Sie sind auf Seite 1von 15

Combustion, Explosion, and Shock Waves, Vol. 41, No. 1, pp.

2034, 2005

Combustion Models for Energetic Materials


with Completely Gaseous Reaction Products
L. K. Gusachenko1 and V. E. Zarko1

UDC 536.46

Translated from Fizika Goreniya i Vzryva, Vol. 41, No. 1, pp. 2440, JanuaryFebruary, 2005.
Original article submitted January 23, 2004.

From a brief review of published combustion models for energetic materials, it follows that considerable modeling efforts have been directed toward a more accurate
description of gas-phase processes. It is often assumed that gas-phase reactions play
a dominant role in burning rate control. At the same time, more and more arguments
have emerged suggesting a dominant role of condensed-phase processes for a number
of the most widely used homogeneous propellants in the rocket range of pressures.
However, serious problems remain in the modeling of such combustion regimes. A solution of these problems is proposed.
Key words: combustion models, nitramines, condensed phase

EXPERIMENTAL DATA ON COMBUSTION


OF HOMOGENEOUS PROPELLANTS
Below we consider energetic materials (EMs) capable of self-sustained combustion that melt near the
burning surface, partially decompose, and then partially
vaporize from the surface [ammonium perchlorate (AP),
RDX, HMX, ammonium dinitramide (ADN), hydrazine
nitroformate (HNF), etc.]. For such materials, as a rule,
empirical dependences of the mass burning rate m on
the pressure p and initial temperature T0 are known over
wide ranges [1, 2]. The dependences of m on the gas flow
velocity and mass forces as well as electrical and acoustic noises and burning rate responses have been studied
less extensively. However, appropriate techniques are
available, and, in principle, the necessary amount of
data can be obtained. At the same time, there are serious problems in experimental studies of the combustion
wave structure in the so-called rocket range of pressures
415 MPa, which is of particular applied interest. The
effectiveness of contact methods is limited by difficulties in miniaturizing measuring elements, for example,
thermocouple junctions or sampler probes. Chemicalcomposition measurements in the subsurface reaction
zone are not yet feasible. For a number of EMs, species
concentration profiles in the gas-phase zone of combus1

Institute of Chemical Kinetics and Combustion,


Siberian Division, Russian Academy of Science,
Novosibirsk 630090; zarko@kinetics.nsc.ru.

20

tion waves [3, 4] (up to 40 MPa for ADN) and temperature profiles in the entire combustion wave [59] have
been found (usually up to 9 MPa, and for HMX, up
to 50 MPa). The errors of thermocouple measurements
have been the subject of special studies. A detailed
discussion and accuracy estimates are given in [10, 11].
Knowledge of concentration profiles in the gas
phase is especially important if it is known that combustion is dominated by gas-phase reactions. In this
case, information on the profiles allows one to construct
an effective combustion model. Knowledge of the complete temperature profile in a combustion wave allows
one to elucidate which of the phases (gaseous or condensed) plays a dominant role in burning rate control
of the given substance under specified conditions.
Let the condensed phase be located on the right of
the burning surface (x = 0). In steady-state combustion, the heat flux qs,c = (dT /dx)s,c into the depth of
the condensed phase from the surface together with the
heat release qc due to condensed-phase reactions should
compensate all heat losses:
qs,c + qc = mH.

(1)

The quantity qs,c generally differs from the heat flux qg


from the gas to the surface because of heat loss effects
on the surface (for example, because of heat consumption in vaporization; see below). The quantity H is
the change in the enthalpy due to the heating of the
condensed phase from the temperature T0 to the sur-

c 2005 Springer Science + Business Media, Inc.


0010-5082/05/4101-0020

Combustion Models for Energetic Materials

21

face temperature Ts . If the heat capacity c is considered constant, H = c(Ts T0 ) + Qm , where Qm is the
heat of melting. In the unsteady case, the term mH
in (1) should be replaced by the heat loss (T /x)
through the cold boundary of the reaction zone.
The burning rate is controlled by subsurface
condensed-phase reactions if the heat flux qs,c from the
surface can be ignored, i.e., if qs,c  qc , which is equivalent to
 dT 
mH.
(2)
qs,c 
dx
In the opposite case, where qc  qs,c or qc 
(dT /dx) mH, the processes involving heat
transfer from the surface to the condensed phase are
dominant. These can be exothermic reactions in the
gas phase or on the surface (heterogeneous reactions).
The quantity qs,c can hardly be measured; therefore, to verify the dominant role of the condensed phase
in burning rate control, one has to use another condition
that a priori ensures that relation (2) is satisfied. To
obtain this condition, we use the surface-heat balance
expression
qs,c = qg + Qs Ws L[m(1 ) Ws ].

(1 )

Here qg = g (dT /dx)s,g +qr , g is the thermal conductivity of the gas, qr is the radiant flux absorbed by the
surface [as a rule, qr  g (dT /dx)s,g ; the transmitted
part of the radiant flux is also ignored], Ws and Qs are
the rate and heat effect of the heterogeneous reaction
on the surface (if such a reaction occurs), respectively,
L is the heat of vaporization, is the mass fraction of
the original substance that has decomposed under the
surface and left the condensed phase (because of the diffusion of individual molecules or motion of continuous
gas volumes in the form of bubbles or jets; see below)
without additional energy consumption. Next, we consider the case of no heterogeneous reactions (Ws = 0);
then, qs,c < qg and inequality (2) is a priori satisfied if
the following condition is valid:
qg = qs,c + Lm(1 )  mH.

(2 )

The approximate inequality (2 ) is preferred for estimation.


Thus, to answer the question of the role of the condensed phase in burning rate control, instead of the
complete profile it suffices to know the surface temperature Ts and the gas-phase temperature gradient
(dT /dx)g near the surface. The  sign indicates
that the concept of condensed-phase controlled combustion has an asymptotic nature: the smaller the ratio
qg /mH, the more this concept is justified (and the use
of appropriate Zeldovichs asymptotic formula).

Of course, to check condition (2 ), in addition to


Ts and (dT /dx)s,g one needs to know the parameters
qr , g , c, and Qm .
From results of thermocouple measurements [59],
it can be concluded that the condensed phase performs
burning rate control in the rocket range of pressures
for pure RDX, HMX, and ADN and following mixtures:
RDX (or HMX) + HTPB; RDX (or HMX) + GAP (glycidyl azide polymer). This conclusion is based on the
fact that, according to experimental data, the right side
of (2 ) exceeds the left side by several times (as a rule,
by an order of magnitude). For the indicated compositions, this ratio increases with pressure. In [12], the
surface heat balance components for HNF were calculated from results of thermocouple measurements only
for pressures of 0.04 and 0.1 MPa. It turned out that
at p = 0.04 MPa, the right side of (2 ) is smaller than
the left side and at p = 0.1 MPa, it is larger than the
left side. One might expect that in the rocket range of
pressures, the dominant role of the condensed phase in
burning rate control of HNF will be pronounced. For
CL-20 (hexanitrohexaazaisowurtzitane), such estimates
are not available. As to AP, we are not aware of data on
the hardly measured (for this EM) quantity (dT /dx)g
in the rocket range of pressures but modeling results
[1315] provide indirect evidence for condensed-phase
controlled combustion.

