Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 51, 4, pp. 847-857, Warsaw 2013 ANALYSIS OF DYNAMIC PARAMETERS IN A METAL CYLINDRICAL ROD STRIKING A RIGID TARGET Edward Włodarczyk, Marcin Sarzyński Military University of Technology, Faculty of Mechatronics and Aviation, Warsaw, Poland e-mail: edward.wlodarczyk@wat.edu.pl; marcin.sarzynski@wat.edu.pl The initial-boundaryvalueproblemofone-dimensionaldynamicalplasticdeformationwithin the scope of large strain of ametallic cylindrical rod has been analytically solved in a closed form in this paper.Thedeformationof aflat-ended rodhasbeen causedby its normal impact on a rigid target. The rod material in the deformed part is defined by an incompressible, rigid-plastic, power strain hardening model. Such a model of the material provided better correlationwith experimental data than a perfectly-plasticmodel. Results presented in this paper have applicable values. Derived in the paper closed analytical relations, written by elementary functions, give researchers and engineers insight into interaction of the physical parameters of the rod during of the impact process and post-impact. Key words: dynamics, rod, Taylor impact 1. Introduction Taylor publishedhis theory in 1948 (Taylor, 1948).Originally, one-dimensional analysis ofTaylor was used by Whiffin (1948) to estimate the dynamic yield stress of specimens. There has been much interest in the impact testingandestimating thedynamicyield stress since then.Aselective review of the literature with respect to the Taylor impact test is in the papers by Jones et al. (1987, 1997), Field et al. (2004) and Zuckas et al. (1982), and is not necessary to be discussed in this paper. TheTaylor impact test is a useful experiment for estimatingmaterial behavior at high strain rates (Meyers, 1994). The test is reproducible and is reasonably economical after the initial investment has been made. Jones et al. (1987) asserted that simple engineering theories still have a considerable value. Such theories frequently give investigators insight into the interaction of physical parameters and their relationship with the outcome of the event. Most often, these interactions are difficult to ascertain from complex computer outputs. As a result, simple engineering theories often provide thebasis for designof experiments andare frequently used to refine theareas inwhich computing is to be done. Bearing in mind the above-mentioned opinions, the Taylor problem, for a rigid-plastic ma- terial with power strain hardening, loaded by an impact has been analytically solved in a closed form in this paper. 2. Formulation of the problem Consider a uniform flat-ended metal rod of initial parameters, length L, cross-section area S0 and density ρ, that normally impacts against a rigid flat target. The initial velocity of the rod is denoted by U. Assuming that U is large enough, a portion of the rod placed at the target face will deform plastically. As in other problems of plasticity, it may be permissible to neglect 848 E.Włodarczyk,M. Sarzyński elastic strains in comparisonwithplastic strains, particularlywithin the scopeof the largeplastic strains that occur in the considered here problem. Bearing inmindthis fact, it is assumed that the rodmaterial is rigid-plasticwithpower strain hardening and incompressible into the region of plastic strains. Furthermore, it is assumed that the material behavior is rate-independent, i.e. σ=σ(ε). The process of rod deformation during the impact is schematically shown in Fig. 1. Let x denote a Lagrangian coordinate aligned with the axis of the rod and having its origin at the striking end of the rod.The rigid back part of the rod is approaching to the target with absolute time-dependent velocity u(t).Theplasticwave front is propagated away fromthe impact surface with absolute time-dependent