You are on page 1of 14

Journal of Quantitative Spectroscopy &

Radiative Transfer 84 (2004) 513 – 526


www.elsevier.com/locate/jqsrt

Coupled radiation and turbulent multiphase +ow in an


aluminised solid propellant rocket engine
Rodolphe Duval, Anouar Sou1ani∗ , Jean Taine
Laboratoire d’Energetique Moleculaire et Macroscopique, Combustion (EM2C), UPR 288, du CNRS et de l’ECP,
Ecole Centrale Paris, Châtenay-Malabry Cedex 92295, France

Abstract

We present in this paper numerical simulations of coupled radiative transfer and turbulent +ows at high tem-
perature and pressure, typical of multiphase +ows encountered in aluminised solid propellant rocket engines.
The radiating medium is constituted of gases and of liquid or solid particles of oxidised aluminum. The turbu-
lent +ow of the gaseous phase is treated by using a four equation, low Reynolds number, boundary-layer-type
turbulence model. The distributions of concentrations, temperatures, and temperature +uctuation variances of
particles are calculated from a Lagrangian approach and a turbulence dispersion model. Thermal and mechan-
ical non-equilibrium between the gas and di6erent classes of particles is allowed. A locally one dimensional,
iteratively based, radiative transfer solver is developed to compute wall +uxes and radiative source terms. It
is shown that the thermal boundary layer attenuates signi1cantly the radiative +uxes coming from the outer
regions. Particle radiation is found to be much more important than gas radiation. Turbulent dispersion of
particles in the boundary layer induces a decrease of particle concentration in the region of maximum turbulent
kinetic energy, and then, decreases the attenuation e6ect of wall +uxes due to the boundary layer. The e6ects
of turbulent temperature +uctuations are found to be small in the problem under consideration.
? 2003 Elsevier Ltd. All rights reserved.

Keywords: Gas-particle radiation; Turbulent multiphase +ow; Solid propellant engines

1. Introduction

High speci1c impulse in a solid propellant motor is generally achieved by adding to the oxidant and
to the polymeric binder a metal fuel such as aluminum. Metal combustion increases the temperature
and the pressure inside the chamber. Typically, temperatures up to 3600 K and pressures between
50 and 150 bar are recovered with small velocities in the combustion chamber. The acceleration


Corresponding author. Tel.: +33-141-13-1031; fax: +33-147-02-8035.
E-mail address: sou1ani@em2c.ecp.fr (A. Sou1ani).

0022-4073/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0022-4073(03)00268-1
514 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

through the nozzle leads to temperatures of about 2000 K at the exit and a Mach number between
2 and 5 depending on the ambient conditions.
Due to the propellant composition, combustion products are a mixture of gases (mainly N2 , CO2 ,
CO, H2 O, HCl and H2 ) and of alumina condensed particles. As a result of the complex mechanism
of aluminum combustion [1], particle sizes vary from 0.5 to 150 m with a multimodal distribution.
Particle properties are also a6ected by coalescence, fragmentation phenomena and by solidi1cation,
especially in the nozzle.
In order to protect the structural parts of the motor, several types of thermal protecting materials are
employed. The prediction of convective and radiative +uxes incident on these materials is important
since the +ux level directly a6ects the performance (ablation of the throat, impulse to weight ratio,
etc.) and the safety of the motor. The +ux levels may be of the order of several MW m−2 and the
radiative contribution ranges from practically 100% (internal parts) to about 10% (divergent part
of the nozzle). Radiative +uxes have also an important feedback on the propellant combustion and
burning rates [2,3].
The aim of the present study is to develop a general model for the prediction of radiative +uxes
and powers inside solid propellant propulsion chambers, and to apply this model to representative
boosters such as the MPS-P230 motor of the Ariane5 launcher. The model couples the prediction of
the turbulent multiphase +ow and thermal 1elds (Section 2), and the calculation of radiative transfer
in the emitting, absorbing and scattering gas-particle mixture (Section 3). Attention will be focused
on regions where the walls (not the propellant) are directly exposed to the hot +ows. The multiphase
mixture is thus considered as non-reacting and the interaction with combustion processes will not be
addressed in this study. All through this paper, we will assume that the main gas species, with the
corresponding molar fractions, are CO2 (0.011), CO (0.255), H2 O (0.114), HCl (0.139), N2 (0.08),
H2 (0.337). This composition results from equilibrium calculations at 3400 K under 50 MPa. The
dense phase (alumina) global mass fraction will be assumed equal to 0.32 and a three-modal size
distribution is adopted (spherical particles of 1, 10, and 80 m diameters).