MATHEMATICAL SIMULATION.
CRITICAL REVIEW
A brief review of modeling studies of nitramine
combustion before 1997 is given in [16]. The data [16]
supplemented by data of later studies, including those
for other monopropellants that melt and vaporize during combustion, provide a general insight into the status
of the modeling studies of combustion of such EMs. According to Lavrentevs views on a system approach (see
the introduction to [17]) and the ideas of [18], In modeling combustion of EMs, Russian researchers, as a rule,
try to establish regimes with a dominant role of the gas
or condensed phase in burning rate control; transitions
from one regime to another are also explored. It should
be noted that foreign researchers often underestimate
the practical significance of this part of work.
A brief review of papers on combustion modeling
for some EMs is given below. Primary attention is paid
to accounting for the decomposition and vaporization of
the condensed phase and determining the dominant role
of a particular phase in burning rate control. We believe
that in order that a model be adequate to the process

22
being described, it must reproduce the experimentally
recorded localization of the burning rate control stage
in the combustion wave.
For AP, the assumption of condensed-phase controlled combustion was made as early as in [14]. The
burning rate was calculated using the Zeldovich formula and the equilibrium ratio on the surface for the dissociatively vaporizing melt. Removal of gaseous products from the liquid-phase reaction zone was not considered (which corresponds, for example, to version III of
Appendix B). Later, this approach was used repeatedly
by various authors to model the combustion of some
EMs.
Guirao and Williams [13] modeled AP combustion
with a similar problem
formulation. It was shown that the best fit of
the proposed combustion model to burning-rate experiments at pressures of 210 MPa is achieved under the
assumption that 70% of the total heat of AP combustion is released in the condensed phase. Of course, this
provides only indirect evidence in support of condensedphase controlled combustion for AP. Another indirect
but convincing argument follows from [13]: at p < 2 MP,
the AP surface temperature corresponding to equilibrium vaporization becomes lower than the melting point
and the liquid layer disappears. It is known that at
p < 2 MPa, self-sustained combustion of AP does not
occur. The disappearance of the liquid layer is the most
reasonably associated with combustion extinction if the
burning rate is assumed to be controlled by the condensed (more precisely, liquid) phase.
A more complex model for AP combustion is proposed by Tanaka and Beckstead [19] and Jing et al.
[20]. It includes 107 reactions in the gas phase, 17 in
the liquid phase, and even four reactions in the solid
phase. For the condensed phase, the kinetic constants
are chosen so as to match calculated and experimental data. Unlike in [13], in [19, 20] it is assumed that
100% of the original substance decomposes under the
surface and the gas is practically released in a jet-like
regime, b ub = c uc (see version II of Appendix B) but
the concept of bubbles in the subsurface reaction zone
is used. It is assumed [20] that 30% of the total heat
of AP combustion is released in the solid phase. Calculations of heat release (or absorption) in the liquidphase decomposition of the remaining substance are not
reported in [19, 20]; therefore, it is impossible to determine where the rate-controlling stages of the combustion process are localized. The weak point of the
model is neglect of the possibility of dissociative vaporization (which was taken into account in [13, 14]).
Apparently, the authors of the cited study believe that
if the substance has completely decomposed under the

Gusachenko and Zarko


surface, nothing remains to be vaporized. There are
two objections to this: 100% decomposition is postulated without additional arguments; in the jet regime
there is a branched interface, and, hence, there may
be subsurface vaporization, which in a similar situation
was taken into account, for example, in [21]. In [19, 20],
it is assumed that good agreement between calculated
and experimental concentration profiles is indicative of
the workability of the model. This is indeed so (provided that the experimental data are reliable) but only
in the neighborhood of the pressure p = 0.06 MPa, at
which the experiment was performed. Nevertheless, the
calculations were performed up to p = 12 MPa.
RDX combustion is considered in detail by Liau
and Yang [21]. The detailed kinetics of gas-phase reactions, three reactions in the liquid layer, and vaporization on the surface and in bubbles are taken into account. The release of gaseous products from the liquid
layer is calculated assuming the jet regime (see version
II of Appendix B) but the authors consider bubbles and
depict them in the figure. The reactions in bubbles are
also taken into account although, as is noted in [16], in
the jet regime and at temperatures below Ts , they have
not time to significantly change the composition of the
gas being released. Indeed, in the jet regime, the velocity of the gas leaving the reaction zone is equal to the
velocity at which the gas flows from the surface. The
temperature profiles in [8] show that the distance from
the surface to the end of the zone of intense heat release
in the gas phase is considerably larger than the dimension of the condensed-phase reaction zone. From this it
follows that the residence time of the gas in the jets under the surface is much smaller than the residence time
in the gas-phase reaction zone above the surface. In
addition, the temperature (and the rate of composition
consumption W ) is lower under the surface. From
R this
it follows that the degree of conversion of the gas W dt
is much lower under the surface than above it. In other
words, the homogeneous reactions in the subsurface gas
in the jet release regime can be ignored, and the possibility of accounting for heterogeneous reactions requires
special investigation. Liau and Yang [21] used the experimental data of [22] (the gas-phase species concentration profiles at p = 0.05 MPa) and [5] (the burning
rate and the complete temperature profile in the pressure range 1 < p < 9 MPa). The kinetic constants of
the liquid-phase reactions were chosen so as to match
calculated gas-phase species concentration profiles with
the data of [22] at p = 0.05 MPa. The results were then
compared with the data of [5] at 1 < p < 9 MPa. Good
correlation was obtained between the dependences Ts (p)
and h(p) (h is the thickness of the molten layer). However, in [21], the calculated heat fluxes qs,c and qg are

Combustion Models for Energetic Materials


not compared with each other and with the experimental values obtained in [5]. This does not allow one to
draw unequivocal conclusions on the localization of the
burning rate control stage from the calculation results.
Liau et al. [23] constructed using the problem
formulation of [21] and performed calculations for a
pseudo-propellant based on nitramines. The constants
were chosen so as to match calculated gas-phase species
concentration profiles and temperature profiles with
measurement data [24] at p = 0.1 MPa; the calculations
were performed up to p = 10 MPa. Unfortunately, Liau
et al. [23] do not use for comparison the results of Zenin
and Finjakov [8], who measured the full temperature
profiles in the rocket range of pressures for the pseudopropellant HMXGAP studied in [23]. The information
given in [23] does not allow conclusions on the localization of the rate-controlling stage.
For RDX, Homan et al. [4] used a formulation similar to that of [22, 24] in description of the gas phase.
The condensed phase is described by the single relation
of the type of the pyrolysis law m exp(const/Ts )
taken from [5], and any reactions and heat release under
the surface or on it are ignored. The authors of the cited
paper believe that good agreement between the pyrolysis law and the experiment with nitramines [5] give
reasons to use this law for the case of complete (100%)
vaporization of RDX assumed by the model. This interpretation of the data of [5] seems insufficiently justified. In addition, it should be noted that in a paper [6],
which give experimental data for HMX (a close relative
of RDX) at 103 < T0 < 373 K, the pyrolysis law
[or the universal correlation between the mass burning
rate and the surface temperature m exp(const/Ts )]
are not declared, probably because the measurement
data imply that m = m(Ts , T0 ). In [4], it would be
more reasonable to use the ClapeyronClausius relation on the free surface rather than the pyrolysis law
m = m(Ts , T0 ). Homan et al. [4] did not try to extend
their model to the case of high pressures and obtained
good agreement of calculations at p = 0.10.2 MPa
with the temperature data of [5] and the concentration
profiles obtained from their own optical measurements.
Most likely, the model cannot be used without modifications in the rocket range of pressures for the EMs
explored in [59], where the heat release in the liquid
phase (ignored in [4]) becomes the major source of heating of this phase.
Ward et al [25] proposed a model for HMX taking into account subsurface heat release and using the
MerzhanovDubovitskii relation. Vaporization is ignored, and an analytical solution for the gas phase is
obtained under the assumption of one global reaction
with almost zero activation energy. Excellent agree-

23
ment is obtained between calculation results and the
data of [5] (full temperature profile) at p = 2 MPa. The
problem of the dominant role of a particular zone in
burning rate control was not discussed in [25], but from
the results of processing of this temperature profile in
[5, 6] it follows that the combustion regime is close to
condensed-phase controlled combustion. The question
of whether the model of [25] can reasonably describe the
entire data set of [5, 6] (various values of p and T0 ) without changes in the remaining input parameters remains
open .
The same model was used in [2628] (and taking
into account that the heating zone of the condensed
phase is nonstationary in [29]) for HNF; the results are
in good agreement with experimental data on the burning rate over wide ranges of pressures and initial temperatures. In these papers, too, the problem of the dominant role of a particular zone in burning rate control
was not considered.
Louwers [26] proposes a model for HNF with detailed gas-phase kinetics. Vaporization is taken into account only in the heat balance. Examining the possibility of accounting for vaporization, Louwers [26] rightly
criticizes the approach of [21] (where the combustion
model includes the difference between large close values: the vaporization rate and the condensation rate).
However, instead of the seemingly reasonable use of the
ClapeyronClausius relation in his model, Louwers [26]
calculated the surface temperature in the same manner
as in [25] (using the MerzhanovDubovitskii formula).
Because of the brevity of the discussion in [26], it remains unclear how this approach is combined with the
simultaneous allowance for the liquid-phase reactions in
the thermal-conductivity equation. The concentration
profiles for some gas-phase species calculated using the
model for HNF (and even for 95/5 HNF/graphite and
90/10 HNF/paraffin mixtures) are in good agreement
with experimental data at atmospheric pressure. Unfortunately, the data given in [26] do not allow conclusions on the presence of a burning rate control zone in
the combustion wave.
Sinditskii et al. [30] and Fogelzang et al. [31] analyzed the combustion of ADN and some other EMs.
A detailed combustion model is not given, but the burning rate is estimated by the Zeldovich formula assuming
condensed-phase combustion control, and the possibility of loss of this control is discussed. In [30, 31], the
following important assumption on the ADN combustion mechanism is made: in the liquid phase, this EM
irreversibly decomposes into nitrogen oxide and liquid
ammonium nitrate (AN), which then can dissociatively
vaporize from the surface. From this, it can be concluded that for ADN, the classical model, which as-

