A reactive flow model for heavily aluminized...
Transcript of A reactive flow model for heavily aluminized...
A reactive flow model for heavily aluminized cyclotrimethylene-trinitramineBohoon Kim, Jungsu Park, Kyung-Cheol Lee, and Jack J. Yoh
Citation: Journal of Applied Physics 116, 023512 (2014); doi: 10.1063/1.4887811 View online: http://dx.doi.org/10.1063/1.4887811 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/116/2?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Hot-spot contributions in shocked high explosives from mesoscale ignition models J. Appl. Phys. 113, 233513 (2013); 10.1063/1.4811233 One-dimensional plate impact experiments on the cyclotetramethylene tetranitramine (HMX) based explosiveEDC32 J. Appl. Phys. 112, 064910 (2012); 10.1063/1.4752865 Sparse Partial Equilibrium Tables in Chemically Resolved Reactive Flow AIP Conf. Proc. 706, 1061 (2004); 10.1063/1.1780421 A Reaction Zone Enthalpy Balance Model to Simulate ShocktoDetonation Transition and Unsteady DetonationWave Propagation AIP Conf. Proc. 706, 943 (2004); 10.1063/1.1780392 Shock loading and reactive flow modeling studies of void induced AP/AL/HTPB propellant AIP Conf. Proc. 429, 373 (1998); 10.1063/1.55692
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
A reactive flow model for heavily aluminized cyclotrimethylene-trinitramine
Bohoon Kim,1 Jungsu Park,2 Kyung-Cheol Lee,1 and Jack J. Yoh1,a)
1Department of Mechanical and Aerospace Engineering, Seoul National University, Seoul 151-742,South Korea2Agency for Defense Development, Daejeon 305-600, South Korea
(Received 26 April 2014; accepted 27 June 2014; published online 11 July 2014)
An accurate and reliable prediction of reactive flow is a challenging task when characterizing an
energetic material subjected to an external shock impact as the detonation transition time is on the
order of a micro second. The present study aims at investigating the size effect behavior of a
heavily aluminized cyclotrimethylene-trinitramine (RDX) which contains 35% of aluminum by
using a detonation rate model that includes ignition and growth mechanisms for shock initiation
and subsequent detonation. A series of unconfined rate stick tests and two-dimensional
hydrodynamic simulations are conducted to construct the size effect curve which represents the
relationship between detonation velocity and inverse radius of the charge. A pressure chamber test
is conducted to further validate the reactive flow model for predicting the response of a heavily
aluminized high explosive subjected to an external impact. VC 2014 AIP Publishing LLC.
[http://dx.doi.org/10.1063/1.4887811]
I. INTRODUCTION
We characterize a multi-purpose high explosive (HE)
which is comprised of 50% RDX (cyclotrimethylene-trinitr-
amine, C3H6N6O6) and 35% aluminum powder with 15%
HTPB (hydroxyl-terminated polybutadiene) binder, and its
initial density after pressing is 1.78 g/cc. To simulate the
shock-to-detonation transition (SDT) that is dominantly
driven by pressure, Lee-Tarver1 and JWLþþ (Jones-
Wilkins-Leeþþ)2 models are used in general. Lee-Tarver
considers generation of hotspots and their effect during ini-
tiation. The model requires predetermination of parameters
both empirically and ad hoc, making it difficult to be univer-
sal for generalized energetic material. The JWLþþ model
offers simplification to Ref. 1 by ignoring the initiation step
and using only the single growth step for a detonating explo-
sive. The methodology to determine model parameters are
somewhat qualitative, and the reactive flow model would
need to be recalibrated each time a new explosive is formu-
lated. Once calibrated for one explosive, it does not manifest
as to how to change the unknown parameters for other
explosives.
Souers and Vitello3 provided a useful approach to deter-
mine the unknown parameters of JWLþþ model. They pro-
vided an analytical approach that allows prediction of the
rate constants in terms of the size effect measurements. The
method is restricted to a growth step only, and an empirical
fit to a diameter effect is required to constrain unknown
parameters, indicating that there are still additional
unknowns to be determined.
The present study aims at developing a modified reac-
tive flow model that includes both ignition and growth steps
while attempting to overcome limitations of aforementioned
JWLþþ model. In addition, the unknown parameters
are handled theoretically to minimize ambiguity of the
numerical iterations. In order to validate the reactive flow
response of a heavily aluminized RDX, two distinct experi-
ments, namely, the unconfined rate stick test and a pressure
chamber test have been carried out.
