LA-UR-

;D-{)/53

Approved for public release;

distribution is unlimited.

Title:

Author(s):

Intended for:

Mesoscale Modeling of Metal-loaded High Explosives

John B. Bdzil, DE-9 & Dept. Mech. Sci. Eng., University of

Illinois @ Urbana-Champaign
Brandon Lieberthal, Dept. Mech. Sci. Eng., University of
Illinois @ Urbana-Champaign
D. Scott Stewart, Dept. Mech. Sci. Eng., University of Illinois
@ Urbana-Champaign

14th International Detonation Symposium

April 11-16, 2010
Coeur d'Alene Resort, ID

~Alamos
NATIONAL LABORATORY

- - - EST . 1943 - - -

Los Alamos National Laboratory, an affirmat ive action/equal opportunrty employer, is operated by the Los Alamos National Security, LLC

for the National Nuclear Security Administration of the U.S. Department of Energy under contract DEAC5206NA25396. By acceptance
of this article, the publisher recognizes that the U.S. Government retains a nonexclusive, royalty-free license to publish or reproduce the
published form of this contribution , or to allow others to do so, for U.S. Government purposes. Los Alamos National Laboratory requests
that the publisher identify this article as work performed under the auspices of the U.S. Department of Energy. Los Alamos National
Laboratory strongly supports academic freedom and a researcher's right to publish; as an institution, however, the Laboratory does not
endorse the viewpoint of a publ ication or guarantee its technical correctness.
Form 836 (7/06)

Mesoscale Modeling of Metal-loaded High Explosives

John B. Bdzil', Brandon Liebertha]", D. Scott Stewart
Department of Mechanical Science and Engineering
University of Illinois, Urbana, IL, 61802
Abstract. We describe a 3D approach to modeling multi-phase blast explosive, which is
primarily condensed explosive by volume with inert embedded particles. These embedded
particles are uniform in size and placed on the array of a regular lattice. The asymptotic
theory of detonation shock dynamics governs the detonation shock propagation in the
explosive. Mesoscale hydrodynamic simulations are used to show how the particles are
compressed, deformed, and accelerated by the high-speed detonation products flow.

Introduction
Metal-loaded blast explosives (MBX) and
metalized enhanced-blast explosives (EBX) are of
interest because with the addition of small metal
particles to a high-performance explosive (HE),
such as HMX, RDX or TNT, one can tailor the
profile of "applied pressure"-vs-time (and the total
delivered impulse). Although the explosives
applied pressure is lowered, it is sustained for a
longer time. These metal particles are accelerated
and heated by the primary products of detonation.
This increases the energy transfer from the
explosive to a target, and thus enhances target
damage. When these metal particles are reactive,
such as with aluminum particles, a further
enhanced-blast performance can be obtained
(depending on particle size/explosive size) as the
particles either react with the hot detonation
products or burn in air as the particles are ejected
from the explosives fireball (thermo-baric
explosive) [1,2].
We describe our mesoscale modeling of
interaction of the detonation wave in a
contain ing embedded inert partic les [3, 4].
restrict our attention to situations where: I)

the
HE
We
the

diameter of the metal particles, d, is either

comparable to or larger than the explosives
detonation reaction-zone length, h, (dlh I), 2)
the shock speed in the metal particles is less than
the detonation speed, and either 3) the particles are
inert or 4) the particles react with the hot
detonation product gases on a time scale longer
than the explosives reaction time scale. In this dlh

/ limit and with the above assumptions,

detonation shock dynamics (DSD) provides an
adequate representation of how the detonation
wave in the explosive successfully propagates via
circuitous path in the explosive between the
particles (see Figure I).

Fig. I. A cartoon that depicts how a detonation

wave propagates in a metal-loaded HE. Particles

significantly
affect
propagation speed .

