Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 55, 2, pp. 667-677, Warsaw 2017 DOI: 10.15632/jtam-pl.55.2.667 ROLE OF WAVE REVERBERATIONS IN EXIT VIBRATIONS OF PROJECTILES AND SHELLS Radosław Trębiński Military University of Technology, Faculty of Mechatronics and Aerospace, Warszawa, Poland e-mail: rkt@wat.edu.pl Carlucci et al. (2013) proposed a theoretical model describing exit vibrations of a shell launched by a gun. The model treats the shell as a system of a spring, damper and a rigid mass. In this paper, similarmodels describing vibrations of shells andmonolithic projectiles arecomparedwith some1Dmodels taking intoaccountwavereverberations.Thecomparison proves that themodels taking into account thewavemotion, forecast similar amplitudes and frequencies of vibrations as discrete models. However, details of acceleration time courses differ considerably. The problem of the influence of the friction force between the shell wall and the filling is also discussed. The results of modelling serve as well for assessing if some projectile velocity oscillations detected by making use of the Doppler radar can be solely attributed to the exit vibrations. Keywords: intermediate ballistics, projectile vibrations, wave reverberation 1. Introduction Projectiles and shells moving inside a gun tube undergo compression. After leaving the muz- zle, they experience a sudden drop of pressure acting on their base. This decompression causes vibrations of the projectiles and shells, which can be detrimental for the integrity of their in- ner structure. Ammunition designers take this into account by assuming that there is a step drop of the pressure from themuzzle pressure to the ambient pressure. However, recently some data became available enabling designers to take into account the continuous character of the pressure changes. On the basis of accelerometer records, Carlucci and Vega (2007) proposed an approximation of pressure changes by the exponential decay curve p(t)= pmexp(−βt)= pmexp ( − t tm ) (1.1) Values of the muzzle pressure pm and the coefficient β were published for a 155mm cannon in Carlucci and Vega (2007) and Carlucci et al. (2013). Some theoretical models were considered by Trębiński and Czyżewska (2015a,b), enabling calculation of pressure courses acting on the projectile in the intermediate period. On the basis of the results of calculations for a broad range of launching systems, an approximate formula was proposed for calculating the value of the coefficient β for a launching system of the caliber d β=(0.74+0.24Mm) um d (1.2) Values of themuzzle flowMach numberMm and themuzzle velocity um can be determined on the basis of results of internal ballistics calculations. In Carlucci et al. (2013), a model was proposed describing vibrations of a shell subjected to the action of a pressure pulse (1.1). The motion of the ogive of the shell was modeled in 668 R. Trębiński the reference frame moving with the center of gravity of the shell. The ogive was treated as a rigid mass, whereas the wall of the shell was treated as a spring. Its inertia was taken into account by adding one third of its mass to the mass of the ogive. A damper was added in order to take into account damping of the vibrations. The model predicts a considerable magnitude of acceleration oscillations, which is in qualitative agreement with experimental observations. Considering accelerometers as vibrating systems showed very small influence of their inertia and rigidity on the courses of acceleration, except for the very beginning of the process. The determined values of the period of vibrations of projectiles and shells as well as the characteristic time tm in expression (1.1) are close to the values of the reverberation time of elastic waves moving in projectiles and shells. In such a case, one can expect that the results of modeling with the use of discrete models and models incorporating wave motion may differ considerably. This paper is devoted to analysis of the role of wave reverberations. The main objective of this analysis is to determine to which extent the amplitudes and frequencies of the exit vibrations of projectiles and shells forecasted by discrete models and models taking into account wave reverberations differ. Several 1D models of the wave motion in