Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: We study experimentally and computationally the penetration of a steel sphere into a block of ballistic
Received 27 December 2012 gelatin for developing an improved understanding of the damage caused to human soft tissues when
Received in revised form impacted by a blunt object moving at a moderately high speed. The gelatin is modeled as an isotropic and
20 May 2013
homogeneous elasticeplastic material that exhibits linear strain-hardening and obeys a polynomial
Accepted 4 July 2013
Available online 13 July 2013
equation of state. Pictures taken by a high speed camera help construct the tunnel formed in the gelatin
that is found to compare well with the computed one. Furthermore, computed time histories of the
pressure at a point agree well with the corresponding experimentally measured ones for small times. The
Keywords:
Ballistic gelatin
computed time histories of the temporary cavity size agree well with the corresponding experimental
Temporary cavity ones. These agreements between test findings and computed results imply that the computational model
Penetration can reasonably well predict significant features of the impact event. Effects of impact velocity and sphere
Pressure wave diameter on damage caused to the gelatin have also been studied.
Ó 2013 Elsevier Ltd. All rights reserved.
0734-743X/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijimpeng.2013.07.002
Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151 143
Us ¼ 1:52 þ 2vp (1) Using water as proxy material for soft tissue, An et al. [17] used
the commercial finite element (FE) software LS-DYNA to simulate
for the 10% gelatin. In Eq. (1) Us and vp are the shock and the particle the evolution of a temporary cavity and the pressure developed
speed, respectively. Appleby-Thomas et al. [14] also employed during high velocity impact of a rigid sphere into a body of water.
plate-impact experiments to study the dynamic response of 25% Dyckmans et al. [18] measured material parameters of ballistic soap
gelatin, ballistic soap and lard. All three materials exhibited linear using the split Hopkinson pressure bar (SHPB), and simulated the
Hugoniot Us vp relations. Whereas the gelatin behaved hydro- impact of steel spheres in a soap block using the commercial soft-
dynamically under shock, the soap and the lard appeared to ware AUTODYN.
strengthen under increased loading. Koene and Papy [19] used AUTODYN to study deformations
Mechanisms dominating deformations of solids during impact induced by the penetration of ABS (AcrylonitrileeButadieneeSty-
generally vary with the impact speed. Wilbeck [15] has classified rene) plastic spheres into gelatin at velocities up to 160 m/s. For
deformations of some low strength materials (birds, gelatin and simulating tests of armor impacting gelatin, Shen et al. [20] modeled
RTV rubber) in five regions: elastic, plastic, hydrodynamic, sonic gelatin as nearly incompressible rubber. Cronin employed a visco-
and explosive. There is no single constitutive relation for gelatin elastic material model [21], and a rate-dependent hyperelasticity
that can well describe its mechanical behavior in all these five re- model using tabulated values of stresses and strains [22] to simulate
gimes. For low velocity impact a rate-dependent hyperelastic the mechanical behavior of gelatin. It was found that the viscoelastic
constitutive model is expected to describe well the mechanical material model could adequately capture only the low strain-rate
behavior of gelatin. However, for high velocity impact the hydro- response of gelatin, and the tabulated hyperelasticity model pro-
dynamic response that considers possible phase transformations vided an accurate representation of the gelatin at low and inter-
may be more suitable [16] at least in the initial phase of the mediate strain rates. However, high strain rates of the order of 1000/
penetration event. The strength of the gelatin may play a role once s were not considered. Minisi [23] simulated the projectile-gelatin
the penetrator has considerably slowed down. In the absence of test interaction at high impact velocities with the 10% gelatin modeled
data for characterizing the material response at high strain rates as a hydrodynamic material with the MieeGrüneisen equation of
and temperatures, we adopt an elasticeplastic model for the state at high strain rates and a MooneyeRivlin material at low strain
gelatin and use a polynomial equation of state (EoS) to describe its rates. However, when to transition between the two material
hydrodynamic response. It is hard to quantify the improvement in models and values of material parameters are not listed in the
the computed results by considering strain-rate and thermal soft- report.
ening effects at this time. Here we experimentally and numerically study deformations
induced by a steel sphere moving at a moderate speed and
impacting at normal incidence a rectangular gelatin block. The steel
is modeled as a rigid material and the gelatin as an elasticeplastic
material with linear strain-hardening and a cubic polynomial
relation between the hydrodynamic pressure and the change in
Fig. 2. Photo of the experimental set-up. Fig. 3. Location of the pressure sensor embedded in the gelatin block.