the

mean

detonation

1:/:(, . e .
Fig. 2. A density color palette plot showing an instantaneous (t = 6 .25 J.ls) projection of the density in the
plane, y = 0.5, passing through the center ofa SC array of 5 steel spheres embedded in both the products of
detonation and in fresh ABS explosive [7]. The Cartesian z-axis is horizontal, the x-axis is vertical, and the
y-axis points out of the plane of the paper. The lines are the current boundaries of the explosive cel,ls that
were originally lxlxl-unit cubes. The original sphere radius is r = 0.363; corresponding to an initial
volume fraction of inert of 0.20. The detonation is moving to the right.
heterogeneity in the pressure loading of the
We begin by describing the DSD method, the
particles, the particles are deformed; this
ultimately can lead to the fragmentation of the
scaling laws for our problem and introduce the
loading scenarios.
particles
under some
metal-loaded HE unit cell problem. We construct
our explosive from a collection of these 3D unit
Experiments performed on the shock-and-release
loading of a copper sphere embedded in a matrix
cells, and restrict attention to loadings of 20% by
of I gm/cc polystyrene shows large deformation
volume particles in these particle-loaded
and fragmentation of the copper sphere [6]. Here
explosives. Specifically, here we consider a
we present results from our medium-resolution,
mono-modal particle distribution with the
three-d imens ional hydrodynamic simu lations of
following symmetries: simple cubic array (SCA),
detonation in a highlly symmetric array of 10
body-centered cubic (BCC) and face-centered
spherical particles packed in a SCA and aligned
cubic (FCC). Under the constraint of 20% by
along the z-axis. We consider the shockvolume particles, these changes in unit cell
compression, acceleration and deformation of the
geometry bring with them changes in the particle
particles by a constant speed, programmed-burn
diameter relative to the explosives intrinsic
detonation wave moving in the z-direction.
reaction-zone length. By stacking these unit
These detai led mesoscale simulations show that
cells, we are able to study detonation
the dynamic loading of the particles is divided
propagation in semi-infinite beds of these metalinto two regimes : I) a non-isentropic, shockloaded explosives. By then averaging the results
loading phase, which ends with the particle and
from DSD front propagation calculations, we
explosive products pressures being nearly
develop effective DSD propagation laws for
equilibrated, that is then followed by 2) a nearly
these composite explosives [5] .
isentropic phase. where "form drag" causes the
particles to be further accelerated by the speedier
The embedded particles are compressed and
flow of the detonation products. A plot of
accelerated by the: I) the oblique shock drive
density along a symmetry plane, a plane passing
provided by the passing detonation wave and 2)
through the centers of 5 steel sphere embedded
the drag forces associated with the streaming of
in the SCA unit cells of explosive, is shown at
the detonation products past the particles .
13.5 f.ts in Figure 2. The explosive is modeled
Because the detonation wave speed in the
using our ABS explosive model [7]. We describe
explosive and the shock speed in the particulate
both the loading of low-impedance plastic
material are in general disparate, shock focusing
spheres and high-impedance metal spheres.
and /or diffraction occurs in the loaded particles.
This can generate both regions of high pressures
DSD Detonation Propagation Methodology
and tensions in the particles. Due to the

The DSD-limit applies when an explosives

detonation reaction zone is broadly curved in the
transverse direction when measured on the scale
of the reaction zone in the normal direction .
Then the normal speed of the detonation front,
D'I) is a function of the radius of curvature of the
detonation front's shock, K [8].

are built up by stacking these cells. Given the

constraint of 20% by volume inert particles , the
diameter of the particles is varied by changing
the packing geometry of the spheres. To be
consistent with the assumption that the
detonation reaction-zone length of the HE is
small compared to the particles, the DSD law is
set to Dn = 1 - O.l -i< K; the reaction-zone length is
- 0.1. Examples of the unit cells are displayed in
Figure 4.

DSD front

Simple Cubic Cell (80% HE)

Fig . 3. The boundary condition applied on W at

all edges of the explosive depends on whether W
< Ws or W > w,' where Ws is the value of the
angle corresponding to a son ic flow at the edge
of the explosive. When W < ws , a continuation ,
outflow boundary condition is applied. When w
> ws, then w is set to w = We.
In its simplest instance, this propagation law has
the form Dn(K} = DCJ *(J - aK}, where DC.! is the
Chapman-Jouguet detonation speed , and a is a
constant related to the explosives ' detonation
reaction-zone length . Thus, the wave either
decelerates or accelerates depending on the sign
of the curvature of the front. In addition , the
speed of the DSD front is affected by the
interaction of the front with inert material
objects. This interaction involves a boundary
condition on the DSD front. The angle between
the normal to the detonation and the normal to
the inert material surface, w, takes on a unique
value, We, depending on the properties of the
explosive and the inert. The boundary interaction
is described in Figure 3 .

BCC Cell (80% HE)

Fig. 4. Two examples of the unit cell geometries.

