See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/265972835

Numerical tools for the impact parameters identification of the 9mm

parabellum FMJ bullet.

Conference Paper · July 2010

CITATIONS READS

2 989

3 authors:

Christophe Marechal Francois Bresson

Polytechnic University of Hauts-de-France Institut National de Police Scientifique
44 PUBLICATIONS 289 CITATIONS 24 PUBLICATIONS 312 CITATIONS

SEE PROFILE SEE PROFILE

G. Haugou
Polytechnic University of Hauts-de-France
89 PUBLICATIONS 722 CITATIONS

SEE PROFILE

All content following this page was uploaded by G. Haugou on 25 November 2016.

The user has requested enhancement of the downloaded file.

 ENGINEERING TRANSACTIONS • Engng. Trans. • 59, 4, 263–272, 2011
Polish Academy of Sciences • Institute of Fundamental Technological Research (IPPT PAN)
National Engineering School of Metz (ENIM)

DEVELOPMENT OF A NUMERICAL MODEL OF THE 9 MM

PARABELLUM FMJ BULLET INCLUDING JACKET FAILURE

Ch. M a r é c h a l 1) , F. B r e s s o n 2) , G. H a u g o u 1)

1)
Université Lille Nord de France
F-59000 Lille, France
UVHC, LAMIH
F-59313 Valenciennes, France
CNRS, FRE 3304
F-59313 Valenciennes, France

2)
Institut National de Police Scientifique
Laboratoire de Police Scientifique de Lille
7 bd Vauban, 59000 Lille, France

Even though ballistic experiments are widely accepted as the only reliable way to probe
terminal effects, we demonstrate that computer simulation can be a useful alternative. Par-
ticularly, the high energy projectiles are seldom studied in the field of forensic sciences. That
situation being favorable to computer simulation, a 3D finite element model of the worldwide-
used 9 mm Parabellum bullet has been developed with Abaqus explicit software. A Johnson-
Cook constitutive model, fed with the split Hopkinson pressure bar experimental parameters,
accurately describes the materials’ behavior (lead and brass). Experiments were performed
with a handgun and a hard steel plate target in order to discuss the reliability of the model.
Accurate predictions about bullet deformation and failure were obtained without any post-
calculation adjustment of parameters.
Key words: ballistics, FEM, Johnson-Cook.

1. Introduction

Gunshot investigation is a key activity of any forensic science institute. The

shooter’s position is classically estimated from a post impact examination of
a scene (multiple impacts). The line passing by the center of each impact fig-
ure is materialized with a laser beam or a set of metallic rods. It can also be
calculated from an accurate 3D positioning of each impact. A correction of the
bullet’s path deflection (under the influence of its weight) can be considered in
some particular cases (long distance shooting). As a consequence, if a gunshot
involves only a single impact no information about the position of the shooter
can be given. The second limitation is the lack of knowledge about impact
264 CH. MARÉCHAL, F. BRESSON, G. HAUGOU

phenomena (deflection, velocity loss, bullet’s integrity and stability) which are
known qualitatively but not quantitatively. A specific procedure involving target
examination (gunshot residue scattering, smoke or multi-projectile dispersion,
the Griess test) is also applied to estimate the shooting distance (not the angle),
but besides the specific case of shotgun (multi-projectile), the efficiency range
of these methods is typically limited to a few meters.
In that context, it has already been demonstrated that comparing the bullet
deformation with experiments can bring up some information about the shoot-
ing distance, in the case of a hard target (metal, concrete [1]) or a soft target
(human body [2, 3]). The method used to determine the ballistic parameters
is the experimental way, the main reason being the lack of adequate computer
simulation of material deformation in the ballistic domain. The source of this
inadequetness can be found in the complexity of the problem itself, and also in
the low number of applications. The deformation state under ballistic impact
velocities (from 100 to 1000 ms−1 ) can only be described by rather complex and
multi-parameter laws. Identifying each parameter is a problem, the complexity
of which increases with the impact velocity. The ballistic domain stands between
the Lagrangian (solid) and Eulerian (fluid) formulation. Unifying these formu-
lations is still an issue. On the application side, military requirements (effects
of a given structure of projectile hitting a given structure of target) are quite
easy to investigate with experimental tools. In forensic applications (recovery
of pre-impact information from post impact examination), computer simulation
could be an alternative to experiments when a gunshot cannot be accurately
reproduced (excessive distance, accuracy or cost), but this demand was only
formulated lately.
In the present study, the worldwide- used 9 mm Parabellum FMJ (Full Metal
Jacket) bullet was selected as a representative sample. A 3D finite element model
has been developed with Abaqus explicit software. The constitutive model for
lead and brass is based on the Johnson-Cook relation without failure formula-
tion. The split Hopkinson pressure bar experimental data were used to identify
the missing parameters. The predictions were compared with ballistic experi-
ments performed with a semi-automatic Glock 26 compact pistol and homemade
ammunition with muzzle velocity ranging from 30 to 200 ms−1 .