24
sumes parallel decomposition and vaporization of the
liquid phase of the original substance, requires modification.2 However, the key assumption of [30] that with
a pressure rise there is transition from the condensedphase controlled regime to a gas-phase controlled regime
seems insufficiently justified and requires further investigation. According to [30, 31], for ADN and AP, this
transition occurs at p = 2 and 28 MPa, respectively,
and for other EMs, a trend toward this transition at
elevated pressures is also observed. When these pressures are reached, the heat release in the liquid layer
is no longer sufficient to heat the EM to the surface
temperature. In particular, according to the data of
[30], for ADN at p = 2 MP, the heat effect of the reaction Q = 400 cal/g coincides with the amount of heat
H required to heat ADN to the temperature Ts . Here
H = c(Ts T0 )+Qm , where Qm is the heat of melting
(30 cal/g) and c = 0.59 cal/(gK) is the heat capacity;
in [30], the source of the thermal data is not indicated;
we note that in [7], the heat capacity of ADN is taken
to be 0.3 cal/(g K). From this estimate it follows that
Ts = 920 K. This is apparently the equilibrium temperature of the ADN surface at the saturated vapor pressure above it (more precisely, the pressure of dissociative
vaporization products) equal to 2 MPa. However, the
conditions in question do not contain saturated vapor!
In the calculation of the equilibrium temperature Ts using the ClapeyronClausius relation, the total pressure
above the substance (2 MPa) should be multiplied by
the molar fraction of vapor above the surface. This fraction is small if the original substance almost completely
decomposes in the liquid layer. In addition, the molar
fraction of vapor above the surface can be decreased by
dilution of the gas mixture coming from the surface with
the products diffusing from the gas-phase reaction zone
(for this see below). Therefore, the real surface temperature should be much lower than the boiling point at
the given pressure. Indeed, the measurements of [7] for
AND show that p = 2 MPa at Ts = 585 K. At the same
time, the possibility that with a rather large increase
in pressure, the gas phase begins to play the dominant
role in burning rate control, as predicted in [30], seems
reasonable and deserves a special study. For example,
studying the combustion of EMs in the artillery range
of pressures, Assovskii [33] assumes that with a pressure rise, the subsurface reaction rate is limited, which
suggests that the dominant role passes to the gas phase.
Meredith and Beckstead [34] propose an unsteady
model for HMX combustion. In the liquid phase, two
parallel reactions with heat release and heat absorption are taken into account. On the surface, the va2

However, it is reported in [32] that ADN vapor was found


in the experiments described there.

Gusachenko and Zarko


porization rate is calculated as the difference between
two large close values, but, unlike in [21], the species
concentration discontinuities are ignored, i.e., the long
flame approximation is actually used. The calculations
results are in good agreement with available experimental data on laser ignition of HMX and with the
steady-state data [6] for HMX over wide ranges of
pressures and initial temperatures. The burning rate,
surface temperature, and molten-layer thickness were
compared. Unfortunately, there is no information on
calculated heat fluxes from the gas onto the surface. As
noted above, the data of [6] suggest that in the rocket
range of pressures, the HMX burning rate is controlled
by condensed-phase reaction. In the formulation of [34]
at the same pressures, the dominant role belongs to the
gas phase. For example, at Ts = 773 K (which for
HMX corresponds to a pressure of 6 MPa) of the two
subsurface reactions examined in [34], the endothermic
reaction is faster, i.e., the total heat release in the condensed phase is negative.
Results of calculations using an unsteady combustion model for melting EMs are given in [35, 36]. One
condensed-phase reaction is considered, and vaporization from the surface and two consecutive gas-phase reactions were taken into account. The authors attempted
to describe RDX and HMX combustion, in which, according to the measurements of [5, 6], at low pressures,
the burning rate is controlled by the gas- phase reaction kinetics whereas in the rocket range of pressures,
a condensed-phase controlled regime undoubtedly exists. The model successfully predicts the establishment
of a steady-state combustion regime at low pressures;
at elevated pressures, however, instead of steady-state
combustion, a flash in the subsurface reaction zone occurs in the calculation. Special studies in [36] showed
that in the employed physical formulation, the instability is not due to the computational procedure but
it necessarily arises with a pressure rise as soon as the
condensed-phase reactions become intense enough and
a temperature maximum appears under the surface (see
Fig. 2). On the stability boundary (at a zero temperature gradient on the burning surface from the side of
the condensed phase), the substance that has not yet
decomposed vaporizes from the surface only due to the
heat flux q from the gas phase:
m(1 )L = qg .
(3)
Here is the mass fraction of condensed-phase decomposition of the substance and L is the heat consumption in vaporization of mass unit of the original substance. It is assumed that the subsurface-decomposition
products are only gaseous and pass through the surface
without additional heat consumption. The instability
after the attainment of the temperature maximum is

Combustion Models for Energetic Materials

25

caused by unsteady processes in the subsurface reaction


zone and, hence, cannot be described by the Zeldovich
Novozhilov model, which assumes that the indicated
zone is quasi-steady-state.
The results of [35, 36] make it necessary to critically assess other studies that model the unsteady combustion of EMs with vaporization and subsurface reactions and describe condensed-phase controlled regimes.
This is also true for steady-state combustion models that are implemented numerically with the use of
the establishment method. As regards purely steadystate models (with numerical calculation without using
the establishment method and analytical methods), the
stability of the steady-state regimes described by these
models is questionable if they predict the dominant role
of the condensed phase in burning rate control.

COMBUSTION MODEL FOR A PARTIALLY


VAPORIZING ENERGETIC MATERIALS
WITH THE DOMINANT ROLE OF
CONDENSED PHASE IN BURNING
RATE CONTROL. ZELDOVICHS
APPROXIMATION
The assumption of the dominant role of condensedphase reactions in burning rate control greatly simplifies
the construction of a mathematical model for combustion of melting and vaporizing EMs in the rocket range
of pressures (if it is required to calculate only the burning rate, it is possible, as a first approximation, to do
without gas-phase equations). Such steady-state models are described or used, to a greater or lesser extent,
in [14, 15, 30, 3639].
A widely used combustion model for vaporizing
EMs with the dominant role of condensed-phase reactions in burning rate control is described in [14, 15]:
m2 =

2QkR Ts2 exp(E/RTs )


,
E
c2 (Ts T0 )2
c(Ts T0 ) = Q,


1
L 
= const exp
.
1 + f
RTs

(4)
(5)
(6)

Here f is the ratio of the number of moles in the decomposition and vaporization of the substance [for example, the liquid-phase decomposition of HNF yields
five to seven (according to different data) gas molecules
and the dissociative vaporization of HNF yields two
molecules] and is the molecular weight; the ratio on
the left side of relation (6) is the molar fraction of the vapor of the original EM above the surface. This fraction
is assumed to coincide with the fraction of the molar flux

of this vapor in the total molar flux from the surface (the
so-called long flame approximation). In addition, formulas (4)(6) were derived using assumption (3). The
derivation of Eq. (4) for the mass burning rate is given
in Appendix A. It should be noted that in [14], the
Zeldovich formula is used for a first-order liquid-phase
reaction, whereas in [15] [and above in relation (4)], it is
used for a zero-order liquid-phase reaction. The model
predicts reasonable dependences on pressure and initial
temperature not only for the burning rate but also for
its partial derivatives with respect to these parameters.
In [14, 15], it is shown that the model describes the
possibility of occurrence of a minimum in the pressure
dependence of the burning rate temperature sensitivity. Such minima indeed take place for some EMs [1, 2]
and are not described by the other known combustion
models. In [38, 39], the ZeldovichNovozhilov stability
criterion was calculated using this model and regions of
instability for AP and ADN were obtained that are in
satisfactory agreement with the real regions. In addition, we note that relations (5) and (6) imply a positive
dependence of on pressure and a negative dependence
on the initial temperature. Indeed, at T0 = const, relation (5) implies a positive dependence of Ts on , which
after substitution into (6) yields a positive dependence
of p on and, hence, a positive dependence of on
p. At p = const, relation (6) implies a negative dependence of Ts on , which after substitution into (5)
yields a negative dependence of T0 on , and hence, a
negative dependence of on T0 . The latter result is far
from being obvious, and Zenin et al. [6], who found this
effect in experiments with HMX, called it an interesting
and unexpected. Later, they obtained the same result
for ADN and RDXHTPB and HMXHTPB mixtures
[79]. Of course, one should bear in mind that the model
of [15] has an asymptotic nature: its predictions are the
better the smaller the ratio qs,c /qc , whose value should
be estimated using models that take into account the
gas phase.
There is seeming inconsistency between the good
workability of the model of [14, 15] in the rocket range of
pressures3 and a series of seemingly rough assumptions
included in the model:
1) a zero-order liquid-phase reaction is assumed (in
[15]) although the degree of decomposition can be rather
high;
2) all thermal characteristics in the liquid phase
are taken for the original substance, which may seem to
be inconsistent with the above-mentioned high degree of
3

