You are on page 1of 20

47th AIAA Aerospace Sciences Meeting Including The New Horizons Forum and Aerospace Exposition AIAA 2009-805

5 - 8 January 2009, Orlando, Florida

DNS Simulation of Erosive Burning


in Planar Periodic Rockets
Amir H. G. Isfahani∗
Department of Mechanical Science and Engineering,
University of Illinois at Urbana-Champaign, Urbana, IL 61801

Ju Zhang† and Thomas L. Jackson‡


Computational Science & Engineering,
University of Illinois at Urbana-Champaign, Urbana, IL 61801

We examine the combustion response of a flame anchored to two 1/4-spaces of solid


fuel and oxidizer, a configuration relevant to the combustion of heterogeneous solid propel-
lants. A time-periodic shear flow is applied to model the shear that can be generated by
the presence of acoustics or turbulence in a rocket chamber. To estimate the magnitude
and frequency of the shear for the case of a turbulent flow, we present DNS results of a
planar periodic rocket, a configuration that has its roots in a multiscale analysis. Such a
configuration allows for the determination of the shear parameters as functions of motor
geometry and downstream location. The response of the flame to this shear, the heat flux
to the surface, and the burning rate are calculated numerically. Significant enhancement
to the burn rate, commonly known as erosive burning, is found.

I. Introduction
This paper is concerned with erosive burning of ammonium percholate (AP)-based composite solid pro-
pellants, a common propellant used for solid rocket propulsion for more than five decades. Erosive burning
is defined as the modification of the local burn rate from its nominal value by some mechanism. The nominal
burn rate is measured using strands in an enclosed bomb at some predefined pressure, or measured using the
ballistic evaluation motor (BEM), which is a standardized small rocket motor with high port/throat area
ratio. Erosive burning is a well known phenomenon first reported in [1] and subsequently in a number of
independent studies and review articles (e.g., [2-11]). The most important process that has been observed
that leads to a modification of the nominal burn rate is high crossflow velocity. Thus, erosive burning refers
to the sensitivity of the solid composite propellant burning rate to crossflow velocity. For composite propel-
lants, studies show that the nominal burn rate increases with increasing crossflow velocity, an increase in
the coarse grain AP diameter, type of binder, and temperature in the unburnt propellant. Factors which do
not appear to modify the burn rate are crossflow temperature, crossflow composition, and metal addition.
Erosive burning has also been observed for homogeneous propellants, but here we are only concerned with
composite propellants. The increase in burning rate from the nominal value can vary from one (no erosive
burning) for low pressures and low crossflow velocities, to three or more at higher pressures and higher
crossflow velocities. As an example, Figure 1 of [6] shows the burn rate for an AP/polyester propellant as a
function of pressure and crossflow velocity. The data is taken from [2]. It can be seen from the figure that
the burning rate is significantly increased as the pressure increases with the crossflow velocity held fixed, or
as the crossflow velocity increases with fixed pressure. For example, at 80 atm, the burn rates are roughly
0.4 cm/s (no crossflow), 0.6 cm/s (at a crossflow velocity of 49 m/s), 0.8 cm/s (at 103 m/s), and 1.2 cm/s
(at 214 m/s).
∗ ResearchAssistant, 1324 Mechanical Engineering Laboratory, Email: ghalayan@illinois.edu, Member AIAA
† Visiting
Scholar, 3243 Digital Computer Laboratory, Email: juzhang@illinois.edu, Member AIAA
‡ Senior Research Scientist, 2256 Digital Computer Laboratory, Email: tlj@csar.uiuc.edu. Corresponding author. Associate

Fellow, AIAA.

1 of 20

Copyright © 2009 by the authors. Published by the AmericanAmerican


Institute of Aeronautics
Institute ofand Astronautics, and
Aeronautics Inc., Astronautics
with permission.
It is generally agreed that erosive burning results from an increase in heat transfer from the gas-phase
to the propellant surface. With an increase in heat flux, the propellant evaporates more quickly, adding
more fuel and oxidizer to the flame zone. A number of theories have been proposed to explain the increased
heat flux to the surface as a function of crossflow velocity. These include heat transfer from the chamber
flow (where, it is argued, that the heat flux in the core flow is increased due to an increase in crossflow
velocity, and that this heat flux passes through the combustion zone thus providing additional heating to
the propellant surface), “flame bending” (it is argued that the chamber flow bends the final diffusion flame
toward the propellant surface, moving the point of heat release closer to the surface), and turbulence. In
the latter case the thought is that the thermal conductivity and the diffusion coefficients in the gas-phase
are increased as turbulent eddies are convected closer to the surface with an increase in crossflow velocity.
The models used in the turbulent simulations presuppose a particular turbulence closure model and that
the model is valid through the combustion layer down to the propellant surface. However, it is not yet
possible to perform a direct numerical simulation (DNS), which resolves all relevant length scales, to validate
any of the turbulent models used in the studies, and so any conclusions based on a particular turbulence
closure model must be considered preliminary. We comment here that all of the theories (heat transfer,
flame bending, and turbulence) are based on assumptions about the nature of the interaction between the
chamber flow and the combustion layer near the propellant surface, and none have been verified numerically
or experimentally. As a result, a number of empirically based models have been developed, one popular
example being Lenoir and Robillard’s model [12]. Unfortunately, none of the empirical models fit all the
data. What is required, therefore, is a more robust framework, a framework that can couple the chamber
flow dynamics with combustion of heterogeneous propellants. Over the past decade we have been engaged in
developing a variety of mathematical and numerical tools for the study of composite propellant combustion,
and those tools are now sufficiently mature to address erosive burning.
Propellants of solid rocket motors are subject to intense time-dependent shear flows, and the response
of the combustion field to these flows is of fundamental interest. In addition, the flow perturbations that
are generated can modify the acoustic field within the chamber and, because of the two-way coupling, lead
to violent instabilities. There are a number of possible mechanisms that can give rise to time-dependent
shear flows, namely, acoustics, surface irregularities, and turbulence. It has been shown that interaction
between axial acoustic waves and the vortical mean flow, the Taylor solution [13, 14], can generate large
time-dependent shear flows in the vicinity of the propellant surface [15, 16, 17]. An estimate of the shear
rate in the presence of acoustic waves is 2.7 × 105 s−1 [16]. It should be noted, however, that little effect of
the shear upon the combustion process is reported in [16], but the flame there is supported by a double-base
propellant. Such a flame is locally one-dimensional, with a structure that only varies in the direction normal
to the surface, so that a flow parallel to the surface can be expected to have no effect. Thus, it is crucial to
consider appropriate models for heterogeneous propellant combustion.
The flames supported by the combustion of heterogeneous propellants are complicated, and so in earlier
work [18] we examined the role played by the edge-structure of a single diffusion flame between regions of
ammonium perchlorate (AP) and binder when subject to a time-dependent shear. The geometry was that
of two quarter-planes of solid propellant. The response of the single edge structure to an oscillating flow
was discussed in detail. In particular, we showed that the heat flux to the surface was significantly increased
when the shear blows binder gases over the AP and yet remains relatively unchanged when the AP gases are
blown over the binder. This asymmetry in the heat flux, when averaged over one period in time, resulted in
a net increase in the surface averaged burn rate. This in turn will increase the burn rate, commonly referred
to as erosive burning.
The model reported in [18] was necessarily simple since many modeling ingredients and parameter values
were not known at that time, and a numerical framework had not yet been developed for heterogeneous
solid propellants. For example, only a single diffusion flame was taken into account, yet we now know the
importance of a three-step mechanism that takes into account AP decomposition and two diffusion flames,
one between virgin AP gases and binder and one between AP decomposition gases and binder [19]. Also, in
[18] the gas-phase was decoupled from the solid phase and thus the surface temperature had to be prescribed
along the entire propellant surface. Only the heat flux to the surface could be calculated; nothing could be
said about the burning rate. We now have the numerical tools to simulate the coupling of the gas-phase
to the solid-phase and therefore, the surface temperature and the burn rate are determined as part of the
numerical procedure.
We are not aware of any work that addresses the shear rate for surface irregularities and turbulence. As