144 Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151
Diameter Impact Kinetic Maximum Time of Perforation 3.1. Material model and verification
(mm) speed energy (J) temporary maximum
(m/s) cavity temporary We used the commercial FE software, LS-DYNA, to simulate the
diameter (mm) cavity (ms)
impact experiment described above. The gelatin was modeled as an
4.8 721 118 62.1 2.1 No elasticeplastic material with the polynomial EoS (Eq. (5)) and the
4.8 728 120 63.7 2.2 No
yield strength sy given by [24]
4.8 731 121 63.9 2.2 Yes
4.8 947 203 76.2 2.7 Yes
4 717 67 48.7 1.7 No sy ¼ s0 þ Eh 3 p (2)
3 659 24 37.5 1.4 No
wheres0 is the initial yield strength, 3
p
the effective plastic strain,
3 5
Experiment 4 Experiment
Numeric Numeric
2
3
Pressure (MPa)
Pressure (MPa)
2
1
1
0 0
-1
-1
0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25
Time (ms) Time (ms)
Fig. 4. Time histories of the computed and the experimental pressures at the point (0, 50 mm, 50 mm) for the 4.8 mm diameter sphere impacting at normal incidence the gelatin
block at 728 m/s (left) and 947 m/s (right).
Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151 145
Fig. 5. Computed (left) and experimentally (right) observed temporary cavity profiles for impact velocity of 728 m/s (top) and 947 m/s (bottom).
Table 2 Table 3
Values of material parameters for the gelatin. Values of material parameters for the steel sphere.
r (kg/m3) E (kPa) Et (kPa) s0 (kPa) C0 (GPa) C1 (GPa) C2 (GPa) C3 (GPa) Mass density (kg/m3) Young’s modulus (GPa) Poisson’s ratio
Fig. 6. Discretization of the sphere and the gelatin block into 8-node brick elements. (a) Sectioned view of steel sphere, (b) FE grids on surfaces of the gelatin block, and enlarged
view of the FE mesh in the impacted area.
for quasistatic uniaxial compression tests and finding the secant Using the finest FE mesh, it can be concluded from values
modulus for the point (296 kPa, 0.349). The value of Et is listed in Table 4 that for critical values of the effective plastic
estimated. strain equal to 0.7, 0.9 and 1.1, the computed penetration depths
It was noticed that the impacting spheres underwent very little and the maximum cavity diameters differed from each other by
deformations during the penetration process. We thus assume the less than 3% and 12%, respectively. Subsequently, we use 0.9 as
sphere to be rigid during numerical simulations and use for it the the critical value of the effective plastic strain. We note that
material model (MAT_RIGID) in LS-DYNA. Even though values of Cronin and Falzon [7] analyzed an axisymmetric problem and
material parameters for the steel are listed in Table 3, only the value found the optimized value of the erosion strain to be between
of mass density is needed for the numerical work. 0.738 and 0.755 which increased with a decrease in the mesh
The 3-dimensional (3-D) geometries of the sphere and the size. Assuming that all of the plastic working is converted into
gelatin were discretized into 1048 and 595,200 8-node brick el- heating and deformations are locally adiabatic, temperature rise
ements, respectively. One such discretization is exhibited in Fig. 6. in an element just before it is deleted equals approximately
The gelatin has small elements (0.31 mm 0.31 mm 0.6 mm) in 45 C. We realize that neglecting thermal softening and strain-
the cylindrical region encompassing the impacted area, and the rate dependence in the material model makes the analysis
element size increases as one moves away from this zone. The approximate. However, we neither could find test data in the
element size in the sphere is less critical for computing the open literature nor we could generate it ourselves to determine
penetration depth provided that the sphere geometry can be values of material parameters for quantifying these effects. There
adequately reproduced. Two other FE meshes, shown in Fig. 7, may be phase transformations induced because of the tempera-
with element sizes of 0.63 mm 0.63 mm 0.6 mm and ture rise and accounting for the latent heats of phase trans-
0.42 mm 0.42 mm 0.6 mm were used to study the depen- formations will complicate the analysis. Effects of phase
dence of computed results upon the FE mesh used. Results transformations have been considered by Zhu and Batra [27]. In a
computed with the three discretization of the gelatin region commercial code these effects are usually incorporated in the
should help delineate the effect of the FE mesh size on the EoS. However, no such EoS is available for the gelatin in the open
computed results. literature.