a projectile or a shell, when subjected to action of the pressure pulse described by (1.1), are considered. Conclu- sions concerning the potential influence of wave processes on the indications of accelerometers embedded in the projectiles and shells are formulated. Doppler radar measurements of the velocity changes of a projectile leaving the muzzle of a 30mm launching system, presented in Leciejewski et al. (2013), showed oscillations of the velocity value (Fig. 1). The question arose whether these oscillations can be attributed solely to the vibrations of the projectile. Finding the answer to this question was the secondary objective of the analysis presented in this paper. Fig. 1. Results of measurements of the projectile velocity changes outside the muzzle of a 30mm launching system (Leciejewski et al., 2013) 2. Models 2.1. Models of a monolithic projectile The sense of physical models considered can be explained by referring to Fig. 2. Inmodel 1, the projectile is considered as a system of two rigid masses m1 and m2 connected by a spring. The stiffness of the spring is determined as ks = AE l0 A= πd2 4 (2.1) Role of wave reverberations in exit vibrations of projectiles and shells 669 Fig. 2. Schemes of models 1 and 2 Themathematical model corresponding to physical model 1 is given by the following initial value problem m1 d2x1 dt2 −ks(x2−x1− l0)=Apme − t tm m2 d2x2 dt2 +ks(x2−x1− l0)= 0 x1(0)= 0 x2(0)= l0− m2Apm (m1+m2)ks dx1 dt ∣ ∣ ∣ t=0 = dx2 dt ∣ ∣ ∣ t=0 =um (2.2) Dividing the first equation bym1 and the second one bym2 and summing up the equations, we obtain d2(x2−x1) dt2 + ( ks m1 + ks m2 ) (x2−x1− l0)=− Fp(t) m1 (2.3) Defining a new variable w=x2−x1− l0 (2.4) we obtain the following initial value problem d2w dt2 +ω2w=− Apm m1 e − t tm ω2 = ks ( 1 m1 + 1 m2 ) w(0)=− m2Apm (m1+m2)ks dw dt ∣ ∣ ∣ t=0 =0 (2.5) The solution to the problem is given by w(t) =− Apm m1ω2(1+ω2t2m) [ ωtm sin(ωt)+cos(ωt)+ω 2t2me − t tm ] (2.6) Making use of this,we obtain fromthe second equation of (2.2) an expression for acceleration of the forward part of the projectile a(t)= Apm m(1+ω2t2m) [ ωtm sin(ωt)+cos(ωt)+ω 2t2me − t tw ] m=m1+m2 (2.7) 670 R. Trębiński Integrating (2.7) from 0 to t, we obtain the following formula for the velocity increase of the forward part of the projectile ∆u(t)= Apmtm m(1+ω2t2m) [ 1− cos(ωt)+ 1 ωtm sin(ωt)+ω2t2m ( 1− e − t tm )] (2.8) In model 2, the projectile is treated as an elastic cylindrical rod with the same caliber d andmassm as the real projectile and length l0 =4m/(πd 2). Thematerial properties of the rod are determined by its density ρ and Young’s modulus E. It is assumed that the force Fp(t) is produced by the pressure uniformly distributed at the left end of the rod. The pressure changes in time are described by (1.1). The loading produces a uniaxial stress state in the rod. It is assumed that there is no force at the right end of the rod. This assumption is justified by the observation that the pressure acting on the forward part of the projectile is much less than the pressure acting on its base in the whole phase when the projectile is accelerated. We assume that at the beginning of the process all parts of the rod have the same velocity and acceleration. Themathematical model corresponding to physical model 2 is given by the following initial- boundary value problem in the Lagrangian coordinates ρu,t=σ,x σ,t=Eu,x u(x,0)=um σ(x,0)= pm (x l0 −1 ) σ(0, t) =−pme − t tm σ(l0, t)= 0 (2.9) Initial condition (2.9)3,4 for the stress σ follows from the assumption that the whole rod has initially the same acceleration. Solving the problem, we obtain the following expressions for the velocity increase and acce- leration of the forward part of the projectile ∆u(t)=∆um [ t Tw −2N+2 N ∑ j=1 exp ( − t− (2j−1)Tw tm ) ] ∆um = pm ρc N =E (n+1 2 ) n=0,1,2, . . . Tw = l0 c c= √ E ρ a(t)=∆um [ 1 Tw − 2 tm N ∑ j=1 exp ( − t− (2j−1)Tw tm ) ] (2.10) 2.2. Models of the shell The physical object considered and its physical models are illustrated in Fig. 3. In model 3, the ogive and the base of the shell are treated as rigid masses. The wall of the shell of length l0 is treated as an elastic hollow cylinder, whereas the filling (explosive) is treated as an elastic cylindrical rod with density ρf and velocity of sound cf. It is assumed