2 of 20

American Institute of Aeronautics and Astronautics


a consequence we focus only on turbulence-induced time-periodic shear. To address the turbulence issue we
also present here a DNS study of a planar periodic rocket, a configuration that has its roots in a multiscale
analysis [20-23]. Such a configuration allows for the determination of the shear parameters, amplitude and
frequency, as a function of motor geometry and downstream location.
This paper is organized as follows. Section 2 presents the mathematical model for 1/4-spaces of solid
fuel and oxidizer. The configuration is drawn in Figure (1). In Section 3 we present results for a planar
periodic rocket and derive estimates for the shear rate at the propellant surface. Section 4 gives results for
the quarter-plane problem subject to a time-dependent shear. Results include the flame structure, surface-
averaged heat flux, and burn rate as a function of shear parameters. Finally, concluding remarks are given
in Section 5.

Time
periodic Primary Diffusion
applied Flame
shear

AP Monopropellant Flame

Propellant flux

x
Binder AP

Figure 1: Sketch of the configuration. The binder lies in the region: x < 0, y < 0 and AP in the region:
x > 0, y < 0. The solid propellant turns to gas at the surface x = 0, with the gas-phase region defined by
y > 0. Shown are the primary diffusion flame and the AP decomposition flame. The final diffusion flame
lies beyond the leading edge of the primary diffusion flame.

II. The Mathematical Model


A full description of the numerical model can be found in our earlier work [24], but here we sketch its
essential details. The computational strategies are described in [25].
Gas-phase Equations
Gas-phase processes are rooted in the zero Mach number Navier-Stokes equations for a variable density
gas, the energy equation, species equations, and suitable chemical kinetics. Although our code can include
the Navier-Stokes equations, in this work we use the Oseen approximation which significantly reduces the
computational burden, and calculations for sandwich propellants attest to its accuracy [26]. The Oseen
approximation sets the mass flux nominally normal to the burning surface to a constant and the mass
flux nominally tangent to the surface to zero, thus satisfying the continuity but jettisoning the momentum
equation. An alternative and essentially equivalent formulation is to set the density to a constant, so that
the equation of state, Charles’s law, is jettisoned; and a uniform velocity field u = 0 and v = constant is
adopted, which satisfies both the continuity equation and the momentum equation. Then the mass flux ρv is
constant, and the model is merely a variation on that proposed by Burke and Schumann in 1928 [27]. This
approximation has proven to be adequate and suffices for our purposes here.
We use a simple three-step kinetics scheme, represented by
R
AP(X) →1 decomposition products(Z), (1)
R2
βX + binder(Y ) → final products(P ), (2)
R3
βZ + binder(Y ) → final products(P ). (3)

3 of 20

American Institute of Aeronautics and Astronautics


where

R1 = D1 Po n1 X exp{−E1 /Ru T },
R2 = D2 Po n2 X 3.3 Y 0.4 exp{−E2 /Ru T },
R3 = D3 Po n3 Y Z exp{−E3 /Ru T }.

The reaction rate R1 corresponds to AP decomposition, R2 is the so-called primary diffusion flame of the
Beckstead-Derr-Price model (BDP) [19] and takes into account reaction between virgin AP gases and binder
gases, and R3 is the final diffusion flame. Because the reaction rates do not correspond to real reactions but
are representative of many, the usual relations between the stoichiometric coefficients, the pressure exponents,
and the exponents of the reacting species do not have to be satisfied. The interested reader is referred to
[24] for a complete discussion of the kinetics model.
Note that in our earlier work [18] only the primary diffusion flame, represented by the reaction rate
R2 (and hence with different parameter values from the ones listed below in Table 1), was used. Below
we compare the results of the three-step mechanism with that of a two-step (R2 = 0) and a one-step
(R1 = R3 = 0), the latter being similar to the work of [18].
The corresponding dimensional equations for the species X, Y, Z, and the temperature T are

L(X, Y, Z) = (−R1 − βR2 , −R2 − R3 , R1 − βR3 ), (4)

L(T ) = (Qg1 R1 + Qg2 R2 + Qg3 R3 )/cp , (5)


where    
∂ ~ · λg ∇
~ −∇ ~ .
L ≡ ρg + ~u · ∇ (6)
∂t cp
The two-dimensional velocity vector is given by ~u = (u, v). Since an Oseen approximation is used, ρg v =
constant. We have assumed that all Lewis numbers are equal to 1, but temperature-dependent transport is
accounted for by
λg = 1.08 × 10−4 T + 0.0133 W/mK, (7)
when T is assigned in degrees Kelvin and cp is assumed to be constant. The parameter β in (4) is the overall
mass-based AP/binder stoichiometric ratio: β kg of AP (X) is required for the stoichiometric consumption
of 1 kg of binder (Y ).
Solid-phase Equations
In the condensed phase we solve the heat equation
∂T ~ ),
~ · (λc ∇T
cp ρ c =∇ (8)
∂t
where we have assumed, for simplicity, that the specific heat of the solid is the same as that of the gas. The
values of ρc and λc are piecewise constant and are assigned according to whether a point is located in binder
or in an AP particle. When the binder is replaced by a homogenized AP/binder blend, λc is defined by the
homogenization formulas in [28].
Propagation of the Surface
The surface regresses normal to itself with a speed rb (> 0), where we assume that rb is defined by simple
pyrolysis laws, viz. (
rb,AP = AAP exp{−EAP /Ru TAP,s }, in AP
rb = (9)
rb,B = AB exp{−EB /Ru TB,s }, in binder
where T,s is the surface temperature. These are commonly used as surrogates for the poorly understood
condensed phase reactions. If the binder is replaced by a blend the homogenization formulas from [28] must
be used for parameters Ablend and Eblend .
It is convenient to represent the surface by

y = f (x, t), (10)

