all unknowns are found and the calculation can continue to the next time increment. Multi-ply targets are investigated by calculating the thickness of each individual plate and with varying degrees of interaction between the plates. This interaction is bounded: first by having a completely compressible plate, which allows plate interaction only after the projectile has travelled through the thickness of the preceding ply of material or by having an incompressible plate, which allows interaction of the plates as soon as the first plate is impacted (see Figure 2-9 and Figure 2-10). Lomov (1996) created a numerical model to predict the deformation and rupture of a woven multi-ply target. The multi-ply fabric system properties are the sum of the properties of the undamaged individual layers. The individual layers of the armour system are damaged and fail sequentially, thereby altering the multi-ply system properties as the ballistic event unfolds. The rupture of the individual yarns in the panel is controlled by the deformation of the yarn within the woven structure. It is unclear how Lomov allows the individual layers to interact with each other throughout the ballistic event. Lomov claims good success in modelling the V50 performance and deformation of the target. The U B C Composites Group Zhang et al. (1998) and Shahkarami (1999) developed a Roylance (1980a) based numerical model modelling the fabric as a net-like structure made of discrete string-mass components. It models the impact of a blunt cylindrical projectile on a rectangular fabric panel. It is capable of modelling multi-ply and hybrid targets, with different boundary conditions. The details of the code are discussed in Chapter 5. 17 Literature Review Ting et al. (1998) presented a model that models multi-ply targets with a selection of springs to model the interaction between layers. This model is an extension of the model created by Roylance and Wang (1980a) discussed earlier and includes interaction between the warp and fill yarns in a fabric structure, allowing the direct incorporation of decrimping in the model. The bending stiffness of the yarns is also incorporated with the inclusion of a torsional spring at the nodes. The different layers interact through compression elements (see Figure 2-11). There are three choices in the model: 1) rigid element, 2) linear spring, and 3) nonlinear spring. The layers are not allowed to pass through each other and once the deformation has exceeded the spacing the nodes displace together. The plies are laterally decoupled, therefore friction and slipping between layers is not included. It is concluded that transverse coupling affects the modelled fabric behaviour. Billon (1998) published a report which details a model he had created to investigate ballistic impact on soft armours. This model is very similar in formulation and implementation to that developed by Roylance (1980a). In this report Billon attempts to model multi-ply targets by assuming that there is no interaction between the layers (i.e. they are infinitely spaced). His model predicts the ballistic limit with reasonable success. Billon (2001) published a paper outlining updates to the numerical model he created in 1998. In the model, for multi-ply targets, Billon assumes the impact process is confined to the immediate vicinity of the impacted node and all other interactions between nodes are ignored. Further details of how the multi-plies are modelled in the code are not revealed in the paper. Due to the limitations in the numerical model Billon creates a simple analytical model in order to predict the ballistic limit of fabric armours constructed as hybrids of two or more materials. This analytical model is based on a determination of the loss of kinetic 18 Literature Review energy of the projectile per unit penetration depth into the fabric pack. A fundamental assumption in this analytical formulation is that the rate of change of kinetic energy of the projectile is constant with increasing penetration depth. This assumption is in conflict with the experimental results obtained for the fabric panels tested at U B C (Starratt, 1998). Nonetheless, from his analytical formulation Billon concludes it is not possible to increase the ballistic limit of a fabric armour system of constant areal density by replacing some of the fabric layers with a material with a lower single ply ballistic limit, i.e. the highest ballistic limit of a given armour system occurs when it is purely made of the material with the best single ply ballistic limit. Billon applies this model to hybrid cases with panels of ascending sonic velocity to avoid the problems associated with modelling interaction of the deformation cones. The ballistic limit of these hybrids is modelled with reasonable accuracy. With the stacking sequence of ascending sonic velocities, in this case Nylon impacted before H M W P E , the ballistic limit of a Nylon and H M W P E hybrid armour system seems to be strongly dependent only on the total number of layers, and only weakly on the proportion of each. 19 Literature Review 250 T S 200 • 00 0.2 0.4 0.6 OJ AREAL DENSITY (g/cm2) Figure 2-1: Absorbed energy as a function of areal density for a 1500 denier Kevlar 29 fabric. The line indicates the predicted absorbed fabric energy and the dots represent the measured values; the difference is attributed to system effects. The line is not a linearly increasing line because a 1-ply Vs-Vr curve is used to calculate this line: therefore the curvature inherent in the Vs-Vr curve is present here.(Cunniff, 1992). 0.0 0.1 0.2 0.3 0.4 0.5 0.6 AREAL DENSITY (g/cm2) 0.7 Figure 2-2: For a given areal density the lower denier panels absorb a greater amount of energy. These curves are derived from experimental data. (Cunniff, 1992) 20 Literature Review 64-Grain O-Deg. 12-ply Kevlar KM2 1000 1 800 o 600 u 5 400 3 8 200 420 m/s V50 2 3 . p i y = 501 m/s 400 600 800 Striking Velocity (m/s) 1000 Figure 2-3: The V50 of a 23-ply pack is found by finding the strike velocity that yields a residual velocity equal to the V50 for a 12-ply pack. This residual velocity is then the Vsoof the 23-ply pack. (23-ply data is used due to the lack of 24-ply data) (Cunniff, 1999) 2200 2000 O 1800 LU CO P 1600 o i n > 1400 1200 1000 1500D 1000D 850D 600D Kevlar 29 Kevlar 29 Kevlar KM2 Kevlar KM2 Figure 2-4: Comparison of Vso performance for different armour systems, of similar areal densities. Superior performance is seen in the lower denier armour systems. (Tejani, 2002). 21 Literature Review T: 3 . 0 T 3 B C3 "5. (N C u I m o c .o & o X) < i» e 2.5 2 . 0 1.5 1.0 0.5 0 . 0 A (2.48) ° ( Z 4 2 ) O (2.33) • (1.25) — Average Hemispherical Flat Ogival Projectile Shape Conical Figure 2-5: Ratio of maximum energy absorbed in 1 and 2-ply systems for various projectile shapes. The projectiles are not necessarily the same mass, implying also that the ballistic limit velocities are different. The fabric samples are clamped on opposite edges, and free on the other two opposite edges. (Lim et al., 2002) 22 Literature Review Figure 2-7: Idealization of a point impacted fabric panel made up of pin-jointed tension members. (Roylance and Wang, 1980a) 0.20 • 0.18 • ? 0.16 • f 0.14 o I 0.12 | & 0.10 I — 0.08 fcooooaaeaoooeaoaoaeooooooooooooaoaaaooo p o o o o o o o o o o a o o e e a o a a a a e e e o o o o a a o o o a o o o c c 0.06 p o o o o o e a a o o o a a o a a o a a o o e o a a a o a a a a a a a a e a o ^ B o o o o o a o o o e o a a e a e e a a o a a a a a o o a e a a a e o a a a a O 0.04 - a o o a o o o o o o o o o o o o a o a a o o a o o a e a e e a o e a o a a a e T p o o o o o o o o o o o o o e e e a a a o o e a a a o o a a o a o a a a a e 0.02 • B o o o o o o o o o o o o o o o e o o o o o o a o a a a a o e a a a a o e a - a o o o o o o o o o o o a a a a a a o o a o e a o a a e a a o e e a a a o a 0.00 —>. 1 <~ ....... ^ — *- ^ •- — 0 1 2 3 4 5 6 7 8 9 10 1L Distance From Impact Point (cm) Figure 2-8:Plot of out-of-plane positions along the main load bearing fibre for a 9-panel Kevlar target. (Ting et al., 1993) 23 Literature Review Layer 3 Layer 1 Layer 2 Figure 2-9:Projectile impacting multi-ply fabric panel modelled using plate theory, in this case the plate has 100% compressibility. (Taylor and Vinson, 1998) Layer 2 Figure 2-10:Projectile impacting multi-ply fabric panel modelled using plate theory, in this case the plate has 0% compressibility. (Taylor and Vinson, 1998) 24 Experimental Procedure & Results: Entire Impact Event 3 Experimental Procedure & Results: Entire Impact Event 3.1 Objectives The objective of this experimental study was to characterize and understand the behaviour of ballistic Nylon fabric panels, as well as 2 and 4-ply hybrid systems of Nylon and Kevlar. This chapter presents the methodology, equipment used and results of experiments carried out on fabric panels. 3.2 Experimental Set-up A l l ballistic tests reported here were conducted using the powder gun experimental facility and the E L V S measurement system available at U B C . 3.2.1 Powder Gun To fire projectiles at the fabric panels at the desired velocities a powder gun propulsion system was used. The components of the powder gun are the remote firing switch, solenoid, universal receiver, the barrel, the blast deflector and the catchment chamber (see Figure 3-1). The remote firing switch is the device used to trigger the solenoid thereby firing the system. It is attached to the solenoid by an 8-foot cable allowing the system to be fired from outside of the lab for safety reasons. The solenoid is the sparking device used to activate the instantaneous burn of gunpowder that propels the projectile. The gunpowder was Hodgdon H450, a lower energy powder, which allows a more stable burn and hence more repeatable velocities. The universal receiver is a tool that allows the use of varied barrel sizes, and 25 Experimental Procedure & Results: Entire Impact Event therefore different projectile calibres, with the solenoid-firing device. The barrel is used to obtain a controlled flight of the projectile and to aim the projectile at the desired target. In our case a 5.58 mm (22-calibre) diameter Remington 22-250 riffle barrel was used. The blast deflector is used to catch the majority of the unburned gunpowder or other debris that follows the ejection of a projectile from the barrel. The catchment chamber collects projectiles that perforated the fabric armour targets. The ELVS system and the test fixture are located between the blast deflector and the catchment chamber. These are discussed in more detail later in this chapter. 3.2.2 Projectiles The projectiles used are .22-inch diameter right circular cylinders (see Figure 3-2). They have a mass of approximately 3 grams and a length of approximately 42 mm. They are made from .22-inch ground steel rod, with the core hollowed out from the tail to attain the desired mass (see Figure 3-2). Experience has shown that these projectiles show no sign of deformation when impacting the multi-ply fabrics studied here. Plastic deformation of the projectile is therefore discounted as an energy absorption mechanism in the ballistic event, which is useful when the quantity of interest is the energy absorbed by the fabric. 3.2.3 Test Fixture A testing jig was designed and custom-built at U B C to grip the fabric panels as they were being impacted. This fixture was adapted from the NATO-STANAG-2920 (1996) standard test frame. It consists of two circular aluminium plates 444 mm in diameter that have a central 203 mm by 203 mm square opening where the material is exposed (see Figure 3-3). This is smaller than the N A T O standard, but is necessitated by the existing powder gun set-26 Experimental Procedure & Results: Entire Impact Event up. The two circular plates are clamped together with the use of 16 12.7mm coarse thread, counter bored Allen bolts. These are placed symmetrically around the opening with four bolts on each side. Removable square rods are placed in a staggered pattern (one on the front plate, two on the back plate) along the boundaries to provide the necessary gripping force to hold the fabric panel in place during the impact event (see Figure 3-4). The square rods are rounded on the leading edge contact surface to eliminate any cutting of the material at the clamping surface. The clamping pressure, and thereby the gripping force, can be adjusted by changing the torque applied to the Allen bolts via a torque wrench. Slots are machined into the front and back plates to allow the E L V S laser sheets to be located as close as possible to the target (see Figure 3-3). 3.2.4 Boundary Conditions Boundary conditions have a major affect on fabric panel behaviour (Cepus et al, 1999). Depending on the clamping pressure there is the potential for material to be pulled in at the boundaries during the ballistic event, as seen in Figure 3-5. This can be a major energy absorption mechanism affecting the behaviour of the fabric panel (Cepus, 2002). A mark was made around the visible section of the fabric panel before and after the impact event to measure the magnitude of this effect. For single material Nylon, at the clamping pressure of 20 ft-lb no pullout was observed for any of the tests. For the hybrid panel tests small amounts of pullout were noted under an applied clamping pressure of 50 ft-lb. 3.2.5 ELVS System 3.2.5.1 Hardware The enhanced laser velocity sensor (ELVS) measurement technique is a novel method of continuously measuring the location of a projectile in a ballistic impact event (Starratt et al., 27 Experimental Procedure & Results: Entire Impact Event 2000). The concept of it is quite simple. A laser sheet is emitted which passes very near the front of the fabric panel. This laser sheet is perpendicular to the fabric sheet and directly in the path where the projectile passes through to impact the fabric panel. As the projectile obscures the laser sheet a photo detector registers the change in light intensity as a drop in voltage. This voltage change is then calibrated to a displacement thereby allowing a constant measurement of the location of the projectile's tail. In the dual E L V S system now used at U B C , another laser is located near the rear face of the fabric panel allowing a continuous measurement of the target displacement or projectile's tip location on the other side of the fabric panel. This rear laser also allows tracking of the projectile after it perforates the fabric panel. A schematic of the E L V S system is shown in Figure 3-6. For more details on the ELVS system see Starratt et al. (2000), Sanders (1997) and Cepus (2002) 3.2.5.2 Data Acquisition and Processing The data captured from the E L V S system is a voltage-time signal. The voltage signal is captured with an 8 bit 4-channel digital Techtronix oscilloscope. A typical front laser voltage-time graph is shown in Figure 3-7. For an illustration of what each point on the voltage-time graph means in terms of the location of the projectile refer to Figure 3-8. The voltage-time signal is then imported into the ELVS software where a calibration file converts the captured data into a displacement-time graph. The software then calculates the projectile velocity and acceleration by successive numerical differentiation of the displacement-time curve. The processed data is then exported to a spreadsheet program for further graphing and data manipulation. 28 Experimental Procedure & Results: Entire Impact Event 3.3 Material Systems 3.3.1 Kevlar Experimental results from the testing of 1, 2, 3, 4, 8 and 16-plies of Kevlar are used extensively throughout this thesis. The ELVS results for these systems are presented and reviewed by Cepus (2002). The Kevlar examined is Kevlar 129, 840 denier plain-woven panels with a 28 by 28 yarns per inch weave structure. The areal density of the panels was found by recording the mass of a number of panels, measuring the average area of these panels then dividing the area per panel by the mass per panel. This yields a value with the units of mass per unit area. The measured areal density of 1-ply Kevlar was found to be .01922 g/cm2, the calculated areal density of an n-ply Kevlar 840d is therefore .01922 * n g/cm2. The quoted value of areal density for the specified fabric panel is .0204 g/cm2. In this work the calculated value of areal density is used. The engineering material and panel properties for the Kevlar tested are listed in Table 3-1 and Table 3-2, respectively. 3.3.2 Nylon The specifics of the material tested are Nylon Lot Brookwood 5761, 1050 denier 15oz coated 45 by 45 yarns per inch, 2x2 basket weave. The coating is a thin plastic present on one side of the panel. The effect of the coating on nylon panel behaviour is unclear, but it unquestionably increases the areal density of the panel. The areal density of the Nylon was measured to be .05207 g/cm , using the same method as that described for the Kevlar. No quoted value of areal density was available for the Nylon. The engineering material and panel properties for the Nylon tested are listed in Table 3-3 and Table 3-4, respectively. 29 Experimental Procedure & Results: Entire Impact Event 3.4 Single Material Panels Single material panel testing was geared towards achieving similar strike energies to allow easy comparison of Nylon and Kevlar results. The iso-energy results are overlaid on a single graph for easy comparison. For a list of all shots on 1-ply Nylon undertaken in this thesis refer to Table 3-5. Nylon's behaviour is compared with that of 1, 2 and 4-plies of Kevlar. Multi-ply Kevlar tests are used in order to bound the areal density of Nylon (i.e. to compare Kevlar tests with areal densities greater and less than 1-ply Nylon). Strike energy is calculated using: Strike Energy -^mVs2 (3.1) where m is the mass of the projectile and Vs is its strike velocity. For the testing carried out in this thesis the projectiles have very similar masses therefore iso-energy also means iso-impact velocity. Perforating events are not examined due to a lack of overlapping strike energies for the different systems. 