SC contains I-sphere and BCC contains 2spheres .
The DSD boundary angle parameters used were
Ws = 0. 785 for the HE and w, = 0. 785 for plastic
spheres and w, = 1.374 for heavy-metal spheres.
An example of SCA periodic packing assembly
of spheres in fresh explosive is shown in Figure
5. In the calculations, symmetry boundary
cond itions are appl ied along I) the two large
square planes that make up opposite sides of the
explosive and 2) the two smaller rectangular
surfaces that make up the bottom and left side of
the explosive.

Periodic Loadings of Spherical Particles

To reduce the size of the calculations, we
consider that the particle-loaded HE consists of a
3D, periodic array of uniform sized , cubic unit
cells, of length L = 1 on a side. The test samples

Fig. 5. The array geometry of SCA cells used to

study the expansion of an initially cylindrically
symmetric wave (with axis along the x-direction

and y = 0, Z = 0) into an explosive sample of

infinite extent.
DSD Detonation Propagation Results

T a bl e I < D0> vs. geometry an db oun d ary ang e.

OOc

0.785
1.374

se

Bee

<Do>=O.78
<Do>=0.86

<Do>=O.71
<Do>=0.80

FCC
<Do>=0.66
<Do>=O.77

Two classes of initial-value problems were

considered; a planar input along the bottom
surface of the geometry shown in Figure 5 and a
cylindrical wave input, whose axis is coincident
with the lower left-hand edge of the Figure 5
geometry. From data for the detonation
propagation problem with the initial, planar wave
input, we determined the average, planar
detonation propagation velocity, <Do>, displayed
in Table I.

Fig. 6. Time Ih = 5 isosurface for a detonation

front, initially a cylindrical wave, with axis along
the x-axis and r = 0.5, spreading into an infinite
SCA inert sphere-loaded HE composite (wc =
0.785) as shown in Figure 5.
Given the solution data from the propagation
problem with the initial, cylindrical wave input,
we were also able to determine the "effective"
mean curvature dependence of the average
detonation propagation law, <Do> = 0.803+/0.003 (0.868+/-0.003) and <0> = 0.156+/-0.004
(0.124+1-0.004) for the SC geometry and for the
cases Wc = 0.785 (1.374). These values of <Do>
are essentially identical to what is displayed in
Table 1 for identical cases.

3D Mesoscale Hydrodynamic Simulations

What might be described as the standard

approach to modeling a multiphase problem such
as metal-loaded exp losive is two-phase,
continuum mixture theory [1,9,10]. In that
approach, the loading particles and the explosive
are assumed to coexist at each spatial location,
each phase is separately governed by its
independent set of compressib Ie flow equations
(Euler equations, typically), and the interaction
between the phases is handled through a set of
source terms to the equations (exchange terms)
that describe the exchange of momentum and
energy between the phases. A Ithough the
modeling of those exchange terms is at the heart
of the problem, the forms adopted for those
exchange terms is carried over from simpler
flows, like from the theory of incompressible
low-speed flow over rigid particles [I]. That is
particularly the case for the momentum
interaction between the phases, wh ich is
essentially modeled with some variant of a
Stokes drag law. In the same (mesoscale) spirit
as that we used to describe the detonation-front
propagation problem, our objective here is to
perform detailed interaction modeling of the
high-speed, compressible detonation flow with
the defined particles, and from those results, to
develop
an
"average"
mean-flow
then
description for the problem. That is, we seek to
create the database needed in order to develop
exchange terms appropriate for high-speed,
compressible flow. To study the hydrodynamic
loading of the embedded particles, we couple our
detonation front propagation to the threedimensional hydrodynamic flow simulation
capability provided by the LLNL code, ALE3D
[I I].
As In our modeling of detonation
we
limit
consideration
to
propagation,
geometrically simple inert particle geometries;
here the SCA geometry described above. We do
this because the problem is fundamentally threedimensional, and the numerical resolution of the
flow must be adequate enough to resolve the
details of the flow. Our geometry consists of a
stack of I x I x I-unit cubes of explosive, stacked
one upon another in the z-direction. At the center
of each of these explosive cubes is a hollowed
out spherical space containing a sphere of inert