The ERODING_SURFACE_TO_SURFACE contact definition was For the impact speed of 728 m/s, time histories of the numeri-
used to simulate the interaction between the sphere and the cally computed penetration depths with the three FE meshes are
gelatin. The viscous hourglass control algorithm with hourglass compared with the corresponding experimental one in Fig. 8.
coefficient ¼ 0.01 was employed. All bounding surfaces of the Penetration depths computed with the coarse mesh have large
gelatin block and the steel sphere except those contacting each deviations from those found experimentally. The difference be-
other are taken to be traction free. The contacting surfaces are taken tween the computed and the test values of penetration depth de-
to be smooth. The block is assumed to be initially at rest and stress creases with a decrease in the size of smallest element in the FE
free. An element of the gelatin is assumed to fail when the effective mesh for the gelatin. At t ¼ 2 ms, the penetration depth with the
plastic strain in it equals a critical value, and the failed element is finest FE mesh differs from the test value by less than 7%. Results
deleted from the analysis domain. presented below are with the finest FE mesh.
Fig. 7. Partial enlarged views of the coarse (left), medium (center) and fine (right) FE mesh for the gelatin.
Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151 147
Table 4 350
For the impact velocity of 728 m/s, computed penetration depth and the maximum
cavity diameter for three values of the critical effective plastic strain. v947 experiment
300
v947 numeric
Critical value of the Penetration Maximum cavity v728 experiment
effective plastic strain depth (mm) diameter (mm)
100
3.2. Numerical results and discussion
50
A phenomenon of interest in the penetration of gelatin is the
formation of the temporary cavity. The kinetic energy of the sphere
0
transferred to the gelatin accelerates the medium surrounding the 0.0 0.5 1.0 1.5 2.0
path of the sphere and moves gelatin away from the sphere both Time (ms)
radially and axially thereby creating a tunnel, called temporary
cavity, behind the sphere. As should be clear from time histories, Fig. 9. Time histories of the computed and experimentally determined penetration
exhibited in Fig. 9, of the computed and the experimental depths of depth.
penetration for impact speeds of 728 and 947 ms/, the numerical
results agree well with the experimental data and the computed dissipation than that evidenced by the computed results. The dif-
penetration depth is less than that measured experimentally by at ference between the numerical and the experimental results for
most 10%. later times is more for the 947 m/s impact speed than that for the
For the two impact speeds, Fig. 5 exhibits the numerically 728 m/s impact speed. It seems that a strain-rate dependent ma-
computed and experimentally observed cavity profiles at 0.3, 1 terial model for the gelatin may be more appropriate for capturing
and either 2.2 or 2.7 ms after impact. For the impact speed of 728 this dissipation in the gelatin.