3.4.1 Non-Perforating Examining the results shown in Figure 3-9 to Figure 3-12 leads to the conclusion that 1-ply Nylon is the most compliant armour system, with the Kevlar getting progressively less compliant as more layers are added (note the numbers following the graph caption label refers to the test number). Although no multi-ply Nylon data exists it is presumed that it would get progressively less compliant as additional layers are added. Nylon undergoes 30 Experimental Procedure & Results: Entire Impact Event greater final deformation (df) than any Kevlar system, as shown in the non-perforating 19 and 28 Joule comparisons of velocity-displacement curves shown in Figure 3-9 and Figure 3-11. If we consider the velocity-time graphs shown in Figure 3-10 and Figure 3-12, we see that Nylon has a greater time to maximum deformation (tj) than the 1, 2 or 4 ply Kevlar systems. It should be noted that 1-ply Kevlar is not shown in the 28 Joule comparison as it is perforated at this strike energy. A collated representation of the final displacements and corresponding times (df and tj respectively), as well as the approximate critical velocities and energies (Vc and Ec respectively) for the systems described here are listed in Table 3-6. The calculation of Vc is described in the next section. Ec for this thesis is defined as the arithmetic mean of the highest energy non-perforating shot and the lowest energy perforating shot. 3.4.2 Vs-Vr A comparison of the Vs-Vr curves of 1-ply Kevlar and 1-ply Nylon is displayed in Figure 3-13. The Vs-Vr curve for 1-ply Kevlar was taken from the data generated by Cepus (2002). The curves display typical Vs-Vr shapes for Nylon and Kevlar with 1-ply Nylon and 1-ply Kevlar having Vcs of approximately 145 m/s and 116 m/s, respectively. Vc in this case was calculated by taking the arithmetic average of the velocity of the highest non-perforating shot and lowest perforating shot. This Vc value is valid only for this specific testing set-up and projectile configuration. 31 Experimental Procedure & Results: Entire Impact Event 3.5 Hybrid Panels In comparing the different hybrid panels with the single material systems perforating cases are not examined. This is due to there being no overlapping strike energy events to allow a relevant comparison. 3.5.1 2-ply Combinations of Kevlar and Nylon The first hybrid panel to be discussed consists of one layer of Kevlar, and one layer of Nylon (see Table 3-5 for shot details). The testing was carried out with both possible stacking sequences: i.e. Nylon-Kevlar (NK) and Kevlar-Nylon (KN). The areal densities of both hybrid systems are .07129g/cm (i.e. the sum of the areal densities of 1-ply Kevlar and 1-ply i Nylon). For a listing of the Vc and Ec values of these systems as well as the df and t/ values for the 15 Joule and 27 Joule strike energy events, see Table 3-7. 3.5.1.1 Non-perforating Similar to that done previously for the single material systems, the iso-strike energy E L V S results of both hybrid systems are overlaid to compare the behaviours of the different systems. First a non-perforating event at nominally 27 Joules striking energy (see Figure 3-14 and Figure 3-15) is examined. Note that on the graphs the underlined text indicates a penetrated fabric layer. As shown in these figures the front face results for both hybrid systems are very similar: This implies that from a macroscopic standpoint, the early impact event behaviour of both hybrid panels is similar. This region of the impact event is explored in much greater detail in Chapter 4. The back-face data on the other hand shows that the Nylon-Kevlar system exhibits smaller df and tf values compared to those of the Kevlar-Nylon system. 32 Experimental Procedure & Results: Entire Impact Event In exploring the Kevlar-Nylon system it is interesting to investigate how it achieves greater maximum back face deformation. Initially, it behaves very similarly to the Nylon-Kevlar system, but at a displacement of approximately 22 mm there is a dramatic change in the slope of velocity-displacement curve (see Figure 3-14). A similar result can be seen in the velocity-time curves shown in Figure 3-15, with the change in the rate of velocity decrease occurring at a time of approximately .3 milliseconds. Post-mortem observations of the panels offer an explanation. For non-perforating tests on the Kevlar-Nylon hybrid panel it was found that the Kevlar layer was penetrated, with the only exception being very low strike energy events. The only event to fit this latter criterion in the testing undertaken was Kevlar-Nylon test number 06 for which the velocity-displacement curve is shown in Figure 3-16. The velocity-displacement curve of this event is smooth without a sudden change in displacement. For all other velocities below the Vc of the Kevlar-Nylon system the Kevlar panel is penetrated whereas the Nylon panel is not. When the Kevlar panel is penetrated some of the Kevlar fibres are stamped-on the Nylon panel. The visual presentation of this phenomenon is seen in Figure 3-17 and Figure 3-18, which are photos of the Kevlar and Nylon panels from a Kevlar-Nylon hybrid after a non-perforating ballistic event. The wispy nature of the stamped Kevlar fibres suggests that they underwent significant elongation before failing. The orthogonal yarns of the Kevlar panel show clear signs of being displaced from their initial position as seen in Figure 3-17. The Nylon panel shown in Figure 3-18 shows little sign of damage aside from a slight permanent billow deformation and the presence of the stamped-on Kevlar fibres. 3 In Table 3-5 this shot is listed as having bad data due to problems with the strike velocity, however, the qualitative description used in this analysis is still valid for the data with a strike energy of 15 Joules. 33 Experimental Procedure & Results: Entire Impact Event As opposed to the Kevlar-Nylon, the Nylon-Kevlar hybrid in this comparison has no clear demarcations in its behaviour to distinguish it from a single material system. For this stacking sequence either there is perforation or no penetration at all (i.e. partial penetration does not occur). For non-perforating events there is a small rounded indentation on both panels but no broken fibres as observed in the Kevlar-Nylon set-up. 3.5.1.2 Perforating A 42 Joule strike energy event is used next to investigate the perforation behaviour of the 2-ply hybrids. The velocity-displacement and velocity-time graphs in Figure 3-19 and Figure 3-20 respectively indicate that the Kevlar-Nylon set-up reduces the residual velocity of the projectile more than the Nylon-Kevlar set-up. This seems to be related to the Kevlar-Nylon set-up having a slightly greater Vc than Nylon-Kevlar, this is discussed at greater length in the next section. The Kevlar-Nylon system also interacts with the projectile over a greater distance and longer time frame, reducing the projectile velocity more than the Nylon-Kevlar system. This is observed in Figure 3-19 and Figure 3-20 where the Nylon-Kevlar system plateaus (i.e. the projectile perforates the panel) at a smaller displacement and shorter time than the Kevlar-Nylon set-up. 3.5.1.3 VrVr Curves The calculated F c's for Nylon-Kevlar and Kevlar-Nylon hybrids were found to be 159 m/s and 164 m/s, respectively. These values are greater than the Vc of a 1-ply Nylon panel (145 m/s) and a 1-ply Kevlar panel (116 m/s) but less than a 2-ply Kevlar panel (180 m/s). The Vs-Vr graphs are overlaid in Figure 3-21 for easy comparison. There is a window of strike velocities just after Vc where the Kevlar-Nylon system absorbs more energy (i.e. the residual Experimental Procedure & Results: Entire Impact Event velocity of the projectile is smaller) than the Nylon-Kevlar system. At a velocity of -190 m/s the two curves begin to converge to similar paths. 3.5.2 Comparison between 2-ply Hybrids and Single Material Systems 3.5.2.1 Non-perforating The hybrid panels are compared here with the relevant single-material systems at iso-strike energies. The material systems compared are 1-ply Nylon, 2-ply Kevlar and the two hybrid systems. Table 3-7 lists the Vc and Ec of these systems, as well as the df and t/ for these systems at the comparison strike energy of 27 Joules. The 2-ply Kevlar system is used for three reasons: first it is the Kevlar test with the proper strike energy, second it allows a better areal density comparison and lastly it is a two layer system which allows a better direct comparison with the hybrid systems. Each hybrid system is examined separately to avoid overcrowding the graphs. They are compared at an iso-strike energy of approximately 27 Joules. First the Kevlar-Nylon hybrid is compared with the single material systems (see Figure 3-22 and Figure 3-23). As shown in Figure 3-22, the 2-ply Kevlar system undergoes the smallest deformation, the 1-ply Nylon the largest and the hybrid somewhere in-between. The 2-ply Kevlar takes the least time to defeat the projectile (see Figure 3-23), followed by the Nylon and then lastly the hybrid. Akin to the behaviour discussed in Section 3.5.1.1 the Kevlar-Nylon hybrid behaves similarly to the 2-ply Kevlar panel initially, then at a distinct point there is a significant reduction in the slope of the velocity-displacement and velocity-time graphs, which leads to a sharp increase in the df and tf of the Kevlar-Nylon panel. Experimental Procedure & Results: Entire Impact Event The Nylon-Kevlar system however, results in the smallest time and displacement to defeat followed by the 2-ply Kevlar and then 1-ply Nylon (see Figure 3-24 and Figure 3-25). 3.5.3 4-ply Combinations of Kev lar and Nylon Next in the testing regime are 4-ply hybrids consisting of 2-layers of Kevlar, and 2-layers of Nylon. The testing was carried out with four possible-stacking sequences: Nylon-Nylon-Kevlar-Kevlar (NNKK), Kevlar-Kevlar-Nylon-Nylon (KKNN), Kevlar-Nylon-Kevlar-Nylon (KNKN) and Nylon-Kevlar-Nylon-Kevlar (NKNK). The areal density of these 4-ply hybrid systems is .1426 g/cm (i.e. the sum of the areal densities of 2-ply Kevlar and 2-ply Nylon). For a summary of all tests carried out on this set-up refer to Table 3-5. In all the subsequent figures underlined letters in the text on the graphs indicates penetrated plies. Table 3-8 provides a listing of Vc and Ec, of these systems as well as the //and djof these systems for a nominally 60 Joule strike energy event. K N K N 02 and K N K N 06 correspond to different tests performed on the same stacking sequence. It is worth noting that despite their similar strike energies, their behaviour is very different. This will be explained further later in this section. 3.5.3.1 Non-perforating The 4-ply hybrids are compared at a nominal strike energy of 57 Joules, which is near the Vc of the systems and in the zone of mixed results (see Figure 3-26 and Figure 3-27). Based on the front face E L V S data the 4-ply hybrids behave very similarly early in the event. Again it must be remembered that this analysis is global in nature and the early behaviour of the hybrids is looked at in greater detail in Chapter 4. The back face ELVS data shows a visible divergence in the behaviour of the 4-ply hybrids. In a similar fashion to what was observed 36 Experimental Procedure & Results: Entire Impact Event in the Kevlar-Nylon 2-ply hybrid, the K K N N and K N K N hybrids have significantly greater df values (see Figure 3-26) and t/ values (see Figure 3-27) than those of the N N K K or N K N K hybrids. As in the 2-ply case there is a distinct change in the slope of the velocity-displacement and velocity-time curves for K K N N and K N K N systems. Post mortem examination suggests this change in slope is due to penetration of individual plies in the system. Examination of the post mortem damage in the individual layers of the K K N N system reveal the front 2 layers of Kevlar to be penetrated, whereas the two distal Nylon layers are undamaged. In the case of K N K N 06 its front 3-plies are penetrated leaving only the final Nylon ply undamaged. In a similar fashion to what was observed in the Kevlar-Nylon 2-ply hybrid, the 4-ply hybrids have some Kevlar fibres stamped onto the first non-penetrated Nylon ply. This is seen on the final ply of Nylon in K N K N 06, on the first ply of Nylon in K K N N , as well on the second ply of Nylon in N K N K . Figure 3-28 is a picture of the stamped-on Kevlar fibres from a K K N N setup, these stamped-on fibres look very similar to that observed in the 2-ply hybrid system shown in Figure 3-18. The wispy nature of the Kevlar fibres stamped on the Nylon panel once again suggests significant elongation of the Kevlar fibres before failure. The final panels of Nylon in the K K N N and the final panel of Nylon in K N K N 02 show no visible damage aside from a slight permanent tent formation near the impact point. Comparing K N K N 06 (strike energy of 57 Joules) with K N K N 02 (strike energy of 67 Joules) the effect of individual ply failure on the behaviour of the whole system becomes clear. In Figure 3-29 and Figure 3-30 we see that despite K N K N 02 suffering the greater strike energy, it has smaller values of df and tf. Individual ply failures offer an explanation 37 Experimental Procedure & Results: Entire Impact Event of this phenomenon: in K N K N 02 only the initial Kevlar ply is failed whereas in K N K N 06 both plies of Kevlar as well as the first ply of Nylon are failed. One disparity between the 2 and 4-ply hybrids is that in the 4-ply hybrid case Nylon plies are penetrated in a non-perforating event for the system. This is seen in the N N K K , N K N K and K N K N 06 systems that were involved in the 57 Joule strike energy comparison, photographic evidence is seen in Figure 3-31. In the 2-ply hybrid case there was either no damage to the Nylon panel or a perforation of the entire system. When Nylon panels are penetrated there appears to be some melting of the fibre ends (this is described in more detail later in Section 3.5.3.2) and some of the material from the front layer is cut away, as shown in Figure 3-31. In the N N K K case the cut Nylon fibres are stamped onto the second Nylon panel (see Figure 3-32). These stamped-on Nylon fibres show no significant signs of elongation, only signs of being cut from the Nylon panel. The final two layers of Kevlar in the N N K K system show significant signs of displacement and elongation of the orthogonal yarns (see Figure 3-33). The projectile also left a permanent deformation in the Kevlar plies, approximately the size of its presented area. Only the third layer (first Kevlar ply) is displayed, as the fourth layer (second Kevlar ply) looks similar. Aside from the cut material the Nylon panels do not show any signs of permanent deformation due to elongation of the orthogonal yarns. The cutting failure of the Nylon plies is in agreement with evidence of a cutting (shearing) mechanism of failure observed in Nylon panels by Prosser (1988b). 38 Experimental Procedure & Results: Entire Impact Event 3.5.3.2 Perforating The perforation of the 4-ply hybrid systems is examined with the comparison of 90 Joule strike energy events. In examining the velocity-displacement (Figure 3-34) and the velocity-time (Figure 3-35) graphs little insight can be gained from them, however, they are included here for completeness. As mentioned earlier one interesting observation is that Nylon appears to be sensitive to heat generated from the high-speed impact of a projectile. This was noticed in events where the Nylon plies are penetrated, particularly when the whole pack is perforated, and there are Nylon layers adjacent to each other. When the Nylon plies were pulled apart for the first time after the event, there was clearly a bond developed between the plies at the impact point. This may be partly due to mechanical interlock between broken fibres from each layer as well as the interaction of melted Nylon fibre ends (see Figure 3-31). There has been some previous work done by Prosser et al. (2000) and Hall (1968) on the topic of heat released and melting of Nylon under ballistic loading. Their conclusions are that the energy dissipated into heat is small and difficult to quantify, with inconsequential effect on the behaviour of a ballistic fabric panel. Nonetheless this is a phenomenon that must be accounted for in the failure of a Nylon panel. 3.5.3.3 Vs-Vr Curves Considering the Vs-Vr curves shown in Figure 3-36 the Vc's of the 4-ply hybrids is in the range of 204 to 215 m/s depending on stacking sequence (see Table 3-8 for details). The systems with the Kevlar on the impact side have a slightly (5%) higher Vc. There is a lack of data density in this zone and the difference under discussion is not large, but observation over 39 Experimental Procedure & Results: Entire Impact Event the six types of impacted hybrids (including the 2-ply systems) seems to suggest the existence of this trend. 3.5.4 Comparison between 4-ply Hybr ids and Single Mater ia l Systems 3.5.4.1 Non-perforating In this section E L V S results of the 4-ply hybrids are compared with 4-ply Kevlar results. The 4-ply hybrids are almost 2 times heavier than 4-ply Kevlar with areal densities of .1426 g/cm2 and .07688 g/cm2, respectively. The Vc of 4-ply Kevlar and the 4-ply hybrids are approximately 250 m/s and 210 m/s, respectively (see Table 3-8 for more details). In these comparisons there is a small difference in the strike energies. 