II. UNCONFINED RATE STICK TESTS
A series of unconfined rate stick tests for heavily alumi-
nized RDX is performed, where the detonation velocities
were measured as a function of cylindrical charge diameter.
The steady detonation wave is observed by utilizing a charge
length sufficiently long enough to damp out any booster
overdrive or initiation transients which can persist for up to
five times the charge diameter. A charge length-to-diameter
ratio of 10 is chosen to adequately capture the fully devel-
oped detonation velocities in all tests.
The samples are prepared by compacting the heavily
aluminized RDX into a precisely machined mono-cast
Nylon. Electronic pins were utilized for measurement of the
propagation velocity of detonation wave. If the installed pin
is exposed to a high temperature-pressure shock wave, the
pin breaks and signal generator transmits a signal. Then a
high speed counter, a Time-to-Digital Converter (TDC)
measures the arrival time of the shock wave, sensing an
ascending (or descending) waveform of input pulses. We
used pin CA-1041 (DYNASEN, Inc.) and a high speed coun-
ter Lec-4208 (LeCroy, Inc.) as the wide range real-time
TDC. The TDC is designed to cover the time measurements
performed in real time, and requiring a wide dynamic range
with high resolution.
A rate stick packed in a cylindrical mono-cast Nylon of
90 mm diameter is shown in Fig. 1. 16 different diameters
used were 13.5 mm, 14.0 mm, 15.0 mm, 16.0 mm, 17.0 mm,
18.0 mm, 20.0 mm, 25.0 mm, 30.0 mm, 35.0 mm, 40.0 mm,
45.0 mm, 50.0 mm, 70.0 mm, 90.0 mm, and 110.0 mm. We
considered a wide range of diameters to secure enough dataa)Email: [email protected]. Tel.: 82-2-880-9334.
0021-8979/2014/116(2)/023512/9/$30.00 VC 2014 AIP Publishing LLC116, 023512-1
JOURNAL OF APPLIED PHYSICS 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
to establish the size effect behavior of heavily aluminized
RDX.
The propagation of the detonation front along the length
of each charge was measured using five equally spaced
sensor probes. The TDC with a bandwidth of 1 GHz or fre-
quency resolution of 1 ns (10�9 s) was used during each test.
The measured probe data for 30 mm diameter test are shown
on a plot of probe position versus trigger time in Fig. 2. The
time-distance diagram proves that all initiation transients
have in fact dampened and that the self-sustained propaga-
tion velocity is constant by the end of a charge with
L/D¼ 10 because the time dependent slope of the pin posi-
tion is steadily maintained. The probe data minimizing the
influence of initiation transients were used to get a fully
developed detonation velocity.
The chemical kinetic information comes from the size
effect data where the change in steady state or fully devel-
oped detonation velocity, Us, is given in terms of radius, R0.
The detonation velocity is plotted as a function of the inverse
radius, 1/R0, and the curve is extrapolated back to zero to
obtain the infinite radius detonation velocity, D.
III. A REACTIVE FLOW MODEL
The rate of product mass fraction consists of ignition
and growth terms4 such that
dkdt¼ I 1�kð Þb g�að Þx
þG1 1�kð ÞckdPyþG2 1�kð ÞekgPz; g¼ qq0
�1:(1)
Here, k is the burned mass fraction, and constants I, b, a, x,G1, c, d, y, G2, e, g, and z are the unknown parameters. k is
reaction progress variable (k¼ 0 unreacted state and k¼ 1
reacted state). P is the pressure, t is the time, and q0 and q are
the initial and current densities, respectively. The Mie-
Gruneisen equation of state (EOS)5 in Eq. (2) is for the
unreacted solid HE, and the isentropic JWL EOS6 in Eq. (3)
is used for the reacted gaseous product
Punreacted ¼ PH þ Cqðe� eHÞ; (2)
Preacted ¼ Ae�R1ðq0=qÞ þ Be�R2ðq0=qÞ þ Cðq0=qÞ�ð1þxÞ: (3)
Here, PH and eH are pressure and internal energy of a
reference state that follows the Hugoniot curve, and C is
the Gruneisen gamma. PH and eH can be expressed as
follows:
PH ¼ C20
1
q0
� 1
q
� ��1
q0
� S1
q0
� 1
q
� �� �2
; (4)
eH ¼ C20
1
q0
� 1
q
� �2�2
1
q0
� S1
q0
� 1
q
� �� �2
: (5)
C0, S are the bulk sound speed and linear Hugoniot slope
coefficient, respectively. Also the shock speed relations are
S ¼ dUshock=dUparticle; (6)
C0 ¼ ð@P=@qÞ1=2; (7)
Ushock ¼ C0 þ SUparticle; (8)
where Ushock is the shock wave velocity and Uparticle is the
material particle velocity.