velocity v(t) leaving the material behind it at rest since elastic recovery is neglected. S and σ are, respectively, the cross sectional area and nominal stress just behind the plastic wave front, and S0 and σsd just ahead of it; σsd is the dynamic yield-point stress. The remaining symbols, shown in Fig. 1, denote: xp = xp(t) momentary length of the plastically deformed part of the rod, xs = xs(t) momentary length of undeformed portion of the rod, xt =xt(t) momentary displacement of the rear end of the rod, Xt =Xt(t) momentary total length of the rod, D0 initial diameter of the rod, DL maximum diameter of the deformed section of the rod, L initial lengthof the rod, xLagrangian coordinate.Additionally, the index k denotes the final values of the quoted above parameters of the post-test, e.g. xpk is the final length of the deformed part of the rod, and so on. Fig. 1. Forms of deformation of the rod: (a) during the striking, (b) post-striking Neglecting compressibility in the plastic deformedmaterial, the equation of continuity takes the form (u+v)S0 = vS (2.1) This leads to the following expression for the longitudinal compressive strain behind the plastic wave front ε= S−S0 S = u u+v (2.2) Defining xs as the time-dependent undeformed length of the rod at any instant, we have dxs dt =−(u+v) (2.3) Momentum considerations across the wave front yield the equation ρ(u+v)u=σ−σsd (2.4) The equation of motion of the rigid part of the rod is ρxs du dt =−σsd (2.5) Analysis of dynamic parameters in a metal cylindrical rod striking a rigid target 849 The stress-strain curve of the rodmaterial is approximated by the rigid-plastic with a power strain hardeningmodel in the following form σ−σsd =Ewε m (2.6) where the coefficient Ew and exponent m are obtained experimentally for a given material. Such a formulated problem, completed by initial and boundary conditions (see Sec. 3) has been solved in an analytical form in the next section of this paper. 3. Analytical solution of the problem Eliminating the differential dt fromEqs. (2.3) and (2.5), and using relationship (2.2), we obtain d(ρu2)= 2σsdε dxs xs (3.1) From formulae (2.2), (2.4) and (2.6) it follows that ρu2 =Ewε m+1 (3.2) Then, from Eqs. (3.1) and (3.2), we have dxs xs = m+1 2 Ew σsd εm−1dε (3.3) At the moment t = 0, the rod is striking against the rigid target with velocity U. Then xs(0)=L, and u(0)=U. Taking into account this fact, from expression (3.2), we obtain ε(xs =L)= εL = (ρU2 Ew ) 1 m+1 = (U c ) 2 m+1 c= √ Ew ρ (3.4) Integration of Eq. (3.3) at initial conditions (3.4), gives ε= [ 2m m+1 σsd Ew ln xs L + (U c ) 2m m+1 ] 1 m = [y(xs)] 1 m (3.5) where y(xs)= 2m m+1 σsd Ew ln xs L + (U c ) 2m m+1 (3.6) In turn, straightforward algebraic transformations of relationships (2.2), (3.2) and (3.5), yield u(xs)= cε m+1 2 = c[y(xs)] m+1 2m u(xs)+v(xs)= c[y(xs)] m−1 2m v(xs)= 1−ε ε u= c(1−ε)ε m−1 2 = c { [y(xs)] m−1 2m − [y(xs)] m+1 2m } (3.7) In agreement with Eqs. (2.3), (3.5) and (3.7)2, differentiation of the function ε[xs(t)] with respect to t gives formulae for the strain rate, namely dε dt = 2 m+1 σsd Ew [y(xs)] 1−m m 1 xs dxs dt =− 2 m+1 σsd Ew [y(xs)] 1−m m u+v xs =− 2 m+1 σsd Ew c xs [y(xs)] 1−m 2m (3.8) 850 E.Włodarczyk,M. Sarzyński From Eqs. (3.7)2 and (2.3) after transformation and separation of the variables xs and t, we obtain [y(xs)] 1−m 2m d (xs L ) =−d (ct L ) (3.9) The inverse function to the function y(xs), in agreement with notation (3.6), has form xs L =exp [ − m+1 2m Ew σsd (U c ) 2m m+1 ] exp (m+1 2m Ew σsd y ) (3.10) Differentiation of Eq. (3.10) gives d (xs L ) = m+1 2m Ew σsd exp [ − m+1 2m Ew σsd (U c ) 2m m+1 ] exp (m+1 2m Ew σsd y ) dy (3.11) Integrating Eq. (3.9) and using relation (3.11) as well as noting that for t=0 is xs =L and y(L)= yL, we obtain − ct L = m+1 2m Ew σsd exp [ − m+1 2m Ew σsd (U c ) 2m m+1 ] y(xs) ∫ yL ξ 1−m 2m exp (m+1 2m Ew σsd ξ ) dξ (3.12) where yL = (U c ) 2m m+1 y(xs)= 2m m+1 σsd Ew ln xs L +yL y(xs)¬ ξ¬ yL (3.13) So that the current length of the rigid part of the rod xs(t) has been obtained in form of an implicit function of t for any value of the coefficient m. In some cases, one can define the time-dependent function xs(t) explicitly (see example). In turn, the function xp(t), in agreement with Fig. 1a, is defined by the formula xp(t) L =1− [xs(t) L + xt(t) L ] where xt L = ct/L ∫ 0 u[xs(τ)] c d (cτ L ) (3.14) This quantity xt/L, after making use of relations (3.7), (3.9) and (3.11) has been transformed to xt(xs) L = m+1 2m Ew σsd exp ( − m+1 2m Ew σsd yL ) yL ∫ y(xs) ξ 1 m exp (m+1 2m Ew σsd ξ ) dξ (3.15) In agreement with expressions (2.2), (3.7)3 and (3.7)2 the current diameter of the plasticly deformed part of the rod Dp has been determinated by the formula Dp(xs)=D0 √ u+v v = 1 √ 1− [y(xs)] 1 m D0 (3.16) Thus have been defined all the dynamic parameters of the rigid-plastic rodwith power strain hardeningbymeans of the implicit function of xs. The quantity xs is a time-dependent function defined by Eqs. (3.12) and (3.13). In the example (next Section) these parameters are presented in formof an elementary time-dependent function during the striking and their final post-impact values. Analysis of dynamic parameters in a metal cylindrical rod striking a rigid target 851 4. Example Uniform annealed (500◦C, 1h) cooper (Cu-ETP) rods of the initial dimensions: length L = 48mm and diameter D0 = 12mm were used in the examinations. The mechanical pa- rameters of cooper are: density ρ=8900kg/m3, engineering static yield stress R0.2 =84MPa, Young’s modulus E =130GPa. Stress-strain curves in compression of cooper for the true sys- tem σ=σ(ϕ) and for the nominal system σ=σ(ε) are depicted in Fig. 2, where ϕ is the true (logarithmic) strain, and ε is the nominal (engineering) strain, where ε=1− exp(−ϕ). Fig. 2. Stress-strain curves in compression of annealed cooper for the true system σ=σ(ϕ) and the nominal system σ=σ(ε) The rods were driven by a firing gun to the initial speeds within the range from 60m/s to 220m/s. The rods impacted perpendicularly against the flat rigid target. The pictures of deformed rods after impact are shown in Fig. 3. Fig. 3. Pictures of deformed rods Figure 2 shows that the curve of the nominal stress-strain for ε < 0.4 can be well enough approximated bymeans of a straight line (m=1), namely σ−σsd =1100ε [MPa] (dashed line in Fig. 2). The approximation error does not exceed several percent. In agreement with the general solution of the problem presented in Section 3, for m=1, we obtain x= ct xs L =1− ct L xt L = (U c − σsd Ew )ct L − σsd Ew ( 1− ct L ) ln ( 1− ct L ) xp L = ( 1− U c + σsd Ew )ct L + σsd Ew ( 1− ct L ) ln ( 1− ct L ) Xt L = xp L + xs L (4.1) 852 E.Włodarczyk,M. Sarzyński and ε= u c = U c + σsd Ew ln ( 1− ct L ) εL = εmax = U c dε dt =− σsd Ew c xs =− σsd Ew c L− ct σ=σsd+Ewε σL =σmax =σsd+Ew U c u=U+ c σsd Ew ln ( 1− ct L ) v= c−U −c σsd Ew ln ( 1− ct L ) u+v= c= const Dp =D0 [ 1− U c − σsd Ew ln ( 1− ct L )] − 1 2 DL =Dpmax =D0 ( 1− U c ) − 1 2 (4.2) The end of the striking process occurs at the instant when ε(tk) = u(tk) = 0. From this condition and Eq. (4.2)1, it follows that tk = L c [ 1− exp ( − Ew σsd U c )] (4.3) Then the final values of the parameters of the rod, in agreementwith the above quoted formulae, one may define by the following expressions xsk L =exp ( − Ew σsd U c ) xtk L = U c − σsd Ew ( 1− xsk L ) xpk L = ( 1− xsk L )( 1+ σsd Ew ) − U c Xtk L =1− xtk L (4.4) From Eq. (4.2)1 and condition ε(tk) = 0 as well as formula (4.3), after transformations, we obtain σsd = Ew(U/c) ln(L/xsk) (4.5) The simple Taylor formula for σsd has form (Taylor, 1948) σsd = L−xsk 2(L−Xtk) ρU2 ln(L/xsk) (4.6) So that