2. Multiphase ow and temperature elds

As a result of the small volume fraction of the dense phase, the multiphase +ow is considered in the
diluted regime, i.e. particle–particle interactions are neglected and particle properties are calculated
from their trajectories in the turbulent gaseous +ow, although the retro-action of particles on gaseous
+ow can be taken into account. We present 1rst the model used for gaseous +ow and temperature
predictions, and then, the way to obtain particle concentration and temperature 1elds.

2.1. Turbulent gaseous phase

In order to take into account the e6ects of gas and particle temperature +uctuations on radiative
transfer, a four additional equations turbulence model is adopted for the gas phase. The gas +ow
is modeled by using a boundary layer type, two-dimensional (in average), low Reynolds number
model in which, in addition to the two components uK and vK of the Favre averaged velocity and to
the mean temperature T , the transport equations of the kinetic energy k of turbulence, its dissipation
rate , the variance of temperature +uctuations , and its rate of dissipation  are solved [4]. The
R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526 515

variations of thermophysical properties with temperature are taken into account and the e6ects of
gas compressibility (especially in the the transport equation of ) are modeled through the dilatation
dissipation model of Zeman [5]. The detailed description of this turbulence model is out of the scope
of this paper and can be found in Refs. [6,4,7].

2.2. Dispersed phase

For radiative transfer prediction, the most important particle properties are their concentration
or volume fraction distribution, their mean and +uctuating temperature 1eld and, of course, their
radiative properties. The latter point is discussed in Section 3.1. A Lagangian approach is used here
for the computation of particle concentrations and temperature 1elds from the gaseous turbulent gas
phase 1eld. The inlet section of the computational domain is subdivided into elementary surfaces
and individual particles are injected through these surfaces. Consecutive positions Xp , velocities Vp
and temperatures Tp of a given particle of diameter dp are calculated from:
dVp 3 K Cd
= Vg − Vp (Vg − Vp ) + g; (1)
dt 4 p dp
dXp
= Vp ; (2)
dt
dTp Pr (dp )
mp Cpp = d2p hT + : (3)
dt N (dp )
In these equations, K; p ; Vg and g are the intrinsic gas and alumina densities, the instantaneous
local gas velocity vector, and gravity acceleration, respectively. mp and Cpp are the mass of the
particle and its speci1c heat, N (dp ) is the number density of particles of diameter dp , Pr (dp ) is the
radiative power per unit volume exchanged between particles of diameter dp and their environment;
T is a temperature di6erence accounting for the di6erence between gas and particle kinetic energies:
 
Vg − Vp 2
T = (Tg − Tp ) + 0; 5 Pr 1=3 ; (4)
CK p
where Pr is the Prandtl number, Tg the instantaneous and local gas temperature, and Cp is the gas
speci1c heat. Cd in Eq. (1) and h in Eq. (3) are the drag and heat transfer coeOcients obtained
from
Rep 6 1; Cd = 24=Rep ;

1 6 Rep ; Cd = 28=Rep0:85 + 0:48;



Nup = 2 + 0:6 Rep Pr 1=3 ; (5)
where Rep and Nup are the particle Reynolds ( Kdp Vg − Vp =g ) and Nusselt (hdp =g ) numbers.
The above particle trajectory equations require the instantaneous gas velocity and temperature at
position Xp . A turbulent dispersion model is used to generate these quantities from the average gas
velocity VK g and temperature T g and from turbulent parameters such as k; ; , and  . A time
step Pt is used to integrate Eqs. (1)–(3). The particle remains associated to a turbulent eddy until,
either the elapsed time nPt becomes greater than the interaction time de1ned as the minimum of
516 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