A, B, C, R1, and R2 are the material dependent JWL
parameters with x being the Gruneisen coefficient of
Eq. (3). Equations (2) and (3) are combined into a single
expression Eq. (9) using the product mass fraction (k) and
FIG. 1. A rate stick packed in a cylindrical mono-cast Nylon tube.
FIG. 2. Captured position versus trigger time plotted for a 30 mm diameter
case.
023512-2 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
reactant depletion (1� k). The standard mixture rule applies
to the internal energy and volume ratio such as
Ptotal ¼ ð1� kÞPunreacted þ kPreacted; (9)
etotal ¼ ð1� kÞeunreacted þ kereacted; (10)
�total ¼ ð1� kÞ�unreacted þ k�reacted: (11)
The Lee-Tarver model is comprised of (i) ignition term
that represents formation of the hotspots by the rapid compres-
sion, and (ii) first growth term that describes the effect of the
propagation of the reacting waves in the substance and second
growth term that represents completion due to a detonation
transition. For determination of 12 unknowns here, a curve fit-
ting method with optimization technique is usually needed.
If the ignition term involving the initial density ratio is
omitted, a JWLþþ model2 is generally recovered, and its
simple form is
dkdt¼ G 1� kð Þ Pþ Qð Þb: (12)
Here, G is the growth constant, b is the pressure sensitivity,
and Q is the artificial viscosity. The determination of the
unknowns is relatively easier, and thus, the usability of a
simplified form of the rate law is greater if the process is
mostly growth dependent.
An alternative to these two limiting forms of the reactive
flow equation can be casted into a following form such that:
dkdt¼ I 1� kð Þga þ G 1� kð ÞPb; g ¼ q
q0
� 1: (13)
Here, the model resembles a full Lee-Tarver, but it now
includes an initiation step which is missing in JWLþþ. Four
unknown parameters having the major significance in view
of detonation are kept, namely, I, a, G, and b.
The initiation step requires defining the ignition constant,
I, while the reactant depletion (1� k) is time-resolved in the
governing equation with a compression g¼q/q0� 1. The
ignition of high explosive occurs by compression because of
the shock wave propagation. Hotspots are formed in a shocked
high explosive, leading to a void collapse. Any void or gas
bubble that exists in high explosives may provide a potential
site for local adiabatic compression that leads to the localized
heating well beyond the activation energy for detonation.
Wackerle et al.7,8 reported that hotspot formation due to
void collapse depends on the shock pressure, Ps, and the
reaction rate is proportional to P2 which is known experi-
mentally. To investigate the shock-pressure-squared depend-
ence of ignition, compression sensitivity, a, must be set to 4
because the relationship between pressure and density is
parabolic based on the Rankine-Hugoniot argument. Once
again, the pressure squared term is dominant in the shock
induced ignition, and the pressure is proportional to com-
pression squared. As such the compression sensitivity being
4 is the best choice for most of the SDT problems.
Souers and Vitello3 provided a useful approach to deter-
mine the unknown parameters of JWLþþ model. They ana-
lytically linked the growth term with the unconfined rate
stick data. The rate stick test provides detonation velocity
versus the inverse radius of a cylindrical charge. With such
size effect information, the unknowns of the growth term are
determined. That is, the relationship between non-
dimensional detonation velocity and a non-dimensional
inverse radius is derived using a detonation velocity equation
of unconfined rate stick following Eyring et al.9 as:
Us
D¼ 1� ð0:4=½1� Us=Dð Þ2�0:8Þ�1 teUs
R0
; (14)
where Us is the detonation velocity for radius R0, and D is
the detonation velocity at infinite radius. te is the time to
cross the reaction zone during which Eq. (14) is valid.