4 of 20

American Institute of Aeronautics and Astronautics


where f , the surface location, is a single-valued function of x. Thus, y is nominally measured perpendicular
to the propellant surface and x is parallel to the surface. Surface kinematics lead to the following equation
p
ft + rb 1 + fx2 = 0. (11)

In some of the results presented below for the quarter-plane geometry we turn off surface corrugation and
so explicitly set fx = 0.
Connection Conditions at the Propellant Surface
The propellant surface is an interface between the condensed phase and the gas phase, and certain conditions
are imposed there which relate the solution in one phase to that in the other. These include continuity of
tangential velocity (zero in the solid phase, and therefore zero in the gas) and continuity of normal mass flux
(ρc rb in the solid) [29]. Energy conservation at the surface has the form

~ ] = −Qs M,
[λ~n · ∇T (12)

and the species flux condition is  


λ ~
M [Yi ] = ~n · ∇Yi . (13)
cp
In all of these formulas [·] ≡ (·)g − (·)c ; Yi refers to the species X, Y or Z; M is the mass flux normal to the
surface; Qs > 0 (< 0) represents an exothermic (endothermic) process; and both rb and Qs take on different
values according to whether the surface is AP or binder. Finally, the temperature is continuous across the
interface, [T ] = 0.
It is computationally convenient to deal with the unevenness of the surface by use of the mapping

x, y, t → x, η = y − f (x, t), t (14)

so that the surface is flat in the new coordinates. This is used to transform both the field equations and the
surface conditions.
Boundary Conditions
In addition to the connection conditions across the propellant surface, we assume zero normal derivatives at
x = ±∞, zero normal derivatives at y → ∞, and a supply temperature T0 as y → −∞.
Kinetic Parameters
The parameters are given in Table 1. A complete discussion of how the parameters were determined can be
found in [24].

5 of 20

American Institute of Aeronautics and Astronautics


Parameter value
AAP 1.45 × 105 cm/s
AB 1.036 × 103 cm/s
cp 0.3 kcal/kgK
D1 4.11 × 10 g/cm3 s barn1
D2 2.35 × 104 g/cm3 s barn2
D3 9.5 × 10 g/cm3 s barn3
EAP /Ru 11000 K
EB /Ru 7500 K
E1 /Ru 3000 K
E2 /Ru 8500 K
E3 /Ru 8500 K
MW 26
n1 2.06
n2 2.06
n3 1.60
Pr 1
Qg1 410 kcal/kg
Qg2 7403 kcal/kg
Qg3 4396 kcal/kg
Qs,AP −80 kcal/kg
Qs,B −66 kcal/kg
Ru 1.9859 kcal/kmol K
T0 300 K
β 7.33
λAP 0.405 W/m K
λB 0.276 W/m K
ρAP 1950 kg/m3
ρB 920 kg/m3
Table 1: Parameter values.

6 of 20

American Institute of Aeronautics and Astronautics


III. Estimate of Shear
Consider a linear shear of the form
u = ky, (15)
where k (s −1
) is the shear rate. The shear rate is then taken to be
k = a sin(2πf t), (16)
which is the shear rate used in [18]. Here, a (s−1 ) is the amplitude and f (Hz) the frequency. For the
quarter-plane geometry considered here, when k > 0 the binder gases are blown over the AP, and when
k < 0 the AP gases are blown over the binder.
We need an estimate of the amplitude a and the frequency f . As mentioned in Section 1, Roh et al.
[16] report a value on the order of 105 s−1 for the shear rate. We are not aware of any estimates based on
turbulence. However, there is a need to be able to estimate the shear parameters as a function of motor
size and downstream location without resorting to the expense of fully time-dependent three-dimensional
turbulent simulations. And so we present a strategy based on the periodic rocket analysis of [20-23]. Since
our mathematical model described above is two-dimensional, we consider only a planar periodic rocket
(hereafter, PPR). The extension to axisymmetric is reported in [30].

III.A. Multiscale Formulation


Consider the two-dimensional, nondimensionalized incompressible equations
∂ui
= −ǫU, (17)
∂xi
∂ui ∂ui ∂p π2 1 ∂ 2 ui
+ ǫu(ui − δ2i Ui ) + uj =− + δ1i + , i = 1, 2. (18)
∂t ∂xj ∂xi 4ǫ Reinj ∂xj ∂xj
Lengths are scaled by h, the chamber half-height, velocities by the injection velocity Vinj ∗
, and time by
h/Vinj

. Here, the superscript ∗ denotes a dimensional quantitya . The Reynolds number is based on the
injection velocity, Reinj = Vinj

h/ν. The parameter δ2i is the Dirac delta function (i.e., δ21 = 0 and δ22 = 1).
Also, Ui = (U, V ) denote the nondimensional mean velocities. These equations are derived from a multi-
scale analysis with the small parameter ǫ = Vinj ∗
/Um∗
being the ratio of the injection velocity to the average
streamwise velocity. The mean velocities Ui can be prescribed either from numerical simulations, as was
done in [20-23], or by using Culick’s analytical solution [14]. Culick’s solution was shown to be a reasonable
approximation for the mean velocities [20]. Since our goal is to develop a framework to estimate the shear
parameters, we use Culick’s solution for the mean velocities, given by
π x∗
U ∗ = Vinj

cos(πy ∗ /2h), V ∗ = −Vinj

sin(πy ∗ /2h), (19)
2 h
or, in nondimensional form,
π π
x cos(πy/2) ≡
U= cos(πy/2), V = − sin(πy/2). (20)
2 2ǫ
Note that the average dimensional streamwise velocity is Um∗
= Vinj

x∗ /h. Also note that 1/ǫ = x can be
interpreted as the nondimensional streamwise location for a fixed Vinj . An additional pressure gradient of

2
− π4ǫ due to the multiscale analysis is added to the x1 -momentum equation. This value is again based on
Culick’s solution. The interested reader should consult [20-23] for details.
It should be pointed out that the mechanism for the generation of shear used in this study is turbulence
and vortex shedding (eddy-turnovers) in the incompressible limit, as opposed to acoustics [16]. Even though
the estimated shear rate and frequencies may actually be quantitatively similar, the flow features are qualita-
tively different as can be easily imagined. For example, the flow here is characterized by near-wall large-scale
coherent structures (eddies) inclined at an angle opposite to the mean flow direction, whereas that driven by
acoustics is characterized by standing waves with strong near-wall variations. The frequencies in acoustic-
driven flows correspond to those of various longitudinal acoustic modes, whereas here for turbulence the
frequencies are determined by the magnitude of the rms velocities and the eddy size. The current framework
is therefore an excellent place to examine the shear induced by turbulence and nonacoustics mechanisms.
a In the discussion presented here, the equations are cast in nondimensional terms. This should not be confused with the

dimensional formulation presented in Section 2.