We imply, first of all, the agreement of the model with the


experimentally established fact of condensed-phase kinetic
control in combustion of some EMs in the rocket range of
pressures.

26
its decomposition and the possible formation of a foamy
layer;
3) relation (3) (or the equivalent relation qs,c = 0)
is inadequately justified physically.
The first two inconsistencies are eliminated by the
assumption that the gas is released from the liquidphase reaction zone by version III of Appendix B, i.e.,
by diffusion. If an appreciable fraction of the gas is released in the form of bubbles or jets, the model needs
to be improved.
To justify relation (3) at moderate pressures, we
discuss the data of experiments [59] taking into account the analysis in [36]. For all EMs explored in [59],
the subsurface part of the gas-phase temperature profile has an upward convexity, which is easiest to explain
by distributed heat release from the global reaction
with a low effective activation energy. On the subsurface part of the temperature profiles in the rocket
range of pressures, a convex upward region is not seen.
However, thermocouple measurements at such pressures
combined with an analysis of the surface heat balance
[59] suggest a considerable heat release in the liquid
phase of the examined EMs. Hence, a convex upward
region probably exists on the T (x) profile in this case,
and it is not seen most likely because its dimension (in
the x direction) is comparable to the thickness of the
thermocouple thickness (see the estimate in Appendix
B). As noted above, the data of [59] for the examined substances definitely indicate that in the rocket
range of pressures the burning rate is controlled by the
condensed-phase reaction kinetics but they do not allow one to draw a conclusion on the localization of the
burning rate control zone at lower pressures. Hence, at
low pressures, relation (3) can have the < sign, which
corresponds to heat loss from the surface to the liquidphase reaction zone. As the pressure increases, the situation becomes the opposite. First, a weak temperature maximum arises in the reaction zone, and then,
according to [31], oscillatory instability occurs (of high
frequency with the oscillation period corresponding to
the thermal relaxation time of the reaction zone). Figure 1 illustrates the simplest explanation of the instability of the temperature profile with a maximum. An
occasional fast displacement of the burning surface for a
small distance to the left produces a vertical section on
the temperature profile (very considerable heat supply
from the liquid to its new surface, on which vaporization
occurs), which causes further motion of the surface to
the left at a high velocity, and so up to point D. The profile changes from ABCDE through ABFCDE to ABDE.
Region BD vaporizes almost instantaneously. A random jump for an even small distance to the left from D
has an inverse effect (great heat loss from the surface

Gusachenko and Zarko

Fig. 1. Occurrence of instability.

into the liquid and temporary decay) or, in other words,


negative feedback that stabilizes the process. A more
rigorous proof of the combustion instability (investigation of the thermal problem with reactions using the
small perturbation method) is given in [36].
We assume that this instability somehow acts, according to Le Chateliers principle, against the factor
that causes it and does not permit the process to extend into the region of instability. Then, with further
increase in pressure, the following relation holds:
0 < m(1 )L/qg 1  1.

(7)

This condition ensures condensed-phase kinetic control


over a wide range of pressures (for elevated values of p)
and the applicability of model (4)(6) at the expense of
the presence of weak high-frequency oscillations. They
are not necessarily synchronous on the entire burning
surface and can exist in the form of flickering spots or
running waves. Ananev et al. [40] report that oscillations of this type are observed over a wide range of pressures for many EMs, including the region of Zeldovich
Novozhilov combustion stability. We note that in running waves, mass, heat, momentum, and even information are not transferred along the surface. Oscillations
at different sites of the burning surface differ in phase,
and the observed wave is a phase change wave.4 It is
known that the phase velocity can exceed the velocity
of light, and in this limiting case, it is easiest to understand that the phase change wave does not transfer
anything real.
The concepts of asynchronous (incoherent) oscillations of the burning rate at different sites of the burning
surface allow the formulation of one of the mechanisms
of stabilizing action with negative feedback (Le Chateliers principle) in the process considered. The abovementioned asynchronous (random) oscillations of the
4

This term as applied to the processes related to combustion


was used, for example, in [41] for ignition waves.

Combustion Models for Energetic Materials

27

flow velocity generate gas-phase turbulence, which increases the effective coefficients of heat and mass transfer, and, hence, the heat flux from the gas phase onto
the surface. Turbulence generated by combustion near
the burning surface was considered in [42, 43], where
gas-phase perturbations due to the heterogeneity of the
original EM were studied. Here we discuss a different
factor responsible for turbulence that also acts in the
combustion of an ideally homogeneous EM. The higher
the overheat of the condensed phase in the fluctuation,
the more effectively the turbulence increases the average heat flux from the gas. This favors an increase in
the burning rate, and, hence, a decrease in the residence
time of the substance in the liquid-phase reaction zone.
Therefore, the overheat of the subsurface reaction zone
decreases.
The increase in the subsurface transfer coefficients
(and, hence, the heat flux from the gas) can be estimated qualitatively. Let a zero-order reaction proceeds
in the liquid: c /t = W (T ). As the pressure increases, the values of qs,c decrease and, beginning with
qs,c = 0, self-oscillations occur with the characteristic
time
tosc = c /W (Ts ).

(8)

These oscillations, disordered in phase along the surface, generate subsurface turbulence with the mixing
coefficient
Deff = const(vg )2 tosc ,

(9)

where vg is the fluctuation of the gas outflow velocity. Thus, negative feedback occurs that limits the selfoscillation amplitude: a random increase in the amplitude m (and hence, vg ) enhances the mixing and
increases the average value of qg , resulting in a decrease
in m.
The other possible stabilization mechanism is related to the existence of the limiting nonequilibrium
solubility of the gaseous products of liquid-phase decomposition in the original liquid. When this limit is
reached, an explosion-like gas release occurs. In this
case, part of the heat is expended in changing the state
of the substance and the other part is carried away from
the subsurface reaction zone in the form of the enthalpy
of the released gas or dispersed condensed products (if
dispersion occurs).
In the classification of [18], condensed-phase kinetically controlled regimes should be called flame detachment regimes. In the classical case [18], the second
detached reaction zone is carried away by the flow and
burns at a certain distance from the first zone in a thermal explosion regime, so that the second reaction zone
moves at the same velocity as the first relative to the
original substance. In our case, the pattern is changed

somewhat by the instability described above. Instead


of a permanent thermal explosion, the propagation of
the second zone is accelerated (to the velocity of the entire combustion wave) by turbulence, which moves the
second zone closer to the liquid surface, so that this
zone could transfer heat to the surface to vaporize the
unreacted liquid in accordance with Eq. (7).