material having a radius, r = 0.363. The lower

most explosive cube contains only explosive. A
two-dimensional plane cut of the geometry is
shown in Figure 2. To have this stack of sphere
containing explosive cubes in the z-direction act
as a part of a semi-infinite sample (as in the
previous sections), we apply symmetry boundary
conditions along the bounding planes, x = 0, x =
I, Y = 0 and y = I. Additionally, to maintain
symmetry, we assume that an established, planar
detonation wave enters our test sample along the
plane z = 0, simultaneously at t = o.
The initial set of problems we solve and
report on here, restrict the number of length
scales in the problem to just the unit length of a
sphere edge and the fixed in itial radius of the
sphere. We do this by assuming that the
detonation
is
a
Chapman-Jouguet
(CJ)
detonation, with a zero th ickness reaction zone
(the fu II detonation energy release is triggered by
the passage of the detonation shock) and with the
detonation being fully supported at the CJ-state.
This then corresponds to a CJ-state inflow
boundary condition being applied at z = O. Then
with the ratio of the explosive cell-edge length
fixed, the sphere radius fixed and no explosive
length scale, the problem time scales with /lD cl .
The results we describe apply equally well to any
system with this same ratio of dll = 0.363, which
corresponds to a particle volume fraction loading
of20%.

Our explosive detonation model is the ABS

model [7]. This model employs a constant
gamma, y = 3, equation of state, an in itial
explosive density of Po = 2 gm /cc and a
,
, ThIS
detonation energy of q = 0.04 cm-/!-!s-.
gives DCl = 0.8 cm/~ls and provides detonation
characteristics like those in a solid , highperformance explosive. We use the Lund
detonation lighting algorithm in ALE3D to
propagate the zero-th ickness reaction-zone
length detonation . At some point, we will
introduce a detonation reaction-zone length scale
and use a DSD lighting algorithm for these
calculations. The currently implemented ALE3D
DSD algorithm does not have a proper DSD
boundary cond ition treatment, and thus would
find no difference in the explosive confinement
effect between a plastic, like Lucite, and a heavy
metal, like tantalum.

Mesoscale Hydrodynamic Simulation Results

We consider four inert materials for the
spheres, which span the range of loading
material densities that could be considered of
interest for applications : I) Lucite, 2) aluminum,
3) steel and 4) tantalum . In all cases, the
materials do not react with the explosive
products. The constituent response was taken
from the ALE3 0 KO/SGEOS material library .

Fig. 7. Four density palettes (different rainbow palettes for each case) are shown at t = 12.0 !-!s for the
SCA I 0 problem. The Iines are the deformed boundaries of the original I x I x I-unit cells of explosive. From

top to bottom, the original inert spheres are Lucite plastic (rho o = 1.182 gm/cc), aluminum (rho o = 2.785
gm/cc), steel (rho o = 7.90 gm/cc) and tantalum (rho o = 16.69 gm!cc). In all cases, the detonation first
reaches the right boundary, z = II , at t = 13.75 ).l.S. (See the Figure 2 legend for more display details.

The SGEOS material model numbers are 37, 89,

28 and 77, for Lucite, aluminum, steel and
tantalum, respectively [II]. A footnote on the
ALE30 simulations we performed . We tried to
import an interface conforming mesh/geometry
into ALE30 from a Cubit hexahedral mesh,
written as a genesis file. Our goal was to run the
problem in ALE30's pure lagrangian mode.
However, due to the large mesh distortions that
occur in th is problem, it was necessary to switch
to the A LE mode. Thus, there is little advantage
to having a geometry conforming mesh. Here all
the spheres were "shaped" into the geometry,
and the material interface description was via
ALE30's volume of fluid (VOF) mixed cell
interface treatment approach. The explosive was
burned using a programmed burn approach, with
the total explosive energy release being over 1-2
computational cells. The programmed burn
detonation state is very close to the exact
computed detonation state for this model of PC! =
0.32 Mbar, uC) = 0.2 cm/).l.s, PC) = 2.667 gm/cc.
Our goal is to use the simulation results to
aid in the development of exchange laws for
continuum mixture-type, two-phase models that
are appropriate for metal-loaded explosives. To
that el1d, we devised some diagnostics with
which to measure spatial-averaged properties of
this multi-material flow. Those diagnostics are
based on doing instantaneous volume averages
over the active parts of the flow (regions with
pressure > 0.00 I Mbar) for each cell. We do
separate bookkeeping for the explosive and inert
material regions in each cell. Those diagnostics
include the cell-volume averaged flow-involved
volume, pressure, z-direction particle velocity
(detonation propagation direction), and in the
inert material region on Iy, the net d iss ipated
energy (rate of entropy generation). We use the
monikers "abscell#" and "insphere#" to
separately label the explosive products and inert
region parts of each cell. From left to right, the