7 of 20

American Institute of Aeronautics and Astronautics


III.B. Numerical Method
The equations (17-18) are solved using a high-order incompressible flow solver [32]. In particular, the
convective terms are discretized using the 5th-order weighted essentially nonoscillatory (WENO) scheme
of [33], making it possible to resolve difficult applications with strong shear or turbulence. The diffusion
terms are treated using a 6th-order compact scheme of [34]. The method makes use of the fractional-step
scheme by [35] in conjunction with the optimized two-step alternating 4-6 low-dissipation and low-dispersion
Runge-Kutta (LDDRK) scheme of [36] to improve temporal accuracy of the scheme.

III.C. Grid and Domain Size Adequacy


Detailed discussion of the result associated with 2D PPR can be found in [31]. Here, we focus on addressing
the issues that were not elaborated in [31]. For reader’s convenience, the typical vorticity field at statistically
steady state is rehashed here and again given in Figure (2). It can be seen that the flow features characterized

Figure 2: Vorticity field corresponding to Reinj = 400 and ǫ = 0.04 for the planar periodic rocket.

by near-wall large-scale coherent structures, the parietal vortex shedding (PVS), inclined at an angle opposite
to the mean flow direction seen in [20] was reproduced [31]. It was also shown in [31] that the vortex is
stronger for smaller ǫ (further downstream in the full-scale configuration) and weaker for larger ǫ, and the

60 14

50 12

10
40

8
urms
U

30
6
∆ywall = 2.88e-3; ∆ysym = 2.84e-2
20 ∆ywall = 7.81e-3 ∆ywall = 2.88e-3; ∆ysym = 2.84e-2
∆ywall = 1.56e-2 4 ∆ywall = 7.81e-3
∆ywall = 1.56e-2

10
2

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
y/h y/h

Figure 3: Time-averaged (left) and rms (right) velocity profiles along wall normal direction for an ǫ = 0.025
invscid PPR with different grid resolutions.

main frequencies found for velocity time series are identified as being associated with how fast the vortex is
being shed. The grid convergence was shown for finite Reynolds number cases there whereas in this work,
we will try to examine the inviscid case in greater detail. To this end, we performed inviscid ǫ = 0.025 PPR
simulations with three different type of grid points, two of which are uniform and have 1282 and 2562 grid
points, respectively, while the remaining one has a mesh of 1282 grid points that is stretched in the wall
normal direction. The stretched grid has the best resolution in the wall normal direction near the wall. None

8 of 20

American Institute of Aeronautics and Astronautics


of the three sets of meshes is expected to be able to resolve the smallest scale in the inviscid flow, yet what
we attempt here is to examine whether the shear estimated at the wall can still somehow converge as long
as the rms velocity profile near the wall is reasonably resolved. The convergence of the rms velocity near
the wall is ensured as seen in the profiles shown in Figure (3) for the different meshes. The time-averaged
streamwise velocity are actually converged across the entire channel. With these convergence properties, the
shear rates based on the velocity at the first grid point away from the wall are indeed shown to be reasonably
converged in Figure (4). The magnitudes of the shear are also close to the values estimated based on the
Reynolds number of 1000 PPR justifying our claim of Reynolds number insensitivity of estimated shear in
[31].

1000

500

0
shear

-500

-1000

∆y = 2.88e-3; ∆ysym = 2.83e-2


-1500
∆y = 7.16e-3
∆y = 1.56e-2

-2000

0.5 1 1.5 2
t

Figure 4: Time history of the shear estimated based on the streamwise velocity at the first grid point away
from the wall for the different resolutions shown in Figure (3).

One natural question we also brought up in [31] is, “What determines the streamwise domain size?”
It is suggested here that it is nominally related to the turbulence driving scale, i.e., the chamber half-
width h. By employing a domain of sufficiently large size in the streamwise direction, one can first identify
the spatial wavelength of the parietal vorticies. A domain with adequate streamwise domain size should,
therefore, contain at least one such periodic feature along each side of the wall. If the domain size is too
small, an artificial spatial wavelength will be incorrectly imposed. For example, the typical nondimensional
streamwise domain size in our simulation is 4π, and we identified three vorticies along each wall so the spatial
wavelength is nominally 4π/3 as seen in Figure (5). If we further increase or decrease the domain size by
2π, the total number of vortices is indeed roughly proportional to the streamwise domain length as also can
be seen in Figure (5). This also indicates that there is, in fact, an length scale in the problem, i.e., the
turbulence driving scale, h, that dictates the spatial wavelength, λ, of the parietal vorticies. Note that this
spatial wavelength can be defined only nominally as a domain size of noninteger times of λ can also result in
periodic parietal vorticies due to the periodic boundary condition imposed in the streamwise direction. The
vorticies are simply slightly “shrinked” or “stretched” as shown by comparing the cases with domain sizes
of 3.5π, 4π and 4.5π in Figure (5). In the remaining discussion the streamwise domain length is set to 4π.

III.D. Dimensional Insights


Dimensional insights can be obtained by considering representative motors of various sizes. Table 2 gives
typical parameter values for four different solid rocket motors, ranging in size from a small motor to a 5-
segment booster. The 5-segment motor is taken here to be identical to the 4-segment Space Shuttle booster
with an additional segment. These values are meant to be representative only, and can be found elsewhere
(e.g., [41]). Also shown in the table are typical values for a slot, a region of constricted flow that may

9 of 20

American Institute of Aeronautics and Astronautics


Figure 5: Vorticity field in a Re = 1000 and ǫ = 0.04 PPR for different streamwise domain sizes. From top
to bottom: 2π, 3.5π, 4π, 4.5π and 6π.

occur, for example, in star-aft grains. The Reynolds number is defined above, i.e., Reinj = Vinj ∗
h/ν where
2
ν = µ/ρ = 1.43 × 10 m /s, and its value is found by using Sutherland’s law and assuming a chamber
−5

temperature and pressure of 3600 K and 6 MPa, respectively. Using the parameters in Table 2 as the
reference scales, i.e., h as the reference length scale and Vinj

the velocity scale, and defining the reference
time scale as tr = h/Vinj , we obtained the dimensional shear rates and frequencies listed in Table 3. The

ranges next to them show the effect of variation in the chamber pressure that will be discussed below. These
dimensional values are based on the nondimensional values obtained in [31] and summarized in Table 4.
Note that Table 4 is based on the results for Reinj of 1000 since we just demonstrated that the estimated
shear rates do not differ much between the inviscid limit and the Reinj = 1000 case. This Reynolds number
independence is also attributed to the fact that there is no boundary layer at the wall because it is blown-off
by injection.
To assess the effect of chamber pressure on the reference scales, Vinj

in particular, used in Table 2, we
assume that the propellant burn rate has the following empirical dependence on pressure,
rb = apn , (21)
with n < 1. The injection velocity can be obtained by the connection condition at the propellant surface
through
Vinj = (ρc − ρg )rb /ρg ≈ ρc aRT pn−1. (22)
Thus, for a particular motor, the higher the pressure, the lower the injection velocity and, thus, the lower
the reference velocity scale and higher the reference time scale, tr = h/Vinj , which imply a lower shear rate
and frequency. For a reasonable approximation, take, for example, the values of n of 0.5 and ρc rb = 25.14
kg/m2 -s at a chamber pressure of 6 MPa. Then the injection velocities are ∼2.25, 4.10 and 5.80 m/s for
chamber pressures of 20, 6 and 3 MPa, respectively. Therefore, Vinj varies within a factor of only ∼3 for
typical operation pressures from 3 to 20 MPa. Assuming the same propellant mentioned in the above example
for all the motors, the ranges of the variation in shear rate and frequency due to the variation in chamber
pressure can be derived and are given in Table 3 with the minimum and maximum corresponding to 20 and
3 MPa chamber pressures, respectively. It is also worthwhile to note that for a particular motor, the higher
the injection velocity Vinj , the higher the shear rate which leads to a more pronounced erosive burning as
will be seen in the next section.
Finally, we note that the gas-phase combustion zone is less than 1 mm thick, and this value is much
smaller than the thickness of the oscillating flow region, yth∗
. It is therefore expected that the dynamics
within the oscillating flow region can couple with the combustion, a subject investigated in the next section.