POSSIBILITIES OF IMPROVING
THE MODEL
The model can be improved in the following ways.
1. Abandonment of the long flame approximation,
i.e., accounting for diffusion in the gas phase. This
changes the content of the concept of the dominant
role of the condensed phase in burning rate control:
although in the modified model, as in the model of
[14, 15], the gas phase does not supply heat to the subsurface reaction zone, gas-phase processes can affect the
burning rate via the degree of EM decomposition .
This leads to the necessity of modeling the gas phase
and, hence, to the problem of accounting for the abovementioned turbulence. It is the easiest (if the turbulence is ignored) to use the result of [25] and to assume
that because of dilution with the final products, the
molar fractions x1 and x2 (see Sec. 5 below) of vapor
that passed through the surface should be multiplied
by Ys = Hf /Qg . Here Hf and Qg are the change
in the enthalpy and heat effect of the gas-phase reaction. Condition (3) implies satisfaction of the energy
balance Qg = Hf + qg = Hf + (1 )L1 + m L2 ,
where m is the mass fraction of liquid in the subsurface
decomposition products, so that
Ys = 1

(1 )L1 + m L2
.
Qg

2. Accounting for the nonlinearity of the dependence H(T ): generally, c(Ts T0 ) in expressions (4) and
(5) should be replaced by H H(Ts )H(T0 ), so that
in view of melting and for c = c0 +c1 T , relations (2) and
(3) include H Qm + c0 (Ts T0 ) + 0.5c1 (Ts2 T02 )
instead of c(Ts T0 ). Strunin et al. [14] used this in
their subsequent studies.
3. If the variability of L is taken into account,5 the
RTs
exponent in (6) should have the form L(T )/RT 2 dT ,
which follows from the general formula d ln ppart /dTs
= L/RTs2 .
5

For example, for RDX and HMX, the dependence L(T ) can
be approximated using the pairs of values L and T given
in [44] for the melting and boiling points and the critical
temperature.

28

Gusachenko and Zarko

4. For more complex condensed-phase reaction kinetics, Eq. (4) becomes


2

(mH) = 2

ZTs

(4 )

(T )dT.

Here (T ) is the heat-release rate in unit volume of the


liquid phase. For the single zero-order reaction

assumed to be controlled by subsurface reactions, then


Eq. (5), as before, includes the subsurface degree of
decomposition but in Eq. (6) it should be replaced
by the total degree of decomposition 1 = + Ws /m.
Since this model assumes no heat flux from the surface
to the liquid, which is equivalent to the dominant role
of subsurface reactions, instead of (3), the following relation holds:

(4 )

(T ) = Qk exp(E/RT ),

Eq. (4 ) implies (4). In the case of several parallel reactions, the right side of (4 ) contain several similar
terms. In the case of several consecutive reactions, it
is necessary to distinguish the rate-controlling reaction
and then we again obtain an expression for m of the
form (4). The arguments given above suggest that the
gaseous intermediate and final products practically do
not participate in the subsequent liquid-phase reactions
(for example, because of the small residence time).
5. Accounting for the fact that the molar fraction
of the subsurface decomposition products is a liquid
that should then vaporize from the surface. In this case,
according to [45], instead of (6), the following relation
should hold:
 L  1
X
1  1
i i
= .
(6 )

xg,i exp
R
T
T
p
s
boil,i
i
Here the subscripts i = 1 and 2 refer to the original
substance and the liquid part of its subsurface decomposition products, respectively, x is the molar latent
fraction of vapor above the surface, L is the molar
heat of vaporization, and Tboil is the boiling point at
atmospheric pressure. If a molecule of the original substance decomposes into rr parts in subsurface decomposition and dissociates into rv parts in vaporization,
then in the flow leaving the surface, one mole of the
original substance produces (1 )rv moles of vapor
and rr moles of products, including rr moles of the
vapor of the initially liquid product. In the long flame
approximation, assuming that the concentrations above
the surface are in the same ratio as the molar fluxes, we
have
1
f
rr
x1 =
, x2 =
, f= .
1 + (f 1)
1 + (f 1)
rv
6. Accounting for the possible reaction on the surface. The question is the decomposition reaction on the
liquid surface that is used above in balance (1 ):
2

Ws = Bs act (p/patm ) exp(Es /RTs ) [g/(cm sec)].


(1 )
Here act is the molar fraction of active components
that diffuse from the gas-phase reaction zone to the liquid surface. In this case, as before, the burning rate is

qg = Lm(1 1 ) QWs .

(3 )

It should be noted that the measurement data of [59]


can be treated from a view point of the dominant role
(for some EMs in the rocket pressure range) of processes
both in the condensed phase and on its surface. Since in
real EM, the conventional thick thermocouples cannot detect a region with a weak dependence T (x) within
the very thin zone of subsurface reactions, the existence
of this region for the rocket pressure range remains a
hypothesis, though rather plausible. An alternate hypothesis is that in the rocket pressure range, the examined substances do not have the above-mentioned region
and the combustion is dominated by a heterogeneous
exothermic reaction of type (1 ). We are not aware
of any studies on this topic. The form of (1 ) implies
that the model with the burning rate control reaction
(1 ) should nevertheless depend appreciably on gasphase processes. Indeed, without subsurface reactions,
the surface heat balance has the form
qg + Qs Ws = mc(Ts T0 ) + (m Ws )L.
According to the data of [5, 7], qg  mc(Ts T0 ); Then,
the following relation holds:
m = Ws

Qs + L
.
c(Ts T0 ) + L

(4 )

From this, in view of (1 ), it follows that the


model should incorporate a mechanism that ensures
a negative dependence of act on pressure (otherwise,
dlnm/dlnp > 1, which is inconsistent with experimental data). In addition, the form of expression (4 ) implies that for Qs < c(Ts T0 ), the burning rate cannot
be controlled by surface reactions. The latter statement
has a simple physical meaning and can be formulated
immediately without using (4 ).
7. In the model with subsurface reactions, accounting for the possibility qs,c > 0 (heat supply through the
burning surface to the subsurface reaction zone) with
the use of the MerzhanovDubovitskii formula (a combination of the models of [15, 25]); see formula (A7) in
Appendix A.

Combustion Models for Energetic Materials


CONCLUSIONS
1. At present, for homogeneous EMs capable of
vaporization and exothermic condensed-phase decomposition, there are no comprehensive combustion models that describe combustion over a wide range of pressures assuming the possibility of condensed-phase or
gas-phase regimes with transition from one regime to
another.
2. There are arguments in favor of the dominant
role of condensed-phase reactions in burning rate control for a number of the most widely used EMs in the
rocket pressure range. For these EMs, a stable steadystate combustion regime is not observed locally anywhere on the burning surface but it is possible to speak
of the steadiness of the parameters averaged along the
surface.
3. Concepts were formulated according to which an
oscillatory regime occurs at each local site of the burning surface in the rocket pressure range. The oscillations
are not described by the ZeldovichNovozhilov model
and exist on the entire burning surface, for example,
in the form of waves of change in the condensed-phase
decomposition rate that move along the surface. The oscillation amplitude of the parameters in the subsurface
reaction zone is limited by nonlinear effects. The feedback limiting the oscillations is produced by the positive
dependence of the gasification rate (which consumes
the overheated region) on the value of its overheat.
The authors are grateful to A. A. Paletskii for useful remarks. The support of INTAS (Grant No. 03-535203) is greatly acknowledged.
APPENDIX A
Derivation of the Zeldovich
and MerzhanovDubovitskii Formulas
and Expressions for the Reaction-Zone
Thickness and Dissolved-Gas Concentration
We consider the thermal-conductivity equation in
the condensed phase with zero-order reactions:
[mc(T T0 ) T 0 ]0 = QW (T ),
(A1)
T () = T0 , T (0) = Ts .
The prime denotes differentiation with respect to x.
In the case of a sharp dependence W (T ), an asymptotic approach is suitable and the convective term can
be ignored in the reaction zone. We denote T 0 = ;
then T 00 = 0 = (d/dT ) (dT /dx) = d(2 /2)/dT and
Eq. (A1) implies the relation
ZTs
2
2
( ) (s,c ) = 2Q W (T )dT.
(A2)
T

29
The subscript s, c and superscript asterisk refer to the
hot and cold boundaries of the reaction zone, respectively. If the quantity (s,c / )2 can be ignored in
comparison with 1, then, using the heat balance of the
heating zone
= m[H(T ) H(T0 )]
m[H(Ts ) H(T0 )] = mH,
we obtain the Zeldovich formula [46]
RTs
ZTs
2Q W (T )dT
RTs2
2
W (T )dt Ws
,
.
m =
2
(H)
E
In this case,
H = Q,

(A3)

(A4)

(A5)

where is the degree of decomposition of the substance


under the surface. Taking into account the heat transfer
to the reaction zone from the surface, it is convenient to
use the heat balance for the reaction zone in the form
s,c = Qm.