4-ply Kevlar, N N K K , K K N N and N K N K have strike energies of 63, 56, 57 and 56 joules respectively while K N K N is compared at both 56 and 67 Joules, corresponding to K N K N 06 and K N K N 02. The hybrid strike energies are approximately ten percent different from the 4-ply Kevlar. This is reasonable for the type of whole event (gross response) comparison we are performing. Each hybrid set-up is compared with the Kevlar 4-ply system individually, in order to make the graphs easier to read and understand. Global behaviour observations are made in this section, while the detailed behaviour of the panels early in the event is examined in Chapter 4. No Kevlar plies were penetrated in the 4-ply Kevlar test used for comparison. First K K N N is compared with 4-ply Kevlar. Initially the systems behave similarly, as shown in the front E L V S data in Figure 3-37 and Figure 3-38. The back face E L V S data shows that K K N N undergoes significantly larger //and Rvalues. This is linked to the penetration of the front layers of Kevlar in the hybrid discussed earlier. Once the Kevlar layers are penetrated the Nylon, being significantly more compliant than Kevlar, undergoes more deformation before defeating the projectile. 40 Experimental Procedure & Results: Entire Impact Event The N N K K system behaves similarly to the 4-ply Kevlar throughout the whole event. The 4-ply hybrid allows slightly smaller values of tf and df, as shown in Figure 3-39 and Figure 3-40, respectively. The N K N K also behaves similarly to the 4-ply Kevlar. Once again the hybrid allows a slightly smaller slightly smaller values of (/-and df, as shown in Figure 3-41 and Figure 3-42, respectively. For K N K N we have two comparisons to make, K N K N 02 and K N K N 06 (Figure 3-43 and Figure 3-44). A l l three systems are overlaid on the same graph to highlight behavioural differences. Figure 3-43 illustrates that although K N K N 02 has a greater strike velocity it undergoes only a slightly larger df than the 4-ply Kevlar whereas K N K N 06 undergoes a significantly larger df. Figure 3-44 illustrates that although K N K N 02 has a greater strike velocity its tf is similar to that of the 4-ply Kevlar whereas tf for K N K N 06 is significantly higher. The effect of ply penetration on the behaviour of the hybrid system is evident. 3.6 Chapter Summary Ballistic experiments were carried out on 1-ply Nylon and various stacking sequences of 2 and 4-ply Kevlar Nylon hybrids. The behaviour of these systems was compared using relevant iso-strike energy event E L V S results. Post-mortem observation of the hybrid systems revealed ply penetrations that correlated well with the panel behaviour as measured by the E L V S system. It was found that hybrid stacking sequence has a small effect on the Vc performance of the hybrid and a significant effect on the back face deformation behaviour of the system. Experimental Procedure & Results: Entire Impact Event Table 3-l:Material Properties for Kevlar 129 (Shahkarami, 1999) Tensile Elastic Modulus N/tex* (GPa) 66.8 (96) Tensile Strength N/tex (MPa) 2.35 (3378) Strain to Failure (%) 3.3 Fibre density g/cm3 1.44 *tex is the mass in grams of 1000 m of the material Table 3-2:Properties of Kevlar 840d panel tested. Weave Plain l x l Count (yarns/cm) 11 Areal Density (g/m2) 192.2 Yarn Crimp (%) Taken to be 0 for the work in this thesis. Yarn Density, dtex** (denier) 930 (840) **dtex is the mass in grams of 10 000 m of the material Table 3-3:Material Properties for Nylon (Smith et al., 1956) Tensile Elastic Modulus N/tex (GPa) 5.03 (5.7) Tensile Strength N/tex (MPa) .668 (760) Strain to Failure (%) 14.7 Fibre density g/cm3 1.14 Table 3-4: Properties of Nylon 1050d panel tested. Weave Basket 2x2 Count (yarns/cm) 18 Areal Density (g/m2) 520.7 Yarn Crimp (%) Taken to be 0 for the work in this thesis. Yarn Density dtex 1166(1050) (denier) 42 Experimental Procedure & Results: Entire Impact Event Table 3-5: A l l shots taken for 1-ply Nylon, 2-ply and 4-ply Kevlar-Nylon hybrids. Filename* Plies Test # Notes Strike Velocity [m/s] Residual Velocity [m/s] Mass Proj. [gram] Strike Energy Ul Residual Energy [J] Outcome Ballistic Nylon 01 1 1 63 0 2.92 5.8 0.0 stopped Ballistic Nylon 09 1 9 104.5 0 2.95 16.1 0.0 stopped Ballistic Nylon 02 1 2 114.5 0 2.96 19.4 0.0 stopped Ballistic Nylon 03 1 3 125.4 0 2.96 23.3 0.0 stopped Ballistic Nylon 10 1 10 138 0 2.95 28.1 0.0 stopped Ballistic Nylon 08 1 8 154.3 100 2.93 34.9 14.7 failed Ballistic Nylon 04 1 4 164.2 127 2.98 40.2 24.0 failed Ballistic Nylon 05 1 5 165.5 130 2.96 40.5 25.0 failed Ballistic Nylon 07 1 7 219.5 180 2.95 71.1 47.8 failed Kevlar Nylon 06 2 6 94.7 0.0 2.98 13.4 0.0 stopped Kevlar Nylon 05 2 5 128.7 0.0 2.96 24.5 0.0 stopped Kevlar Nylon 03 2 3 133.7 0.0 2.96 26.5 0.0 stopped Kevlar Nylon 08 2 8 156.6 0.0 2.97 36.4 0.0 stopped Kevlar Nylon 04 2 4 171.6 65.0 2.96 43.6 6.3 failed Kevlar Nylon 02 2 2 181.8 95.0 2.95 48.8 13.3 failed Kevlar Nylon 07 2 7 187.3 133.0 2.97 52.1 26.3 failed Kevlar Nylon 01 2 1 257.4 215.4 2.95 97.7 68.4 failed bad Nylon Kevlar 01 2 1 shot bad Nylon Kevlar 03 2 3 shot Nylon Kevlar 02 2 2 104.7 0.0 2.98 16.3 0.0 stopped Nylon Kevlar 06 2 6 128.7 0.0 2.94 24.3 0.0 stopped Nylon Kevlar 04 2 4 135.2 0.0 2.97 27.1 0.0 stopped Nylon Kevlar 07 2 7 149.0 0.0 2.94 32.6 0.0 stopped Nylon Kevlar 08 2 8 168.3 104.0 2.95 41.8 16.0 failed Nylon Kevlar 05 2 5 187.9 131.0 2.95 52.1 25.3 failed Nylon Kevlar 09 2 9 240.2 205.0 2.95 85.1 62.0 failed K K N N 03 4 3 194.1 0.0 2.98 56.1 0.0 stopped K K N N 08 4 8 199.5 0.0 2.97 59.1 0.0 stopped K K N N 02 4 2 222.7 125.0 2.98 73.9 23.3 failed K K N N 10 4 10 234.9 140.0 2.97 81.9 29.1 failed K K N N 09 4 9 243.1 145.0 2.97 87.8 31.2 failed K K N N 01 4 1 388.9 355.0 2.98 225.4 187.8 failed bad K K N N 04 4 4 data data K K N N 05 4 5 lost bad K K N N 06 4 6 data bad K K N N 07 4 7 data K K N N 11 4 11 data 43 Experimental Procedure & Results: Entire Impact Event Test # Strike Residual Mass Strike Residual Filename* Plies Notes Velocity Velocity Proj. Energy Energy Outcome [m/s] [m/s] Igraml [J] UI lost K K N N 12 4 12 bad data N N K K 05 4 5 160.1 0.0 2.95 37.8 0.0 stopped N N K K 04 4 4 192.5 0.0 2.97 55.0 0.0 stopped N N K K 07 4 7 211.6 135.0 2.97 66.5 27.1 failed N N K K 02 4 2 242.0 185.0 2.97 87.0 50.8 failed N N K K 01 4 1 bad data N N K K 03 4 3 bad data N N K K 06 4 6 bad data K N K N 01 4 1 bad data K N K N 06 4 6 195.4 0 2.98 56.9 0.0 stopped K N K N 02 4 2 212.7 0 2.97 67.2 0.0 stopped K N K N 04 4 4 216.4 95 2.98 69.8 13.4 failed K N K N 05 4 5 235.5 140 2.98 82.6 29.2 failed K N K N 03 4 3 252.4 185 2.98 94.9 51.0 failed N K N K 04 4 4 189 0 2.98 53.2 0.0 stopped N K N K 07 4 7 196 0 2.95 56.7 0.0 stopped N K N K 06 4 6 211.9 120 2.95 66.2 21.2 failed N K N K 05 4 5 228.8 150 2.95 77.2 33.2 failed N K N K 03 4 3 252.7 180 2.98 95.1 48.3 failed N K N K 01 4 1 259.2 185 2.98 100.1 51.0 failed N K N K 02 4 2 bad shot *Nylon, Kevlar, N(Nylon) and K(Kevlar) refer to the material used and stacking sequence with impact face on the left, 01 refers to the test number. 44 Experimental Procedure & Results: Entire Impact Event Table 3-6:Comparison of results for 1-ply Nylon and 1,2,4-ply Kevlar. Note: There is no data available for 1-ply Kevlar for the 28 Joule shot because the panel is perforated at this velocity. Material 19 Joule Strike Energy 28 Joule Strike Energy Approx. Vc Approx. Ec (Joules) Time (//) (milliseconds) Displacement (df**) (mm) Time (tf) (milliseconds) Displacement (df**) (mm) (m/s) 1-ply Nylon 1-ply Kevlar .54 .44 30 22.2 .5 N / A 33.2 N / A 146 116 32 20 2-ply Kevlar .41 19.9 .42 24.5 182 49 4-ply Kevlar .46 16.1 .33 16.6 250 91 *tf is the time at final (maximum) displacement **df is the final (maximum) displacement of the armour system Table 3-7: Comparison of results for Kevlar-Nylon and Nylon-Kevlar comparison Areal Density (g/m2) 15 Joule Strike Energy 27 Joule Strike Energy Approx. Approx. Material Time (tf*) (milliseconds) Displacement (df*) (mm) Time (//) (milliseconds) Displacement (df*) (mm) Vc |m/s] £ C [ J | 1-ply Nylon 520.7 N / A N / A .5 33.2 146 32 2-ply Kevlar 384.4 N / A N / A .42 24.5 182 49 Kevlar-Nylon Nylon-Kevlar 712.9 712.9 .31 .30 20.7 18.7 .6 .33 26.3 20.85 164 159 40 37 *tf is the time at final (maximum) displacement **df is the final (maximum) displacement of the armour system Table 3-8: Comparison of results for 4-ply hybrids and 4-ply Kevlar. Material Areal Density (g/m2) Failed Nominally 60 Joule strike Event Approx. Vc Approx. Plies Time (//) (milliseconds) Displacement (df**) (mm) |m/s| EC[J\ K K N N 1426 K K .489 29.5 213 67 K N K N 02 1426 K .35 24.6 215 68 K N K N 06 1426 K N K .489 28.2 215 68 N K N K 1426 N K .295 23 204 61 N N K K 1426 N .323 23.5 205 62 4-ply Kevlar 768.8 none .369 24.1 250 91 *tf is the time at final (maximum) displacement **d f is the final (maximum) displacement of the armour system 45 Experimental Procedure & Results: Entire Impact Event Catchments C h a m b e r Test F ix ture Front E L V S i / / Blast Deflector Ba r r e l 1 Figure 3-1 :The powder gun set-up. 1 0 2 0 3 0 4 0 S U 6 4 Figure 3-2:Projectile used in experimental testing. Left standing projectile shows the front face and the right standing projectile shows the tail end. 46 Experimental Procedure & Results: Entire Impact Event holes for 12.7 mm (1/2") alignment pins (x 2) 7.9 mm ( V ) square rods <x4) mm (17 1/ 2") diameter inside surface 12.7 mm (V) N/C coarse thread, counter bored alien screws (x 16) slot for laser sheet: 15.9 mm (5/8") deep 4.8 mm ( V ) wide 31.8 mm d V ) FRONT PLATE 7.9 mm ( V ) square rods (x 8) 12.7 mm (1/2") alignment pins (x 2) 279 mm (11") inside surface slot for laser sheet: 15.9 mm deep 4.8 mm wide BACK PLATE Figure 3-3: Schematic view of testingjig used to hold fabric panels during testing.(Starratt, 1998) Fabric Target Front Plate Back Plate Clamping rods Figure 3-4:Clamping mechanism used to grip the fabric panel in the testingjig. (Starratt, 1998) 47 Experimental Procedure & Results: Entire Impact Event < Difference Between Initial and Final Figure 3-5: Photo displaying how the boundary pull out was monitored and quantified, in this case for a 3-ply Kevlar shot. 3 r 1 1 1 5 t II i Figure 3-6:Simple schematic of the E L V S system. 1-Laser diode 2-Horizontal collimating Lens 3- Aperture device 4-Collector lens 5-Photo-voltaic Detector (Sanders, 1997) 48 Experimental Procedure & Results: Entire Impact Event Figure 3-7:A Voltage-Time output from the E L V S system. (Starratt, 1998) Laser Diode Lens Projectile at various points in time Photo Detector Lens Figure 3-8:Schematic displaying the position of the projectile relative to the voltage output shown in Figure 3-7 (Starratt, 1998) 49 Experimental Procedure & Results: Entire Impact Event Displacement [mm] Figure 3-9:Prqjectile velocity-displacement graph of 1-ply Nylon and 1,2,4 ply Kevlar for a 19 Joule strike energy, non-perforating event. Note the number following the material name refers to the test number. Kevlar data taken from Cepus (2002). (0 E u o > Time [msec] Figure 3-10:Projectile velocity-time graph comparing 1-ply Nylon and 1,2,4-ply Kevlar for a 19 Joule strike energy, non-perforating event. Kevlar data taken from Cepus (2002). 50 Experimental Procedure & Results: Entire Impact Event Figure 3-11 :Projectile velocity-displacement graph of 1-ply Nylon and 2 and 4-ply Kevlar for a 28 Joule strike energy, non-perforating event. Kevlar data taken from Cepus (2002). Figure 3-12:Projectile velocity-time graph of 1-ply Nylon and 1,2,4-ply Kevlar for a 28 Joule strike energy, non-perforating event. Kevlar data taken from Cepus (2002). 51 Experimental Procedure & Results: Entire Impact Event Figure 3-13: Vs-Vr curve for 1-ply Nylon and 1-ply Kevlar. Kevlar data taken from Cepus (2002). 200 i — r - — r - — r - — ^ -150 Displacement [mm] Figure 3-14:Projectile velocity-displacement graph of both 2-ply hybrids for a 27 Joule strike energy, non-perforating event. Note: Kevlar text underlined indicates a penetrated ply. 52 Experimental Procedure & Results: Entire Impact Event 200 - i 1 1 150 -50 1 1 Time [msec] Figure 3-15:Projectile velocity-time graph of both 2-ply hybrids for a 27 Joule strike energy, non-perforating event. Note: Kevlar text underlined indicates a penetrated ply. 140 120 Displacement [mm] Figure 3-16: Projectile velocity-time graph of Kevlar-Nylon hybrid for a 15 Joule strike energy, non-perforating event. Note the front Kevlar panel is not penetrated. 53 Experimental Procedure & Results: Entire Impact Event Line to monitor boundary pull out 5cm Figure 3-17:Picture of Kevlar panel from Kevlar-Nylon 03. This Kevlar panel has been penetrated, but the adjacent Nylon panel has not. This is the front panel from the 27 Joule event examined in Figure 3-14 and Figure 3-15. Figure 3-18: Nylon panel from Kevlar-Nylon 03 hybrid pack where the Kevlar panel has been penetrated. Note the wispy Kevlar fibres stamped on the panel at the point of projectile impact. This is the rear panel from the 27 Joule event examined in Figure 3-14 and Figure 3-15. 54 Experimental Procedure & Results: Entire Impact Event 200 180 0 5 10 15 20 25 30 35 40 45 Displacement [mm] Figure 3-20:Projectile velocity-time graph of both 2-ply hybrids for a 42 Joule strike energy, perforating event. 55 Experimental Procedure & Results: Entire Impact Event 250 j 1 --(evIar-Nylon Jylon-Kevlar k • * — * k - i 50 100 150 Vstrike (m/s) 200 250 300 Figure 3-21 :Comparison of the Vs-V, curves for both 2-ply hybrid systems. 200 Displacement [mm] Figure 3-22:Projectile velocity-displacement graph of Kevlar-Nylon hybrid, 2-ply Kevlar and 1-ply Nylon for a.27 Joule strike energy, non-perforating event. 2-ply Kevlar data taken from Cepus (2002). 56 Experimental Procedure & Results: Entire Impact Event 200 150 Time [msec] Figure 3-23: Projectile velocity-time graph of Kevlar-Nylon hybrid, 2-ply Kevlar and 1-ply Nylon for a 27 Joule strike energy, non-perforating event. 2-ply Kevlar data taken from Cepus (2002). 200 -i 1 1 1 1 1 1 1 150 -100 J 1 1 1 1 Displacement [mm] Figure 3-24:Projectile velocity-displacement graph of Nylon-Kevlar hybrid, 2-ply Kevlar and 1-ply Nylon for a 27 Joule strike energy, non- perforating event. 2-ply Kevlar data taken from Cepus (2002). 57 5 Experimental Procedure & Results: Entire Impact Event 200 150 Time [msec] Figure 3-26:Projectile velocity-displacement graph of all four 4-ply hybrids for a nominal 57 Joule strike energy, non-perforating event. Note the underlined text indicates a penetrated layer. 58 Experimental Procedure & Results: Entire Impact Event 250 -100 —I— Time [msec] Figure 3-27:Projectile velocity-time graph of all four 4-ply hybrids for a nominal 57 Joule strike energy, non-perforating event. Note the underlined text indicates a penetrated layer. Figure 3-28:First Nylon panel in K K N N 08 impacted at 59 Joules. Note the wispy nature of the stamped on Kevlar fibres. 59 Experimental Procedure & Results: Entire Impact Event 250 -50 Displacement [mm] Figure 3-29:Projectile velocity-displacement graph of K N K N 02 and K N K N 06 for strike energies of 67 Joule and 57 Joule, respectively. Note the underlined text indicates a penetrated layer. 250 Time [msec] Figure 3-30: Projectile velocity-time graph of K N K N 02 and K N K N 06 for strike energies of 67 Joule and 57 Joule, respectively. Note the underlined text indicates a penetrated layer. 60 Experimental Procedure & Results: Entire Impact Event Figure 3-31 :Front layer of Nylon in N N K K hybrid impacted at 57 Joule strike energy. Note the melted fibre ends at the projectile impact point. Figure 3-32:Second layer of Nylon in N N K K hybrid impacted at 57 Joule strike energy. Note the weave of the fibres from initial layer of Nylon have been cut from the front panel and stamped onto this layer of Nylon at the projectile impact point. 61 Experimental Procedure & Results: Entire Impact Event Figure 3-33:First Kevlar layer in N N K K set-up impacted at 57 Joule strike energy. The projectiles impact point is quite evident. The arrows indicate the strained orthogonal yarns. 300 15 20 25 30 Displacement [mm] 35 40 45 Figure 3-34: Projectile velocity-displacement of all four 4-ply Hybrids for a 90 Joule strike energy, perforating event. They are not labelled because there is no discernible difference. 62 Experimental Procedure & Results: Entire Impact Event 300 0.05 0.1 0.15 Time [msec] 0.2 0.25 Figure 3-35: Projectile velocity-time of all four 4-ply Hybrids for 90 Joule strike energy perforating event. They are not labelled because there is no discernible difference. 400 350 300 «" 250 1 2 •a '35 2 > 150 100 50 0 : : - • - K N K N - • - N K N K -W-NNKK : : H - * -K I (NN . ; 71 50 100 150 200 250 Vstrike (m/s) 300 350 400 450 Figure 3-36:Vs-Vr curve for all four 4-ply hybrids. Note both hybrids that have initial layers of Kevlar have slightly greater x-axis intercepts. 63 Experimental Procedure & Results: Entire Impact Event 250 200 150 E 100 "5 o 0) > 50 -50 K K NN 03 4-ply K e v l a r 0 6 -0 J i 1 0 i 5 * 3 0 3 Displacement [mm] Figure 3-3 7: Projectile velocity-displacement graph of 4-ply Kevlar and K K N N hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002). 250 200 „ 1 5 0 "Jo i 2? 100 o o > 50 -50 K K NN 03 0 0 2 / 0 4-ply Kev lar 06 4 ^ V V s ^ s ^ Time [msec] Figure 3-38:Projectile velocity-time graph of 4-ply Kevlar and K K N N hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002). 64 Experimental Procedure & Results: Entire Impact Event 250 Figure 3-39: Projectile velocity-displacement graph of 4-ply Kevlar and N N K K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002). 250 -i 1 1 1 1 1 -50 1 1 Time [msec] Figure 3-40:Projectile velocity-time graph of 4-ply Kevlar and N N K K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002). 65 Experimental Procedure & Results: Entire Impact Event 250 Displacement [mm] Figure 3-41: Projectile velocity-displacement graph of 4-ply Kevlar and N K N K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002). 250 200 150 •52 E 100 o O 50 Q) > -50 -100 4-ply Kevlar 06 0 0 NKNK 07 Time [msec] Figure 3-42: Projectile velocity-time graph of 4-ply Kevlar and N K N K hybrid for a nominally 60 Joule strike energy, non-perforating event. 4-ply Kevlar data taken from Cepus (2002). 