Now, all k’s are collected on the left of the growth term
in Eq. (13), and the equation is integrated such thatð1
1� kð Þ dk ¼ð
GPbdt ! �ln 1� kð Þ ¼ GPbte: (15)
Finally, Eq. (16) specifies the relationship between
dimensionless detonation velocity and the charge inverse
radius
RD
R0
¼ �0:4 1� Us=Dð Þ2h i�0:8
ln 1� Us=Dð Þ2� � Us
D
� �2b�1
1� Us
D
� �: (16)
The non-dimensional and generalized radius (RD) and
growth constant (G) are given by
RD ¼ DðGPbÞ�1;G ¼ DðRDPbÞ�1: (17)
This means that the Eyring equation can be expressed in
terms of the detonation velocity and the charge size, defined
through a pressure term b such that the size effect can be
derived by changing the value b in Eq. (16).
The curve fitting procedure for a heavily aluminized
RDX is performed to determine the pressure sensitivity term,
b. The optimal size effect curve of the analytical solution is
drawn based on the pressure sensitivity set to b¼ 0.7 with
maximum detonation velocity at zero inverse radius being
7.387 mm/ls, and the generalized radius equals 0.75 mm.
The analytical line which is an optimal fit is characteristi-
cally high for RD/R0< 0.05 and low for 0.05<RD/R0< 0.12.
Now, the Ignition constant is defined by the ratio of igni-
tion to growth, namely,
dkdtI
Ignition
¼ I 1� kð Þga;dkdtG
Growth
¼ G 1� kð ÞPb: (18)
Here, tI and tG are the total time for completion. The dimen-
sionless ignition and growth (IG) takes the form
IG ¼ Iga
GPb� 0:1 or I ffi 0:1GPbg�a; (19)
where the four parameters (I, a, G, and b) decide the intrinsic
characteristics of initiation and detonation. By assigning 0.1
to this number, ignition constant I can be calculated from
Eq. (19), suggesting that 10% of the full reaction time is
023512-3 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
spent for growth. Thus, if IG is zero, the model reduces to a
JWLþþ rate law where the rapid growth step is of only rele-
vance in the reactive flow process.
The constants of ignition I and growth G are set to
3.177� 108 s�1 and 0.7045 s�1 Pa�b, respectively. The pres-
sure sensitivity of the explosive is 0.7, and the compression
sensitivity is 4. Table I shows comparison of the pressure sen-
sitivity values and the literature values. The sensitivity of each
energetic material was derived by the theoretical argument
without any speculations based on numerical trials, which is an
improvement from the earlier reactive flow models.
IV. RATE STICK SIMULATION
The multi-material numerical simulation of shock-
to-detonation transition is conducted with a hydrocode.13 A
level-set based reactive Ghost Fluid Method (GFM) that
imposes exact boundary conditions at material interfaces is
developed to reproduce the experimental size effect data of
the unconfined rate stick tests.
The conservative laws of mass, momentum, and energy
equations in two-dimensional axisymmetric coordinate are
defined by
@ ~U
@tþ @
~E
@zþ @
~F
@r¼ ~S ~Uð Þ; (20)
where the vectors represent the conservative variables U, spa-
tial fluxes in axial and radial directions E and F, and the source
term S that represents a variety of multi-material loading con-
ditions. In the Eulerian form, each vector is given as follows:
~U ¼
q
quz
qur
qE
266664
377775 ~E ¼
quz
qu2z þ p
quruz
uz qEþ pð Þ
266664
377775 ~F
¼
qur
quzur
qu2r þ p
ur qEþ pð Þ
266664
377775~S ¼ �u
1
r
qur
quruz
qu2r
ur qEþ pð Þ
266664
377775; (21)
where u¼ 0, 1 for rectangular and cylindrical coordinates,
respectively. Here, q is the density, and uz and ur are velocity
components in axial and radial coordinates. E ¼ eþðu2
z þ u2r Þ=2 is the total energy per unit mass with the spe-
cific internal energy e, and p is the hydrostatic pressure,
respectively. The governing equations are solved by a third-
order Runge-Kutta in time, and convex ENO (essentially
non-oscillatory) method discretizes the spatial fluxes.14
The rate of burnt mass production is governed by
@ qkið Þ@tþ@ qkiuj
�@xj
¼ wi; (22)
where wi is the reaction rate and ki is the reaction progress
variable or burned mass fraction. The model is closed by
specifying the equation of state, and the form for the rate
law. Both Mie-Gruneisen EOS and JWL EOS are used to
provide closure to the formulation. The parameters related to
these constitutive relations are given in Table II.