10 of 20

American Institute of Aeronautics and Astronautics


Parameter Small Tactical Space Shuttle 5-Segment Slot
L (m) 0.6 2.03 35.1 41.0 1.0
h (m) 0.025 0.102 0.700 0.700 0.01
Vinj (m/s) 2.5024 4.5322 3.1487 3.1487 3.1487
Reinj 4.37e3 3.23e4 1.54e5 1.54e5 1.54e5

Table 2: Parameter values for several representative motors.

ǫ x∗ (m) U ∗ max (m/s) Shear (s−1 ) Frequency (Hz) yth



(cm)
Small 0.1 0.25 40 7.17e3 (6.45e3–1.66e4) 320 (288–742) –
0.04 0.625 100 5.68e4 (5.12e4–1.32e5) 530 (477–1230) 0.17
Tactical 0.1 1 72 3.18e3 (1.59e3–4.07e3) 142 (71–182) –
0.04 2.5 181 2.52e4 (1.26e4–3.23e4) 235 (118–3001) 0.69
5-Segment 0.1 7 50.4 3.22e2 (2.29e2–5.92e4) 14.4 (10.2–26.5) –
0.04 17.5 126 2.56e3 (1.82e3–4.71e3) 23.9 (17.0–44.0) 4.76
0.025 28 195 4.79e3 (3.40e3–8.81e3) 36.9 (26.2–67.9) 3.78
0.0171 41 290 8.52e3 (6.05e3–1.11e4) 50.9 (36.1–93.7) 3.43
Slot 0.1 0.1 50.4 2.26e4 (1.60e4–4.16e4) 1008 (715.7–1855) –
0.04 0.25 126 1.79e5 (1.27e5–3.29e5) 1670 (1186–3073) 0.068
0.025 0.4 195 3.35e5 (2.38e5–6.16e5) 2583 (1834–4753) 0.054
0.0171 0.586 290 5.97e5 (4.24e5–1.10e6) 3560 (2528–6550) 0.049

Table 3: Shear rates, maximum streamwise velocities, frequencies and thickness of oscillating flow region
at the surface for Reinj = 1000 at different streamwise locations (x∗ ) and for different rocket motor sizes;
2562 uniform mesh. Except for ǫ = 0.0171, the Space Shuttle booster and the 5-Segment are identical. The
estimated shear rates are based on the maximum positive values of the near wall velocity variations [31].
The numbers in parentheses correspond to a range of values for pressures of 20 MPa (the first number in the
range) to 3 MPa.

ǫ f a Shear rate yth Ū


0.1 3.2 0.14 71.68 – 10
0.04 5.3 0.87 568.32 0.068 25
0.025 8.2 3.45 1063.68 0.054 40
0.0171 11.3 8.16 1894.4 0.049 58.6

Table 4: Nondimensional main frequency f , corresponding amplitude a, shear rate, oscillating flow thickness
yth and mean velocity Ū as a function of ǫ.

11 of 20

American Institute of Aeronautics and Astronautics


IV. Numerical Results
The configuration is the same as that of [18]; i.e., the binder lies in the quarter-plane x < 0, y < 0 and
AP lies in the quarter-plane x > 0, y < 0. See Figure (1) for the configuration. For the following cases the
grid is 240 points in the x-direction (uniform) and 301 points in the y−direction. A stretched grid is used to
cluster points near y = 0 to adequately resolve the combustion field in the gas-phase and the thermal layer
in the solid-phase. A grid resolution study was performed and the results presented here represent converged
solutions. The pressure is set to 30 atm.
We first consider the case where the region x < 0, y < 0 is binder. Since the binder does not burn as
x → −∞, surface corrugation is turned off (fx = 0) and we assume that the propellant regresses uniformly
in x. The case of a blend, where the region x < 0, y < 0 is replaced by a homogeneous blend of AP and
binder, will be discussed last.
Figure (6) shows the gas-phase temperature and reaction rate R3 for a = 60 1/s and f = 1000 Hz (see
equation 16) over a single period. Note that both the temperature and reaction rate are strongly influenced
by the presence of the shear. Similar results are obtained for f = 100 and 10 Hz, though the hot gases have
more time to respond to the shear with lower frequencies. To make sure the size of the domain is sufficiently
large (to imitate infinity at the left and right boundaries) we examined three different domain sizes of 3 × 3,
6 × 3 and 12 × 6 mm2 . The results were essentially the same for all these different domain sizes. Also, the
surface distribution of the heat flux is relatively unaffected by the resolution in the x-direction, which results
in a small negligible difference in surface-averaged heat fluxes. For example, quadrupling the grid resolution
in the x-direction results in about 0.7% change in the surface-averaged heat flux for both a = 0 (no shear)
and a = 3000 s−1 . Figure (7) plots the reaction rate R3 as a function of shear k (constant shear, see equation
15) and shows how the position of the final diffusion flame changes with k.
Figure (8) plots the surface-averaged heat flux and the corresponding surface-averaged burn rate as a
function of time for f = 1000 Hz and for various amplitudes a. The surface-averaged heat flux is defined by
+ L2x
1
Z
1 + fx2 Ty − fx Tx dx,
  