66 Experimental Procedure & Results: Entire Impact Event 250 Figure 3-43: Projectile velocity-displacement graph of 4-ply Kevlar and K N K N hybrid for a nominally 60 Joule strike energy, non-perforating event. Because of the differences in behaviour, the results of both K N K N 06 and K N K N 02 events are displayed. 4-ply Kevlar data taken from Cepus (2002). Figure 3-44: Projectile velocity-time graph of 4-ply Kevlar and K N K N hybrid for a nominally 60 Joule strike energy, non-perforating event. Because of the differences in behaviour, the results of both K N K N 06 and K N K N 02 events are displayed. 4-ply Kevlar data taken from Cepus (2002). 67 Experimental Results: Early Impact Behaviour 4 Experimental Results: Early Impact Behaviour 4.1 Introduction Early event behaviour here refers to the response of the panels before the longitudinal strain wave induced by the projectile impact returns to the impact point after reflection from the boundary. In other words, the following analysis is valid for trebound ~ (4-1) weave where L is the distance from the impact point to the edge boundary (half the panel size), CWeave is the longitudinal strain wave velocity of the woven fabric (discussed at more length later in this chapter) and treb0und is the end point of the analysis (i.e. the time taken by the longitudinal strain wave to arrive at the impact point). In order to understand the mechanics of how fabric panels are absorbing energy we must return to one of the seminal papers presented by Smith et al. (1958). From the equations developed for the behaviour of transversely impacted single yarns we can develop equations to help us understand fabric panels. Making some simplifications and reasonable assumptions we can develop a tractable analytical method of examining the early behaviour of fabric armour panels. 4.1.1 Theoretical Background The development and a more detailed derivation of the relationship presented in this section can be found in Cepus (2002). 68 Experimental Results: Early Impact Behaviour 4.1.1.1 Relation Between Strain, Strain Wave Velocity and Projectile Velocity Smith et al. (1958) developed the following equation relating the velocity of the projectile (projectile) to the strain f i n a single fibre: Vpro)eclik = J(\ + e)2Ua-[{\ + s)U'-wY (4.2) where LP is the transverse wave velocity, which is the velocity of the out of plane deformation of the yarn, and w is the in-plane velocity of the yarn. In the form it is written the transverse wave velocity is in the Lagrangian frame of reference and is therefore moving with the yarn. It can be shown that (Smith et al., 1958): U~C^^h (43) where Cf,bre is the longitudinal strain wave velocity in the fibre given by: Cf^=ff- (4-4) where Ef,bre and p are the Young's modulus and density of the fibre, respectively. We can define the in-plane velocity of the yarn, which is the flow of material behind the strain wave in towards the impact point as: ^> = Cflbn-s (4.5) Substituting Equations 4.3 and 4.5 into Equation 4.2, making some reasonable assumptions and simplifications and evaluating the equation at the instant of impact where t=0 (VProjectiie=Vsirike) we are left with the following relationship: 69 Experimental Results: Early Impact Behaviour strike s = - C \ ^ fibre J This leaves us with a simple equation for strain in terms of the strain wave velocity of the material and the strike velocity of the projectile. This equation was originally derived by Ringleb (1957) in his research on arrestor cables. Writing the equation for the velocity of the transverse wave in terms of a fixed (laboratory) reference we have the following: Substituting Equation 4.3 into 4.7, simplifying and making the assumption that the strain for the time frame we are looking at is small (Cepus, 2002) results in: 4.1.1.2 Panel Strain Wave Velocity The quantity Cweave is the strain wave velocity measured in the fabric panels. This was done experimentally at U B C by the author and Cepus. A more detailed discussion of the experimental method and results can be found in Cepus (2002). It was found that the strain wave velocity in the fabric panel was different from that expected theoretically for a single yarn. This has to do with weave factors such as the type and density of the weave (tightness of the weave based on the yarn spacing). Roylance (1980b) proposed a factor be included in the theoretical calculation of strain wave velocity to account for the yarn crossovers in a fabric panel. From the experimental determination of Cweave we can calculate directly what this factor should be. This gives us insight into the effect of weave architecture on the performance of a fabric panel. The equation for Cweave can therefore be written as: U = (l + sp'--w (4.7) U = C fibred (4.8) 70 Experimental Results: Early Impact Behaviour E fibre C fibre C, weave 4a (4.9) ap where a >1 is a single parameter that encompasses the weave effect. 4.1.2 Energy Absorption Mechanisms A fabric panel absorbs the impact energy via three major mechanisms: strain energy, in-plane kinetic energy, and out-of-plane kinetic energy. Each will be described mathematically in this section. It is important to note that we are evaluating the absorption of energy over the interval 0 < t < treb0Und- Over this interval the slope of the experimental energy-time graph was observed to be fairly linear. Since the velocity of the projectile changes over this interval an average projectile velocity should be used in these calculations. But the velocity change can be considered to be small enough that we can replace the average projectile velocity with the strike velocity. 4.1.2.1 Strain Energy The strain energy absorbed is the internal work done in extending the fibres. This is found by assuming linear-elastic constitutive behaviour for the fabric in the time frame of interest. The area under the stress-strain curve is then calculated and multiplied by the volume of material being strained to result in: Making certain assumptions on the regions of the fabric being strained (Cepus, 2002), the volume of material can be written in the following proportional form: E ,bres2Volume (4.10) Volume oc C weave (4.11) 71 Experimental Results: Early Impact Behaviour where C w e a v e is the strain wave velocity in the woven fabric panel, t the time after impact, Ad the areal density of the fabric panel and p is the density of the fibres. Substituting Equations 4.6 and 4.11 into Equation 4.10 and simplifying we are left with: E az A , t - ^ - C e a ^ weave strike c \ weave J (4.12) Differentiating with respect to time and taking the average we arrive at the following expression for the average rate of strain energy absorption: V dt j avg * fibre >—i 0 0 A d ^ weave P v strike V ^weave J (4.13) 4.1.2.2 In-plane Kinetic Energy In-plane kinetic energy corresponds to the flow of material towards the impact point in the wake of the strain wave. This material has a mass and therefore it requires energy to be displaced. It is calculated using the following equation for kinetic energy: EIP --^massxw2 (4-14) Where mass can be defined as being proportional to: massccCweajAd (4.15) Substituting Equations 4.15, 4.6 and 4.5 into Equation 4.14 and simplifying we are left with: E,p oc A,tC IP a weave {v ^ r strike V ^ weave J (4.16) 72 Experimental Results: Early Impact Behaviour Differentiating with respect to time and computing the average: ( y dt a weave strike V weave J (4.17) 4.1.2.3 Out-of-Plane Kinetic Energy The out-of-plane kinetic energy is the energy required to transversely displace the mass of material in the fabric panel involved in the tent formed by the impact of the projectile. Assuming that the transverse velocity of the tent is constant and equal to the instantaneous projectile velocity (Roylance, 1977). We can write the out-of-plane kinetic energy of the tent as: E(3P 2 mClSStransverse^projectile (4.18) where the mass is proportional to: maSS,ransverse X A d t U (4.19) Substituting Equations 4.19, 4.8 and 4.6 into Equation 4.18, equating the transverse velocity of the material with the projectile velocity and simplifying we are left with: EQP x AdtCweave strike c V weave J (4.20) After time differentiation and averaging: dE0I, dt v.. trlke r1 V weave J (4.21) 4.1.2.4 Rate of Energy Absorption Summing the energy absorption mechanisms, the total rate of energy absorption becomes: 73 Experimental Results: Early Impact Behaviour <dE ^ Ur/total dt avg -(E,+E„,+E„r) cc A„ avg V strike V ^ weave J •'fibre P c c c,.,„„„„ + 3 weave 3 ' weave + (4.22) 4.1.3 Early Event Ballistic Efficiency, B Rewriting Equation 4.22 with the knowledge that Cf,bre, Ef,bre, p are material properties and a is a weave parameter we are left with: fdE^ V dt Javg = AdB(vslnkey (4.23) where the constant B takes into account the material properties and weave architecture of the system being impacted. Finally, we can rearrange the equation to result in: UdE} A,, \ dt =B (vstnkey (4.24) *d V U l Javg This normalises the equation with respect to areal density allowing for direct comparison of early event behaviour of different material systems (including hybrids). The calculation of B is discussed in the next section. 4.1.3.1 Rate of Energy Absorption In Equation 4.24 the areal density (Ad) of the panel and the projectile strike velocity (Vstrike) are known, the remaining unknowns are the constant B and rate of energy absorption (dE/dt). The calculation of B is discussed in the.next section. The rate of energy absorption is found from the energy-time graphs obtained from the ELVS outputs. The hybrid panels under 74 Experimental Results: Early Impact Behaviour consideration appear to have a linear rate of energy absorption over an interval of approximately 50 microseconds for Nylon impact faced systems and 35 microseconds for Kevlar impact faced systems (the difference is due to Kevlar's greater longitudinal strain wave speed). A linear regression over this interval was performed on the energy-time plots of the panels, where the slope of the regression line yields an average rate of energy absorption of the system. This linear line was checked graphically to make sure it passed through the measured energy-time graph with an acceptable fit (see Figure 4-1 for an illustration of this procedure). 4.1.3.2 Calculation of B The slopes calculated from all the tests for a particular material system are then normalised for areal density and plotted versus the strike velocity on a log-log graph. A least squares best-fit line is then fitted through these data points. Using Equation 4.24 we force the slope of this line to be proportional to 8/3, and then calculate the intercept, which is related to B in Equation 4.24. In this fashion the only unknown parameter, B, in Equation 4.24 is found. Figure 4-2 and Figure 4-3 display the normalised dE/dt versus strike velocity plots for the 2 and 4-ply hybrid systems, respectively. Both the best fit line and the data points are shown on these log-log plots. The good agreement indicates that indeed the rate of energy absorption is proportional to the strike velocity to the power of 8/3. Figure 4-4 displays the calculated B parameter for all material systems that have been tested in this study. 4.1.3.3 Calculation of B Based on Rule of Mixtures Analysis When examining the experimentally obtained B values for the hybrids, one would first ask, given the B values of Kevlar and Nylon are known, what would we expect the B value of the hybrid to be? The most straightforward method of conducting this analysis is an application 75 Experimental Results: Early Impact Behaviour of the rule of mixtures. The hybrid B is calculated by adding together the B values for each material, weighted with respect to areal density, and then dividing by the total areal density of the hybrid panel. Mathematically this can be written as: n _ (^d) kevlar & kevlar + i ^ d ) nylon &'nylon , . . . . °hybrid ~ 7V\ . / A s l 4 - ^ ) ( A d )kevlar + ( A d )nylon When calculating the Bhybrid for 2-ply hybrids the B value for 2-ply Kevlar is used. There is no experimental data available for a B value for 2-ply Nylon; therefore the reduction seen in B for Kevlar from 1 to 2-ply (in percentage terms) is applied to the B value for 1-ply of Nylon. A similar procedure was followed to calculate the Bhybrid for the 4-ply hybrids. The 4-ply Kevlar B value was used, along with a similar reduction procedure to calculate the 4-ply Nylon B value from the 1-ply Nylon B value. The results of the rule of mixture analysis are shown in Table 4-1. It is clear from these results that the rule of mixtures gives a reasonably good estimate of B for the hybrids. The rule of mixtures, however, cannot differentiate between the different stacking sequences. The experimental results provide no conclusive evidence that stacking sequence affects the value of B (see Figure 4-5 and Figure 4-6). For the 2-ply case (Figure 4-5), no discernable difference is seen between either hybrid set-up. For the 4-ply case (Figure 4-6) K N K N has a larger B value, but the remaining 3 stacking sequences have very small differences. It is unclear whether more testing data would lower the B value for K N K N therefore making the effect of stacking sequence on B minimal. 76 Experimental Results: Early Impact Behaviour 4.1.3.4 The significance of the Constant B Quantification of the constant B allows us to compare systems with different materials, material sequences and ply counts, providing an idea of which system absorbs energy at a superior rate. This does not necessarily give an indication of which system has a greater ballistic limit (Vc), but it does help to provide insight into the performance of the system under evaluation. More importantly this gives us a starting point towards the understanding of the behaviour of fabric panels. By focusing on the early event behaviour we have effectively eliminated the structural response of the system and the effect of imperfect boundary conditions and are evaluating only the material response. With this understanding of the experimental behaviour early in the impact event, we can now proceed to use a numerical model to study the observed fabric panel behaviour. In the next chapter a numerical model, T E X I M , is used to compute B and develop an understanding of the factors that influence it. 4.2 Chapter Summary The early event behaviour of the 2 and 4-ply Kevlar Nylon hybrids are found to fit the relationship developed by Cepus (2002); namely that the initial rate of energy absorption of the armour systems is proportional to V8''\ The stacking sequence is found to have no significant effect on the early rate of energy absorption of the armour systems as seen in the calculated B values. It was also found that a rule of mixtures analysis gives a reasonable estimate of the expected value of B. ll Experimental Results: Early Impact Behaviour Table 4-1: Comparison of hybrid B values calculated from Rule of Mixtures (Equation 4.25) and experimental results. Hybrid System B Calculated from the Rule of Mixtures B Calculated Experimentally Ratio (^rw/e of mixture^ $experimental} K N .33 .28 1.18 N K .33 .27 1.22 K K N N .20 .20 1.00 K N K N .20 .28 .71 N K N K .20 .20 1.00 N N K K .20 .22 .91 78 Experimental Results: Early Impact Behaviour 0.3 0.4 Time [msec] 0.6 Figure 4-1 :Calculating the rate of energy absorption over the first 35 microseconds of impact for a Kevlar-Nylon hybrid system struck at 27 Joules of energy (non-perforating event). The graph on the left displays the whole event, whereas the graph on the right displays the first 35 microseconds. 1000000 E 5 UJ 100000 1000000 100 1000 Vswke [m/s] E 5 LU •o 100000 100 1000 : [m/s] Figure 4-2: Log-Log normalised rate of energy absorption versus strike velocity graph for the 2-ply hybrid systems, along with the fitted regression line. 79 Experimental Results: Early Impact Behaviour 1000000 UJ •o 100000 1000000 100 [m/s] 1000 5 UJ 100000 100 1000 Vstrike [m/s] 1000000 5 UJ T3 100000 1000000 100 1000 "strike [m/s] O) UJ •o 100000 100 1000 ' strike [m/s] Figure 4-3: Log-Log normalised rate of energy absorption versus strike velocity graph for the four 4-ply hybrid systems, along with the fitted regression line. 80 Experimental Results: Early Impact Behaviour 0.9 0.8 0.7 0.6 j 0.5 I CO 0.4 4 CO 0.3 0.2 0.1 0 0.79 0.64 0.26 0.28 0.47 0.28 0.20 0.20 Kevlar Nylon Kevlar KN NK Kevlar NKNK KNKN NNKK KKNN 840d 1050d 840d 840d 1ply 1-ply 2ply 4ply Figure 4-4: Experimentally calculated values of B for the material systems relevant to this thesis. The single and multi-ply Kevlar results are taken from Cepus (2002). 1000000 O J TJ < T J UJ T J 100000 - L -100 1000 Vstrike [m/s] Figure 4-5:Comparing normalised dE/dt curves for 2-ply hybrids. 81 Experimental Results: Early Impact Behaviour 10000000 Figure 4-6:Comparing normalised dE/dt curves for 4-ply hybrids 82 Numerical Model and Results 5 Numerical Model and Results 5.1 Introduction Numerical modelling of the fabric panels is carried out here to further investigate the ballistic performance of single and multi-ply fabric armours. The codes used to conduct the numerical studies are T E X I M and L S - D Y N A , both of which are explicit finite element codes. T E X I M is a program developed in-house and as such is easy to set up and run with different material configurations. Alternatively, L S - D Y N A is a more sophisticated general-purpose commercial finite element code, for which data preparation and creation of new meshes is a more time consuming effort. For the purposes of this study most of the numerical simulations are carried out using T E X I M . However, benchmark problems are analysed using both codes to verify the capability of TEXIM. Subsequently simulations are carried out using T E X I M to investigate and explain observed experimental behaviour. 