The control of the time step increment is determined
automatically, where the step size is given by
DtCFL ¼ CFLmin Dz;Drð Þ
max uþ c; u; u; u� c½ �: (23)
Here, the sound speeds for Mie-Gruneisen EOS (unreacted)
and JWL EOS (reacted) are given by
c2unreacted ¼
1
qq0
1� gCg
PH þ Cq e� eHð Þ�
þ 2S
1þ g 1� Sð Þ PH � CqeHð Þ�; (24)
TABLE I. Comparison of the pressure sensitivity of selective energetic
materials.
Energetic materials
Pressure sensitivity
b - Present b - Others
LX-17 1.7 0.67 (Ref. 10)
ANFO-K1 1.38 1.3 (Ref. 11)
ANFO emulsion with 30% HMX 1.7 1.5 (Ref. 12)
Creamed TNT 1.0 1.75 (Ref. 12)
Urea nitrate (0.69 g/cm3) 1.0 1.0 (Ref. 12)
HANFO 0.95 0.8 (Ref. 3)
TNT 1.0 …
HMX emulsion 1.7 …
Non-ideal AN emulsion k1a 1.45 1.5 (Ref. 3)
Extremely non-ideal AN 0.8 0.8 (Ref. 3)
ANFO prill 1.65 1.5 (Ref. 3)
Potassium chlorate/sugar 1.05 0.8 (Ref. 3)
TABLE II. Equation of state coefficients and rate law parameters for the
heavily aluminized RDX.
Model parameter Value
Reactant C0 (mm/ls) 2.60
S 1.86
C 0.99
Product A (GPa) 2633
B (GPa) 8.59
C (GPa) 1.09
R1 6.68
R2 1.11
w (J/g K) 0.09
Chemical kinetics I (s�1) 3.18� 108
A 4.0
G (s�1 Mbar�b) 3.53� 107
B 0.7
TABLE III. Chapman-Jouguet values of the heavily aluminized RDX.
C-J condition Pcj (GPa) 16.14
qcj (g/cc) 2.2
ecj (kJ/cc) 1.54
Tcj (K) 4410
c 4.24
023512-4 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
c2reacted ¼
q0
q2AR1 exp �R1
q0
q
� ��
þBR2 exp �R2
q0
q
� �þ C
1þ x
q0=qð Þ2þx
#: (25)
The combined sound speed is then calculated by using
c2total ¼ ð1� kÞc2
unreacted þ kc2reacted: (26)
To evaluate the predictability of a modeled SDT, the
unconfined rate stick experiment is conducted. A series of
cylindrical charges with 0.1 m in length and inverse radii
being 0 mm�1 to 0.125 mm�1 is used and shock-initiated by
an impact on one side with 300 m/s. The prescribed impact
was sufficient for pressure rise above the threshold value of
the initiation.
The C-J conditions for heavily aluminized RDX are
given in Table III obtained by using a thermo-chemical
code.15 Most of the aluminum particles do not participate in
the reaction during the detonation phase16 as these particles
in micron size requires a significantly long time to be oxi-
dized. Since the reaction time is ultra-short, these metal par-
ticles do not have enough time to participate in the reaction.
The RDX would still follow its sole reaction path without
the particle effect. For a different particle size distribution,
however, smaller particles are ignited in a relatively shorter
time, thus providing extra heat release of the reaction.
Figure 3 shows computational domain. Two distinct
materials which are void and explosive were initially brought
to a contact by a zero level set represented by a black line in
the figure. Six different radiuses are considered in the simu-
lation, which were carried out on a uniform mesh of size
1/120 mm.
Figure 4 is a plot of the detonation wave structures illus-
trated according to grid resolution. The lines are represented
by dotted (h¼ 0.05 mm), dashed (h¼ 0.0167 mm), and solid
(h¼ 0.0083 mm). As seen from the figure, optimal mesh size
of h¼ 0.0083 mm seems to support convergence of a detona-
tion wave.
The length of the reaction zone of heavily aluminized
RDX is obtained from the detonation pressure profile and the
burned mass fraction in Fig. 5, and it is approximately
4.0 mm. At that time, pZND¼ 35 GPa, pCJ¼ 16 GPa, and
pZND=pCJ¼ 2.1875. For reference, the length of the reaction
zone of a pure RDX is 0.4 mm–0.5 mm.17,18 This suggests
FIG. 3. Computational domain of a 2-D cylindrical duct for unconfined rate
stick simulation.