(947) m/s, the numerically computed maximum diameter of 63.7 For the impact speed of 728 m/s, results plotted in Fig. 4 reveal that
(76.2) mm compares well with the experimental value of 60.3 the first three wave peaks with pressures of nearly 2.54, 1.8 and
(73.6) mm with the difference between the computed and the 1.6 MPa occur at the gage location. The second and subsequent peaks
measured values being 5.5 (3.5)%. The temporary cavity looks like are caused by the interaction between the incident wave and waves
a slender cone that with the passage of time expands in both reflected from boundaries of the gelatin including from the free
radial and axial directions. After reaching the maximum size it surface of the cavity formed in the wake of the sphere. Computed
begins to collapse as elastic deformations of the gelatin are pressure profiles at 0.1 and 0.2 ms after impact are exhibited in Fig.10.
recovered.
Upon impact a very high pressure is generated in the region
around the impacted face that propagates both radially and axially.
This initial phase of the pressure pulse can be divided into two
parts: penetration shock wave and pressure fluctuations. For the
two impact speeds, the time histories exhibited in Fig. 4 of the
computed and the experimentally measured pressure at the loca-
tion (0, 50 mm, 50 mm) of the pressure gauge in the gelatin suggest
that initially the two sets of results are very close to each other.
However, after 0.05 ms the experimental results show more
350
Experiment
300
Fine mesh
Medium mesh
Penetration depth (mm)
200
150
100
50
0
0.0 0.5 1.0 1.5 2.0
Time (ms)
Fig. 8. Comparison of computed and experimental time histories of the penetration Fig. 10. Contours of pressure (105 MPa) in gelatin at 0.1 (top) and 0.2 ms (bottom) after
depths. impact (contour level 2.6-05 means pressure of 2.6 MPa).
148 Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151
Fig. 11. At t ¼ 0.5, 2, 3.5 and 5 ms after impact, contours of effective plastic strain in the impacted face (left) and of effective stress in 105 MPa (right) in the gelatin around the
temporary cavity.
We have displayed in Fig. 11 contours of the effective plastic stored energy in the gelatin has been dissipated. However, we could
strain and the effective stress in the gelatin. The size of the plasti- not numerically reproduce this pulsation of the cavity surface due to
cally deformed region increases around the cavity surface irre- the use of the Lagrangian formulation of the problem, and not being
spective of whether the cavity expands or contracts as the kinetic able to adaptively refine the FE mesh in the gelatin.
energy of the gelatin is converted into the plastic energy of defor-
mation. About 0.2% of the gelatin mass was lost due to erosion of 3.3. Effect of impact parameters of steel sphere
elements.
Pictures taken by the high-speed camera displayed in Fig. 12 The numerical results computed with three different sphere
reveal that for the 728 m/s impact speed the cavity expands and diameters and several different impact velocities are listed in
contracts about seven to eight times until all of the kinetic and the Table 5. Three simulations labeled A, B and C in Table 5 employed
Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151 149
Table 5
Numerical results with different impact velocities and sphere diameters.
No. Diameter of Impact Kinetic Maximum temporary Time of temporary Maximum penetration Initial expansion velocity
steel sphere (mm) velocity (m/s) energy (J) cavity size (mm) cavity (ms) resistance (N) of temporary cavity (m/s)
the same sphere that has different initial velocities and hence The temporary cavity profiles at t ¼ 0.1, 0.5, 1.5, 3.0 and 4.5 ms
different initial kinetic energies. Computed results show that the for different diameters of the impacting sphere but having the same
maximum temporary cavity size, maximum impact resistance and initial kinetic energy (simulations C, F and G) impacting gelatin are
the initial expansion velocity of the cavity monotonically decrease compared in Fig. 14. The 4.8 (3) mm sphere produces the maximum
with a decrease in the initial kinetic energy of the impacting sphere. (minimum) penetration depth although its impact velocity is the
Taking results for simulation B as the reference, þ20% and 20% lowest of the three spheres. It suggests that the sphere mass is the
change in the impact speed alters, respectively, the maximum primary factor that determines the penetration depth.
penetration resistance by þ40% and 22% and the maximum
temporary cavity size by þ13% and 12%.