labeling for the explosive is abscell21, abscellll ,

abscelll2, . .. , abscell20, while the labeling for
the inert is insphere I, insphere2, insphere3 . . .. ,
insphere 10. Thus , abscel121 is the left-most
explosive-only cell, abscelill /insphere I is the
following first explosive-inert cell, etc.
A graphical summary of our results is
displayed in Figure 7. These base-level
simu lations used 50x50x550 computational cells.
(We performed simulations at both half and
double this resolution . Those results show only
small quantitative differences.) As we go from
top to bottom , the density increases, going from
Lucite at the top to tantalum at the bottom. As
expected, the Lucite "spheres" are more easily
entrained in the flow than are the tantalum
"spheres." What is clear is that once the
detonation passes over a sphere, the spheres are
deformed into other shapes; a convex shape for
the Lucite and a concave shape for the tantalum.
A Iso, note that the left boundary of each cell
(being coincident with the original cell material
boundary) mirrors the nature of the explosive
products flow. Appropriate drag laws must
reflect these facts . There are some interesting
comments to be made with regards to acoustic
impedance and sound speed in the inert materials
and shock/detonation speed , but we reserve those
for another time.
Summary of Mesoscale Simulation Results

In Figures 8, 9 and 10 we display the

detai led cell-averaged results for the cell
comprised of abscell13 /insphere3, composite
ceH3 from the left.

ABS-Iucite
0 .8

ABSlucite
0.35 .----,.-..,---;-,-----,----,--.,.....---,

0,7

0.3

0,6

...-..l.-_ _..:----'

0.00015

"H' 0.5

c: 0.25

os

(I)

E
>

.----.-.---,r--.-...--,- -, 0.0002

0.2

p-abscel1J-vmean

p-lnsphere3-vmean

~
o 0.4

0,0001

.J-..

E 0.15

03
.
o
> 0.2

.L

'l:i: 0.1

0.1
0.05

!(f'S)

Fig. 8. The volumemean (vmean) pressures in

ce1l3 are compared in th is figure. The average
over time of these pressures is nearly the same
and slightly below PCJ . The Lucite pressure
shows significant oscillations.
ABSlucite
0.3 .----...-..,--r---r--,--,....---,
0.25
c:

::: 0.2

>
';;;'01 5
::>.

E
~
'"::>

'\

'\
I1'",1 ~ ,....,.~
I
II -- - --::------j

0.1
005

1'0

12

14

t(f's)

Fig. 9. The vmean of the zdirection explosive

products particle velocity is essentially UCJ of the
neat explosive. The vmean of the Lucite particle
begins by being above UCJ, reflecting the
explosive drive into the central portion of this
relatively low shock-impedance inert, and it then
drops to near UCJ , as the particle deforms and its '
edges are pulled back (moved to the left) by the
drag of the exp losive products.

Fig. 10. This figure shows that: I) the overall

volume of the cell is decreased, 2) the volume
fraction of the Lucite is decreased due to the
compression of the Lucite and 3) all of the
entropy generated in the cell occurs during the
time of shock/detonation passage through the
cell. Thus , the oscillations in the Lucite pressure
shown in Figure 8 are isentropic.
The most striking observation in this example is
the non-classical drag signature shown in Figure
9. The vmean particle velocity of the Lucite
remains nearly constant and above UCJ until the
drag forces applied to the particle, principally at
its far outer surface, begin to deform the particle
and slow it in the process. Thus, there is a delay,
significant in length compared to the particle
shock passage time, for which the vmean particle
velocity of the Lucite remains constant before
slowing in something that might be referred to as
flow-induced drag. Also, it is noteworthy that all
of the entropy generation in the Lucite occurs
during shock/detonation passage . Otherwise, the
flow is isentropic.
The case for aluminum particles is similar to
that for Lucite. Because of the difference in the
shock impedance (higher shock impedance for
the alum inurn), the vmean particle velocity in the
aluminum is on a horizontal plateau below the
UCJ speed of the detonation products, before it
moves towards UCJ seven microsecond s after the
detonation passage time. Otherwise, the
pressures are nearly in a non-osci llatory
equilibrium, at a value that is slightly above PCJ .
The volume fraction graph shows the same
features as are shown in Figure 10 for Lucite.