FIG. 4. Convergence of detonation wave front shown with grid resolution.
FIG. 5. Structure of detonation wave representing the reaction zone of
heavily aluminized RDX.
FIG. 6. Density contours for uncon-
fined rate stick simulation of 0 (top),
0.075 mm�1 (middle), and 0.125 mm�1
(bottom) inverse radius cases with den-
sity range 0.1–2.4 g/cc.
023512-5 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
that the reaction zone length of a heavily aluminized RDX is
10 times that of a pure RDX.
Figure 6 depicts density contours that account for the
detonation front propagation and expansion of exhaust gas
outwards in the cases of infinite (0 mm�1), 13.33 mm
(0.075 mm�1), and 8 mm (0.125 mm�1) radius. Behind the det-
onation front, most of reactants are consumed and high pres-
sure product gas gives rise to an extended reaction zone length.
The reaction starts at the shock front and ends along the sonic
point (C-J state). The compression region occurs just ahead of
the reaction zone in both cases. Results show that the detona-
tion velocity is strongly dependent on the size of the charge.
Figure 7 describes propagation of detonation wave for
0 and 0.125 mm�1 unconfined rate sticks. A series of pres-
sure profiles shows the transition and stabilization of a von
Neumann spike. The figure indicates that there are two dis-
tinct regions: (i) run up region that represents fluctuation of
detonation pressure, and (ii) steady region that describes the
convergence of detonation flow.
There is a difference in the detonation velocity of 0 and
0.125 mm�1 rate sticks. The mean Zeldovich-von Neumann-
FIG. 7. Propagation of detonation front in 0 and 0.125 mm�1 unconfined
rate sticks.
FIG. 8. Size effect curve for the unconfined rate stick tests for heavily
aluminized RDX.
FIG. 9. (a) Schematic of a charge detonation simulation and (b) calculated
pressure history at each gauge.
TABLE IV. Comparison of the fully developed detonation velocity for dif-
ferent charge radiuses.
Inverse radius (mm�1)
Detonation velocity (m/s)
Numerical Experimental
0 7403.3 7387 (est.)
0.025 7348.9 7322 (est.)
0.050 7207.1 7197
0.075 7018.3 7044 (est.)
0.100 6829.9 6848
0.125 6544.2 6565
023512-6 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
Doering (ZND) pressures in the steady region for these
rate sticks are calculated as being pZNDjr¼0 mm�1 ¼ 35 GPa,
pZNDjr¼0:125 mm�1 ¼ 27 GPa, respectively. This suggests that
pressure gradient determines the detonation velocity of a rate
stick as such that the chemical energy pumping in the
shocked region is decreased as the stick diameter is reduced.
The size effect curves showing detonation velocity ver-
sus inverse radius are shown in Fig. 8. The hydrocode simu-
lation is shown to reproduce the detonation velocities of six
sample radii of unconfined rate sticks (0 mm�1, 0.025 mm�1,
0.050 mm�1, 0.075 mm�1, 0.100 mm�1, and 0.125 mm�1).
Table IV shows comparison of the fully developed
FIG. 10. Detonation of RDX followed by the aluminum afterburning at different times: (a) Pressure, (b) RDX-product mass fraction, and (c) Al-product mass fraction.
023512-7 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
detonation velocity. The size effect data using the present
rate law and parametric study are in good agreement with
the experimental data, suggesting that the model effectively
reduced steps in the parameter determination and solved the
over-simplification issue related to JWLþþ model.
V. PRESSURE CHAMBER TEST OF A HEAVILYALUMINIZED RDX AND ITS SIMULATION
Another validation of the model is pursued using the
pressure chamber test. A spherical charge of 3.5 cm radius
with 1.78 g/cc density is placed in a chamber maintained at
1 bar and detonated. Two pressure gauges are placed at two
radial distances at 1.6 m and 1.9 m from the initial position
of a charge. The purpose of the test was to obtain a primary
detonation signal (pressure) followed by the sequent pressure
peak representative of the aluminum afterburning. In order
to simulate both chemical processes of detonation and aero-
bic reaction of aluminum, we use a multi-ignition reaction
scheme for combined RDX and aluminum.
As before, we assume that none of the aluminum
particles reacted during the detonation of RDX. The after-
burning of the aluminum is modeled following the work of
Ref. 19, which was originally applied for aluminum combus-
tion in a gas phase. The aluminum ignition mechanism at
high pressure condition is utilized to predict the afterburning
of a heavily aluminized RDX.