all dynamical parameters of the annealed cooper (Cu-ETP) rod loaded according to the Taylor test have been defined explicitly by elementary time-dependent functions during the striking, and by closed form formulae in post-impact. Bearing in mind this fact, we will make an analysis only for some of them. The experimental data obtained from measurements of the deformed post-impact cooper rods and the results of theoretical calculations for several values of the impact velocity are listed in Table 1. One ought to notice that time of the impact action is very short, about 0.13ms in the whole range of variation of the impact velocity (60m/s–220m/s). It corresponds with the literature data (Whiffin, 1948; Jones et al., 1987, 1997). From analysis of the variations of quantities listed in Table 1 with respect to the values impact velocity, it follows that the parameters σsd,Xtk,xpk and εL vary approximately linearly with respect to an increase in the impact velocity within the range 60m/s–220m/s, and one may themwell enough extrapolate bymeans of the following expressions: — for formula (4.5) σsd =65+0.712(U −68) (4.7) Analysis of dynamic parameters in a metal cylindrical rod striking a rigid target 853 Table 1. Comparison of the experimental data obtained from Taylor tests with theoretical results calculated for annealed cooper (Cu-ETP) U [m/s] 68 125 146 180 214 σsd [MPa] Paper (4.5) 65 111 122 142 169 Taylor (4.6) 61 89 96 112 133 tk ·10 3 [s] 0.13200 0.13315 0.13400 0.13457 0.13458 xsk =xsk/L theoretical 0.03733 0.02904 0.02325 0.01869 0.01860 experimental 0.03752 0.02910 0.02292 0.01877 0.01873 |∆xsk| [%] difference 0.5 0.2 1.5 0.9 0.2 Xtk = Xtk/L theoretical 0.863 0.741 0.691 0.613 0.539 experimental 0.901 0.790 0.744 0.682 0.622 |∆Xtk| [%] difference 4.3 6.2 7.2 10.1 13.3 xpk =xpk/L theoretical 0.825 0.712 0.668 0.594 0.521 experimental 0.864 0.757 0.721 0.663 0.603 |∆xpk| [%] difference 4.5 5.9 7.4 10.4 13.7 DL/D0 theoretical 1.12 1.25 1.31 1.44 1.60 experimental 1.11 1.31 1.43 1.65 1.97 |∆(DL/D0)| [%] difference 1.06 4.97 8.58 13.03 18.48 εL theoretical 0.19 0.36 0.42 0.51 0.61 experimental 0.18 0.42 0.51 0.63 0.74 |∆εL| [%] difference 5.6 14.3 17.6 19.0 17.8 — for Taylor formula (4.6) σsd =61+0.493(U −68) (4.8) — theoretical expressions Xtk = Xtk L =0.002219(214−U)+0.539 xpk = xpk L =0.002082(214−U)+0.521 εL = U c c= √ Ew ρ =351m/s (4.9) — experimental expressions Xtk = Xtk L =0.001911(214−U)+0.622 xpk = xpk L =0.001788(214−U)+0.603 εL =0.003836(U −68)+0.18 (4.10) where, in the quoted above relationships values of the impact velocity U ought to substitute in m/s. The absolute value of the differences between the experimental and theoretical results of the studied parameters increase from several to over a dozen percent together with the increase in the impact velocity from 60m/s to 220m/s. The cause of this fact is the influence of the radial spreading (inertia) of the tested specimen on its longitudinal compressive strain. The spreading increases together the increasing impact velocity, especially as related to the annealed cooper (Cu-ETP), which is very sensitive to plastic deformation (Jones et al., (1997). In case of alloy steels (e.g. nickel-chrome steel), the differences are considerably smaller (Włodarczyk et al., 2012). The quoted above facts show that the presented here one-dimensional theory, with 854 E.Włodarczyk,M. Sarzyński respect to the Taylor impact problem, is permissible for materials with strain hardening and moderate values of impact velocities. Variations of the relative speeds v/c and u/c versus the Lagrangian coordinate x/L= ct/L for selected values of the impact velocity U are presented in Fig. 4. Fig. 4. Variations of relative speeds v/c (solid line) and u/c (dashed line) versus Lagrangian coordinate ct/L for selected values of the impact velocity U