(A6)

Then, using (A3) and (A6), from (A2) we obtain the


MerzhanovDubovitskii formula [47]
RTs
W (T )dT
2
.
(A7)
m =
(H Q/2)
To estimate the reaction-zone thickness xr from the
condition
W (T (xr ))
1
= ,
(A8)
Ws
e
we use the Frank-Kamenetskii expansion W Ws
exp((T Ts )E/RTs2 ) and the parabolic (because of
s,c = 0) approximation T Ts (x2 /2)Ts00 (Ts00 =
QWs /). Then,
s
2 RTs2
.
xr
Ws Q E
Substituting Ws Q from (A4) and (A5) into the above
relation, we obtain
2 RTs2
2
RTs2
xr
=
.
(A9)
mQ E
u E(Ts T0 )
Here u is the linear burning rate and is the thermal diffusivity. To estimate the concentration (mass
fraction Y ) of the dissolved gas, we assume that it is
released only by diffusion and the concentration profile
is parabolic with the maximum Ymax near the cold
boundary of the reaction zone. The product flow is
equal to m; on the other hand, for such a profile, under Ficks law, it is equal to D2Ymax /xr . Taking into
account (A9), we obtain
Ymax

RTs2

.
D E(Ts T0 )

(A10)

30

Gusachenko and Zarko

APPENDIX B
Release of Gaseous Products
from the Subsurface Reaction Zone
All models that take into account the subsurface
decomposition of ESs should include the mechanism of
release of gaseous products through the surface. The
products can be removed by diffusion of individual
molecules or by motion of continuous gas volumes in
bubbles or jets. The release of the products in the form
of bubbles or jets is considered in [48, 49]. The general
case can be described in a coordinate system attached
to the burning surface by the equations
m = b ub
+ (1 )c [(1 Y )uc DdY /dx],
= (1 )c + b ,
d[b ub (1 )c DdY /dx] = W dx.

(B1)
(B2)
(B3)

Here the subscripts b and c refer to bubbles (or jets) and


to the liquid, u are the velocities of the components relative to the surface (the gas velocity in bubbles or jets
and the velocity of the original liquid), D is the diffusion coefficient of the gas dissolved in the liquid, is
the density, is the porosity (the volume fraction of
bubbles or jets), Y  1 is the mass fraction of the dissolved gas in the liquid, and W is the rate of chemical
conversion of the original liquid to the gas. The fraction [b ub DdY /dx]/m of the product flow in the
total mass flow can be called the degree of decomposition of EMs. The decomposition reaction redistributes
the flows of both components in the total mass flow,
and the degree of decomposition increases approaching
the surface. To close the system in order to find the
distributions of the component velocities and densities,
one needs additional data or assumptions on the mechanism and driving forces of the gas transport to the
surface and on the value of Y . Most often, combustion models assume one of the following three versions
to describe the release of products from the subsurface
reaction zones.
I. Multilayer Foam on the Surface [50]. The liquid exists in the form of interbubble membranes and
moves together with the bubbles (uc = ub ); diffusion in
the liquid is complicated by the tortuosity of the path
to the surface along the membranes, and we can set
D = 0. The modeling of combustion of EMs with foam
on the surface involves the following difficulty: with distance from the interface between the continuous liquid
and foam, the longitudinal dimensions of the bubbles
increase sharply and the question of their shape and
the possibility of coalescence arises. In describing the

heat-transfer process, it is useful to estimate the contribution of the following mechanism: an extended bubble
that grows in an nonisothermal medium due to intake
of dissolved gas and vaporization actually represents a
miniature heat pipe. However, models with a foamy
layer has yet ignored these effects.
II. Gas Release in the Form of Individual Bubbles
or Jets Ignoring Diffusion. Margolis et al. [48] and Li et
al. [49] suggest that the action of the Marangoni effect
on gas bubbles in the reaction zone is taken to be the
driving force. This action undoubtedly takes place, but
in the practically important case of combustion with the
burning rate controlled by condensed-phase reactions,
the temperature gradient in the reaction zone is low
and the Marangoni effect does not work. This conclusion follows from the definition of a condensed-phase kinetically controlled combustion regime. In this regime,
heat transfer from the surface to the liquid-phase reaction zone is much lower than the heat release in the
reaction zone. For the heat balance (1), this implies
that qg /qc  1. A zone of sharp decrease in the liquid
temperature gradient with approach to the liquid surface can be seen in thermocouple oscillograms only in
rare cases (when intense heat release in the condensed
phase occurs even at a low pressure; see [51, 52]). The
reason is most likely that in the examined regime, the
reaction-zone thickness is comparable to the thermocouple thickness (especially taking into account weak
irregular changes in the slope of the plane of rolled thermocouples). The thickness of the liquid-phase reaction
zone for HMX at a pressure of 60 atm is estimated by
formula (A9) of Appendix A, assuming that, according
to [5, 6], = 0.001 cm2 /sec, Ts = 770 K, T0 = 300 K,
u = 0.9 cm/sec, and E = 21,000 cal/mole:
RTs2
2
= 2.5 m.
xr
u E(Ts T0 )
Since, according to [6], thin strip thermocouples
3540 thick and unrolled (round) thermocouples 10 m
thick have almost identical readings, the resolution of
the instrument was not higher than 10 m. Therefore, thermal inhomogeneities of such size are simply
not recorded on the temperature profile.
We propose to consider another gas-release mechanism: nonisotropic growth of bubbles. A bubble that
grows very rapidly (because of the supersaturation of
the solution with the gas) in the liquid near its surface,
is expanded primarily in the direction of the minimum
inertial resistance, i.e., toward the liquid surface, and
its evolution is similar to the ejection of a short-lived
jet. In the case of the extreme velocity (an underwater
explosion), this effect is known and is used in blasting. After the discharge of each bubble, surface tension flattens the surface, under which convective liquid

Combustion Models for Energetic Materials

Fig. 2. Representation of the jet regime.

flow arises. As a result, even for a small coefficient


of diffusion (by a molecular mechanism), the effective
convection-enhanced diffusive transport of dissolved gas
to the surface can become sufficient to limit the number of bubbles. At each time, the fraction of the surface
occupied by bubbles can be small [5356], but because
of the high frequency of the process, visually (and in
motion-picture recording at insufficient speed), it looks
as if the surface is covered with a continuous foam layer.
The liquid convection produced by the discharged bubbles also enhances the effective heat transfer in the reaction zone. In addition, the above-mentioned mechanism of heat-transfer enhancement, similar to that in
heat pipes, is also possible.
The indicated similarity of nonisotropically expanding bubbles to jets allows a more justified use of the
assumption of a jet regime of the gas release, for which
relation b ub = c uc holds, in addition to (B1)(B3).
This relation was employed in some of the two-phase
models discussed above. It was noted that it provides a
better fit between calculations and experiment than the
relation ub = uc , which corresponds to the regime with
a true (multilayer) foam. The regime b ub = c uc is
called the jet regime because it is consistent with the
concept of the partitioning of an EM sample into a set
of cylindrical elements, in each of which the EM approaching the surface at a velocity ub instantaneously
becomes a gas jet at a certain temperature (Fig. 2).
In different elements, these temperatures are different (therefore, in different elements, the sudden phase
change occurs at different distances from the surface).
In practice, there is a limited (per unit area) number of
jets that transport the gas to the surface. The gaseous
reaction products dissolved in the liquid are supplied to
each of these jets from the adjoining volume by diffusion.
III. Diffusive Release of the Dissolved Gas. One
might expect that in a certain range of parameters of the
EM and ambient conditions, combustion regimes occur