5.2 Numerical Codes 5.2.1 TEXIM This numerical code has been developed in-house by the U B C Composites Group Zhang et al. (1998) and Shahkarami (1999). It is written in Fortran and the source code is readily available. The model is a net-like pin-jointed model, similar in formulation to the model developed by Roylance and Wang (1980a). While the general formulation and pertinent features of the model are discussed here, more details on the workings of the model can be found in Shahkarami (1999) and Zhang et al. (1998). 83 Numerical Model and Results 5.2.1.1 Mass-String Model T E X I M models the fabric panel as a two-dimensional assembly of pin-jointed yarns resulting in a net-like structure (see Figure 5-1). Discrete nodal masses that sum up to the mass of the fabric panel are located at the crossings of the yarns. The nodal masses can translate in three directions (x, y, and z) and are connected to the four neighbouring nodes via cable elements that simulate the structural behaviour of the yarns. The cable elements have no compression or bending resistance and can only transmit tensile loading. 5.2.1.2 Constitutive Model The material constitutive model relating tension and strain in the cable elements is based on a simple visco-elastic formulation. A single Kelvin model is considered with a large spring representing stiffness, and a parallel dashpot representing the viscous behaviour of the fibres. Mathematically this can be written as: T = KS + TJS where T is the tension in the spring, K is the tensile stiffness (Young's modulus multiplied by cross sectional area), r] is the viscous parameter, s is the strain and t is the strain rate. 5.2.1.3 Projectile T E X I M models a blunt-nosed right circular cylinder (RCC) projectile impacting the centre of the fabric panel at 90 degrees (i.e. normal impact). It assigns the mass of the projectile to the nodes of fabric located within the contact area of the projectile, in what is known as patch loading (see Figure 5-2). The patch nodes and the associated masses are then given an initial velocity, which is equal to the projectile strike velocity. 84 Numerical Model and Results 5.2.1.4 Boundary Conditions Different boundary conditions can be imposed on the model including any mixture of fixed and free edges. In our case all runs employ fixed-fixed boundary conditions on both outer edges, with the inside boundaries being symmetric (see Figure 5-3). Symmetry implies that translation in the direction perpendicular to the line of symmetry is prevented. As an example, along the x-axis of symmetry, motion in the y direction is not permitted. 5.2.1.5 Multi-Ply Targets & Interlayer Contact T E X I M also models multi-ply targets. The panels are modelled as discrete plies, with a specified gap spacing placed between the layers. The adjacent panels are not considered until they are engaged. Subsequent layers (those other than the initially impacted fabric ply) are engaged when the transverse displacement of the preceding panel nodes have exceeded the gap value separating the plies. When this gap is exceeded the corresponding nodes in the adjacent fabric ply are then fused to those of the front ply. The principle of conservation of momentum is applied to the contacting nodal masses and projectile, assuming inelastic collision. Sliding between the nodes of the different layers is not permitted as they are fused to those that they are initially adjacent to. The size of the gap is an adjustable input parameter. It is therefore possible to study the effect of ply spacing on the behaviour of a multi-ply pack. In this thesis for multi-ply targets the gap is assumed to be .5mm, unless otherwise stated. 5.2.1.6 Failure The code calculates the strain in each element at each time step. This calculated strain is then compared with an input threshold value (failure strain). If the calculated strain 85 Numerical Model and Results exceeds the threshold value the element under evaluation is considered failed. The force in the failed element is then set to zero and the element is eliminated from the subsequent calculations. On a more global scale, when the force resisting the projectile is calculated to be zero the panel as a whole is considered failed. In a 1-ply case the code then outputs the residual velocity and in a multi-ply case the failed ply is then eliminated from the calculations. In most studies in this thesis an instantaneous failure criterion is used. According to an instantaneous failure criterion if the instantaneous value of strain exceeds the threshold value that element is considered to have failed. Due to numerical oscillations in the strain-time curves this can sometimes lead to premature failure. In order to overcome this problem, two options are available. One is averaging the strain of the element over the excitation period of the element, and then comparing this average strain to the threshold value. In this study, when strain oscillations were a concern the latter technique was used to prevent premature failure. The second method is to average the strain over an interval of time, in this case over the previous 5 time steps. This averaged strain value is then compared with the threshold value. Typically these numerical oscillations occur later in the event; or when the panels are impacted at relatively high velocities (typically greater than 350m/s). These velocities begin to approach the critical velocity of the yarns, defined as the smallest strike velocity at which the amplitude of the initial strain wave imparted into the fabric yarns is greater than the failure strain of the yarn. For Kevlar 29, which has properties very similar to the Kevlar 129 examined in this thesis, the transverse critical velocity for single fibres was found experimentally to be approximately 570m/s (Roylance, 1977). For 1, 2, 3, 4, 8 and 86 Numerical Model and Results 16-plies of Kevlar 840d, 1-ply Kevlar 1500d and 2 and 4-ply Kevlar-Nylon hybrids examined, strike velocities used in the numerical simulations were significantly below the critical velocity value and corresponded with experimental strike velocities. For the 24, 35 and 50-ply Kevlar 840d targets it was found that the maximum velocity without instantaneous failure was approximately 500 m/s for targets with no gap between the layers, 450 m/s for a gap of .5 mm and 400 m/s for a gap value of 1 mm. Greater ply interaction (i.e. systems with smaller gap values) seemed to have reduced numerical oscillations allowing greater strike velocities without causing instantaneous failure. Higher strike velocities for the 24, 35 and 50-ply systems were examined because they are the threat more likely to be faced by these systems. Strike velocities greater than these were beyond the scope of this study. 5.2.1.7 Time Discretisation The numerical stability of T E X I M is similar to that necessary for a wave propagation problem using an explicit time-integration. Courant, Friedrichs and Levy discussed the stability of this problem and stated that the strain wave cannot travel farther than the length of the smallest element during a time step, lest numerical stability become compromised. Written in equation form: A,<L C where, / is the length of the smallest element and c is the longitudinal stress wave speed. Therefore for a given material the grid size and time step size are inter-related. A stability parameter X, called the CFL number, is defined as follows: 87 Numerical Model and Results cAt where X must be less than 1 for numerical stability. For an undamped linear elastic case A,=l results in a stable solution. For our case the interaction between the transverse (tent formation) and longitudinal waves causes some numerical instability. Therefore, viscosity is used to dampen the numerical oscillations that occur. In this work 5% viscosity is used in all simulations, as this has been found to be the most effective in reducing the instability (Shahkarami, 1999). For Kevlar 129, this criterion was found to be sufficient for numerical stability. In modelling single plies of Nylon, an additional time step reduction was required in order to stabilise the solution. 5.2.2 LS-DYNA T E X I M is the primary numerical code used in this thesis but in order to verify its outputs a commercially available finite element code, L S - D Y N A was used. In this section some of the pertinent features of the L S - D Y N A model are briefly discussed. For more detail refer to Shahkarami etal. (2000, 2001, 2002a, 2002b). The L S - D Y N A model is a net-like structure made up of beam elements, with no bending or compressive strength, only tensile loading of the beam elements is allowed. Nodes are located at the crossing of the yarns and the mass of the fabric is lumped at these nodes: Yarns are assumed to be pin-jointed at the crossover points. The projectile is discretised with 8 noded solid elements. The addition of the solid elements tends to increase the run time for each simulation, but it allows consideration of different projectile shapes, 88 Numerical Model and Results projectile-fabric interaction and energy absorption via projectile deformation. The visco-elastic model used to simulate the yarn material is similar to that employed in T E X I M (Section 5.2.1.2). The symmetry and boundary conditions for the L S - D Y N A model are the same as those for T E X I M (Section 5.2.1.4). L S - D Y N A can also model multi-ply targets; this is achieved with meshes of discrete layers separated by a small spacing. Shell elements control contact and are permitted to slide over, but not penetrate each other. The shell elements are placed between the beam elements in the net-like structure. There are no structural properties assigned to these shell elements and they are introduced strictly for better simulation of contact. 5.3 Model Verification and Validation The first step in a numerical study is to carry out the analysis of a set of benchmark problems to gain confidence in the model's predictive capability. This was approached via two methods. The first method is to compare the outputs of two different numerical codes (verification) and the next is to compare the numerical results with the corresponding experimental results (validation). The material parameters used for Nylon and Kevlar 840d in the numerical models are displayed in Table 3-1, Table 3-2, Table 3-3 and Table 3-4. Kevlar 1500d was also modelled. The engineering material properties of Kevlar 1500d can be found in Table 3-1, and its panel properties can be found in Table 5-1. In our case a quarter section of the fabric panel of size 200 mm by 200 mm (as used in the experiments) was modelled. 89 Numerical Model and Results 5.3.1 Numerical Verification For each numerical code, L S - D Y N A and T E X I M , the known material properties, geometries and initial conditions are independently inputted and the results compared. This comparison ascertains that the results of numerical simulations carried out with different codes are as close as possible for a given set of identical material parameters and impact conditions. For Kevlar 840d the results compared are the velocity-time and energy-time outputs of the projectile. L S - D Y N A was run with both a deformable projectile as well as a lumped mass model (patch) for impact to allow direct comparison with T E X I M . As can be seen in Figure 5-4 and Figure 5-5 these results agree very well for the 1-ply Kevlar 840d material system tested. Since multi-ply Kevlar 840d targets are also of interest in this thesis, the effect of the different contact algorithms used in LS-D Y N A and T E X I M are also compared. A 2-ply Kevlar 840d case is compared in Figure 5-6, where it is seen that the results again agree very well. This comparison is for the case where the impact is modelled as a patch load. A n interlayer gap of .5 mm was assumed in both cases. The velocity-time graph for a 1-ply Nylon case was also compared with good agreement as shown in Figure 5-7. 5.3.2 Experimental Validation An essential step in validating the numerical results is to compare them with the corresponding experimental data. This indicates whether or not the numerical model is capable of capturing the observed physical behaviour of the fabric system. In this study we are primarily concerned with modelling the early impact event behaviour of the fabric panel. Therefore, in confirming the numerical results we compare them during the first 35 microseconds of the impact event (this is the theoretical treb0Und for a single Kevlar 129 90 Numerical Model and Results layer 200 mm by 200 mm in size). E L V S results from Cepus (2002) are used in this section to confirm the validity of the numerical model for Kevlar 840d. Once again it is the velocity-time and energy-time graphs that are compared for Kevlar 840d. Figure 5-8 and Figure 5-9 illustrate that for the limited time that we are interested in, T E X I M predicts the results of Kevlar 840d very well. Figure 5-10 shows a comparison between the T E X I M predictions and the experiments for a complete Kevlar 840d event, displaying the divergence of the two curves after the arrival of the reflected strain wave. The energy-time graphs are also compared for the case of Nylon in Figure 5-11. The results for the Nylon do not compare as favourably, with the slope of the numerical output being greater than the corresponding experimental result. 5.4 Analysis of the Entire Impact Event As discussed above, T E X I M does not model the whole event very well when evaluated in absolute terms. But, from a qualitative standpoint T E X I M can give insight into interesting phenomena and trends observed in the E L V S results. A noteworthy point is the numerical evaluation of hybrid packs. The model is capable of providing insight into the order of failure of separate plies in a hybrid pack. In Figure 5-12 a velocity-time curve is displayed showing that for a Kevlar-Nylon hybrid pack T E X I M detects the failure of the initial Kevlar layer. This is very similar to the measured velocity-time curve shown in Figure 5-13. Quantitatively, in terms of time of failure, the results do not match, but qualitatively the numerical model predicts the trends quite well. 91 Numerical Model and Results 5.5 Modelling of the Early Impact Event 5.5.1 Background When the velocity-time outputs of T E X I M are compared with the appropriate experimental results, it is seen that for Kevlar the results agree very well during the early stages of the impact event (see Figure 5-8 and Figure 5-9). The numerical Nylon results do not agree as well as seen in Figure 5-11, and therefore Nylon is not discussed here. A closer look at the Kevlar results indicates that the numerical and experimental results begin to diverge after the first reflection of the strain wave (see Figure 5-10). To refresh some of the ideas introduced in Chapter 4, we can attribute this divergence between the numerical and experimental results to a host of issues, foremost among them being the difficulty in modelling the material versus the structural response of the system. In the Section 3.2.4 of this thesis it was explained that during the experiments the yarns at the test jig boundary experienced a certain amount of pullout. This is evidence that the boundaries in the experimental jig are not perfectly clamped but rather somewhere between the perfectly free and perfectly fixed boundaries. The type of boundary condition has a significant effect on the magnitude and sign of the strain waves reflected back towards the impact point. Since the strain term is embedded in all three energy absorption mechanisms (see Section 4.1.2), it is clear that correctly modelling the magnitude of strain in the fabric panel is fundamental to correct modelling of the behaviour of the fabric panel. This makes the results from the numerical model, which uses perfect fixity at the boundary, wholly useful before the first reflection and only qualitatively so after. As a point of clarification, it should be noted that we are tracking the center of the panel (the projectile) and that the demarcation between before and after 92 Numerical Model and Results the first strain wave reflection is taken as the point in time when the reflected strain wave arrives at the center point of the panel. A summary of all simulation runs for Modelling of the Early Impact Event is listed in Table 5-2. 5.5.2 Calculation of B In keeping with the analysis performed on the experimental data, the predicted rate of energy absorption of the armour systems before the first arrival of the reflected strain wave at the impact point was examined. The procedure followed was to first plot the T E X I M predicted energy-time curve. The linear portion of this curve was located and a slope calculated (see Figure 5-14 for an example). Due to strain oscillations, the results of the energy-time curves must be examined closely to find a somewhat linear portion. This linear portion is located in a region where both the strain oscillations have died out and the strain has attained a relatively constant value (see Figure 5-14 for an illustration). The element used to monitor the strain oscillations is located just outside the patch area (see Figure 5-2). This is also the element used in all strain-time discussions in this dE chapter. This calculated slope is the rate of energy absorption (—) of the system. dt These slopes, which are calculated using a linear regression analysis are then normalised with respect to the areal density and plotted versus the strike velocity on a log-log plot. Figure 5-15 shows an example of such a plot for 1-ply Kevlar in a similar fashion to that done in the experimental chapter (Section 4.1.3). Recall Equation 4.24: 93 Numerical Model and Results 1 dE\ -— — =B(Vslnke? (5.1) A V dt )avg As previously defined, B is the ballistic efficiency parameter that takes into account different materials and weave architectures. From our log-log plot of the rate of energy absorption versus the strike velocity we fit a line to the data points, forcing the slope to be proportional to 8/3. In performing this analysis we calculate the B parameter for each material system. This allows us to conduct the same analysis as was carried out experimentally, with the numerical model giving us further insight into the behaviour of panels which have not necessarily been tested. The B values calculated using T E X I M are shown in Figure 5-16. Figure 5-16 illustrates that in keeping with the experimental findings our measure of ballistic efficiency B diminishes with the number of plies in a multi-ply pack, and with increased denier. It should be noted that we force the fitted line through the data points to have a slope of 8/3. This was found to be quite good for the 1, 2, 3, 4, 8 and 16 plies of Kevlar 840d, 1-ply Kevlar 1500d, 1-ply Nylon 1050d, 2-ply and 4-ply Kevlar-Nylon hybrids as shown in Figure 5-17, Figure 5-18, Figure 5-19, Figure 5-20 and Figure 5-21 (note the gap value 8/ for the multi-ply systems was .5 mm). Although the F / 3 relationship fits all simulations quite well there is poor agreement between the T E X I M predictions and experimental results in absolute terms in the 1-ply Nylon. This poor agreement seems to propagate through the 2 and 4-ply hybrids where Nylon is a component material (see Section 5.6.3 8/ for more details). Although the Vn relationship works reasonably well, if a power regression analysis is performed on the T E X I M results the exponent value is not exactly 94 Numerical Model and Results 8/3 for all cases. In particular for multi-ply systems, as the gap is increased (i.e from .5 mm to 1 mm) the exponent value deviates towards a greater value (see Table 5-3 for a list of exponent values for the various systems investigated). This reveals that indeed there are simplifications made in the analysis in Chapter 4. However, it also reveals that these assumptions are reasonable and the error introduced is quite small for the window of analysis investigated in this study. 