As can seen, the propagation speed of the plastic wave front v intensively increases in the surroundings of the fore-front of the undeformed portion of the rod, and at x=L−xsk reaches the maximum value v = υmax = c= √ Ew/ρ. On the contrary, the velocity of the undeformed part of the rod u intensively decreases here, and is equal to 0 at x=L−xsk. The sum of these velocities u(ct/L)+v(ct/L)≡ c is constant. This fact was noticed by Lee andTupper (1951) in their computational scheme. InTaylor (1948) andWhiffin (1948) investigations, it was assumed that v(t)≡ const and du/dt≡ const. It is an erroneous assumptionwhich gives unreal results. The graph of variation of the longitudinal compressive strain behind the plastic wave front ε with respect to the Lagrangian coordinate x/L= ct/L is the same as the graph of u/c (ε=u/c – see (4.2)1). Figure5 showsthat theabsolutevalueof the strain rate intensively increases at the immediate surroundingsof theplasticwave front fromthe target sideat thefinal stage of the impactprocess. It reaches maximum values at the instant t = tk which is determined by the impact velocity, e.g. [ε̇(tk)] = [ ε̇]max = 11502s −1 for U = 68m/s, on the other hand [ε̇]max = 59845s −1 for U = 214m/s. One ought to notice that in excess of 90%, the specimen length is deformed quasi-statically. The absolute value of the maximum strain rate |ε̇|max [s −1], and the absolute value of the average strain rate |ε̇|mid [s −1] are determined by formulae |ε̇|max = σsd Ew 1 1−ηk c L ηk = ctk L |ε̇|mid = 1 ηk σsd Ew ηk ∫ 0 dη 1−η = σsd Ew | ln(1−ηk)| ηk (4.11) From analysis of formula (4.11)3, it follows that the quantity |ε̇|mid monotonically increases from |ε̇|mid =0.202s −1 for U =68m/s to |ε̇|mid =0.647s −1 for U =214m/s. Accordingly, the assumption that σ=σ(ε) at the formulation of the problem is permissible. Analysis of dynamic parameters in a metal cylindrical rod striking a rigid target 855 Fig. 5. Variation of the absolute values of the strain rate |ε̇| versus Lagrangian coordinate x/L= ct/L for selected values of the impact velocity U Figure 6 shows, among other things, graphs of the parameter (σ−σsd)/Ew in the deformed part of the rod. As seen, the stress is constant in the whole deformed part of the rod at the given instant t, and is equal to the stress in the fore-front of the plastic wave from the target side at the same time t. The stress decreases together with propagation of the wave front down to value σsd and suddenly decreases to 0 at the final impact process – the rod separates from the target. Fig. 6. Graphs of the parameter (σ−σsd)/Ew in the deformed portion of the rod at the instants: t=0.2L/c, 0.6L/c and 0.9L/c for the impact velocity U =214m/s 5. Final conclusions Themain conclusions from the above theoretical and experimental investigations may be briefly summarized as follows: • The analytical solution of the dynamics of the one-dimensional rigid-plastic cylindrical rod striking against a rigid target has been found in this paper. It is an extended original Taylor perfectly-plastic model for materials with power strain hardening. 856 E.Włodarczyk,M. Sarzyński • The extended theory takes into account the step change of the nominal stress across the plastic wave front in the equation of momentum balance, because this front is a source of strong discontinuity in physical parameters. In the Taylor model, it was assumed that the stress is continuous across this front. • Thedynamical parameters of the annealed cooper (Cu-ETP) rodwere examined bymeans of the Taylor test. • The static stress-strain curves of this cooper rod for the true system, σ = σ(ϕ), and for the nominal system, σ=σ(ε), were well enough approximated by the following functions σ−σsd =409ϕ 0.36MPa, and σ−σsd =1100εMPa (see Fig. 2). • The absolute velocity of propagation of the plastic wave front