12037

NUMERICAL MODELING OF ROCKET WARHEAD DETONATION

AND FRAGMENTATION

Joseph D. Baum1, G. Daniel Williams1, G. Daniel Williams ,Orlando A. Soto1, Fumiya Togashi,
and Rainald Lhner2,
1
SAIC, 1710 SAIC Drive, McLean, VA 22102, USA
2
George Mason University, 4400 University Dr, Fairfax, VA 22030, USA

ABSTRACT
Numerical models provide a safer and more cost-effective method to evaluate the performance of military warheads
and weapons systems. These simulations are increasingly employed for military research and development. The
scientific community must be confident in the accuracy of these computations, and thus analysts need to continually
validate numerical models. This paper presents results from a numerical study of the detonation and fragmentation
of a rocket warhead commonly found throughout the world. The simulation utilizes a coupled nonlinear finite
element structural and fluid dynamics codes to model the detonation of the high explosive, the case break up, and
the mass, velocity, and trajectory of the resulting fragments. The analytic research program focuses on differing
impact angles and velocities, and an assessment is provided of the ability to numerically represent the behavior of
the rocket using simplifying assumptions.

.INTRODUCTION
Rockets are frequently employed throughout the world in areas of military conflict and armed
internal resistance. While they come in various shapes, sizes, and material composition, they all
pose serious lethal threats to both civilians and military personnel. Thus, military research
engineers and scientists need to accurately assess the lethality of these weapon systems to design
countermeasures to protect human life. While experimental testing is often employed for such
purposes, numerical simulations provide a cost-effective alternative. Modern finite element
programs are able to accurately characterize shock propagation and material response; however,
inaccurate material models can generate erroneous solutions without proper validation, and
analysts can select inappropriate assumptions that yield misguided conclusions.
Accordingly, researchers should know the adequacy of both the selected numerical approach and
the assumptions employed. This paper addresses the influence of two common assumptions
implicit in modeling any rocket or warhead system. The first is the inclusion of rocket velocity
prior to detonation. Fragmentation data for both stationary and moving rocket models are
presented. This comparison will examine the relative influence of weapon velocity on the
fragmentation and the need to include weapon velocity in numerical models. The second
assumption examined in this paper is the inclusion of rocket impact angle at the time of
detonation. Results from rocket models oriented at 0 degrees, 50 degree and 90 degrees are
presented to examine the differing fragmentation patterns and to determine the necessity of
modeling differing impact angles. The discussion and results presented in this paper will provide
military researchers with valuable insight into the most effective and expeditious methods to
model rockets and warheads.
 12037

Numerical Models
Over the last several years we have developed a numerical methodology that couples state-of-
the-art Computational Fluid Dynamics (CFD) and Computational Structural Dynamics (CSD)
methodologies [1,2]. FEFLO98 is the CFD code used while SAICSD handles the CSD portion.
FEFLO98 solves the time-dependent, compressible Euler and Reynolds-Averaged Navier-Stokes
equations on an unstructured mesh of tetrahedral elements. SAICSD solves explicitly the large
deformation, large strain formulation equations on an unstructured grid composed of bricks and
hexahedral elements.
Mesh generation for both CSD and CFD is performed using FRGEN3D [3], an unstructured grid
generator based on the advancing front method. The CFD mesh is composed of triangular
(surface) and tetrahedral (volume) elements. The CSD mesh includes beams, triangular or quad
shells and bricks for the solid
The flow solver employed is FEFLO, a 3-D adaptive, unstructured, edge-based hydro-solver that
-solves the Arbitrary Lagrangean-Eulerian (ALE) formulation of the Euler and Reynolds-
averaged turbulent, Navier-Stokes equations. The code includes a large variety of state-of-the art
numerical shock-capturing schemes, from FCT, exact or approximate Rieman, to ENO to HLLC,
and from second-order to eight-order accuracy, a choice that is continuously updated as new
schemes are developed. The spatial mesh adaptation is based on local H-refinement, where the
refinement/deletion criterion is a modified H2-seminorm [4] based on a user-specified criteria.
FEFLO supports various equations of state including real air, water, SESAME and JWL with
afterburning. Particles are treated as a solid phase, exchanging mass, momentum and energy with
the fluid.
The structural dynamics solver used was SAICSD [5,6]. This code solves the continuous
mechanics equilibrium equation. The weak formulation (virtual work principle) is written in the
spatial configuration (actual configuration) and it is discretized in time using an explicit second-
order central difference scheme. In space, the virtual work equation is solved by using stable
finite element types. The most used elements are: a full integrated large-deformation Q1/P0 solid
element (hexahedra with an 8 nodes interpolation scheme for the cinematic variables and
constant pressure) which does not present hourglass modes and it does not lock for
incompressible cases. Several 3-node and 4-node large-deformation shell elements (Hughes-Liu
shell, Belytschko shells, MITC shells, ASGS stabilized shells) which are formulated using
standard objective stress update schemes (Jaumann-Zaremba, co-rotational embedded axis, etc,),
are fully integrated to avoid hourglass spurious modes. Finally, some objective truss and beam
elements (i.e. Belytschko and Hughes-Liu beams) have also been implemented. Many different
material models have been included into the code. The most used are: a plasticity model which
relies on a hyper-elastic characterization of the elastic material response for the solid elements,
and a standard hypo-elastic plasticity model for the shell, beam and truss elements. The most
often used failure criterion is based on the maximum effective plastic strain and the stress tensor
inside the element. The fracture may be simulated by element erosion and/or node disconnection.
RESULTS
Weapon Details
The specific rocket dimensions and velocities are excluded for security purposes. Additionally,
the size, mass, and velocity of the resulting fragmentation patterns are also excluded. The
resulting fragmentation is discussed in terms of dimensionless variables. The rocket mass is
 12037