5.6 Parametric Studies In Loss of Ballistic Efficiency B If we compare systems with different yarn deniers and different numbers of plies on a per unit mass basis, it has been found that the higher denier and greater ply count systems have lower efficiencies than the lower denier and single ply systems, respectively (Cunniff, 1992). Understanding why this occurs is fundamental to understanding fabric armour systems. If we confine our attention to the early impact behaviour, using T E X I M we can develop an understanding of why there is a decrease in ballistic efficiency. 5.6.1 Areal Density Effect 5.6.1.1 Background The decrease in ballistic efficiency, due to increased denier was measured not only in terms of the traditional ballistic measurement V50 (Cunniff, 1992), but also in terms of B, the early event material efficiency factor (Cepus, 2002). It must be noted here that increased denier typically goes hand in hand with increased panel areal density. T E X I M also detects this loss of efficiency due to increased denier when considering the early impact event B factor. This allows us to dissect the early event behaviour and determine why there is a decrease in efficiency. Table 5-4 lists all simulations performed to study the effect of areal density. 95 Numerical Model and Results T E X I M has three parameters that allow the user to change the weave structure of the fabric panel. It allows the alteration of the yarn linear density (denier), the yarn count and the areal density of the panel. These three inputs can each be changed independently, but in terms of physical relevance they are interrelated. This is illustrated by the following equation: 2 x Linear Density x Length x Yarn Count — Areal DensityQ. + crimp) The factor of 2 is due to the presence of yarns in orthogonal directions, and crimp is the percent difference in length of the yarn in the fabric panel versus the length of a non-woven yarn under a small amount of tension. In all numerical runs reported in this thesis, the crimp was assumed to be zero. 5.6.1.2 Results In terms of the B factor for a single ply of material the only parameter that changes the efficiency of the system is the areal density. As previously mentioned typically increased deniers go hand in hand with increased areal density. It is surmised that a significant portion of the loss of efficiency in greater denier panels is in fact due to their increased areal density. Figure 5-22 and Figure 5-23 show cases where the yarn-count, and linear density are held constant. In all cases the only factor that affects the rate of energy absorption is the change in areal density. In Figure 5-24 we can see that in the case where the areal density is constant the energy absorption behaviour is exactly the same. In order to understand why this is the case, we need to look at the strain in the system. Using the element shown in Figure 5-2 we monitor the strain in the yarn for each case. Close examination shows that the fabric panel with the lower areal density has a slightly 96 Numerical Model and Results higher level of strain (see Figure 5-25 and Figure 5-26), whereas when the areal densities are the same the magnitude of strain is found to be identical (see Figure 5-27). Since all three energy mechanisms are strain dependent (see Section 4.1.2) the system with the higher strain for a single ply will have absorbed more energy on a per areal density basis. Essentially the fabric panel material is used more effectively when greater strains are developed in it. 5.6.2 Multi-ply Effects 5.6.2.1 Background Another area of great interest in studying fabric armour packs is the multi-ply behaviour of armour systems. Understanding why there is a loss of efficiency when more layers are added would be an important contribution to the basic understanding of fabric armour systems as well as how to hybridise these armour systems effectively. Table 5-5 provides a list of simulations run to investigate multi-ply systems. The numerically calculated values of B show a clear loss of efficiency (see Figure 5-16 and Figure 5-28) as additional plies are added to a fabric armour system. This is also observed directly in the energy-time graphs as shown in Figure 5-29. In conducting this analysis it is important to make the comparisons equivalent. If there were no multi-ply system effects (losses of efficiency) then it would be correct to make the statement that 2 plies of material absorb twice as much energy at twice the rate of a single ply of material. The experiments as well as the numerical runs show this is not the case as illustrated by comparison of specific E50 and more importantly for this thesis, the B values (Cepus, 2002). As more plies are added the multi-ply systems do absorb more energy, and at a greater rate, but the increase in the rate of energy absorption is not a whole number 97 Numerical Model and Results multiple, i.e. a 2-ply system is not twice as effective and a 4-ply system is not four times as effective as a single ply. Therefore, when comparing numerical runs with different numbers of plies we must normalise the relevant outputs with their areal density to determine the loss of efficiency due to system effects. The numerical testing of multi-ply systems was carried out on Kevlar 840d. This is because Kevlar 840d is the material that has been tested the most, both numerically and experimentally, and therefore it is the material for which the greatest understanding exists. For this set of numerical testing a strike energy of 18 Joules (strike velocity of 110 m/s) was used. This was nominally chosen so as to be able to compare non-perforating events (at this strike velocity the 1-ply Kevlar panel does not fail). 5.6.2.2 Effect of Spacing The logical parameter to alter when conducting numerical experiments on a layered system is the spacing or gap between the layers. The magnitude of this gap has a major effect on the behaviour of the multi-plied system. If the gap is assigned a zero value, all layers are in contact at the start of the event and the system behaves as a single layer with an increased areal density. When the size of the gap is increased, the energy absorption for a given time decreases as shown in the energy-time graph in Figure 5-30, and in terms of B in Figure 5-31. This is due to the involvement of less material in the absorption of energy early on in the event. The larger the gap the further the projectile has to penetrate into the system to encounter the increased resistance of the additional plies. Therefore the rate of energy absorption for larger spaced systems is lower, and conversely the B for that system is also lower. Additional plies also reduce the magnitude of strain in the preceding plies, as shown in Figure 5-32. This is due to the additional plies being 98 Numerical Model and Results strained, once they are engaged. The strain in these plies reduces the strain in the initial ply, which means that the material in the initial ply is not used as effectively as in the 1 -ply case. As discussed in the examination of the areal density effect, the smaller strains means less energy is absorbed by all three energy absorption mechanisms. This small difference makes a significant difference when considered for the whole system. The existence of a gap also produces a strain gradient into the thickness of the pack as shown in the strain-time histories of an 8-ply system in Figure 5-33. This strain gradient means that the rear panels early in the event experience smaller strains and therefore absorb less energy. This is due to the reduced impact energy of the projectile by the time it reaches the distal plies, as the front impacted plies have already absorbed some of the initial projectile energy. Therefore the system with the smallest gap will absorb energy at the greatest rate and therefore have the greatest B value. The addition of gap also smoothes the resultant strain wave reflections. This can be seen in Figure 5-30 where the 4-ply case with no gap has a very distinct elbow in the energy-time graph. As the size of the gap is increased, this kink is smoothed out and its exact location along the curve is more difficult to identify. 5.6.2.3 Effect of Contact As the gap is increased the multi-ply systems also begin to behave similarly to individual plies acting alone instead of acting together as a system. It is difficult to isolate a specific gap size at which there is a change from a system to an individual layer behaviour as it is more of a transition than a clear cut-off. This was studied by running a multi-plied system with a nominal gap value of .5 mm. The time at which each layer becomes engaged is noted. The velocity of the projectile at this time is also noted. This is carried 99 Numerical Model and Results out for each layer. Each layer is then run individually with the strike velocity being the velocity determined in the multi-plied system at the instant of engagement. The absorbed energy of each layer is then added up at time 34 microseconds (time begins when the initial layer is engaged). 34 microseconds relates to the moment just before the strain wave reflection returns to the center point of the initially impacted layer. When these absorbed energy values are added together and compared to corresponding values obtained for the case when the plies act as a system the effect of layer contact becomes apparent. Table 5-6 summarises the details of the results for a 4-ply system with gap values of .5 and 1 mm. Though the effect is not large, contact between the layers does decrease the ability of the armour systems to absorb energy early in the event. 5.6.3 Validation of Numerical B Values The insights and analysis discussed in the previous sections are based on the assumption that the numerical model is accurately modelling the early impact behaviour of the fabric panels. Comparing the experimental and numerically calculated B values, good agreement is obtained as shown in Figure 5-34. The numerical model does a very good job with the single and multi-ply Kevlar material systems. The model follows the trends of the Nylon and the hybrid panels correctly, but the absolute numbers are slightly greater for the numerical model. It is felt that this discrepancy is due to two factors: (1) the Nylon tested experimentally was coated on one side making it not a true fabric due to the coating's addition of mass to the areal density and the effect it has on the displacement of the Nylon yarns which make up the panel, and (2) the elastic modulus of Nylon is known to be strain-rate dependent (Meredith, 1954), much more so than Kevlar and perhaps the 100 Numerical Model and Results visco-elastic model employed in T E X I M needs to be altered to model the Nylon material more accurately. 5.7 Practical Design Insight T E X I M models the behaviour of fabric panels within the first strain wave reflection with an acceptable accuracy. We can extend some of the insights found to the construction of an actual vest. Computations with increasingly thicker packs were conducted and B values were calculated. These simulations were run with no gap, .5mm gap and 1mm gap (see Figure 5-35). The effect of the presence of a gap begins to become pronounced at approximately 10 plies. For the velocities examined, when the system has a gap between the layers, there is a ply count at which additional plies have no effect on the rate of energy absorption (the normalised rate of energy absorption continues to decrease due to the increased areal density). This is due to the inability of the projectile to displace sufficiently to initiate contact with the distal layers of the system. This means that even though the distal plies contribute to the areal density of the pack they have no effect on the absorbed energy (at least in the early event time interval we are commenting on). Therefore compared to the system with no gap, where the areal density is the same but all the layers are engaged in energy absorption, the ballistic efficiency B is lower. The numerical results therefore suggest that it would be beneficial to have the least amount of gap possible to achieve the best rate of energy-absorption with the hope that the pack can translate that into greater V50 performance. From the study of the effect of areal density the numerical model also suggests that it would be beneficial to design with 101 Numerical Model and Results lower areal density panels to allow superior absorption of projectile impact energy early in the event, thereby making best use of the armour material available to defeat the projectile. 5.8 Chapter Summary The numerical code T E X I M is used to model the early behaviour of single and multi-ply Kevlar systems. The results of the T E X I M simulations compare favourably with the applicable experimental results. T E X I M detects the loss of ballistic efficiency in the early behaviour of a greater denier single ply system as well multi-ply systems. Parametric studies were then carried out to find which factors affect the behaviour of fabric panels early in the impact event. Reducing the areal density reduces the efficiency loss in single panels. Reducing the gap between layers reduces the efficiency loss in multi-ply systems. From these results practical design recommendations for fabric armour systems are made. 102 Numerical Model and Results Table 5-1: Properties of Kevlar 1500d panel tested. Plain l x l 9 294.7 Taken to be 0 for the work in this thesis. 1670(1500) Table 5-2:List of simulations of the early impact event Material Ply Gap Denier Areal Strike Normalised Rate of Svstem Count Imml Densitv Velocitv Enersv Absorption Kevlar ! N / A 840 192.2 95 99196 Kevlar 1 N / A 840 192.2 110 156375 Kevlar 1 N / A 840 192.2 127 244002 Kevlar 1 N / A 840 192.2 154 445347 Kevlar 1 N / A 840 192.2 184 776882 Kevlar 1 N / A 840 192.2 247 1949057 Kevlar 2 0 840 384.4 109 144472 Kevlar 2 0 840 384.4 143 330920 Kevlar 2 0 840 384.4 200 919619 Kevlar 2 0 840 384.4 234 1479660 Kevlar 2 0.5 840 384.4 109 137367 Kevlar 2 0.5 840 384.4 143 318286 Kevlar 2 0.5 840 384.4 200 895706 Kevlar 2 0.5 840 384.4 234 1447291 Kevlar 2 1 840 384.4 109 130394 Kevlar 2 1 840 384.4 143 306131 Kevlar 2 1 840 384.4 200 871622 Kevlar 2 1 840 384.4 234 1415954 Kevlar 3 0 840 576.6 113 153152 Kevlar 3 0 840 576.6 144 316950 Kevlar 3 0 840 576.6 225 1190712 Kevlar 3 0 840 576.6 272 2064329 Kevlar 3 0.5 840 576.6 113 140014 Kevlar 3 0.5 840 576.6 144 296278 Kevlar 3 0.5 840 576.6 225 1149848 Kevlar 3 0.5 840 576.6 272 2015427 Kevlar 3 1 840 576.6 113 126096 Kevlar 3 1 840 576.6 144 273845 Kevlar 3 1 840 576.6 225 1103384 Kevlar 3 1 840 576.6 272 1957733 Weave Count (yarns/cm) Areal Density (g/m2) Yarn Crimp (%) Yarn Density, dtex (denier) 103 Numerical Model and Results Material Svstem Ply Count Gap Imml Denier Areal Densitv Strike Velocitv Normalised Rate of Enerev Absorption Kevlar 4 0 840 768.8 113 145614 Kevlar 4 0 840 768.8 144 297669 Kevlar 4 0 840 768.8 205 830591 Kevlar 4 0 840 768.8 272 1844620 Kevlar 4 0.5 840 768.8 113 128319 Kevlar 4 0.5 840 768.8 144 271938 Kevlar 4 0.5 840 768.8 205 791062 • Kevlar 4 0.5 840 768.8 230 1110670 Kevlar 4 0.5 840 768.8 272 1803715 Kevlar 4 1 840 768.8 113 103754 Kevlar 4 1 840 768.8 144 239930 Kevlar 4 1 840 768.8 205 738211 Kevlar 4 1 840 768.8 272 1739298 Kevlar 8 0 840 1537.6 121 145235 Kevlar 8 0 840 1537.6 168 356065 Kevlar 8 0 840 1537.6 247 991805 Kevlar 8 0 840 1537.6 318 1947898 Kevlar 8 0.5 840 1537.6 121 109644 Kevlar 8 0.5 840 1537.6 168 323251 Kevlar 8 0.5 840 1537.6 247 983076 Kevlar 8 0.5 840 1537.6 318 1901192 Kevlar 8 1 840 1537.6 121 74334 Kevlar 8 1 840 1537.6 168 240332 Kevlar 8 1 840 1537.6 247 875702 Kevlar 8 1 840 1537.6 318 1855342 Kevlar 12 0 840 2306.4 150 209820 Kevlar 12 0 840 2306.4 250 830711 Kevlar 12 0 840 2306.4 350 1967759 Kevlar 12 0 840 2306.4 450 4018827 Kevlar 12 0.5 840 2306.4 150 156560 Kevlar 12 0.5 840 2306.4 250 794510 Kevlar 12 0.5 840 2306.4 350 1959877 Kevlar 12 0.5 840 2306.4 450 3852187 Kevlar 12 1 840 2306.4 150 96765 Kevlar 12 1 840 2306.4 250 584960 Kevlar 12 1 840 2306.4 350 1691383 Kevlar 12 1 840 2306.4 400 2461210 Kevlar 16 0 840 3075.2 118 95350 Kevlar 16 0 840 3075.2 220 