v and the absolute speed of displacement of the rear of the rod u are individually intensively time-dependent functions (see Fig. 4). On the other hand, their sum,u(t)+ v(t) = c is constant. In Taylor (1948) andWhiffin (1948) investigations, it was assumed that v(t)≡ const and du/dt≡ const. These are far-reaching simplifications contradictory to reality. • The absolute value of the strain rate |ε̇| intensively increases only in the immediate surro- unding of the plastic wave front at the final stage of the impact process, and reaches the maximum value at the instant t = tk. On the contrary, in excess of 90%, the specimen length is deformed quasi-statically. The average absolute value of the strain rate |ε̇|mid increases from |ε̇|mid = 0.202s −1 for U = 68m/s to |ε̇|mid = 0.647s −1 for U = 214m/s. Accordingly, the influence of the strain rate on the stress-strain curve for annealed cooper (Cu-ETP) is negligible. • The stress is constant in the whole deformed section of the rod at the given moment t, and is equal to stress in the fore-front of the plastic wave from the target side at the same time t (see Fig. 7). • All final parameters of the deformed rods post-impact, together with the dynamic yield stress for the examinedcooper, are approximately linear functionsof the impactvelocity U, see formulae (4.7)-(4.10). • The absolute values of differences between the experimental and the theoretical results of the studiedparameters increase fromseveral to a dozen percentwith the increasing impact velocity (see Table 1). These differences are caused by the radial spreading during impact, especially with respect to annealed cooper, which is very sensitive to plastic deformation. • The above quoted facts show that the presented here one-dimensional theory, with respect to the Taylor impact problem, is permissible for materials with strain hardening and moderate impact velocities. • According to the authors, the results presented in this paper have applicable and cognitive effects. The derived in a closed form formulae, which are written by elementary functions, give investigators and engineers insight into the interaction of the physical parameters of the rod during the impact process and post-impact. References 1. Field J.E., Walley S.M., Proud W.G., Goldrein H.T., Siviour C.R., 2004, Review of experimental techniquesofhighratedeformationandshockstudies, International Journal of Impact Engineering, 30, 725-775 2. Jones S.E., Gillis P.P., Foster J.C. Jr., 1987, On the equation of motion of the undeformed section of a Taylor impact specimen, Journal of Applied Physics, 61, 2 Analysis of dynamic parameters in a metal cylindrical rod striking a rigid target 857 3. Jones S.E., Maudlin P.J., Foster J.C. Jr., 1997, An engineering analysis of plastic wave propagation in the Taylor test, International Journal of Impact Engineering, 19, 2 4. Lee E., Tupper S., 1951, Analysis of inelastic deformation in a steel cylinder striking a rigid target, Journal of Applied Mechanics, 21, 1 5. Meyers M.A., 1994, Dynamic Behavior of Materials, John Wiley and Sons, Inc. New York- Chichester-Brisbane-Toronto-Singapore 6. Taylor G., 1948, The use of flat-ended projectiles for determining dynamic yield stress. I. The- oretical considerations,Proceedings of the Royal Society, Series A, London 7. WhiffinA.C., 1948,Theuseofflat-endedprojectiles fordeterminingdynamicyield stress. 2.Tests on variousmetallic materials,Proceedings of the Royal Society, Series A, London 8. WłodarczykE., Jackowski A., SarzyńskiM., 2012,Dynamic behavior of ametallic cylinder striking a rigid target,Biuletyn Wojskowej Akademii Technicznej, 61, 4 9. Zukas J.A., Nicholas T., Swift H.F., Greszczuk L.B., Curran D.R., 1982, Impact Dyna- mics, Willey-Interscience, NewYork Manuscript received Decembr 7, 2012; accepted for print January 28, 2013