31
in which the dissolved product of liquid decomposition
is transported to the burning surface by molecular diffusion without bubble formation, so that a concentration
(mass fraction) profile of the product with a maximum
Ymax inside the melted layer is formed. It is known
that in liquids, the diffusion coefficient D, as a rule,
increases with temperature (under an exponential law)
and is very small. For many years, the smallness of D
has seemed a convincing reason for ignoring the diffusive
mechanism of the gas release. However, diffusion even
with a very small coefficient D can ensure an effective
release of dissolved gas from a rather thin liquid film.
In the experiment of [57], bubbles did not form in the
polymer film on the surface of a disk heated (to 430 C)
if the film thickness was smaller than 1020 m.6 Since
the thickness of the liquid-phase reaction zone of burning EMs can be the same as or smaller than this value
(an estimate was made above), the existence of a combustion regime with purely diffusive gas release is not
ruled out. Let us elucidate how this possibility depends
on pressure. We use formula (A10) of Appendix A.
For polymers (and, in particular, for rocket ballistite
propellants), the diffusion coefficient of dissolved gases
included in this formula depends strongly on the temperature Ts , according to [59]. Hence, in view of the
ClapeyronClausius relation, the dependence of the coefficient D on pressure is power law. An analysis of the
data of [60] confirms the existence of a similar dependence for nonpolymeric liquids: water and some organic
compounds. Apparently, the same dependence holds
for melts of nonpolymeric individual EMs. Since the
value of the left side of (A10) is limited from above by
the solubility limit, a positive dependence D(p) implies
that at a low pressure, equality (A10) cannot be satisfied, which suggests the formation of bubbles or jets,
and a pressure rise leads to an increase in the probability that the (A10) is satisfied. For HMX at p = 60 atm,
we set = 0.001 cm2 /sec, Ts = 770, T0 = 300 K,
E = 21,000 cal/mole, and = 0.5. Under these conditions, let the limiting mass fraction of the dissolved gas
be 0.05. Then, for the absence of bubbles (or jets), it
is necessary that Ymax < 0.05 or, according to (A10),
D > 1.2 103 cm2 /sec. Because of the lack of data
on the diffusion of dissolved gases in HMX, we compare available data for other liquids to clarify whether
this coefficient of gas diffusion in liquids is realistic at
770 K. According to [58], at T = 300 K, nearly all liquids have D 105 cm2 /sec and a dependence of the
form D T /, where is the dynamic viscosity. For ,
tables are given in [58]. For example, as the water tem6

Diffusion of volatiles from thinner (to 1 m) films of liquid nitroglycerin powder at temperatures up to 210 C was
studied in [58].

32
perature rises from 300 to 600 K, the viscosity decreases
by a factor of 20 and, hence, D increases by a factor of
40 and DH2 O (600 K) = 4 104 cm2 /sec. For a number of organic liquids, the changes in D are of the same
order, including for liquids with a molar mass comparable to those of HMX. The comparison problem now
reduces to the question: can the value of D for HMX
at T = 770 K be three times the value of D for water
at T = 600 K? The answer is anyway not obvious. The
diffusive mechanism of release of the products can be
considered a probable mechanism, and the conditions
of its occurrence require an additional study.
If this mechanism operates, this implies that
= 0 in system (B1)(B3). Then, Eq. (B1) becomes
(1 Y )c uc + Y yc ug = const = m. From this it is
seen that for a small value of Y with approach to the
surface, an increase in the degree of decomposition of
Y c ug /m leads to a decrease in uc , i.e., deceleration of
the liquid.
In some of the papers discussed above, no assumptions are made on the mechanism of gas release and
variations in the density and thermal parameters of the
substance in the reaction zone are ignored [1315, 61].
Mathematically, this implies that = 0 in system (B1)
(B3). We note that this approach corresponds to the
case of rather effective diffusive release of products and,
hence, it is not as rough as seemed earlier.

REFERENCES
1. A. I. Atwood, T. L. Boggs, P. O. Curran, et al., Burn
rate temperature and pressure sensitivity of solid propellant ingredients, in: Combustion Instability of Solid
Propellants and Rocket Motors, AGARD Consulting
Mission, Politechnico di Milano, MI, Italy, June 1618
(1997).
2. A. I. Atwood, T. L. Boggs, P. O. Curran, et al., Burning rate of solid propellant ingredients, J. Propuls.
Power, 15, No. 6, 740747 (1999).
3. A. G. Tereshchenko, O. P. Korobeinichev, P. A.
Skovorodko, et al., Probe method for sampling solidrocket propellant combustion products at temperatures
and the pressures typical of a rocket combustion chamber, Combust. Expl. Shock Waves, 38, No. 1, 8191
(2002).
4. B. E. Homan, M. S. Miller, and J. A. Vanderhoff,
Absorption diagnostics and modeling investigations of
RDX flame structure, Combust. Flame, 120, 301317
(2000).
5. A. A. Zenin, HMX and RDX: Combustion mechanism
and influence on modern double-base propellant combustion, J. Propuls. Power, 11, No. 4 (1995).

Gusachenko and Zarko


6. A. A. Zenin, V. M. Puchkov, and S. V. Finyakov, Characteristics of HMX combustion waves at various pressures and initial temperatures, Combust. Expl. Shock
Waves, 34, No. 2, 170176 (1998).
7. A. A. Zenin and S. V. Finjakov, Physics of GAP combustion, AIAA Paper No. 20001032 (2000).
8. A. A. Zenin and S. V. Finjakov, Burning wave
structure and combustion mechanism of glycidylazide/nitramine mixtures, in: Energetic Materials:
Analysis, Diagnostics and Testing, Proc. of the 31st
Int. Ann. Confer. of ICT, Karlsruhe, FRG (2000),
pp. 132(1)132(13).
9. A. A. Zenin and S. V. Finjakov, Physics of combustion of HTPB/nitramine compositions, in: Energetic
Materials: Ignition, Combustion, and Detonation, Proc.
of the 32nd Int. Ann. Confer. of ICT, Karlsruhe, FRG
(2001,), pp. 8(1)8(24).
10. A. A. Zenin, Errors of thermocouple measurements of
flames, Inzh.-Fiz. Zh., 5, No. 5, 6774 (1962).
11. A. A. Zenin, On heat exchange of thermocouples in a
solid-propellant combustion wave, J. Appl. Mech. Tech.
Phys., No. 5, 125131 (1963).
12. V. P. Sinditskii, V. V. Serushkin, S. A. Filatov, and
V. Yu. Egorshev, Flame structure of hydrazinium nitroformate, in: K. K. Kuo and L. T. DeLuca (eds.),
Combustion of Energetic Materials, Begell House, New
York (2002).
13. C. Guirao and F. A. Williams, Model of ammonium perchlorate deflagration at a pressure of 196
981 N/cm2 , AIAA J., 9, No. 7, 13451356 (1971).
14. V. A. Strunin, G. B. Manelis, A. N. Ponomarev, and
V. L. Talroze, Effect of ionizing radiation on the combustion of ammonium perchlorate and mixed systems
based on it, Combust. Expl. Shock Waves, 4, No. 4,
584590 (1968).
15. G. B. Manelis, G. M. Nazin, Yu. I. Rubtsov, and
V. A. Strunin, Thermal Decomposition and Combustion of Explosives and Propellants [in Russian], Nauka,
Moscow (1996).
16. N. E. Ermolin and V. E. Zarko, Modeling of cyclicnitramine combustion, Combust. Expl. Shock Waves,
34, No. 5, 485501 (1998).
17. M. A. Lavrentev and G. V. Shabat, Problems of Hydrodynamics and Their Mathematical Models [in Russian],
Nauka, Moscow (1977).
N. Rumyanov, Exothermic reac18. B. I. Khaikin and E.
tion regimes in a one-dimensional flow, Combust. Expl.
Shock Waves, 11, No. 5, 573578 (1975).
19. M. Tanaka and M. W. Beckstead, A three phase combustion model of ammonium perchlorate, AIAA Paper
No. 962888 (1996).
20. O. Jing, M. W. Beckstead, and M. Jeppson, Influence
of AP solid phase decomposition on temperature profile
and sensitivity, AIAA Paper No. 98-0448 (1998).