514025 Kevlar 16 0 840 3075.2 360 1911084 Kevlar 16 0 840 3075.2 398 2504237 104 Numerical Model and Results Material Ply Gap Denier Areal Strike Normalised Rate of Svstem Count Imml Densitv Velocitv Enerav AbsorDtion Kevlar 16 0 840 3075.2 455 3559634 Kevlar 16 0.5 840 3075.2 118 54023 Kevlar 16 0.5 840 3075.2 220 418840 Kevlar 16 0.5 840 3075.2 360 1641462 Kevlar 16 0.5 840 3075.2 398 2204509 Kevlar 16 0.5 840 3075.2 455 3167807 Kevlar 16 1 840 3075.2 118 30308 Kevlar 16 1 840 3075.2 220 284902 Kevlar 16 1 840 3075.2 360 1376362 Kevlar 16 1 840 3075.2 398 1821189 Kevlar 24 0 840 4612.8 150 151400 Kevlar 24 0 840 4612.8 250 592926 Kevlar 24 0 840 4612.8 350 1480982 Kevlar 24 0 840 4612.8 450 2881558 Kevlar 24 0 840 4612.8 550 4783459 Kevlar 24 0.5 840 4612.8 150 78280 Kevlar 24 0.5 840 4612.8 250 411421 Kevlar 24 0.5 840 4612.8 300 692615 Kevlar 24 0.5 840 4612.8 350 1047215 Kevlar 24 0.5 840 4612.8 450 2087379 Kevlar 24 1 840 4612.8 150 48494 Kevlar 24 1 840 4612.8 250 293510 Kevlar 24 1 840 4612.8 300 531478 Kevlar 24 1 840 4612.8 350 839202 Kevlar 24 1 840 4612.8 400 1230605 Kevlar 35 0 840 6727 150 127026 Kevlar 35 0 840 6727 250 495525 Kevlar 35 0 840 6727 350 1190448 Kevlar 35 0 840 6727 450 2369887 Kevlar 35 0 840 6727 550 3883335 Kevlar 35 0.5 840 6727 150 53491 Kevlar 35 0.5 840 6727 250 280770 Kevlar 35 0.5 840 6727 350 718090 Kevlar 35 0.5 840 6727 450 1431346 Kevlar 35 1 840 6727 150 37250 Kevlar 35 1 840 6727 250 200558 Kevlar 35 1 840 6727 ' 350 579903 Kevlar 35 1 840 6727 400 843843 Kevlar 50 0 840 9610 150 111454 Kevlar 50 0 840 9610 250 407939 Kevlar 50 0 840 9610 300 642496 Kevlar 50 0 840 9610 350 963066 105 Numerical Model and Results Material Ply Gap Denier Areal Strike Normalised Rate of Svstem Count Imml Densitv Velocitv Energv AbsorDtion Kevlar 50 0 840 9610 450 1841579 Kevlar 50 0 840 9610 550 3468061 Kevlar 50 0.5 840 9610 150 37574 Kevlar 50 0.5 840 9610 250 197482 Kevlar 50 0.5 840 9610 300 328062 Kevlar 50 0.5 840 9610 350 498203 Kevlar 50 0.5 840 9610 450 1011668 Kevlar 50 1 840 9610 150 23277 Kevlar 50 1 840 9610 250 140885 Kevlar 50 1 840 9610 300 255109 Kevlar 50 1 840 9610 350 398763 Kevlar 50 1 840 9610 400 595713 Kevlar 1 N / A 1500 294.7 65 29660 Kevlar 1 N / A 1500 294.7 120 193589 Kevlar 1 N / A 1500 294.7 139 304608 Kevlar 1 N / A 1500 294.7 180 675769 Nylon 1 N / A 1050 520.7 63 25227 Nylon 1 N / A 1050 520.7 105 110091 Nylon 1 N / A 1050 520.7 113 136747 Nylon 1 N / A 1050 520.7 153 306377 Nylon 1 N / A 1050 520.7 219 799869 Nylon 2 0.5 1050/840 712.9 128 144869 Nylon 2 0.5 1050/840 712.9 133 162601 Nylon 2 0.5 1050/840 712.9 163 301364 Nylon 2 0.5 1050/840 712.9 184 432901 Kevlar 2 0.5 840/1050 712.9 133 158629 Kevlar 2 0.5 840/1050 712.9 155 254984 Kevlar 2 0.5 840/1050 712.9 186 445414 Kevlar 2 0.5 840/1050 712.9 254 1148336 K K N N 4 0.5 840/1050 1425.8 190 373151 K K N N 4 0.5 840/1050 1425.8 199 428529 K K N N 4 0.5 840/1050 1425.8 233 684349 K K N N 4 0.5 840/1050 1425.8 242 764422 K N K N 4 0.5 840/1050 1425.8 190 389240 K N K N 4 0.5 840/1050 1425.8 211 531099 K N K N 4 0.5 840/1050 1425.8 216 569145 K N K N 4 0.5 840/1050 1425.8 235 729907 N K N K 4 0.5 840/1050 1425.8 195 416648 N K N K 4 0.5 840/1050 1425.8 210 517273 N K N K 4 0.5 840/1050 1425.8 227 646350 N K N K 4 0.5 840/1050 1425.8 251 859349 106 Numerical Model and Results Material Ply Gap Denier Areal Strike Normalised Rate of Svstem Count Imml Density Velocity Enersv Absorption N N K K 4 0.5 840/1050 1425.8 159 233837 N N K K 4 0.5 840/1050 1425.8 191 396126 N N K K 4 0.5 840/1050 1425.8 210 516999 N N K K 4 0.5 840/1050 1425.8 242 767237 Table 5-3:Exponent values from power regression analysis performed on numerical normalised dE/dt versus Vslrikl, data for different systems. Material Denier Plies Exponent Value No Gap .5 mm gap 1 mm gap Kevlar 840 1 3.12 - -Kevlar 840 2- 3.05 3.08 3.12 Kevlar 840 3 2.96 3.04 3.12 Kevlar 840 4 2.89 3.01 3.08 Kevlar 840 8 2.68 2.95 3.34 Kevlar 840 12 2.68 2.91 3.2 Kevlar 840 16 2.68 3 3.29 Kevlar 840 24 2.67 3 3.2 Kevlar 840 35 2.64 3 3.2 Kevlar 840 50 2.62 3 3.2 Kevlar 1500 1 3.07 - -Nylon 1050 1 2.77 Kevlar-Nylon 840/1050 2 - 3.06 -Nylon-Kevlar 1050/840 2 - 3.02 -N N K K 1050/840 4 - 2.83 -N K N K 1050/840 4 - 2.87 -K N K N 840/1050 4 - 2.96 -K K N N 840/1050 4 - 2.97 -Table 5-4:Simulations carried out on 1-ply Kevlar panels impacted at 18 Joules to study the effect of areal density Areal Yarn Denier Density Count |g/mA21 [/cm] 200 213.33 48 200 106.67 24 200 53.33 12 400 213.33 24 800 426.67 24 800 213.33 12 800 106.67 6 1600 426.67 12 1600 213.33 6 1600 106.67 3 107 Numerical Model and Results Table 5-5: Simulations on Kevlar 840d impacted at 18 Joules to study effect of multi-plies. Plies Gap [mm| 1 N / A 2 0 2 0.5 2 1 4 0 4 0.5 4 1 8 0 8 0.5 8 1 Table 5-6:Effect of layer interaction on the energy absorption characteristics of multi-ply fabric armours. These are values predicted for an 18 Joule strike energy on a 4-ply Kevlar 840d system at 34 microseconds after impact, i.e. just before the arrival of the reflected strain wave in the initially impacted layer. Gap [mm] Energy Mechanism 4-ply System [Joules] 4-ply individual Layers [Joules] Percent Difference Total Energy 2.227 2.364 6.15% Strain Energy 1.081 1.131 4.63% .D K.E. Out of Plane .8953 .9590 7.10% K . E . In-Plane .2508 .2738 9.17% Total Energy 1.649 1.694 2.8% 1 Strain Energy .762 .776 1.8% 1 K.E Out of Plane .683 .707 3.5% K . E . In-Plane .204 .212 3.9% 108 Numerical Model and Results Level z-1 Level / Level z'+l Level j+l Figure 5-1: A net-like mass string system (Zhang et al., 1998) Level j=0 - 4- - I - 1 - 4 - 1 I I I I impact — -t —1 — area I I Element where strain is monitored Figure 5-2:Projectile patch configuration for T E X I M with mass of projectile lumped on the indicated nodes. The element used for strain-time evaluations is also indicated.(Shahkarami, 1999) 109 Numerical Model and Results Figure 5-3:Symmetry of the fabric panel (Shahkarami, 1999) 120 -80 -J— Time [milliseconds] Figure 5-4:Projectile velocity-time graph predicted by L S - D Y N A (patch and deformable projectile) and T E X I M for 1 -ply Kevlar fabric, impacted with a strike energy of 18 Joules; non-perforating event. Note all three lines are on top of each other with the L S - D Y N A deformable projectile displaying the small oscillations. 110 Numerical Model and Results 0.05 Time [milliseconds] Figure 5-5: Energy-time graph predicted by L S - D Y N A and T E X I M for 1-ply Kevlar panel, impacted with a strike energy of 18 Joules Energy; non-perforating event. Time [milliseconds] Figure 5-6:Projectile velocity-time graph predicted by L S - D Y N A and T E X I M for a 2-ply Kevlar panel impacted with a strike energy of 18 Joules; non-perforating event. Note the lines are on ton of each other. K 111 Numerical Model and Results 120 -40 Time [milliseconds] Figure 5-7:Projectile velocity-time graph predicted by L S - D Y N A and T E X I M for 1-ply Nylon, impacted with a strike energy of 15 Joules, non-perforating event. Note the lines are on top of each other. 109 T— -T— - | —r- —r- —r- — r — — , 108.5 104 4~ -4— —4— —I— -4— 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 Time [milliseconds] Figure 5-8:Projectile velocity-time graph predicted by T E X I M as well as experimental results for 1-ply Kevlar, impacted with a strike energy of 17 Joule; non-perforating event, experimental data taken from Cepus (2002). 112 Numerical Model and Results Figure 5-9:Energy-time graph predicted by T E X I M as well as experimental results for 1-ply Kevlar, impacted with a strike energy of 17 Joules; non-perforating event, experimental data taken from Cepus (2002). 120 100 80 60 I 40 ~ 20 o o CD 0 -20 -40 -60 -80 - Ap iroximate locatio n of strain wave i eflection X \ \ V ELVS Data-Kevls i r01p05 Back ELVS Data -Kevlar01p05 <|> 0 4 0 B 0 8 1 T E X I M / Time [milliseconds] Figure 5-10:Divergence of behaviour between experimental and predicted T E X I M results after the first strain wave reflection for 1-ply Kevlar, impacted with a strike energy of 17 Joules; non-perforating event, experimental data taken from Cepus (2002). 113 Numerical Model and Results Figure 5-11:Energy-time graph predicted by T E X I M as well as experimental results for 1-ply Nylon, impacted with a strike energy of 15 Joules; non-perforating event. Note the difference in slope between the linear regression lines of the experimental and T E X I M results. Figure 5-12: Projectile velocity-time graph predicted by T E X I M for a 2-ply Kevlar-Nylon hybrid impacted with a strike energy of 27 Joules; non-perforating event. The front Kevlar panel is penetrated. 114 Numerical Model and Results o o <a > 140 120 100 80 60 40 20 0 -20 -40 L j \ Failure o 'initial Kev a r layer Kevlar-N' l<> ii 03 3 0. 2 0. 4 0. 1 2 1 mm \ Time [milliseconds] Figure 5-13: Experimentally measured projectile velocity-time graph for a 2-ply Kevlar-Nylon hybrid impacted with a strike energy of 27 Joules; non-perforating event. The front Kevlar panel i penetrated. 2.5 Reduced strain oscillations lead to a linear flortiojiofthe energy-time graph 0.01 0.009 t 0.008 0.007 0.006 c 0.005 2 CO 0.004 0.003 0.002 0.001 0.01 0.02 0.03 Time [milliseconds] 0.04 0.05 Figure 5-14:TEXIM predicted energy and strain-time graph (strain curve computed for element shown in Figure 5-2). The portion of the energy-time curve used to calculate the rate of energy absorption for calculation of B is indicated. This is for 1-ply Kevlar, impacted with a strike energy of 18 Joules; non-perforating event. 115 Numerical Model and Results 10000000 <~ 1000000 J3 TJ 100000 10000 10 Regression line fit through the data points, with slope of V*8/3 • Data Points from slope of Energy Time graphs, usmg TEXIM 100 Vstrike [m/S] 1000 Figure 5-15:Predicted Normalised dE/dt versus Vslnke for 1-ply Kevlar. 0.9 0.8 0.7 0.6 n«o 0.5 CO 5 •E 0.4 m 0.3 0.2 0.1 Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Nylon KN NK NKNK NNKK KKNN KNKN 840(1 840d 840d 840d 840d 840d 840d 840d 840d 840d 1500d 1050d 1ply 2ply 3ply 4ply 8ply 12ply 16ply 24ply 35ply 50ply 1ply 1ply Figure 5-16:B values calculated from T E X I M for various single material and hybrid multi-layer fabrics. 116 Numerical Model and Results 10000000 •C~ 1000000 E < I 5. 100000 10000 1-ply Kevlar 840d 1 1 • Numerical » Experimental 10000000 10 100 VsMke [m/S] 1000 10000000 , f 1000000 E Ol < 3. 100000 10000 3-ply Ke^ lar840d u • Numerical « Experimental 10 100 V „ r l k . [m/s] 1000000 E O) JC ST 1 a H. 100000 10000 • Numerical • Experimental j y / a i 1 2-ply Key la r840d 100 V.trtke [m/s] 1000 10000000 1000000 E Ol JC J? 5 ill 5. 100000 / • Numerical * Experimental / / « 4-ply Key la r840d 100 V,„ike [m/s] 1000 Figure 5-17:Comparison of numerical and experimental normalised dE/dt versus curves for l , 2, 3, 4-ply Kevlar 840d. Note the good agreement of the numerical results with the experimental results, as well as the V*3 relationship, experimental data taken from Cepus (2002). 117 Numerical Model and Results Figure 5-18:Comparison of numerical and experimental normalised dE/dt versus VsMke curves for 8 and 16-ply Kevlar 840d. Note the good agreement of the Vs'3 relationship for the 8 and 16-ply cases, and the good agreement of the numerical and experimental results in absolute terms for the 8-ply case, experimental data taken from Cepus (2002). 10000000 1000000 £ 1-ply Ke\ larl500d 4 • Numerical • Experimental 10000000 1000000 E H. 100000 1-ply Nyl bn 1050d • / 1 • Numerical Experimental 100 VsWi,s [m/s] 100 Strike [m/S] Figure 5-19:Comparison of numerical and experimental normalised dE/dt versus Vslnke curves for 1-ply Kevlar 1500d and 1-ply Nylon 1050d. Note the good agreement of the numerical and experimental results for Kevlar 1500d, but not for the Nylon. However the V8'3 relationship agrees well for the numerical results of both Kevlar 1500d and Nylon. Kevlar experimental data taken from Cepus (2002). 118 Numerical Model and Results 10000000 , •jj, 1000000 -5 T3 < 2 100000 111 T3 10000 Tr H — 1 Kevlar-Nylon • Numerical • Experimental 10000000 10 100 1000 V s t r i k e [m/s] * 1000000 O) JC T3 < 5 100000 LU T3 10000 10 Nylon-Kevlar • / > • Numerical • Experimental 100 V s t r i k e [m/s] 1000 Figure 5-20:Comparison of numerical and experimental normalised dE/dt versus Vslrike curves for 2-ply Kevlar-Nylon hybrids. Note the good agreement of the V83 relationship, but the poor agreement between the predictions and measurements in absolute terms. 119 Numerical Model and Results 1000000 E L J J T J 100000 1000000 -a a T J 100000 • Numerical • Experimental KNKN 100 1000 100 1000 Vstrike [m/s] Vstrike [m/s] 1000000 £ T J U J T J 100000 • Numerical •f Experimental NKNK 1000000 J5 T J U J T J 100000 I Numerical ^Experimental NNKK 100 1000 100 1000 Vstrike [m/S] Vstrike [m/S] Figure 5-21: Comparison of numerical and experimental dE/dt versus Vslrike curves for 4-ply Kevlar-Nylon hybrids. Note the good agreement of the V*3 relationship, but the poor agreement between the predictions and measurements in absolute terms. 120 Numerical Model and Results 0 0.005 0.01 0.015 0.02 0.025 0.03 0 035 Time [milliseconds] Figure 5-22 Normalised energy-time graphs for 1-ply Kevlar fabric with a constant yarn count of 12 yarns per cm, for an 18 Joule Strike Energy event. The lower areal density panel absorbs normalised energy more quickly. 0 4-0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 Time [milliseconds] Figure 5-23: Normalised energy-time graphs for 1-ply Kevlar fabric with a constant linear yarn density of 800d, for an 18 Joule Strike Energy event. The lower areal density panel absorbs normalised energy more quickly. Numerical Model and Results 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 Time [milliseconds] Figure 5-24: Normalised energy-time graphs for 1-ply Kevlar fabric with a constant areal density of 213g/m2, for an 18 Joule Strike Energy event. A l l three lines of outputs are directly on top of each other, indicating energy absorption at exactly the same rate. 0.004 0.0035 0.003 0.0025 c S 0.002 CO 0.0015 0.001 0.0005 0 /V 200d Ad=.53 g/mA2 I I 800d Ad=2 13g/mA2 1600d Ad=42 7 g/mA2 1 0.005 0.01 0.015 0.02 0.025 Time [milliseconds] 0.03 0.035 Figure 5-25: Strain-time graphs for 1-ply Kevlar fabric with a constant yarn count of 12 yarns per cm, for an 18 Joule Strike Energy event. The lower areal density panel has greater strains. 122 Numerical Model and Results 0.005 0.01 0.015 0.02 0.025 Time [milliseconds] 0.03 0.035 Figure 5-26: Strain-time graphs for 1-ply Kevlar fabric with a constant linear yarn density of 800d, for 18 Joule Strike Energy event. The lower areal density panel has greater strains. an 0.004 0.0035 0.003 0.0025 c 2 0.002 to 0.0015 0.001 0.0005 0 A T 200d Ad=213 g/m2 400d Ad=213g/m2 800d Ad=213g/m2 0.005 0.01 0.015 0.02 0.025 Time [milliseconds] 0.03 0.035 Figure 5-27: Strain-time graphs for 1-ply Kevlar fabric with a constant areal density of 213 g/m2, for an 18 Joule Strike Energy event. A l l three outputs are directly on top of each other, indicating exactly the same strains. 123 Numerical Model and Results 100000000 lfply Kevlar840d Figure 5-28:Predicted normalised dE/dt versus Vsmke for 840d Kevlar fabric displaying the loss in efficiency as the ply count is increased. Figure 5-29: Normalised energy-time graph of multi-ply Kevlar for an 18 Joule Strike Energy event. The smaller the ply count, the greater the rate of normalised energy absorption 124 Numerical Model and Results 0 0.01 0.02 0.03 0.04 0.05 0.06 Time [milliseconds] Figure 5-30: Normalised energy-time graph for 4-ply Kevlar fabric with a varying gap values, for an 18 Joule Strike Energy event. Note the smaller the gap, the greater the rate of normalised energy absorption. 100000000 10000000 3 1000000 UJ T3 100000 50-ply Kevlar 840d No gap 10000 100 .5 mm gap 1 mm gap 1000 Vstrik. [m/s] Fi gure 5-31 :Predicted normalised dE/dt versus Vstfike f ° r 50-ply Kevlar 840d displaying the loss in efficiency as the gap is increased. Numerical Model and Results 0.004 0.0035 0.003 0.0025 c 2 0.002 GO 0.0015 0.001 0.0005 0 A 840d 1-ply 71 \ 1 / 840d 2-ply gap Smm —I 840d 4-ply gap .5mm i f 1 840d 8-ply | ;ap .5mm I I i / 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 Time [milliseconds] Figure 5-32:Strain-Time curves for the element on the 1s t ply (impact face) in 1,2,4,8-ply Kevlar systems for an 18 Joule Strike Energy event. Note reduced strains in the front ply due to engagement of subsequent plies. 1 s t ply 0.01 0.02 0.03 Time [milliseconds] 0.04 1 T>fldnKrP\ I I 13rd ply K 5 t h ply k 1 j J 6 t h ply 1 1 1 i-,th • 1 I I I j j fH' 0.05 Figure 5-33:Strain-time curves of 1,2,3,4,5,6,7,8th plies in an 8-ply Kevlar system, for an 18 Joule Strike Energy event. Note lower strains in plies towards the rear of the pack. 126 Numerical Model and Results Figure 5-34:Cross-plot of experimental and numerically obtained B values for various materials and ply counts. 0.9 r -0.8 0 10 20 30 40 50 60 Number of Plies Figure 5-35:Plot of B versus number of plies. Note the effect of gap on the calculated B for increased ply count. 127 Conclusions & Future Work 6 Conclusions & Future Work This chapter presents a summary and conclusions of the work reported in this thesis. A framework is also suggested for future work to further understand the phenomenon investigated. 6.1 Summary 1. A review of the experimental work previously completed on multi-ply and hybrid targets as well as an overview of previously developed multi-ply numerical models has been presented. 2. Instrumented ballistic impact tests have been carried out to examine the effect of hybridising Kevlar and Nylon fabric panels. Relevant data for each system has been compared and post-mortem damage to the panels has been examined. 3. The behaviour of the hybrid panels was also examined during the early stages of the impact event, before the reflected stress waves from the boundaries arrived back at the impact point. The analysis technique proposed by Cepus (2002) was employed with early rates of energy absorption of the armour systems being used to compare the performance of various armour systems. 