2. Ballistic experiments

The thrower is a Glock 26 semi-automatic pistol. The target is a 30 mm

thick iron plate. The shooting distance is set to 2 m. Different velocities were
obtained by varying the amount of propellant (Vectan Ba 9) in home-made
9 mm Parabellum cartridges. The bullets are 9 mm Parabellum lead core FMJ
weighting 115 gr (grain), i.e., 7.45 g (actually ranging from 7.39 to 7.49 g in our
 DEVELOPMENT OF A NUMERICAL MODEL. . . 265

experiments). Impact velocities are measured half way between the handgun
and the target with a chronograph (Pro Chrono, Competition Electronics Inc,
USA). Very low velocities (down to 60 ms−1 ) were repeatedly obtained.
The way we chose to describe both the length reduction and diameter ex-
pansion was to define a deformation criteria C as follow:
v " #
u
u 1  ∆l 2  ∆d 2
(2.1) C=t · + ,
2 l0 d0

with ∆l = l − l0 and ∆d = d − d0 . l0 and d0 are respectively the length and

the diameter of the bullet before impact. The nature of C (single value) is more
suitable for comparison. This criteria can theoretically be higher than 1.
For each experiment, the length (l) and the diameter (d) of the bullet were
measured using a micrometer and finally C was calculated. The behavior of C
versus impact velocity is given in Fig. 1. The blue dots are the experimental
data; the red lines are the least square approximation in the two distinctive
areas. In the second area, the last two dots have not been considered for the
least square approximation as they demonstrate a saturation behavior slightly
below fragmentation.
Figure 1 demonstrates that C follows two rather linear behaviors, with
a velocity threshold triggering jacket fracture in-between. The so-called frac-

Fig. 1. Behavior of C versus impact velocity.

266 CH. MARÉCHAL, F. BRESSON, G. HAUGOU

ture threshold is 130 ms−1 in the experimented case. The slope of the curve is
5.6 · 10−3 sm−1 below the fracture threshold. Beyond the fracture threshold, the
slope of the curve is 6 times higher (3.4·10−2 sm−1 ) before reaching a saturation
behavior and then approaching the fragmentation threshold. A core fragmenta-
tion has been observed with an experimental impact velocity of 176 ms−1 . Core
and jacket fragmentation has been observed with an impact velocity slightly
higher (182 ms−1 ). This area cannot be described with C. With an impact ve-
locity of 203 ms−1 , a part of the core remained soldered to the target. Beyond
the fragmentation threshold, the size of the fragments decreases as the velocity
increases. We believe, however, that a very little quantitative prediction can be
gathered from a statistical fragment sizing.
Table 1 exhibits the shape of the deformation and eventually a fracture and
fragmentation of few bullets for impact velocities ranging from 60 to 200 ms−1 .

Table 1. Deformation shape, fracture and fragmentation of the projectile

for various impact velocities.

Impact Velocity 60 80 101 119 141 164 182 203

[ms−1 ]
C (rate) 0.14 0.24 0.35 0.46 1 1.46

The experimental fracture and fragmentation speed thresholds will increase

dramatically when the target cannot withstand ballistic load without deforma-
tion, as in most of the forensic cases. In that situation, any new target, new
material or new target shape actually need a new study, pointing out to the
usefulness of a FEM simulation tool.

3. FEM model

3.1. Materials models

This part describes the determination of the mechanical properties of the
quasi-pure lead extracted from the bullets by heating process (ρ = 11300 kg m−3
and ν = 0.42). The purpose is to determine the true plastic laws from quasi-
static to dynamic loadings under compression loading. Finally, parameters of the
constitutive simplified Johnson-Cook model are calculated without considering
heat transfer, as shown in Eq. (3.1) [4].
  
n ε̇
(3.1) σ = (σ0 + K · ε ) × 1 + D · ln ,
ε̇0
 DEVELOPMENT OF A NUMERICAL MODEL. . . 267

σ0 is the yield stress, K is the hardening parameter, n is the exponent for

the static law and D is the parameter which determine the viscosity effects.
ε̇0 describes the activation of the strain rate effects.
The experimental setup includes the cylindrical samples (∅ = 8 mm, l0 =
7 mm) submitted to the compression loadings from:
• A high-speed hydraulic machine (VHS) with a strain rates ranging from
0.15 to 20 s−1 (Fig. 2a),
• A set of nylon split Hopkinson pressure bars with a strain rate ranging
from 800 to 2000 s−1 (Fig. 2b).

a) b)

Fig. 2. a) VHS setup, b) Hopkinson bars setup.