that there is no friction between the wall and the filling. Because the filling is more compressible than the wall, it can also be assumed that there is no contact between the filling and the ogive. A uniaxial stress state is assumed inside the wall and the filling. The mathematical description of model 3 is given by the following initial-boundary value problem (the subscript f refers to the filling) ρu,t=σ,x σ,t= ρc 2u,x ρfuf,t=σf,x σf,t= ρfc 2 fuf,x u(x,0)=um uf(x,0)=um (2.11) Role of wave reverberations in exit vibrations of projectiles and shells 671 Fig. 3. Schemes of the physical object (a shell) and of models 1, 3 and 4 used for the analysis and σ(x,0)= ρa0(x− l0)− m2a0 Ac a0 = pmA m1+m2+ρl0Ac+ρfl0Af σf(x,0)= (m1+m2+Acρl0)a0−Apm Af +ρfa0x (2.12) and m1u,t (0, t)=Apme − t tm +Acσ(0, t)+Afσf(0, t) m2u,t (l0, t)=−Acσ(l0, t) σf(l0, t)= 0 (2.13) whereAf,Ac mean the cross-sectional area of the filling and the wall. The initial condition for stress (2.12) has been derived on the assumption that at the begin- ning of theprocess thewhole shell has the sameacceleration a0 andon the assumption that there is no contact between the filling and the ogive. The second assumption has also been applied in the formulation of the boundary condition at x= l0. Problem (2.11)-(2.13) has been solved by the method of characteristics. The characteristic grid is shown in Fig. 4. It shows wave trajectories in the wall of the shell. The values of velocity and stress in the grid nodes are calculated using the characteristic relations ∆σ=±ρc∆u ∆x=±c∆t ∆σf =±ρfcf∆uf ∆x=±cf∆t (2.14) In the node i, we have σi = σi+1+σi−1+ρc(ui+1−ui−1) 2 ui =ui−1+ σi−σi−1 ρc σfi = σfL+σfR+ρfcf(ufR−ufL) 2 ufi =ufL+ σfi−σfL ρfcf (2.15) 672 R. Trębiński Fig. 4. Characteristic grid used for calculations for models 3, 4 and 5 The values of velocity and stress inside the filling in the auxiliary nodesL andR (characteri- sticsL−i andR−i correspond to the wave velocity cf) are calculated by a linear interpolation between the nodes i−1 and i+1 qL = qi−1+(qi+1−qi−1) c− cf 2c qR = qi−1+(qi+1−qi−1) c+ cf 2c q= {uf,σf} (2.16) The values of velocity and stress in nodes 1 and n are calculated by making use of an approximate form of boundary conditions (2.13) and relations at characteristics 1 – BL and n –BR uBL =u1+a1e − t tm ( 1− e − 2∆t tm ) +a2(σ1+σBL)+a3(σf1+σfBL) ufBL =uBL a1 = Apmtm m1 a2 = Ac∆t m1 a3 = Af∆t m1 σBL =σ2−ρc(uB −u2) σfBL =σfR−ρfcf(uBL−ufR) uBR =un−a4(σn+σBR) a4 = Ac∆t m2 ufBR =ufL− σfL ρfcf σBR =σn−1+ρc(uBR−un−1) σfBR =0 (2.17) Thevalues ofvelocity andstress inside thefilling in theauxiliarynodesLandRare calculated by a linear interpolation between nodes 1-2 and (n−1)-n qL = qn+(qn−1−qn) 2cf c+cf qR = q1+(q2− q1) 2cf c+ cf q= {uf,σf} (2.18) Calculations are performed in two iterations. In the first iteration, the values of stress are set as σBL =σ1 σfBL =σf1 σBR =σn (2.19) In the second iteration, the values of stress calculated in the first iteration are used. Neglecting friction between the wall and the filling is questionable because the filling is compressed and it exerts pressure on the wall. Instead of introducing the friction force into the model, two limiting cases are considered. The model without friction is the first limiting case. The second limiting case is a model for which it is assumed that the friction force is strong enough to exclude any displacement between the filling and thewall. If bothmodels give similar results, it will be no necessity to take into account the friction force. In the second limiting case, the compressibility of the filling has no influence on the process. Its inertia can be taken Role of wave reverberations in exit vibrations of projectiles and shells 673 into account as an additional mass added to mass of the wall. This can be done by artificially increasing density of the wall material ρa = ρ+ρf Af Ac (2.20) The central part of the shell can be treated as a cylindrical rod with cross-section area Ac and density ρa. This is the sense of model 4. Its mathematical form is not presented because it can be easily derived from problem (2.11)-(2.13) by neglecting all the expressions corresponding to the filling. Also the method of solving the problem is similar to that described for model 3. In themodeling of the vibrations of the shell,model 1 can beused.Themassm1 is calculated as the sumofmass of the base of the shell andmass of thewall and the filling lying between the base and the center of gravity. The mass m2 is calculated as the sum of mass of the ogive and mass of the wall and the filling lying between the ogive and the center of gravity. The stiffness of the spring is calculated by using formula (2.1) withA replaced byAc. Fig. 5. Schemes of the model by Carlucci et al. (2013) (a), (b) andmodels 5 and 6 (c), (d) In order tomake comparisonwith the results of themodel describedbyCarlucci et al. (2013), twomoremodels are considered, as shown inFig. 