defined as m, and all fragment velocities for both models (stationary and moving) are specified
relative to v for comparison purposes. The rocket contains the fragmenting warhead filled with
explosive, as well as the attached, thick-case solid rocket motor. A thick steel plate separated the
explosive warhead from the motor.
Numerical simulations
Five simulations were conducted. The base configuration included the weapon placed
(stationary) at an angle of 50o off the horizon, with the tip several centimeters off the ground.
The base explosive is defined as HE-A. The second simulation replaced HE-A with a more
energetic explosive, termed HE-B. The third simulation modeled the base configuration, but
moving the weapon at a velocity v. Simulation four had the base case weapon and explosive
placed horizontally (parallel to the ground), while the last simulation placed that weapon vertical
to the ground.
The explosive was nose detonated. A sequence of contour plots shown in Fig 1 depicts the
detonation wave propagation within the explosive, the case fragmentation, the solid motor case
deformation and the expanding blast wave and fragment cloud. The simulation was conducted
for the base configuration, though all other simulations are controlled by the identical physical
mechanisms. Column A shows pressure contours on a cut plane of symmetry and on the ground
plane. Column B shows the velocity contours on the plane cut and on the ground. Column C
adds the CSD surface on top of the velocity contours, while column D shows the CSD velocity
contours.
Upon detonation initiation at the nose, the detonation (point initiated) expands within the HE.
Figures 1 at 40s show the expanding detonation front within the explosive, and the expanding
case (reaching maximum case velocity of 0.6vkm/sec). The case cracked soon afterwards, and at
100s (Fig 2) observe the detonation products escape through the opening cracks, impact on the
floor, and fragment expansion at a maximum velocity of about 0.88vkm/sec. At this time we
observe the initial deformation of the steel plate separating the explosive from the rocket motor.
This plate failed soon afterwards, and at 140s (Fig 3) we observe the detonation products
expansion through the motor case. Further detonation products and fragment expansion, as well
as the thick-cased solid motor expansion, are observed in Figs 4 and 5, at 200s, and 300s,
respectively.
Detonation products expansion through the breaking case is a complex phenomenon.
Examination of the pressure and velocity evolution in columns A, B and C shows that as the case
expands and breaks, the pressure within the volume enclosed by the expanding fragments is
significantly higher than outside (Figs 2a through 4a), and is still noted even at a relatively very
late time (Fig 5a). The solution at 200s shows that the pressure inside the expanding fragment
cloud that has expanded about 25 times the initial volume, is about 40 times higher than outside,
while the velocity through the openings cracks accelerates from local subsonic to supersonic.
This complex phenomenon clearly demonstrates the importance of properly modeling the
coupled case breakup-fragmentation and detonation products expansion, which becomes all the
more critical for non-ideal explosives.
Detonation wave evolution and interaction with nearby structures are discussed below. It is
possible to construct many scenarios. We chose to examine fragment and blast loading on a
circular wall at a given stand-off distance. We are focused only on the initial blast loading, i.e.,
maximum pressure and impulse, before reverberating waves within the enclose come into play.
 12037