Figure 3b illustrates the behaviour of the considered materials used for the
determination of the Johnson-Cook parameters. The materials responses are
determined from:
• strain gauges for Hopkinson tests,
• piezoelectric load cells coupled with electro-optical extensometers for VHS
tests.
The raw signals have been recorded using a numerical recorder at adaptive
sampling rates. For the Hopkinson calculations, the visco-elasticity and punching
corrections have been considered [5, 6].
The Johnson-Cook parameters have been determined after keeping plastic
relations at 0.15, 20800 and 1500 s−1 (dotted line in Fig. 3b). A classical calcu-
lation based on least squares interpolation has been used. The results are sum
up in Table 2.
Lead has revealed a visco-elasto-plastic behaviour under compression load-
ings, but the authors have considered only visco-plasticity as a first step.
268 CH. MARÉCHAL, F. BRESSON, G. HAUGOU

a) b)

Fig. 3. a) Initial and final shapes of specimens; b) behaviour laws under compression
loadings from 0.15 to 2000 s−1 .

Table 2. Johnson-Cook parameters for lead samples (R2 = 0.9891).

ESTATN EDYN σ0 K n D
0.41 GPa 1.3 GPa 5.15 MPa 35.35 MPa 0.5 0.0628

3.2. Numerical model

The numerical simulation was developed by Abaqus finite element model
software with the explicit numerical algorithm formulation. The projectile and
the target were designed using a 3D axi-symmetric solid model. The impact
has been considered as an axi-symmetric phenomenon by neglecting spinning
effects of the projectile (spinning energy represents only few percent of the total
kinetic energy). Furthermore, bullets didn’t undergo any twisted deformation in
the experiments. The main interest of the axi-symmetric simulation was to save
computation time without decreasing the accuracy.
The bullet was meshed in 4 nodes elements with reduced integration. The
projectile grid is composed of 6000 elements. The mechanical behaviour of the
projectile is the main issue of the simulation as the steel target has been consid-
ered as a pure elastic solid (E = 210000 MPa, ρ = 7800 kg m−3 , ν = 0.33). The
shape of the bullet was extracted from of picture of a cut-off bullet by a Matlab
routine. Figure 4 presents the finite element model of the bullet (b) and the
picture of the cut-off bullet (a).
The bullet design includes two materials: a lead core and a brass jacket
(a copper alloy also known as tambac, 90% Cu and 10% Zn). A Johnson-Cook
constitutive model has been used for each component. The parameters of lead are
given in Table 2. The parameters of tambac are extracted from the literature
[7, 8]. The model didn’t include fracture, but the fracture threshold criterion
 DEVELOPMENT OF A NUMERICAL MODEL. . . 269
a) b)

Fig. 4. Cut-off picture (a) and FEM (b) of the bullet.

proposed by Cockcroft and Latham [9] has been compared to the maximum
load observed at the experimental fracture threshold (130 ms−1 ). We also made
sure that the plastic work, Wp, level overlaps the critical value proposed by
Børvik et al. [10] for the brass jacket (Wp = 914 MPa).
Contact was considered without friction between the target and the bullet.
Sliding was allowed between the lead core and the brass jacket, according to the
experimental observations presented in Fig. 5.

Fig. 5. Sliding between lead and jacket.

3.3. Correlation with experimental investigation

The model was run with velocities ranging from 60 to 160 ms−1 . The displace-
ment and plastic work data were monitored from impact time (t) to t + 3 m/s
(until the bullet reaches its steady post-impact state). Bullet diameter and
length were extracted to calculate C for each simulation. Table 3 and the Fig. 6
show the results.
The numerical model and experimental results show a very good similarity
until an impact velocity ranging from 130 to 140 ms−1 occurs. The experimental
data exhibited fracture triggering at the same velocity. The fracture, allowing
stress relaxation in the jacket, can not be considered in the model (not allowed in
the axi-symetric configuration). Such a structural weakening increases crushing
and the criterion C. The lack of stress relaxation in the model explains the
270 CH. MARÉCHAL, F. BRESSON, G. HAUGOU

Table 3. Experiment and numerical results.