hf lux =− λ (23)
Lx − L2x

the same as that used in [18]. A similar definition is used for the surface-averaged burn rate. Also plotted
in each sub-figure as a dashed curve is the shear profile sin(2πf t) with f = 1000 Hz. From the figure it can
be seen that for the smaller amplitudes, e.g., a = 3000 1/s, the surface-averaged heat flux is greater when
the shear is blowing AP gases over the binder. This behavior reverses when the amplitude increases to say,
a = 10000 1/s. This transition is also present for lower frequencies of f = 100 and f = 10 Hz (see Figure 9).
For all values of a, integration over one period results in an increase in both the total surface-averaged heat
flux and the burn rate, an erosive burning effect. We note here that the surface heat flux for no shear is 69.5
cal/s-cm and the corresponding burn rate is 0.217 cm/s (both are shown as the dot-dash lines in Figures 8
and 9).
Erosive burning refers to the modification of the local burn rate from its strand burner value by some
mechanism. Let r0 denote the normal burn rate in the absence of a crossflow. This burn rate is usually
measured using strands in an enclosed bomb at some predefined pressure, or measured using the ballistic
evaluation motor (BEM), which is a standardized small rocket motor with high port/throat area ratio. Define
the erosive burn rate coefficient
ǫb = rb /r0 , (24)
as the ratio of the burn rate under crossflow conditions to the normal burn rate without crossflow. If ǫb > 1
we say that erosive burning has taken place, if ǫb < 1 then negative erosive burning has occurred, and if
ǫb = 1 then there is no erosive burning present. Figure (10a) plots ǫb as a function of shear amplitude and
for various frequencies at a fixed pressure of 30 atm, and Figure (10b) plots the burn rate as a function of
shear amplitude and for various pressures for a fixed frequency of f = 1000 Hz. Note that as the pressure is
increased, the burn rate also increases, a trend in agreement with experimental data. In calculating ǫb , rb
and r0 are both surface and time averaged over one period of the sinusoidal shear. As mentioned, r0 = 0.217
cm/s. This figure clearly shows significant erosive burning at the larger amplitudes and as a function of
pressure.
To have a more realistic time-dependence of the imposed shear, the shear evolution in a PPR simulation
may actually be used in the flame simulation here. The time history of the shear from the ǫ = 0.04 and

12 of 20

American Institute of Aeronautics and Astronautics


Re = 1000 PPR simulation in [31] is adapted here with proper dimensional scaling for the 5-Segment case in
Table 2. The resulted burn rate as a function of time is shown by solid line in Figure (11). (The figure shows
a 0.1 s sample of the burn rate.) The average burn rate is 0.254 (cm/s), and appreciable erosive burning
indeed occurs. We can also assess the validity of using a simple sinusoidal shear as we did throughout this
paper by imposing a sinusoidal shear with the amplitude set to the average shear in the shear evolution from
the PPR simulation and the frequency to the main frequency of the evolution. The resulted burn rate is
also shown in Figure (11) where the evolution is seen to be quite representative of that with the actual PPR
shear. This justifies the use of a simple sinusoidal shear in examining erosive burning in the current work.
Note that the average burn rate of 0.243 (cm/s) in the latter case is expectedly under predicted by about
4% since the high frequency features are absent when the main freuqncy is used.
To determine the effect of chemistry we show in Figure (12a) the erosive burn rate coefficient ǫb for the
BDP model (R1 , R2 , R3 are nonzero), when the primary diffusion flame is neglected (R2 = 0), and the case
with only the primary diffusion flame present (R1 = R3 = 0). The latter case was treated previously in [18].
From the figure we see that the primary diffusion flame plays a dominant role in determining the erosive
burn rate coefficient. For example, ignoring the primary diffusion flame underpredicts the erosive effect (i.e.,
compare the solid and the dotted curves in Figure 12a). This is consistent with the observation of Price
(e.g., [42]) that the edge structure of the primary diffusion flame supported by heterogeneous propellants
play an important role in the transfer of heat to the propellant surface, heat that is responsible for the surface
regression, the burning rate. Keeping only the primary diffusion flame, as was done in [18], exaggerates the
erosive burn rate coefficient when normalized by its nominal burn rate value (i.e., compare the solid and
the dash curves). The left panel is when the normalization is done with respect to the burn rates of the
individual cases. Since this normalization can skew the results when the strand burning rate is small, on
the right panel we show ǫb when normalized by the value corresponding to the BDP model (i.e., r0 = 0.217
cm/s).
In modeling the experimental work of Furfaro [7], Wang [43] develops an empirically based erosive burning
model based on Reynolds analogy. Reynolds analogy, which postulates the existence of a relationship between
the heat transfer coefficient and friction, leads to the following equation
c2
∂u
rb − r0 = c1 pc3 . (25)
∂y
This expression is similar to the one presented in [43] (see equations 7-10). Our expression takes into account
the fact that the gas-phase thermal conductivity is temperature dependent, and hence there is expected to
be a slight pressure dependence. Using the data of Figure (10) we solve the nonlinear least squares problem
and obtain the following fit
 0.346
∂u
rb − r0 = 7.995 × 10−3 p−0.083 , (26)
∂y
and
r0 = b + apn , b = 0.028730, a = 0.006034, n = 1. (27)
Note that the pressure exponent n is one, which is an artifact of using the quarter-plane geometry. The
constant b shows that the nominal burn rate does not go to zero as the pressure goes to zero, a nonphysical
limit. Figure (13) plots as symbols the burn rate as a function of pressure for different shear rates and for
a frequency of f = 1000 Hz. The solid lines are the result of equation (26). Note the good agreement with
the empirical fit, based on Reynolds analogy, and the numerical burn rate data, which clearly suggests a link
between shear and burn rate.
Finally, we show in Figure (14) the effect of an AP/binder blend instead of binder in the left quarter-plane
region on the burning rate (14a) and the erosive burn rate coefficient (14b). From the figure we can see that
the effect of a blend is to reduce ǫb . We comment here that surface regression was turned on (fx 6= 0) for
the blend cases. We also compared the no regression case with cases that had surface regression for αw > 0
and essentially no effect was observed on the results.

V. Concluding Remarks
We have examined a flame exposed to a time-periodic shear, and have shown significant effects on the
average heat flux to the propellant surface and the average burn rate. In particular, the erosive burn rate

13 of 20

American Institute of Aeronautics and Astronautics


(a) t = 0 1 2 3
(b) t = 11
T (c) t = 11
T (d) t = 11
T

4 5 6 7
(e) t = 11
T (f) t = 11
T (g) t = 11
T (h) t = 11
T

8 9 10 (l) t = T
(i) t = 11
T (j) t = 11
T (k) t = 11
T
Figure 6: Top panels: gas-phase temperature in the domain X ∈ [−0.3, 0.3] cm, Y ∈ [0.0, 0.3] cm. Bottom
panels: gas-phase reaction rate R3 in the domain X ∈ [−0.1, 0.1] cm, Y ∈ [0.0, 0.1] cm. Here, a = 60 1/s
and f = 1000 Hz in a period of T = 1 ms.

Figure 7: R3 reaction rate contours for different constant shear rates.

14 of 20

American Institute of Aeronautics and Astronautics


260 +1 0.65 +1
Surface-averaged heat flux (cal/s-cm)

sin 2πf t
sin 2πf t
240 0 0.6 0
220 0.55
−1 −1
200
0.5

r̄b (cm/s)
180
0.45
160
0.4
140
0.35
120
100 0.3

80 0.25
60 0.2
0.046 0.047 0.048 0.049 0.05 0.046 0.047 0.048 0.049 0.05
time (s) time (s)
(a) (b)

Figure 8: (a) Surface-averaged heat flux (calculated from the solid side) and (b) surface-averaged burn rate
for a sinusoidal shear of u = ay sin(2πf t) with a frequency of f = 1000 Hz and amplitudes of a = 0 (dash-
dot), a = 500 1/s (triangle), a = 3000 1/s (square), a = 10000 1/s (circle), and a = 60000 1/s (diamond).
The surface-averaged heat flux for the case with no shear is 69.5 cal/s cm and the surface-averaged burn rate
for the case with no shear is 0.217 cm/s. The shear profile is shown as – –, and is presented for reference.