Pmax=1.8E11
s
Fig 1: 40 Vmax=0.94vkm/sec Vmax=0.92vkm/sec Vmax=0.69vkm/sec

Pmax=1.22E10
s
Fig 2: 100 Vmax=1.31vkm/sec Vmax=1.31vkm/sec Vmax=0.92vkm/sec

Pmax=2.56E9

s
Fig 3: 140 Vmax=1.36vkm/sec Vmax=1.36vkm/sec Vmax=1.13vkm/sec

Pmax=1.1E09

s
Fig 4: 200 Vmax=1.68vkm/sec Vmax=1.68vkm/sec Vmax=1.0vkm/sec

Pmax=3.85E08

s
Fig 5: 300 Vmax=1.56vkm/sec Vmax=1.56vkm/sec Vmax=1.0vkm/sec
A. Pressure contours, B. Velocity contours C. Velocity contours + D. CSD velocity
Dyne/cm**2 CSD surface
Figures 1 through 5. Detonation initiation and propagation, case fragmentation and blast
wave evolution at 40s, 100s, 140s, 200s and 300s, respectively. Each showing
pressure contours on a planar cut and the floor, velocity (cut and floor) without and with the
CSD imposed, and CSD fragment velocity.
 12037

Figures 6 through 10 show the velocity evolution on the plane of symmetry for the five
simulations, at 1.0ms, 3.0ms and 5.0ms, respectively. All velocity contours on the same column
(same time) are plotted using the same scale. Hence, the white zone, such as inside the
expanding flow shown in Fig 7a for HE-B, indicates a velocity higher than the maximum value.
The three simulations for the rocket at 500, include HE-A (Fig 6), HE-B (Fig 7), and the moving
HE-A (Fig 8) show essentially the same pattern of blast wave expansion, with a higher
expanding core velocity for the more energetic HE-B, and a slightly more directional velocity for
the moving rocket, as the rocket flight velocity is significantly lower than detonation velocity.
The horizontally placed weapon shows the directed jetting through the slowly-expanding motor
case, while the vertical detonation shows a perfectly symmetric solution.
Next we examined the blast loading on the circumference of this enclosure, by dividing the 1800
segment to four equal surfaces of 450 each, termed surfaces 9, 10, 11 and 12. Hence, surface 9
expands from 00 to 450, where 00 deg is the tail direction (on a plane view), while surface 12
covers the zone from 1350 to 1800 (nose direction). The pressure and impulse values were
obtained by averaging, at any time step, the values observed on all points within that surface.
While this average processing diminish rise and decay time, peak and minima values, it is an
excellent indication for what a structure placed within this zone will observe. Figure 11 shows a
comparison on each surface for the five configuration modeled, while figure 12 shows, for each
configuration, a comparison on the different surfaces. We note that:
a. Pressure and impulse values at the two central surfaces (10 and 11) for HE-B are highest,
as expected for the more energetic explosive;
b. As HE-A and HE-B have the same initial conditions, the solutions on all surfaces relate
similarly, i.e., identical trends, shifted up for the more energetic HE-B. For these two
explosives the highest impulse values were obtained for the side surfaces, between 450
and 1350, while both produced the highest peak pressures on surface 12;
c. Moving the rocket resulted in shifting of energy from the tail (surface 9) to the nose
(surface 12), where it produced the highest impulse;
d. The vertical-placed rocket loading showed no preferential direction;
e. The horizontal-placed rocket produced results identical to those expected from an arena
test: maximum pressure and impulse values at the 90 to 135 quadrant (surface 11);
Another way of viewing the pressure and impulse values exerted on a given surface is shown in
Figs 13 and 14. These results show the maximum over-pressure and impulse values obtained on
the surface at any time during the simulation. Figure 13a for HE-A shows the peak pressure in
the second and third quadrants near the ground. The more energetic explosive B increases the
value of the maximum overpressure observed on all quadrants, as well as increasing the height
on the surfaces on which these higher values are observed. Still, the maximum values are
observed on the second and third quadrants. The rocket movement during detonation (Fig 13c)
shifted the maximum pressure observed to the third and fourth quadrants (when compared to the
base HE-A, Fig 13a). Placing the rocket horizontally would focus most high pressure at a
narrower zone near the center, with higher peak pressures, while placing it vertically will
disperse the load uniformly.
 12037

HE-A, at 50o

HE-B, at 50o

Moving HE-A, at 50o

HE-A, Horizontal

HE-A, Vertical

Velocity Contours at 1.0ms Velocity Contours at 3.0ms, Velocity Contours at 5.0ms,