the turbulent mechanical lifetime of the eddy, its thermal lifetime, and the particle eddy crossing
time, or the distance traveled by the particle becomes greater than the minimum of mechanical and
thermal integral length scales of turbulence [8]. When one of these uncorrelation criteria occurs, the
particle is assumed to be associated with another turbulent eddy; new instantaneous values of Vg
and Tg are then calculated from gas turbulent 1elds and Gaussian random numbers.
In order to compute particle concentrations and temperatures, each trajectory l of a particle of
diameter dm , issued from the elementary injection surface n, is considered as a +ux tube with constant
mass +ow rate ṁm; n of alumina. If m; n designates the total number of trajectories of particles m
issued from the surface n, ṁm; n is given by
1
ṁm; n = fm; n tot Vn · Sn ; (6)
m; n

where fm; n is the prescribed mass fraction of particles of diameter dm at the injection surface n,
tot is the total density of the multiphase mixture, Vn is the injection velocity taken here equal to
the local gas mean velocity Vg; n , and Sn is the unit vector normal to the injection surface n. The
number density of particles m in a current cell (i; j) is given by
 Ptk
Nm (i; j) = ṁm; n (in m−3 ); (7)
n
Vi; j mm
l k

where the summation extends over injection surfaces n, individual trajectories l, and time steps Ptk
during which the trajectory remains inside the cell (i; j). In the above equation, Vi; j is the cell volume
and mm is the mass of a particle of diameter dm . In the same manner, the mean temperature T m (i; j)
of particles m in the cell (i; j) is calculated as

1    ṁm; n Ptk
1
T m (i; j) = [Tp (k + 1) + Tpl (k)]; (8)
mm Nm (i; j) n Vi; j 2 l
l k

where Tpl (k) is the particle temperature corresponding to trajectory l at time step tk . The temperature
standard deviation for particles m in the cell (i; j) is computed from

!m (i; j) = |T 2m (i; j) − T 2m (i; j)| = Tm2 1=2 (i; j); (9)

with

1    ṁm; n Ptk
T 2m (i; j) = 1
[T 2 (k + 1) + Tp2l (k)]: (10)
mm Nm (i; j) n Vi; j 2 pl
l k

This temperature standard deviation is directly used in radiative transfer calculations in order to take
into account the turbulent +uctuations of particle temperatures (see Section 3.3).
It is worthy of notice that the procedure described above for the computation of particle temper-
atures and concentrations from a Lagrangian approach only holds for stationary turbulent +ows as
those considered here.
R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526 517

-1
10

(a)2680 K
(b)2090 K
absorption index k -2
10 (b)2550 K
(c)2375 K, rocket 1
(c)2959 K, rocket 1
(c)2375 K, rocket 2
-3 (c)2959 K, rocket 2
10
(d)3000 K, formula
(e)3000 K, formula

-4
10
0.0 1.0 2.0 3.0 4.0 5.0
wavelength (µm)

Fig. 1. Compilation of some experimental and semi-empirical data for the absorption index k of rocket alumina. Experi-
mental data from pure alumina are not presented. (a): [11], (b): [14], (c): [15], (d): [9], (e): [16].

3. Radiative properties and radiative transfer

3.1. Particle properties

Many studies have addressed the issue of radiative properties of alumina or more generally,
of metallic oxides, in rocket combustion chambers. Reviews of these studies may be found in
Refs. [9,10]. The real refractive index n has been determined by various methods and the result is
1:6 ¡ n ¡ 1:95 for wavelengths between 0.2 and 6 m and temperatures between 300 and 3300 K
[11]. The uncertainties on n induce some uncertainties on the scattering cross-section and phase
function but has only a limited e6ect on the absorption cross-section. The latter is mostly a6ected
by the imaginary part k. Fig. 1 shows a compilation of di6erent experimental or semi-empirical
values of k. The scatter between di6erent results is probably higher than experimental uncertainties
and can be explained by the molten or solid particle phase (the fusion temperature of Al2 O3 is
about 2320 K), by the crystalline phase for solid particles, and mostly by the degree and nature
of impurities [12,13]. Indeed, pure alumina is practically transparent in the visible and near IR, as
discussed by Konopka et al. [15].
We assume in this study that the particles are perfect spheres and use the Mie theory to compute
absorption and scattering cross-sections, and scattering phase functions. The optical constants are
modeled as functions of the wavelength  (in the useful spectral range) and the temperature T
following [9]:
  0:5
2 1:024 1:058 5:281
n= 1+ + +
2 − 0:00376 2 − 0:01225 2 − 321:4
×[1 + 0:0202(T ∗ − 0:473)]; (11)
T
T∗ = ;
1000
k = 0:002(0:062 + 0:7 + 1)exp[1:847(T ∗ − 2:95)]; (12)
518 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