180 +1 +1

sin 2πf t
sin 2πf t

180
Surface-averaged heat flux (cal/s-cm)
Surface-averaged heat flux (cal/s-cm)

0 0
160
−1 160 −1
140
140

120 120

100 100

80 80

60 60
0.14 0.142 0.144 0.146 0.148 0.15 0.8 0.82 0.84 0.86 0.88 0.9
time (s) time (s)
(a) (b)

Figure 9: Surface-averaged heat flux (calculated from the solid side) over one period for a sinusoidal shear
of u = ay sin(2πf t) with frequencies of (a) f = 100 Hz and (b) f = 10 Hz, and different amplitudes of
a = 0 (dash-dot), a = 500 1/s (triangle), a = 3000 1/s (square), a = 10000 1/s (circle), and a = 60000 1/s
(diamond). The surface-averaged heat flux for the case with no shear is 69.5 cal/s cm, the same as Figure
(8). The shear profile is shown as – –, and is presented for reference.

15 of 20

American Institute of Aeronautics and Astronautics


2.6 0.8
f = 1000 Hz p = 15 atm
f = 100 Hz p = 30 atm
2.4 f = 10 Hz 0.7 p = 60 atm

2.2
0.6

rb (cm/s)
2.0
0.5
1.8
0.4
ǫb

1.6
0.3
1.4

1.2 0.2

1.0 0.1
1 10 102 103 104 105 1 10 102 103 104 105
Shear amplitude (1/s) Shear amplitude (1/s)
(a) (b)

Figure 10: (a) Erosive burn rate coefficient as a function of the shear amplitude for different frequencies and
fixed pressure of 30 atm. (b) Burn rate as a function of shear amplitude for different pressures with fixed
frequency of f = 1000 Hz.

0.31
PPR velocity history
PPR average shear rate and main frequency
0.30
0.29
0.28
rb (cm/s)

0.27
0.26
0.25
0.24
0.23
0.22
0.21
0.21 0.20 0.22 0.24 0.26 0.28 0.30
time (s)
Figure 11: A comparison between the surface-averaged burn rate, rb , as a function of time for when the
actual shear history of the PPR is fed into the combustion computations and that of a simplified sinusoidal
shear with the average shear and main frequency. The constant burn rate lines on the plot represent the
time averaged burn rates.

16 of 20

American Institute of Aeronautics and Astronautics


30 3
R1, R2, R3 R1, R2, R3
R1, R3 R1, R3
R2 R2

10 1

ǫb
ǫb

1 0.1
1 10 102 103 104 105 1 10 102 103 104 105
Shear amplitude (1/s) Shear amplitude (1/s)
(a) (b)

Figure 12: Erosive burn rate coefficient as a function of the shear amplitude with a frequency of f = 1000
Hz and for different reaction combinations. In the left panel the reference burn rate r0 is different for each
case: 0.217 cm/s when all R1 , R2 , and R3 are considered, 0.215 cm/s when only R1 and R3 are considered,
and 0.027 cm/s when only R2 is considered. In the right panel all cases have been normalized by r0 = 0.217
cm/s.

1.0
No shear
0.9 a = 500
a = 3000
0.8 a = 60000

0.7
rb (cm/s)

0.6

0.5

0.4

0.3

0.2

0.1
0 20 40 60 80 100 120
Pressure (atm)
Figure 13: The burn rate, rb , averaged over the surface and one period as a function of the pressure for
different shear amplitudes and for a frequency of f = 1000 Hz.

17 of 20

American Institute of Aeronautics and Astronautics


1.8 2.6
No shear a = 60
a = 60 2.4 a = 600
1.6 a = 6000
a = 600
a = 6000 2.2 a = 60000
1.4 a = 60000
2.0
1.2
rb (cm/s)

1.8
1.0

ǫb
1.6
0.8
1.4
0.6 1.2
0.4 1.0
0.2 0.8
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
αw αw
(a) (b)

Figure 14: Burn rate (a) and erosive burn rate coefficient (b) for f = 1000 Hz and different amplitudes for
different AP/binder blends (based on AP mass fraction, αw ) in the left quarter plane.

coefficient, defined as the ratio of the burn rate with shear to the burn rate without shear, was shown
to increase by almost a factor of three as the amplitude of the shear increases. This is in agreement with
experimental data, although a quantitative comparison will wait until three-dimensional packs are considered.
As in [18], the primary diffusion flame is responsible for most of this increase; ignoring this flame can decrease
the effect significantly. The primary diffusion flame is characterized by a region of intense edge reaction, the
source of surface heating, and similar discrete structures are found in three-dimensional periodic packs so
that we expect similar influence of shear in those cases.
To estimate the amplitude and frequency of the shear for turbulent flows, we carried out a DNS of a
planar periodic rocket. This allows for the estimate of the shear parameters as a function of motor geometry
and downstream location. We showed that the parameter values thus obtained were consistent with that
required to generate significant erosive burning effects.
Future work will re-examine the effect of a time-periodic shear but in a three-dimensional configuration.
We will also extend the DNS solver to three dimensions as well as to compressible flows.

Acknowledgments
This work was supported by the US Department of Energy through the University of California under
subcontract B523819, and by the NASA Constellation University Institutes Project under grant NCC3-
989 through the University of Maryland with a subcontract Z634015 to University of Illinois at Urbana-
Champaign, with Claudia Meyer as the project manager.

References
1. Green, L. Erosive burning of some composite solid propellants. Jet Propulsion, 24:9-15 (1954).
2. Marklund, T. and Lake, A. Experimental investigation of propellant erosion. ARS Journal, 30(2):173-
178 (1960).
3. Gordon, J.C., Duterque, J. and Lengelle, G. Solid propellant erosive burning. Journal of Propulsion
and Power, 8(4):741-747 (1992).
4. Bulgakov, V.K., Karpov, A.I. and Lipanov, A.M. Numerical studies of solid propellant erosive burning.
Journal of Propulsion and Power, 9(6):812-818 (1993).
5. Gordon, J.C., Duterque, J. and Lengelle, G. Erosive burning in solid propellant motors. Journal of
Propulsion and Power, 9(6):741-747 (1993).
6. King, M.K. Erosive burning of solid propellants. Journal of Propulsion and Power, 9(6):785-805 (1993).