Vmax = 0.92vkm/sec Vmax=0.53vkm/sec Vmax=0.48vkm/sec

Figures 6 through 10. Velocity evolution on the plane-of-symmetry of a circular enclosure

for: a. HE-A at 500; b. HE-B at 500; c. HE-A at 500 moving at 100m/sec; d. HE-A horizontal;
and e. HE-A vertical position.
 12037

400.0 0.50 600.0 0.60

350.0
Surface 9 0.45 Surface 10
500.0
0.50
0.40
300.0
0.35 400.0
250.0 0.40

Impulse (psi.s)

Impulse (psi.s)
Pressure (psi)

Pressure (psi)
0.30
200.0 300.0

HE-A 0.25 HE-A 0.30

150.0 200.0 HE-A(moving)
HE-A(moving)
0.20
HE-A(vertical) HE-A(vertical)
100.0 0.20
HE-A(horizontal) 0.15 HE-A(horizontal)
100.0
HE-B HE-B
50.0
0.10
0.10
0.0
0.0 0.05 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.010
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.010
-50.0 0.00 -100.0 0.00
Time (s) Time (s)

600.0 0.60 600.0 0.50

0.45
500.0 500.0
0.50 0.40

400.0
Surface 11 400.0 Surface 12 0.35
0.40

Impulse (psi.s)
Pressure (psi)
Impulse (psi.s)
Pressure (psi)

0.30
300.0 300.0

0.30 HE-A 0.25

HE-A
200.0 HE-A(moving)
200.0 HE-A(moving) 0.20
HE-A(vertical)
HE-A(vertical)
0.20 HE-A(horizontal) 0.15
HE-A(horizontal) 100.0
100.0 HE-B
HE-B
0.10
0.10 0.0
0.0
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.010 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.0100.05

-100.0 0.00 -100.0 0.00

Time (s) Time (s)

Figure 11. Averaged pressure and impulse values on surfaces 9 (00 to 450), 10 (450
to 900), 11 (900 to 1350), and 12 (1350 to 1800). 00 is the tail direction, 1800 is the
nose direction.

450.0 0.50 600.0 0.60

400.0
HE-A, 500 0.45
HE-B, 500
500.0
0.50
350.0 0.40

300.0 0.35 400.0

0.40

Impulse (psi.s)
Impulse (psi.s)

Pressure (psi)
Pressure (psi)

250.0 0.30
300.0
200.0 ave_surf. 9 0.25 ave_surf. 9 0.30
ave_surf. 10 200.0 ave_surf. 10
150.0 0.20
ave_surf. 11 ave_surf. 11
0.20
100.0 ave_surf. 12 0.15 100.0 ave_surf. 12

50.0 0.10
0.10
0.0
0.0 0.05 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.010
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.010
-50.0 0.00 -100.0 0.00
Time (s) Time (s)

500.0 0.50 400.0 0.40

450.0 HE-A, 500, moving 0.45 350.0

HE-A, Vertical 0.35
400.0 0.40
300.0
0.30
350.0
0.35
250.0
300.0 ave_surf. 9
Impulse (psi.s)

Impulse (psi.s)

0.25
Pressure (psi)

Pressure (psi)

0.30 ave_surf. 10
250.0 200.0
ave_surf. 11
0.25 0.20
200.0 ave_surf. 9 ave_surf. 12
150.0
ave_surf. 10 0.20
150.0 0.15
ave_surf. 11 100.0
ave_surf. 12 0.15
100.0
0.10
50.0
50.0 0.10

0.0 0.05
0.0 0.05
0.000 0.002 0.004 0.006 0.008 0.010 0.012
0.000 0.002 0.004 0.006 0.008 0.010 0.012
-50.0 0.00 -50.0 0.00
Time (s) Time (s)

400.0 0.50

350.0 HE-A, Horizontal 0.45 Figure 12. Comparison of pressure and

300.0
0.40
impulse values on the four surfaces for
0.35
250.0
each of the five configurations studied.
Impulse (psi.s)
Pressure (psi)

0.30
200.0
0.25
150.0 ave_surf. 9
0.20
ave_surf. 10
100.0
ave_surf. 11 0.15
ave_surf. 12
50.0
0.10

0.0 0.05
0.000 0.002 0.004 0.006 0.008 0.010 0.012
-50.0 0.00
Time (s)
 12037

Figure 13. Maximum overpressure contours on the surfaces. over-pressure ranges

from 0 to 1.15E08 dynes/cm2 (0 to 1653psi).