A two-step Arrhenius-type mechanism for modeling the
subsequent burning of aluminum that follows a detonation of
RDX is:
T > 2000 K; kAl > 0:01;
dkAl
dt¼ A q0 1� kAlð Þ
�d2
p exp � Ea
RT
� �;
Ea ¼ 22:8 kcal=mol; A ¼ 1� 10�2 �
; (27)
T < 2000 K; kAl � 0:01;
dkAl
dt¼ A q0 1� kAlð Þ
�exp � Ea
RT
� �;
Ea ¼ 60 Kcal=mol; A ¼ 6:25� 1010 �
; (28)
qAl0 ¼ aq0; qAl ¼ ð1� kAlÞqAl0 ; (29)
kT ¼ akAl þ bkRDX; (30)
where dp or the particle diameter is 5 lm, while all other con-
stants are referred from Ref. 19. kAl is the final product mass
fraction of aluminum. A is the pre-exponential factor, and Ea
is the activation energy. qAl0 is the initial density of alumi-
num in the mixture, and qAl is the calculated density of
unreacted aluminum. a is 0.35 for aluminum and b is 0.65
for combined RDX/binder, both of which represent the initial
mass percent of the mixture. Then kT represents total final
product mass fraction. Also, the heat release due to alumi-
num reaction is _wAlqAl, which is added to the energy source
S of Eq. (21), where _wAl is the aluminum reaction rate and
qAl is the enthalpy of reaction of aluminum (85 000 kJ/kg).20
As for the EOS of gas phase aluminum, we adapted the
Noble-Abel equation of the form
PAl ¼akAlqRT
1� nAn; nAn :� 0:75ð Þ; (31)
where R denotes gas constant, n is the number of moles per
unit volume, and An is an empirical constant. Also, total
pressure is defined as
PTotal ¼ ð1� kRDXÞPunreacted þ kRDXPreacted þ kAlPAl: (32)
Figure 9 shows calculated pressures taken from gauges
placed at 0.04 m (gauge 1), 1 m (gauge 2), 1.6 m (gauge 3),
and 1.9 m (gauge 4) along the radial distance from a charge
center at zero. The radially detonating RDX in the chamber
shows a rapid decrease in the pressure.
Figure 10 shows a series of timed images of the chamber
test. Aluminum is ignited at 650 ls behind the detonated
RDX gases. The burnt aluminum mass propagates toward
the center as well as radially outward, because the fresh alu-
minum fuel is distributed within the radius of the hot product
gas of RDX. Both primary peak of RDX and secondary peak
of the afterburning are shown in the simulations.
Figure 11 shows a comparison of pressures calculated
and measured at gauge locations 1.6 and 1.9 m. Calculated
RDX peak slightly over-predicts the measurement, while the
FIG. 11. Comparison of pressure calculated and measured (a) at 1.6 m and
(b) at 1.9 m gauge locations.
023512-8 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24
subsequent aluminum peak agrees well with the measure-
ment. From these figures, the primary wave average velocity
is 1730 m/s and the subsequent afterburning velocity is 410
m/s as estimated at gauge 1.9 m. One fourth of the RDX det-
onation wave speed approximates the average velocity of the
afterburning of aluminum.
VI. CONCLUSION
A reactive flow model aimed at quantifying the detona-
tion response of energetic materials subjected to an external
shock impact is presented. A detailed explanation for deter-
mining the free parameters of the model is provided. Two
validation experiments are performed using a heavily alumi-
nized RDX. The unconfined rate stick experiment offered
detonation velocity as a function of the charge radius. A
pressure chamber test offered the time delay between first
and subsequent reaction peak. The hydrocode simulation that
utilized the present chemical mechanism confirmed that the
proposed reactive flow model reproduced the experimental
measurements, and it can overcome the limitations set by the
earlier reactive flow models.
ACKNOWLEDGMENTS
This work was supported by ADD-HH-5 through the
Institute of Advanced Aerospace Technology at Seoul
National University.