18 of 20

American Institute of Aeronautics and Astronautics


7. Furfaro, J.A. Erosive burning study utilizing ultrasonic measurement techniques. AIAA Paper 2003-
4806 (2003).
8. Krishnan, S. and Rajesh, K.K. Erosive burning of ammonium perchlorate/hydroxyl-terminated-polybutadiene
propellants under supersonic crossflows. Journal of Propulsion and Power, 19(4):741-747 (2003).
9. Landsbaum, E.M. Erosive burning of solid rocket propellants - A revisit. Journal of Propulsion and
Power, 21(3):470-477 (2005).
10. Moss, J., Heister, S. and Linke, K. Experimental program to assess erosive burning in segmented solid
rocket motors. AIAA Paper 2007-5782 (2007).
11. Cai, W., Thakre, P. and Yang, V. A model of AP/HTPB composite propellant combustion in rocket-
motor environments. Combustion Science and Technology, in press, (2008).
12. Lenoir, J.M. and Robillard, G. A mathematical method to predict the effects of erosive burning in
solid propellant rockets. Proceedings of the Sixth Symposium (International) on Combustion, pp.663-
667 (1957).
13. Taylor, G.I. Fluid flow in regions bounded by porous surfaces. Proceedings of the Royal Society of
London, Series A 234:456 (1956).

14. Culick, F.E.C. Rotational Axisymmetric Mean Flow and Damping of Acoustic Waves in Solid Propel-
lant Rocket Motors. AIAA Journal, 4:1462-1464 (1966).
15. Flandro, G.A. Effects of vorticity on rocket combustion stability. Journal of Propulsion and Power,
11(4):607-625 (1995).
16. Roh, T.S., Tseng, I.S., and Yang, V. Effects of acoustic oscillations on flame dynamics of homogeneous
propellants in rocket motors. Journal of Propulsion and Power, 11(4):640-650 (1995). Value of shear
rate provided by V. Yang, private communication.
17. Zhao, Q., Staab, P.L., Kassoy, D.R. and Kirkkopru, K. Acoustically generated vorticity in an internal
flow. Journal of Fluid Mechanics, 413:247-285 (2000).
18. Buckmaster, J. and Jackson, T.L. The Effects of Time-Periodic Shear on a Diffusion Flame Anchored
to a Propellant. Combust. Flame, 120:211-221 (2000).

19. Beckstead, M.W., Derr, R.L. and Price, C.F. A Model of Composite Solid-Propellant Combustion
Based on Multiple Flames. AIAA Journal, 8(12): 2200-2207 (1970).
20. Venugopal, P., Najjar, F.M. and Moser, R.D. DNS and LES Computations of Model Solid Rocket
Motors. AIAA-2000-3571, 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit,
Huntsville, AL, July 16-19, 2000.
21. Venugopal, P., Najjar, F.M. and Moser, R.D. Numerical Simulations of Model Solid Rocket Motor
Flows. AIAA Paper 2001-3950, 37th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and
Exhibit, Salt Lake City, UT, July 8-11, 2001.
22. Venugopal, P. Direct Numerical Simulation of Turbulence in a Model Solid Rocket Motor. PhD Thesis,
University of Illinois at Urbana-Champaign (2003).
23. Venugopal, P., Moser, R.D. and Najjar, F.M. Direct Numerical Simulation of Turbulence in Injection
Driven Plane Channel Flows. Physics of Fluids, 20:105103 (2008).
24. Massa, L., Jackson, T.L. and Buckmaster, J. New Kinetics for a Model of Heterogeneous Propellant
Combustion. Journal of Propulsion and Power, 21(5):914-924 (2005).
25. Massa, L., Jackson, T.L. and Short, M. Numerical Solution of Three-Dimensional Heterogeneous Solid
Propellants. Combustion Theory and Modelling, 7(3):579-602 (2003).

19 of 20

American Institute of Aeronautics and Astronautics


26. Hegab, A., Jackson, T.L., Buckmaster, J. and Stewart, D.S. Nonsteady burning of periodic sandwich
propellants with complete coupling between the solid and gas phases. Combustion and Flame, 125:1055-
1070 (2001).

27. Burke, S.P. and Schumann, T.E.W. Diffusion flames. Industrial and Engineering Chemistry, 20(8):998-
1004 (1928).
28. Chen, M., Buckmaster, J., Jackson, T.L. and Massa, L. Homogenization Issues and the Combustion
of Heterogeneous Solid Propellants. Proceedings of the Combustion Institute, 29:2923-2929 (2002).
29. Jackson, T.L. and Buckmaster, J. Heterogeneous Propellant Combustion. AIAA Journal, 40(6):1122-
1130 (2002).

30. Zhang, J., T.L. Jackson, F. Najjar and J. Buckmaster, Multi-physics Numerical Simulation of Erosion
in Rocket Nozzle, 47th AIAA Aerospace Sciences Meeting, Orlando, FL, Jan. 5-8, 2009.
31. A. H. G. Isfahani, Zhang, J. and Jackson, T.L., The Effects of Turbulence-induced Time-Periodic Shear
on a Flame Anchored to a Propellant, Accepted by Comb. & Flame, 2008.
32. Zhang, J. and Jackson, T.L. A High-Order Incompressible Flow Solver with WENO. Accepted by J.
Comp. Phys. (2008).

33. Shu, C.W. and Osher, S. Efficient Implementation of Weighted ENO Schemes. Journal of Computa-
tional Physics, 126:202 (1996).
34. Lele, S.K. Compact finite differences with spectral like resolution. Journal of Computational Physics,
103:16 (1992).
35. Kim, J. and Moin, P. Applications of a fractional-step method to incompressible Navier-Stokes equa-
tions. Journal of Computational Physics, 59:308-323 (1985).

36. Hu, F.Q., Hussaini, M.Y. and Manthey, J.L. Low-Dissipation and -Dispersion Runge-Kutta Schemes
for Computational Acoustics. Journal of Computational Physics, 124:177-191 (1996).
37. Apte, S.V. and Yang, V. A large-eddy simulation study of transition and flow instability in a porous-
walled chamber with mass injection. Journal of Fluid Mechanics, 477:215-225 (2003).
38. Lupoglazoff, N. and Vuillot, F. Parietal vortex shedding as a cause of instability for long solid propellant
motors - Numerical simulations and comparisons with firing tests. AIAA Paper 96-0761 (1996).
39. Lupoglazoff, N. and Vuillot, F. Numerical simulations of parietal vortex-shedding phenomenon in a
cold flow set-up. AIAA Paper 98-3220 (1998).
40. Avalon, G., Ugurtas, B., Grisch, F. and Bresson, A. Numerical computations and visualization tests of
the flow inside a cold gas simulation with characterization of a parietal vortex shedding. AIAA Paper
2000-3387 (2000).

41. Flandro, G.A. and Majdalani, J. Aeroacoustic instability in rockets. AIAA Journal, 41(3): 485-497
(2003).
42. Lee, S.T., Price, E. and Sigman, R. Effect of multidimensional flamelets in composite propellant
combustion. Journal of Propulsion and Power, 10(6):761-768 (1994).
43. Wang, Q. Development of erosive burning models for CFD predictions of solid rocket motor internal
environment. AIAA Paper 2003-4809 (2003).

20 of 20

American Institute of Aeronautics and Astronautics

You might also like