5. The vibrations aremodeled in the reference framemoving with the center of gravity. Figure 5a shows the scheme of the model presented in Fig. 4 in the cited paper. It is erroneous because the forceF(t) should be applied to themassm, not to the spring and the damper. It is the inertial force, so its value should be expressed as F(t)= m ms Fp(t) (2.21) wherems is the total mass of the shell. In Carlucci et al. (2013), the equality of both the forces was assumed. The proper form of the scheme is shown in Fig. 5b. The equations given in the cited paper correspond to this scheme. Figure 5c presents a scheme of model 5 considered in this work. It differs from the scheme in Fig. 5b by the absence of the damper and by the sign of the coordinate x. The mass m in model 5 is taken as the sum of mass of the ogive and 1/3 of mass of the shell wall. Figure 5d corresponds tomodel 6. The spring is replaced by an elastic rod with cross-section area Ac. The mathematical model corresponding to model 5 is given by the following initial value problem m d2w dt2 +ksw=− m ms Apme − t tm w=x− l0 w(0)=− m ms Apm ks dw dt ∣ ∣ ∣ t=0 =0 (2.22) Solving this problem, the following expressions for the velocity increase and acceleration can be obtained ∆u(t)= Am2pm ms(m+kst2m) [ tme − t tm + tm sin(ωt)− ω ks cos(ωt) ] a(t)= Am2pm ms(m+kst2m) [ ωtmcos(ωt)+ ω2 ks cos(ωt)− e − t tm ] (2.23) 674 R. Trębiński Themathematicalmodel corresponding tomodel 6 is given by the following initial-boundary value problem ρu,t=σ,x σ,t= ρc 2u,x u(x,0)=um σ(x,0)=− mApm msAc u(0, t) = 0 m2u,t (l0, t)=− m ms Apme − t tm −Acσ(l0, t) (2.24) The problem is solved by themethod of characteristics in the way described for model 3. 3. Results of the modeling Calculations have been performed bymaking use ofmodels 1 and 2 for a 30mmgun used in the investigations presented by Leciejewski et al. (2012). The following values of parameters have been applied: A = 7.07cm2, l0 = 66mm, m1 = m2 = 183g, ρ = 7850kg/m 3, E = 200GPa, pm = 51.5MPa, tm = 0.0211ms. The value of parameters pm and tm have been determined by the regression of the pressure course calculated by the model described in Trębiński and Czyżewska (2015). They are shown in Fig. 6. The dotted line corresponds to the approximation given by (1.2). Fig. 6. Pressure course for a 30mm gun calculated by the model proposed in Trębiński and Czyżewska (2015b), regression curve and approximation of the course using formulae (1.1) and (1.2) Figures 7a and 7b present a comparison of the results obtained bymodels 1 and 2. Although the character of the velocity changes is different for the twomodels, the amplitude of oscillations is similar. It ismuch less than the amplitude of the velocity oscillations presented in Fig. 1. This suggests that these oscillations cannot be attributed to axial vibrations or wave reverberations in the projectile. We can suppose that these oscillations are caused by yawing of the projectile or by the effect of ionization of gases in strong shock waves arising in the vicinity of themuzzle (description of wave processes accompanying the exit of projectiles are given inKlingenberg and Heimerl (1988)). As far as acceleration is concerned, the differences between the two models are much mo- re pronounced. This means that it is recommended to take into account the wave motion in predicting the acceleration of a projectile leaving themuzzle. Figures 8a and8bpresent a comparison of the results obtained bymodels 5 and6.Thevalues of parameters used in calculations have been the same as in Carlucci et al. (2013):A=188cm2, Ac = 35cm 2, l0 = 508mm, m = 3.75kg, ρ = 7850kg/m 3, E = 199GPa, pm = 68.9MPa, tm = 0.1731ms. For the purpose