Combustion Models for Energetic Materials


21. Y. C. Liau and V. Yang, Analysis of RDX monopropellant combustion with two-phase subsurface reactions,
J. Propuls. Power, 11, No. 4, 729739 (1995).
22. N. E. Ermolin, O. P. Korobeinichev, L. V. Kuibida, and
V. M. Fomin, Study of the kinetics and mechanism
of chemical reactions in RDX flames, Combust. Expl.
Shock Waves, 22, No. 5, 544552 (1986).
23. Y. C. Liau, V. Yang, and S. T. Thunell, Modeling of RDX/GAP pseudo-propellant combustion with
detailed chemical kinetics, in: K. K. Kuo and
L. T. DeLuca (eds.), Combustion of Energetic Materials,
Begell House, New York (2002), pp. 477499.
24. T. A. Litzinger, Y. J. Lee, and C. J. Tang, Experimental study of nitramine/azide propellant combustion, in:
V. Yang, T. B. Brill, and W. Z. Ren (eds.), Solid Propellant Chemistry, Combustion, and Motor Interior Ballistics, AIAA, Reston (2000), p. 2.4; See also: Progress in
Astronautics and Aeronautics, Vol. 185.
25. M. J. Ward, S. F. Son, and M. Q. Brewster, Steady
deflagration of HMX with simple kinetics: A new modeling paradigm, Combust. Flame, 114, 556 (1998).
26. J. Louwers, Combustion and decomposition of
hidrazinium nitroformate (HNF) and HNF propellants,
Ph. D. Thesis, Delft (2000).
27. J. Louwers, G. M. H. J. L. Gadiot, M. Q. Brewster, et al.
Steady-state hidrazinium nitroformate (HNF) combustion modeling, J. Propuls. Power, 15, No. 6, 772777
(1999).
28. H. F. R. Schoyer, W. H. M. Melland-Veltmans, J. Louwers, et al., Overview of the development of hydrazinium nitroformate-based propellants, J. Propuls.
Power, 18, No. 1, 138145 (2002).
29. J. Louwers and G. M. H. J. L. Gadiot, Model for nonlinear transient burning of hydrazinium nitroformate,
J. Propuls. Power, 15, No. 6 (1999).
30. V. P. Sinditskii, V. Yu. Egorshev, V. V. Serushkin, and
A. I. Levshenkov, Chemical peculiarities of combustion
of solid propellant oxidizers, in: Rocket Propulsion:
Present and Future, Abstracts of the 8th Int. Workshop
on Combustion and Propulsion, Pozzuoli, Naples, Italy,
June 1620 (2002).
31. A. E. Fogelzang, V. P. Sinditskii, V. Y. Egorshev, et al.,
Combustion behavior and flame structure of ammonium dinitramide, in: Proc. of 28th Int. Annual Conf.
of ICT, Karlsruhe, FRG (1997), p. 99.
32. A. G. Shmakov, O. P. Korobeinichev, and
T. A. Bolshova, Thermal decomposition of ammonium dinitramide vapor in a two-temperature flow
reactor, Combust. Expl. Shock Waves, 38, No. 3,
284294 (2002).
33. I. G. Assovskii, Theory of propellant combustion of at
high pressures, in: S. S. Bondarchuk (ed.), Computational Gas Dynamics and Combustion of Condensed
Systems [in Russian], Tomsk Polytechnical University,
Tomsk (2001), pp. 1740.

33
34. K. V. Meredith and M. W. Beckstead, Laser-induced
ignition modeling of HMX, in: Proc. of 39th JANNAF Combustion Meeting, Colorado Springs, December
(2003).
35. V. E. Zarko, L. K. Gusachenko, and A. D. Rychkov,
Modeling of transient combustion regimes of energetic
materials with surface vaporization, in: K. K. Kuo
(ed.), Challenges in Propellants and Combustion 100
Years after Nobel, Begell House, New YorkWallingford
(1997), pp. 10141025.
36. L. K. Gusachenko, V. E. Zarko, A. D. Rychkov, Instability of a combustion model with surface vaporization
and overheat in the condensed phase, Combust. Expl.
Shock Waves, 33, No. 1, 3440 (1997).
37. V. P. Sinditskii, V. Yu. Egorshev, M. V. Berezin, and
V. V. Serushkin, Combustion of high-energy caged
nitramine hexanitrohexaazaisowurtzitane, Report for
ICOC2002, Moscow.
38. V. A. Strunin, A. P. Dyakov, and G. B. Manelis, Combustion of ammonium dinitramide, Combust. Flame,
117, 429434 (1999).
39. V. A. Strunin and G. B. Manelis, Stability of the
steady-state stable process of explosive combustion limited by the reaction in the condensed-phase, Combust.
Expl. Shock Waves, 7, No. 4, 427430 (1971).
40. A. V. Ananev, A. G. Istratov, Z. V. Kirsanova, et al.,
Instability in steady-state combustion of propellants
and explosives, in: Chemical Physics of Combustion
and Explosion Processes, Proc. of the XII Symposium
on Combustion and Explosion, Part I, Izd. Ross. Akad.
Nauk, Chernogolovka (2000).
41. I. G. Assovskii and O. A. Kudryavtsev, Method for
determining the ignition velocity of a channel surface in
a propellant, Khim Fiz., 14, No. 7, 122131 (1995).
42. S. S. Novikov, P. F. Pokhil, Yu. S. Ryazantsev, and
L. A. Sukhanov, A model for the permanent transient
state in the combustion zone for mixed condensed systems, J. Appl. Mech. Tech. Phys., 3, 320322 (1968).
43. V. Ya. Zyryanov, V. M. Bolvanenko, O. G. Glotov, and
Yu. M. Gurenko, Turbulent model for the combustion
of a solid fuel composite, Combust. Expl. Shock Waves,
24, No. 6, 562660 (1988).
44. Yu. Ya. Maksimov, Boiling point and enthalpy of vaporization of liquid RDX and HMX, Zh. Fiz. Khim.,
66, No. 2, 540542 (1992).
45. L. K. Gusachenko and V. E. Zarko, The overheating
in the surface layer during combustion of melting energetic compounds, in: Proc. of the Twenty-First Int.
Pyrotechnics Seminar, Inst. of Chem. Phys., Moscow
(1995), pp. 298312.
46. Ya. B. Zeldovich, On the theory of combustion of pro
pellants and explosives, Zh. Eksp.
Teor. Fiz., 12, p. 498
(1942).

34
47. A. G. Merzhanov and F. I. Dubovitskii, On the theory
of stationary propellant combustion, Dokl. Akad. Nauk
SSSR, 129, 153156 (1959).
48. S. B. Margolis, F. A. Williams, and R. C. Armstrong,
Influences of two-phase flow in the deflagration of homogeneous solids, Combust. Flame, 67, No. 3, 249258
(1987).
49. S. C. Li, F. A. Williams, and S. B. Margolis, Effects of
two-phase flow in a model for nitramine deflagration,
Combust. Flame, 80, No. 3, 329349 (1990).
I. Maksimov, and A. G. Merzhanov, Theory of com50. E.
bustion of condensed substances, Combust. Expl. Shock
Waves, No. 1, 2531 (1966).
I. Maksimov, Yu. M. Maksimov, and V. F. Chukov,
51. E.
Investigation of the combustion of DINA, Combust.
Expl. Shock Waves, 7, No. 2, 165170 (1971).
52. V. V. Aleksandrov, A. V. Boldyreva, V. V. Boldyrev,
and R. K. Tukhtaev, Combustion of DINA at atmospheric pressure, Combust. Expl. Shock Waves, 9,
No. 1, 117119 (1973).
53. V. E. Zarko, V. N. Simonenko, and A. B. Kiskin,
Radiation-driven, transient burning: experimental results, in: L. DeLuca, E. W. Price, and M. Summerfield (eds.), Progress in Astronautics and Aeronautics, Vol. 143, Ch. 10 (1992).
54. V. E. Zarko, V. Ya. Zyryanov, and V. V. Chertischev,
Dispersion of the surface layer during combustion of
homogeneous propellants, AIAA Paper 96-0814 (1996).

Gusachenko and Zarko


55. C. L. Thompson (Jr.) and N. P. Suh, Gas phase reactions near the solid-gas interface of a deflagrating
double-base propellant strand, AIAA J., No. 1, 154
159 (1971); See also: Rocket Technique Astronautics,
No. 1 (1971).
56. O. G. Glotov, V. V. Karasev, V. E. Zarko, and
A. G. Svit, Burning of single crystals and pressed
tablets of RDX, in: Proc. of 34th Int. Annual Conf.
of ICT, Karlsruhe, FRG (2003), pp. 47(1)47(15).
57. O. F. Shlenskii and D. N. Yundev, Effect of decrease
in the nucleation rate in thin liquid films and its application to studies of the properties of overheated substances, Teplofiz. Vysok. Temp., 32, No. 1, 139141
(1994).
58. V. V. Aleksandrov and S. S. Khlevnoi, Surface temperature in the flameless combustion of nitroglycerin
propellants, Combust. Expl. Shock Waves, 5, No. 4,
438443 (1970).
59. D. W. van Crevelen, Properties of Polymers, Elsevier,
Amsterdam (1990).
60. I. S. Grigorev and E. Z. Meilikhov (eds.), Physical

Quantities, Handbook [in Russian], Energoatomizdat,


Moscow (1991).
61. A. F. Zhevlakov, V. A. Strunin, and G. B. Manelis,
Combustion mechanism in hydrazonium nitrate and
the influence of alkali metal additives, Combust. Expl.
Shock Waves, 12, No. 2, 162166 (1976).

Das könnte Ihnen auch gefallen