Experiments Numerical
Velocity Diameter Length Velocity Diameter Length WpMax
C C
[ms−1 ] [mm] [mm] [ms−1 ] [mm] [mm] [MPa]
60.3 9.46 11.23 0.14 60 9.08 11.67 0.15 478
80.2 10.8 10.10 0.24 80 10 10.18 0.21 484
89.9 11.35 9.73 0.29 100 11.3 8.7 0.32 447
100.9 12.1 9.12 0.35 120 12.36 7.62 0.42 614
113.4 13.05 8.45 0.43 130 12.9 7.08 0.47 687
118.6 13.4 8.08 0.46 140 13.54 6.5 0.52 2380
130.1 14.2 7.47 0.53 160 14.13 6.03 0.57 5586
141.1 20 4.3 1
150.9 23.50 3.22 1.27

Fig. 6. Experiment and numerical behavior of C versus impact velocity.

difference between the experimental and the numerical results beyond 130 ms−1 .
As explained above, the plastic work level has been used as an evidence to point
out the emergence of fractures in the model. Figure 7 shows the evolution of
the maximum plastic work of the jacket versus an impact velocity. The Børvik
threshold criterion was overlapped with an impact velocity of 130 ms−1 . The
non-linear increase of the plastic work is not realistic because of the enactment
 DEVELOPMENT OF A NUMERICAL MODEL. . . 271

of element in the model when the fracture phenomena appear. Modifications

of the loading mode were observed in the model between 130 and 140 ms−1 .
The jacket, which was in compression until 130 ms−1 , changed into a buckling
mode. This change explains the fracture’s triggering and the very important
plastic work increase (according to a plastic strain increase). The agreement
with experimental data is also very satisfactory.

Fig. 7. Numerical results for maximum plastic work versus impact velocity.

Extra simulations (not reported here) demonstrated that a particular care

must be taken about the jacket thickness variations. A maximum thickness is
clearly visible near but off the tip (c.f., Fig. 4a). The approximation of a uniform
thickness or even a slightly wrong shape induces a dramatic discordance with
the experimental results.

4. Conclusions

This paper demonstrates the good correspondence between the FEM model
using explicit formulation and the experiment in the ballistic field. The sine qua
non condition for such a good behavior of the model was to provide accurate
parameters for the Johnson-Cook constitutive model of each component. The
parameters were extracted from the literature for tambac, from the Hopkinson
bar experiments for the lead, and no extra adjustment were made. In that drastic
situation, a very good concordance has been observed as far as the dynamic load
did not exceed the fracture threshold.
For the explored velocity range, it has not been necessary to consider spin and
heat transfer, but the results pointed out that a very good care about the jacket
thickness is necessary to accuratelly reproduce the shape of the experimental
deformations. Ongoing researches will include fracture and heat transfer in order
to minimize the limitation of the Lagrangian formulation in the ballistic domain.
 272 CH. MARÉCHAL, F. BRESSON, G. HAUGOU

Acknowledgment

The authors would like to thank D. Lesueur for his helpful hand. The present
research work has been supported by International Campus on Safety and In-
termodality in Transportation- the Nord-Pas-de-Calais Region, the European
Community, the Regional Delegation for Research and Technology, the Min-
istry of Higher Education and Research, and the National Center for Scientific
Research. The authors gratefully acknowledge the support of these institutions.

References
1. F. Bresson, O. Franck, Estimating the shooting distance of a 9 mm Parabellum bullet
via ballistic experiment, Forensic Science International, 192, 1–3, 17–20, 2009.
2. K. G. Sellier, B. P. Kneubuehl, Wound ballistic and the scientific background, Elsevier
1994.
3. F. Bresson, O. Franck, Comparing ballistic wounds with experiments on body simulator,
Forensic Science International, 198, 1–3, 23–27, 2010.
4. G. R. Johnson, W. H. Cook, A constitutive model and data for metals subjected to
large strains, high strain rates and high temperatures, Proceedings of seventh international
symposium on ballistics, The Hague, the Netherlands, 1983.
5. G. Gary, V. Degreef, David, Labview Version, Users manual version, LMS Polytech-
nique, Palaiseau, 2008.
6. K. Safa, G. Gary, Displacement correction for punching at a dynamically loaded bar
end, International Journal of Impact Engineering, 37, 4, 371–384, 2010.
7. S. Chocron, C. E. Andersen Jr, D. J. Grosch, C. H. Popelar, Impact of the 7.62
mm APM2 projectile against the edge of a metallic target, International Journal of Impact
Engineering, 25, 23–37, 2001.
8. C. E. Andersen Jr, M. S. Burkins, J. D. Walker, W. A. Gooch, Time-resolved pene-
tration of B4C tiles by the APM2 bullet, Computer Modeling in Engineering & Sciences,
8, 2, 91–104, 2005.
9. M. G. Cockroft, D. J. Latham, Ductility and the workability of metals, Journal of the
Institute of Metals, 96, 33–39, 1968.
10. T. Børvik, S. Dey, A. H. Clausen, Perforation resistance of five different high-strength
steel plates subjected to small-arms projectiles, International Journal of Impact Engineer-
ing, 36, 7, 948–964, 2009.

Received March 29, 2011; revised version July 21, 2011.

View publication stats