of comparison, it has been assumed that m = ms. The two models give very close results. This means that in this case taking into account the wave rever- berations is not necessary. Role of wave reverberations in exit vibrations of projectiles and shells 675 Fig. 7. Comparison of the (a) velocity increase and (b) acceleration courses calculated by the use of models 1 and 2 Fig. 8. Comparison of the (a) velocity increase and (b) acceleration courses of the shell ogive calculated by the use of models 5 and 6 Thepicture is somewhatdifferent ifweconsider thewavemotion in thewhole shell.Figures 9a and 9b show a comparison of the calculated acceleration courses of the shell base and the shell ogive calculated by the use of models 3 and 4. The values of parameters applied in calculations have been as follows: A = 188cm2, Ac = 93cm 2, Af = 95cm 2, l0 = 590mm, m1 = 6kg, m2 = 5kg, ρ = 7850kg/m 3, c = 5035m/s, ρf = 1715kg/m 3, cf = 2470m/s, pm = 68.9MPa, tm =0.1731ms. Fig. 9. Comparison of the (a) velocity increase and (b) acceleration courses of the shell ogive calculated by the use of models 3 and 4 676 R. Trębiński Although models 3 and 4 represent the opposite limiting cases, as far as friction between the wall and the filling is concerned, they give close results. Therefore, simpler model 4 is recommended. The results obtained by making use of it are compared with the results from model 1 in Figs. 10a and 10b. Although the frequency of oscillations of the velocity and the acceleration values arevery close,model 4 forecasts considerably larger amplitudesof oscillations. Fig. 10. Comparison of the (a) velocity increase and (b) acceleration courses of the shell ogive calculated by the use of models 4 and 1 Figure11presents theacceleration courses of the static center of gravity of the shell calculated bymodel 4. These are compared with the acceleration calculated for the shell treated as a rigid body. It can be seen that an accelerometer positioned in the static center of gravity will record acceleration courses that considerably differ from the mean acceleration of the shell. Fig. 11. Comparison of the acceleration courses of the shell static center of gravity calculated by the use of model 4 and for the shell treated as a rigid body 4. Conclusions • Models taking into account wave reverberations generally predict similar frequencies and amplitudes of vibrations to the simple vibrational models. However, the character of time changes of acceleration differs considerably. Therefore, modeling of the wave motion is recommended for interpretation of the signals from accelerometers embedded in the shells. • Close results obtained for models 3 and 4 prove it to be acceptable to assume that there is no displacement between the wall of the shell and the filling. • The predicted amplitudes of velocity oscillations of the projectile leaving the muzzle are much smaller than those which can be deduced from the results of Doppler radar me- Role of wave reverberations in exit vibrations of projectiles and shells 677 asurements. Therefore, the observed oscillations of velocity cannot be solely attributed to vibrations of the projectile. References 1. Carlucci D., Vega J., 2007, Empirical relationship for muzzle exit pressure in a 155mm gun tube, [In:]Computational Ballistics III, C. A. Brebbia, edit., WIT Press, Ashurst, UK 2. CarlucciD.E., FrydmanA.M,Cordes J.A., 2013,Mathematical description of projectile exit dynamics (set-forward),ASME Journal of Applied Mechanics, 80, 031501-1 3. Klingenberg G., Heimerl J.M., 1988, Gun Muzzle Blast and Flash, 139, AIAA Progress in Astronautics and Aeronautics edition 4. Leciejewski Z., SurmaZ.,Torecki S., Trębiński R., CzyżewskaM., 2013,Theoretical and experimental investigations of projectile motion specifity in the intermediate period, Proceedings of the 27th International Symposium on Ballistics, Freiburg, Germany, 333-343 5. Trębinski R., CzyżewskaM., 2015a,Approximatemodel of the intermediate ballistics,Central European Journal of Energetic Materials, 12, 1, 77-88 6. Trębiński R., Czyżewska M., 2015b, Estimation of projectile velocity increase in the interme- diate ballistics period,Central European Journal of Energetic Materials, 12, 1, 63-76 Manuscript received January 5, 2015; accepted for print December 31, 2016