1E. L. Lee and C. M. Tarver, “Phenomenological model of shock initiation
in heterogeneous explosives,” Phys. Fluids 23(12), 2362–2372 (1980).2P. C. Souers, S. Anderson, J. Mercer, E. McGuire, and P. Vitello,
“JWLþþ: A simple reactive flow code package for detonation,”
Propellants Explos. Pyrotech. 25(2), 54–58 (2000).3P. C. Souers and P. Vitello, “Analytic model of reactive flow,” Propellants
Explos. Pyrotech. 30(5), 381–385 (2005).4C. M. Tarver, J. O. Hallquist, and L. M. Erickson, “Modeling short pulse
duration shock initiation of solid explosives,” in Proceedings of the 8thSymposium on Detonation (1985), pp. 951–961.
5J. E. Guilkey, T. B. Harman, and B. Banerjee, “An Eulerian-Lagrangian
approach for simulating explosions of energetic devices,” Comput. Struct.
85, 660–674 (2007).
6E. L. Lee, H. C. Hornig, and J. W. Kury, “Adiabatic expansion of high
explosive detonation products,” LLNL Report No. UCRL-50422, 1968,
pp. 1–21.7J. Wackerle, J. O. Johnson, and P. M. Halleck, “Shock initiation of high-
density PETN,” in 6th Symposium on Detonation (Office of Naval
Research, ACR-221, 1976), pp. 20–29.8J. Wackerle, R. L. Rabie, M. G. Ginsberg, and A. B. Anderson, “A shock
initiation study of PBX-9404,” in Proceedings of the Symposium on HighDynamic Pressures, LA-UR-78-1219 (1978), pp. 127–139.
9H. Eyring, R. M. Powell, G. E. Duffey, and R. B. Parlin, “The stability of
detonation,” Chem. Rev. 45(144), 69–181 (1949).10P. Vitello and P. C. Souers, “The piece wise linear reactive flow rate
model,” AIP Conf. Proc. 845, 507–510 (2006).11K. H. Kim and J. J. Yoh, “A particle level-set based Eulerian method
for multi-material detonation simulation of high explosive and metal con-
finements,” Proc. Combust. Inst. 34, 2025–2033 (2013).12P. C. Souers, S. Anderson, E. McGuire, M. J. Murphy, and P. Vitello,
“Reactive flow and the size effect,” Propellants Explos. Pyrotech. 26,
26–32 (2001).13J. J. Yoh and K. H. Kim, “Shock compression of condensed matter using
Eulerian multi-material method: Applications to multidimensional shocks,
deflagration, detonation, and laser ablation,” J. Appl. Phys. 103(11),
113507 (2008).14D. Enright, R. Fedkiw, J. Ferziger, and I. Mitchell, “A hybrid particle level
set method for improved interface capturing,” J. Comput. Phys. 183(1),
83–116 (2002).15L. E. Fried, W. M. Howard, and P. C. Souers, “Cheetah 2.0 user’s man-
ual,” Report No. UCRL-MA-117541 Rev. 5, Lawrence Livermore
National Laboratory, Livermore, CA, USA, 1998.16K. B. Kim, W. Wilson, S. Peiris, C. Needham, C. Watry, D. Ortley, and
F. Zhang, “Effects of particle damage during detonation of
thermobarics on subsequent reactions,” in 21st International
Colloquium on the Dynamics of Explosions and Reactive Systems
(ICDERS) (2007).17B. G. Loboiko and S. N. Lubyatinsky, “Reaction zones of detonating
solid explosives,” Combust. Explos. Shock Waves 36(6), 716–733
(2000).18T. A. Litzinger, Y. Lee, and C. J. Tang, “Experimental studies of
Nitramine/Azide propellant combustion,” in Solid Propellant Chemistry,Combustion, and Motor Interior Ballistics, Progress in Astronautics andAeronautics, edited by V. Yang, T. B. Brill, and W.-Z. Ren (AIAA, New
York, 2000), Vol. 185, pp. 355–379.19K. Balakrishnan, A. L. Kuhl, J. B. Bell, and V. E. Beckner, “Ignition of
aluminum particle clouds behind reflected shock waves,” in 23rd
International Colloquium on the Dynamics of Explosions and Reactive
Systems (ICDERS) (2011).20A. L. Kuhl, J. B. Bell, and V. E. Beckner, “Heterogeneous continuum
model of aluminum particle combustion in explosions,” Combust. Explos.
Shock Waves 46(4), 433–448 (2010).
023512-9 Kim et al. J. Appl. Phys. 116, 023512 (2014)
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP:
147.46.241.235 On: Mon, 14 Jul 2014 03:54:24