4. A numerical study was undertaken to investigate the early impact behaviour of single- and multi-ply Kevlar fabric armour systems. Two and four ply Kevlar Nylon hybrid fabric armour system were also explored. Parametric studies were 128 Conclusions & Future Work then performed to develop an understanding of the behaviour of the various systems during the early stages of the impact event. 6.2 Conclusions 6.2.1 Experimental These conclusions apply to the testing conditions employed in this study e.g. projectile shape and size, size of test specimen, etc. They also pertain to impact velocities below the instantaneous failure velocities of either Nylon or Kevlar materials. It should also be noted that these results are specific to small ply counts, which may not be realistic for protective vests. 1. A single ply of ballistic Nylon (1050d, 15oz coated 46x46 yarns per inch, areal density 520.7 kg/m ) has a greater critical velocity, Vc (for the testing conditions investigated) than a single ply Kevlar (840d, 28x28 yarns per inch, areal density 192.2 kg/m ), but allows significantly more back-face deformation. 2. Nylon, when impacted by the RCC projectiles used in this study, seems to fail via a cutting mechanism, as was previously seen in the work by Prosser (1988b). 3. For a given hybridising scheme aimed at improving the Vc of the pack the superior material should be placed at the front of the pack, i.e. Vc performance is dependent on the front materials in a pack. It should be noted that the improvement in Vc for the cases investigated in this thesis are small, i.e. the placement of Kevlar on the impact face of the hybrid improves Vc performance by approximately 3 to 5%. Conclusions & Future Work 4. When hybridising to decrease back-face deformation, the material which allows the least amount of back-face deformation in 1-ply tests should be placed at the rear of the pack, i.e. deformation (blunt-trauma) performance is dependent on the distal layers in the pack. 5. Hybrids follow the same early event behaviour as that determined by Cepus (2002) for single-material systems, with the early rate of energy absorption being proportional to the strike velocity to the power of 8/3. 6.2.2 Numerical Note that the conclusions drawn here pertain to the early event behaviour of the fabric panels. 6.2.2.1 Single Plies 1. Because of the attainment of lower strains in the yarns increasing the areal density decreases B4. This is one reason that lower denier panels, which typically have lower areal densities, perform better in terms of energy absorption when normalised with respect to mass and compared to higher denier panels. 6.2.2.2 Multi-Plies 1. Additional plies decrease B, due to increased areal density and therefore lower strains. 2. Increasing the inter-ply gap decreases B due to the involvement of less material early in the event. 4 A calculated parameter representing a measure of the rate of energy absorption of a given armour system. 130 Conclusions & Future Work 3. Additional plies also decrease the magnitude of the strain in the preceding plies resulting in lower B values for multi-ply packs. 4. The multi-ply pack with the greatest B corresponds to the case where no gap is introduced between the layers. The system with no gap was found to have the exact same behaviour as a single ply system with an equivalently increased areal density. This indicates that there is a trade off; to get greater Vc performance additional plies are needed, however the addition of these very plies reduces the ballistic efficiency of the system due to increased areal density. These factors all contribute to what are referred to as system effects and lead to losses of ballistic efficiency when plies of material are placed together to form a pack. From the above analysis, design of protective fabric armour systems should use lower denier panels, and minimise the spacing between the plies in order to maximise the ballistic efficiency, at least early on during the impact. 6.3 Future Work The following tasks are envisaged for future research. 1. The boundary conditions of the numerical model should be calibrated using a rationally developed method in order to take the analysis beyond the early event behaviour. 2. A testing program should be undertaken to investigate if the conclusions drawn from the numerical model are in fact true. Conclusions & Future Work 3. A testing program using early event analysis should be undertaken on a series of panels where one weave parameter (e.g. yarn denier or the number of threads per unit length) is held constant. This will allow the development of an experimentally based understanding of what effect each weave parameter has on the behaviour of a fabric panel. It will also serve to calibrate the numerical model. 4. The post-mortem examinations of the failed nylon and Kevlar panels show significant damage. However, currently the numerical model employs an instantaneous failure criterion. Therefore the inclusion of a more gradual failure criterion for the yarns of fabric should be considered. 132 References Bajaj, P. and Srirami (1997), "Ballistic Protective Clothing: A n Overview", Indian Journal of Fibre & Textile Research, Vol . 22, pp. 274-291. Billion, FL, (1998), " A Model for Ballistic Impact on Soft Armour", DSTO-TR-0730, DSTO Aeronautical and Maritime Research Laboratory, Melbourne Victoria 3001 Australia. Billon, H . and Robinson D.J., (2001), "Models for the Ballistic Impact of Fabric Armour", International Journal of Impact Engineering, Vol . 25, pp. 411-422. Cepus, E., Shahkarami, A . , Vaziri, R., Poursartip, A . , (1999), "Effect of Boundary Conditions on the Ballistic Response of Fabric Structures", Proceeding of the 12th International Conference on Composite Materials (ICCM-12), Paris, France. Cepus, E., (2002) "An Experimental Investigation of the Early Dynamic Impact Behaviour of Textile Armour Systems", PhD Thesis, Department of Metals and Materials Engineering, University of British Columbia. Cunniff, P .M. , (1992), "An Analysis of the System Effects in Woven Fabrics Under Ballistic Impact", Textile Research Journal, Vol . 62, No. 9, pp. 495-509. Cunniff, P .M. , (1996), " A Semiempirical Model for the Ballistic Impact Performance of Textile-Based Personnel Armor", Textile Research Journal, Vol . 66, No. 1, pp. 45-59. Cunniff, P .M. , (1999), "Decoupled Response of Textile Body Armor", Proceedings of the 18th International Symposium on Ballistics, San Antonio, Texas, pp. 814-821. 133 Field, B. and Soar, R., (1998), "Soft Body Armour", US Patent Number 5796028. Hall, I.H., (1968), "The Effect of Strain Rate on the Stress-Strain Curve of Oriented Polymers. II. The Influence of Heat Developed During Extension", Journal of Applied Polymer Science, Vol . 12, pp. 739-750. Harpell, G.A. , Palley I., Prevorsek, D.C., (1987), "Multi-Layered Flexible Fiber-Containing Articles", US Patent Number 4681792. Laible, R., (1980), "Ballistic Materials and Penetration Mechanics", Elsevier Scientific Publishing Company, Amsterdam, Netherlands. Lim,C.T., Tan,V.B.C, and Cheong,C.H., (2002), "Perforation of High-Strength Double-Ply Fabric System by Varying Shaped Projectiles", International Journal of Impact Engineering, Vol . 27, pp. 577-591. Lomov, S.V., (1996), "Oblique High Velocity Impact on a Textile Woven Target: Mathematical Simulation", Personal Armour Systems Symposium, Colchester U.K. , pp. 145-156. Meredith R., (1954), "The Effect of Rate of Extension on the Tensile Behaviour of Viscose and Acetate Rayons, Silk, Wool, and Nylon.", Journal of the Textile Institute, Vol . 45, pp. T30-T43. N A T O Standardization Agreement (1996) : Ballistic Test Method for Personal Armour Materials and Combat Clothing, S T A N A G 2920 Edition 2. Prosser, R.A., (1988a) "Penetration of Nylon Ballistic Panels by Fragment-Simulating Projectiles, Part I: A Linear Approximation to the Relationship Between the Square of the V50 or Vc Striking Velocity and the Number of Cloth in the Ballistic Panel", Textile Res. Journal, pp. 61-85. Prosser, R.A., (1988b) "Penetration of Nylon Ballistic Panels by Fragment-Simulating Projectiles, Part II: Mechanism of Penetration", Textile Res. Journal, pp. 161-165. Prosser, R.A., Cohen, S.H., and Segars, R.A. , (2000), "Heat As a Factor in the Penetration of Cloth Ballistic Panels by .22 Caliber Projectiles", Textile Res. Journal, Vol . 70, No. 8, pp. 709-722. Ringleb, F.O., (1957), "Motion and Stress of an Elastic Cable Due to Impact", Journal of Applied Mechanics, Vol . 24,, pp. 417-425. Roylance, D., (1973), "Wave Propagation in a Viscoelastic Fiber Subjected to Transverse Impact", Journal of Applied Mechanics, pp. 143-148. Roylance, D., (1977), "Ballistics of Transversely Impacted Fibers", Textile Res. Journal, pp. 679-684. Roylance, D and Wang, S.S., (1980a) "Penetration Mechanics of Textile Structures ", in "Ballistic Materials and Penetration Mechanics", edited by Laible, R.C, Elsevier Scientific Publishing Co. pp. 273-292. Roylance, D., (1980b) "Stress Wave Propagation in Fibres: Effect of Crossovers", Fibre Science and Technology, Vol . 13, pp. 385-395. 135 Sanders, T.A., (1997), "Penetration of Composite Laminates by Conical Indenters and Projectiles", M.A.Sc. Thesis, Department of Civil Engineering, University of British Columbia,. Shahkarami, A . , (1999), " A Numerical Investigation of Ballistic Impact on Textile Structures", M.A.Sc. Thesis Department of Civil Engineering, The University of British Columbia. Shahkarami, A. , Williams, K., Vaziri, R., Poursartip, A . , Pageau, G., (2000), "Numerical Modelling of the Ballistic Response of Textile Materials", Personal Armour Systems Symposium Colchester, U K . Shahkarami, A . , and Vaziri, R., (2001), "Development of a Numerical Model in LS-D Y N A to Simulate the Ballistic Impact Response of Fabric Structures", Submitted to the Defence R & D Canada -Valcartier Quebec Canada. Shahkarami, A . , Vaziri, R., Poursartip, A . , Williams, K., (2002a), " A Numerical Investigation of the Ballistic Impact Response of Fabric Panels", Personal Armour Systems Symposium TNO Prins Maurits Laboratory, The Netherlands. Shahkarami, A . , Vaziri, R., Poursartip, A. , Williams, K., (2002b) " A Numerical Investigation of the Effect of Projectile Mass on the Energy Absorption of Fabric Panels Subjected to Ballistic Impact", 20th International Symposium on Ballistics, Orlando, Florida, USA. Shockey, D.A. , Erlich, D.C. and Simons, J.W., (2000), "Improved Barriers to Turbine Engine Fragments", DOT/FAA/AR-00/0, SRI International. 136 Smith, J.C., McCrackin, F.L., and Schiefere, H.F., (1958), "Stress-Strain Relationships in Yarns Subjected to Rapid Impact Loading, Part V : Wave Propagation in Long Textile Yarns Impacted Transversely", Textile Res. Journal, pp. 288-302. Smith, J.C., McCrackin, F.L., Schiefere, H.F., Stone, W.K., Towne, K . M . , (1956), "Stress-Strain Relationships in Yarns Subjected to Rapid Impact Loading, Part IV: Transverse Impact Tests", Textile Res. Journal, Vol . 26, N o . l 1, pp. 821-828. Starratt, D.L., (1998), " A n Instrumented Experimental Study of the Ballistic Response of Textile Materials", M.A.Sc Thesis, Department of Metals and Materials Engineering, The University of British Columbia. Starratt, D., Sanders, T., Cepus, E., Poursartip, A . , Vaziri, R., (2000), "An Efficient Method for Continuous Measurement of Projectile Motion in Ballistic Impact Experiments", International Journal of Impact Engineering, Vol . 24, No. 2, pp. 155-170. Taylor, G.L, (1958), "The Plastic Wave in a Wire Extended by an Impact Load", in "Scientific Papers", Edited by G.K. Batchelor, Cambridge University Press,, pp. 467-479 Taylor, W.J. and Vinson, J.R., (1998), "Modelling Ballistic Impact into Flexible Materials", A I A A Journal, Vol . 28, No. 12, pp. 2098-2103. Tejani, N . (2002), "Kevlar Brand Fiber in Anti-Ballistic Body Armour", Internal Report/ Presentation DuPont Advanced Fiber Systems. Ting, C , Ting, J., Cunniff, P .M. , Roylance, D., (1998), "Numerical Characterization of the Transverse Yarn Interaction on Textile Ballistic Response", Proceedings of Society 137 for the Advancement of Material and Process Engineering (SAMPE) Symposium, Vol . 30, pp. 57-67. Ting, J., Roylance, D., Chi, C .H. , Chitrangad, B., (1993), "Numerical Modeling of Fabric Panel Response to Ballistic Impact", International Society for the Advancement of Material and Process Engineering (SAMPE) Technical Conference, USA, pp. 384-392. U.S.Department of Justice (2001), Office of Justice Programs National Institute of Justice, "Selection and Application Guide to Personal Body Armor - NIJ Guide", Vol . 100-01. Vinson, J.R. and Zukas, J.A., (1975), "On the Ballistic Impact of Textile Body Armor", Journal of Applied Mechanics, pp. 263-268. Von-Karman, T. and Duwez, P., (1950), "The Propagation of Plastic Deformation in Solids", Journal of Applied Physics, Vol . 21, No. 10, pp. 987-994. Wilde, A. , Roylance, D., and Rogers, J .M., (1973), "Photographic Investigation of High-Speed Missile Impact Upon Nylon Fabric, Part 1: Energy Absorption and Cone Radial Velocity in Fabric", Textile Res. Journal, No. 12, pp. 753-761. Zhang Y., Abdel-Rahman, N . , Vaziri, R., Poursartip, A . , (1998), "Simulation of Ballistic Impact of Textile Structures", Submitted to Defence Research Establishment, Valcartier, Quebec Canada, Contract Number-W7701 -6-0761/001-XSK. Zufle T, (1993), "Soft Body Armour", US Patent Number 5180880.
- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- A numerical and experimental investigation of the impact...
Open Collections
UBC Theses and Dissertations
Featured Collection
UBC Theses and Dissertations
A numerical and experimental investigation of the impact behaviour of hybrid and multi-ply fabric structures Novotny, Wesley R. 2002
pdf
Page Metadata
Item Metadata
Title | A numerical and experimental investigation of the impact behaviour of hybrid and multi-ply fabric structures |
Creator |
Novotny, Wesley R. |
Date Issued | 2002 |
Description | The focus of this work is to develop an understanding of the impact behaviour of singleand multi-ply fabric materials used in protective (armour) structures. Hybrid structures are also investigated to explore the effect of mixing different materials on armour system performance. Emphasis is placed on the behaviour of these systems during the early stages of the impact event prior to the return of boundary-reflected strain waves to the impact point. This allows one to separate the material response from the structural response and in so doing investigate the losses of material efficiency inherent in systems with greater areal densities. The latter can arise from increased ply count and/or greater mass per unit length of the yarns which make up the fabric. A series of ballistic impact experiments are carried out on the permutations of stacking sequences of 2- and 4-ply Kevlar Nylon hybrids. Instrumented impact tests as well as post-mortem examination are used to investigate the unique behaviour of these hybrid systems over the entire duration of the impact event. The early event behaviour of the hybrids is also characterised. The significant difference between the various hybrids tested is the increased transverse deformation of specific stacking sequences due to penetrated layers. A finite element code, TEXIM, is used to explore the response of single and multi-ply Kevlar fabric systems during the early stages of impact. The numerical results of single and multi-ply Kevlar are found to be in good agreement with the relevant experimental data. Parametric studies using the numerical model are then carried out to investigate the effect of various weave and stacking parameters. Single panels with lower areal densities (typical of panels made up of yarns with lower mass per unit length ) are found to have superior early event performance due to increased strains in the yarns. Multi-ply systems are shown to perform better in terms of the rate of energy absorption early in the impact event as the inter-ply spacing is reduced. Minimizing the spacing is shown to result in increased fibre strains and greater material involvement in absorbing the impact energy. |
Extent | 16837092 bytes |
Genre |
Thesis/Dissertation |
Type |
Text |
FileFormat | application/pdf |
Language | eng |
Date Available | 2009-10-05 |
Provider | Vancouver : University of British Columbia Library |
Rights | For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use. |
DOI | 10.14288/1.0063679 |
URI | http://hdl.handle.net/2429/13624 |
Degree |
Master of Applied Science - MASc |
Program |
Civil Engineering |
Affiliation |
Applied Science, Faculty of Civil Engineering, Department of |
Degree Grantor | University of British Columbia |
GraduationDate | 2002-11 |
Campus |
UBCV |
Scholarly Level | Graduate |
AggregatedSourceRepository | DSpace |
Download
- Media
- 831-ubc_2003-0020.pdf [ 16.06MB ]
- Metadata
- JSON: 831-1.0063679.json
- JSON-LD: 831-1.0063679-ld.json
- RDF/XML (Pretty): 831-1.0063679-rdf.xml
- RDF/JSON: 831-1.0063679-rdf.json
- Turtle: 831-1.0063679-turtle.txt
- N-Triples: 831-1.0063679-rdf-ntriples.txt
- Original Record: 831-1.0063679-source.json
- Full Text
- 831-1.0063679-fulltext.txt
- Citation
- 831-1.0063679.ris
Full Text
Cite
Citation Scheme:
Usage Statistics
Share
Embed
Customize your widget with the following options, then copy and paste the code below into the HTML
of your page to embed this item in your website.
<div id="ubcOpenCollectionsWidgetDisplay">
<script id="ubcOpenCollectionsWidget"
src="{[{embed.src}]}"
data-item="{[{embed.item}]}"
data-collection="{[{embed.collection}]}"
data-metadata="{[{embed.showMetadata}]}"
data-width="{[{embed.width}]}"
async >
</script>
</div>
Our image viewer uses the IIIF 2.0 standard.
To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0063679/manifest