Figure 14. Maximum impulse contours on the surfaces. Impulse ranges from 16.87
2.84E4 dynes-sec/cm2.
4000
HE-A

The maximum impulse values observed at 3500

HE-A(moving)
HE-A(vertical)

each point show identical trends to those 3000

HE-A(horizontal)
HE-B
Maximum pressure (psi)

indicated by the maximum pressure, and are 2500

shown in Figs 14 for completeness. To better 2000

understand the results observed we
1500
examined the maximum pressure observed
on surfaces at ground level (Fig 15). Please 1000

note that 00 denotes the tail direction (i.e., 500

tail is on left, as opposed to the contours 0

0 20 40 60 80 100 120 140 160 180
plots where the tail was on the right). The Angle (deg.)

base configuration produced a maximum

pressure at about 600. The enhanced Figure 15. Maximum pressure on the surfaces at
ground level.
 12037

explosive, HE-B produced parallel trends, 350.0 0.45

with significantly higher peak pressure. 300.0

0.40

Moving the rocket resulted in lower 250.0

0.35

Impulse (psi.s)
Pressure (psi)
maximum levels, while pushing the peak 200.0 HE-A
0.30

to the third quadrant, at about 1000. The HE-A(moving) 0.25

150.0 HE-A(vertical)
0.20
HE-A(horizontal)
horizontally-placed rocket produced a 100.0
HE-B 0.15

very high peak at about the center, while 50.0

0.10

0.0
the vertically-placed rocket produced an 0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009
0.05
0.010
-50.0 0.00
almost perfectly symmetric peak loading. Time (s)

The results shown so far clearly Figure 16. Time evolution of averaged
demonstrate the peak loading dependence on pressure and impulse (force) on all surfaces.
rocket placement. The question risen was, when
integrating the load on all walls at this range, would directionally effects wash out. Figure 16
shows the averaged pressure on all faces as a function of time, and the averaged impulse, or
force, on the walls. The base HE-A case, stationary or moving, produced highly different
pressure patterns (Figs 12-15), but almost identical force evolution on the walls. The more
energetic HE-B produced similarly-trended but higher force. The interesting results are for the
vertical and the horizontal rockets. Though significantly different in terms of average pressure
evolution, the total force produced was almost identical. This is contrary to expectations as the
walls were placed fairly close to the rocket, and indicates that for both cases, the energy
contained within the walls height is fairly similar and lower than for the 500 angled weapon,
either stationary or moving.

CONCLUSIONS

The study investigated the blast and fragment distribution produced by a rocket on adjacent
surfaces. The study examined a base configuration, a stationary rocket at 500 inclination,
partially filled with HE-A. The variation included a more energetic fill, HE-B, moving the rocket
at a typical impact speed, and investigating angular dependence by placing the rocket at extreme
angular positions: horizontal and vertical. The study demonstrated that when used for vehicular
or personnel survivability, airblast pressure and impulse values greatly varied with angular
position from the rocket, as well as rocket impact velocity and angle.

REFERENCES
1. Baum, J.D., Mestreau, E., Lhner, R. & Charman, C., An Experimental and Numerical study
of Steel Tower Response to Blast Loading; Proc. of the 26th International Symposium on
Shock Waves, Gttingen, Germany, July 15-20 (2007).
2. Baum, J.D., Soto, O.A., Charman, C., Mestreau, Lhner, R., and Hastie, R. Coupled
CFD/CSD Modelling of Weapon detonation and Fragmentation, and Structural Response to
the Resulting Blast and Fragment Loading, NDIA Warhead ballistic conference, Feb 2008,
Monterey, CA.
 12037

3. R. Lhner and P. Parikh - Three-Dimensional Grid Generation by the Advancing Front

Method; Int. J. Num. Meth. Fluids 8. 1135-1149(1988).
4. R. Lhner and J.D. Baum - Adaptive H-Refinement on 3-D Unstructured Grids for Transient
Problems; Int. J. Num. Meth. Fluids 14, 1407-1419 (1992).
5. Soto, O., Baum, J.,D., Lhner, R., Mestreau, E. Luo, H., - A CSD Finite Element Scheme for
Coupled Fluid-Solid Problems. Fluid Structure Interaction 2005. September 19-21. La
Coruna, Spain.
6. Soto, O., Baum, J., Mestreau, E., Lhner, R., - An Efficient CSD FE Scheme for Coupled
Blast Simulations. Ninth US National Congress on Computational Mechanics, San Francisco,
CA, July 22-26, 2007.