Next we show the case for tantalum spheres. The

case for steel is similar to that for tantalum
except the behavior is somewhat muted.

small amounts of tantalum around the outside of

the particle in what is a shearing motion.
ABS-tantalum

ABS-tantalum
0.5 r-- - - - - -,-----,----,---,

0.8

0.0007

0.7

0.0006

0.6

c
m

Q)

E 0.3

~ 0.4

ro

~ 0.3
>

0.2

>
.......

~ 0.2

0.1

0.1

0.0005
- - -s2-insphere3

0.5

(j)

r;. .:.
..., III

li"C

:::r

0.0004
0.0003
0.0002
0.0001

0
t(~IS)

Fig. I \. Although the vmean pressure in the

explosive products and tantalum track one
another, both pressures are seen to increase
beyond PCJ . The pressure osci llations in the
tantalum are significant.

Fig. 13. The entropy production in the tantalum

continues for an extended time, due to a strong
shock-focusing effect in the particles. Th is
results in a heating and a late-time expansion of
the deforming tantalum particle.
Conclusions from the Mesoscale Flow Study

For the case of high shock-impedance tantalum,

the flow in the particles is both non-isentropic
and high ly deformational for the particles.
ABS-tantalum
0.25 r - - - - - - - - . . . , . - r - - r---,

All of our results scale with the cell-edge

particle dimension, I. We are currently
developing an alternative momentum exchange
description metal-loaded high explosives.
References

0.2

'"

Q)

~ 0.15

I.

.......
(j)

::.

0. 1

~
N

Frost, D. L., et a\., "Particle momentum

effects from the detonation of heterogeneous
explosives,"]. Appl. Phys., 101, 113529,
2007.

:::J

0.05

2.
fi-

8-

1'o-

1'2-

f4

t(ps)

Fig. 12. The vmean z-d irection particle velocity

begins well below the z-direction, vmean
explosive products particle velocity, which is
close to UCJ. The very low particle velocity in the
tantalum is a consequence of the very high
density and shock impedance of the tantalum. As
one can see in Figure 7, the action of the
detonation products flow accelerates relatively

3.

4.

Gogulya, M. F. and Brashnikov, M. A.,

"Detonation
of metalized
composite
explosives," in Shock Wave Science and
Technology
Reference
Library:
Helerogeneous Delonalion, pp. 217-286,
Springer- Verlag, Berlin Heidelberg, 2009.
Frost, D. L., and Zhang, F., "Slurry
detonation," in Shock Wave Science and
Technology
Reference
Library:
Helerogeneolls Delonalion, pp. 169-216,
Springer-Verlag, Berlin Heidelberg, 2009.
Ripley, R. c., et a\., "Shock interaction of
metal particles in condensed explosive

5.

6.

7.

detonation", in Shoek Compression 01

Condensed Matter, 2005 (M.D. Furnish, M.
Elert, T.P . Russell, C.T. White, eds .), part I,
pp.499-502 .
Stewart, D. S., et aI., "Modeling unit cell
interactions for the microstructure of a
heterogeneous explosive: detonation
diffraction past an inert sphere," in Shock
Compression of Condensed Matter, 2009
(M. Elert, and W. Buttler, eds.).
Robey, H. F., et aI., "Experimental
investigation of the three-d imensional
interaction of a strong shock with a spherical
density inhomogeneity," Phys. Rev . Lett., 89
(8),085001-1,2002.
Aslam, T. D., et aI., "Level-set methods
applied to modeling detonation shock
dynamics," 1. Comput. Phys. 126,390-409,
1996.

8.

Bdzil, 1. B., and Stewart, D. S. , "The

dynamics of detonation in explosive
systems," Annu. Rev. Fluid Meeh. , Vol. 39,
pp 263-292, Annual Reviews, Palo Alto,
CA,2007.
9. Bdzil, 1. B., et aI., "Two-phase modeling of
detlagration-to-detonation transition in
granular materials: A critical examination of
modeling issues," Phys . Fluids, 11 (2),378,
1999.
10. Zhang, F., et al. "Explosive dispersal of
solid particles," Shock Waves, 10,431,
200 I.
ll. Nichols, A. L. 1[[, et aI., "User manual for
ALE3D: An arbitrary Lagrange/ Eulerian 2d
and 3d code system," LLNL Report, LLNLSM-404490,2009.