In simulations C, D and E, the impact speed is kept constant at
800 m/s but the sphere diameter is varied. Taking results for 100
simulation D as the reference, þ20% and 25% change in the sphere
diameter affects, respectively, the maximum penetration resistance Linear fit of experiment results
80
by þ42% and 45% and the maximum temporary cavity size
by þ20% and 30%.
Simulations C, F and G have the same initial kinetic energy of the 60 y=0.221x+34.63
sphere. The maximum temporary cavity sizes for the three simu-
R =0.956
lations are within 12% of each other. The maximum penetration
resistances for simulations C and F are close to each other, whereas 40
for simulation G it is considerably higher than that for the other two
simulations. 20
In Fig. 13 we have plotted the variation of the maximum tem-
porary cavity size versus the initial kinetic energy of the impacting
sphere. It is clear that the maximum cavity diameter is an affine 0
0 50 100 150 200 250 300 350
function of the initial kinetic energy of the sphere at least in the
range of parameters studied. The data for the initial kinetic energy
of the impacting sphere between 25 J and 250 J lie on the straight Fig. 13. Maximum temporary cavity size versus the initial kinetic energy of the
line whose equation is listed in the figure. impacting sphere.
150 Y. Wen et al. / International Journal of Impact Engineering 62 (2013) 142e151
Fig. 14. Comparison of temporary cavity profiles for three spheres of different diameters but having the same initial kinetic energy (the first number in 4.8v800 stands for the
sphere diameter in mm, and v800 implies sphere velocity ¼ 800 m/s).
[16] Johnson AF, Holzapfel M. Numerical prediction of damage in composite [21] Cronin DS. Ballistic gelatin characterization and constitutive modeling. In:
structures from soft body impacts. J Mater Sci 2006;41:6622e30. 2011 annual conference on experimental and applied mechanics 2011. p. 51e
[17] An B, Jiang JW, Wei JJ, Zheng JH. The numerical simulation of the temporary 5. Connecticut, USA.
cavity formed during the high velocity steel ball penetrating into water me- [22] Cronin DS, Falzon C. Dynamic characterization and simulation of ballistic
dium. Explosion and Shock Waves 1998;18:245e50. gelatin. In: SEM annual conference and exposition on experimental and
[18] Dyckmans G, Ndompetelo N, Chabotier A. Numerical and experimental study applied mechanics 2009. p. 856e61. Albuquerque, USA.
of the impact of small caliber projectiles on ballistic soap. In: 7th international [23] Minisi MD. Gelatin impact modeling. PM-MAS ES-1A-9000. www.dtic.mil/
conference on mechanical and physical behaviour of materials under dynamic ndia/2006 smallarms/minisi.pdf; 2006.
loading 2003. p. 627e32. Porto, Portugal. [24] Hallquist JO. LS-DYNA keyword user’s manual. Version 971. Livermore: Liv-
[19] Koene L, Papy A. Experimental and numerical study of the impact ermore Software Technology Corporation, LSTC; 2012.
of spherical projectiles on ballistic gelatin at velocities up to 160m/s. In: [25] Wang ZL, Konietzky H, Huang RY. Elasticeplastic-hydrodynamic analysis of crater
25th international symposium on ballistics 2010. p. 1573e9. Beijing, blasting in steel fiber reinforced concrete. Theor Appl Fract Mech 2009;52:111e6.
China. [26] Johnson AF, Holzapfel M. Modelling soft body impact on composite structures.
[20] Shen WX, Niu YQ, Bykanova L, Laurence P, Link N. Characterizing the inter- Compos Struct 2003;61:103e13.
action among bullet, body armor, and human and surrogate targets. J Biomech [27] Zhu ZG, Batra RC. Consideration of phase transformations in the study of shear
Eng 2010;132:1e11. bands in a dynamically loaded steel block. J Eng’g Mat’l Tech 1992;114:368e77.