1.0

cumulated phase function 0.8

0.6

0.4 d=80 µm, λ=0.5 µm


d=80 µm, λ=4 µm
d=1 µm, λ=0.5 µm
d=1 µm, λ=4 µm
0.2

0.0
0.0 50.0 100.0 150.0
scattering angle (degrees)

Fig. 2. Cumulated phase function vs. the scattering angle for alumina particles with the optical constants of Dombrovsky
[9]; T = 3000 K.

where  is expressed in m. A three-modal size distribution is adopted (1, 10, and 80 m diameters).
For each class of diameter dp , a Gaussian size distribution around dp is assumed with a standard
deviation equal to dp =10. This size distribution attenuates interference e6ects and makes easier the
angular integration of the scattering source term in radiative transfer calculations. Absorption and
scattering properties have been averaged for each class of particles according to this size distribution
and tabulated vs.  and T .
Fig. 2 shows the cumulated phase function:
 
P() = 2 p( )sin  d (13)
0

vs. the scattering angle . It shows a highly forward peaked phase function for large size parameters.
For instance, for dp =80 m and  =0:5 m, 80% of the scattered energy is concentrated in less than
2◦ . This behavior requires a special numerical treatment in order to preserve energy conservation in
numerical calculations (see Section 3.3).

3.2. Gas mixture properties

Gaseous radiative properties are modeled by using the statistical narrow-band model in the weak
absorption approximation. In this approximation, the band averaged transmissivity #KP$ of a uniform
R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526 519

p=1 atm
p=10 atm
0.50 p=30 atm
Relative error on total emissivity p=50 atm
p=70 atm
0.40

0.30

0.20

0.10

0.00
1 10 100 1000
Column length (cm)

Fig. 3. Discrepancies due to the weak absorption approximation for the prediction of total gas-mixture emissivities at
T = 2900 K and various pressures. Mixture composition is given in the introduction.

column of length l writes


%K 2xplkK
#KP$ = exp − 1+ −1
 %K
K
 exp(−xplk) if 2xplk= K
K %1; (14)
where x is the molar fraction of the absorbing species, p the total pressure, and kK the mean line
intensity to spacing ratio inside the band P$. EM2C band model parameters [17] have been used
for CO2 , CO, and H2 O, while the parameters for HCl have been extracted from Ref. [18]. Fig. 3
shows the errors on uniform column total emissivities, introduced by the weak absorption approxi-
mation, for the gas mixture considered in this study at various pressures. The approximation gives a
reasonable accuracy (less than 10% error) for pressures higher than 20 atm, whatever is the optical
path. However, the errors become important at low pressures and the approximation is not valid for
the study of the far parts in the divergent. The main advantage of this approximation is its com-
patibility with scattering by particles since radiative properties become given in terms of absorption
coeOcients and not only transmissivities.

3.3. Radiative transfer calculations

Fig. 4 presents an example of mixture radiative properties at 50 atm and 3000 K. Particle ab-
sorption coeOcient is typically about 102 m−1 and the radiation penetration depth is very small in
comparison with the typical length scales of the system which can reach several meters. Therefore,
520 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

4
10

absorption and scattering coefficients (m )


particle scattering coefficient
-1 particle absorption coefficient
3 gas absorption coefficient
10

2
10

1
10

0
10
0.0 2.0 4.0 6.0 8.0 10.0
wavelength (µm)

Fig. 4. Comparison between gaseous absorption coeOcient and alumina particle extinction coeOcients for a typical mixture
of combustion products of an aluminized propellant (32% Al2 O3 mass fraction, particles of 1 m diameter (70% in mass)
and 80 m (30%), 50 atm, 3000 K).

we adopt a locally one-dimensional radiative transfer model, i.e., for a given cross-section normal to
a wall at axial abscissa x0 , all the scalar 1elds (temperatures, particle concentrations, etc.) are those
calculated from the aerothermal models at this cross-section x0 , but are assumed independent of x
for radiative transfer calculations.
For radiative transfer in one-dimensional 1elds, the radiative transfer equation (RTE) may be
written (omitting a subscript for the frequency dependence) as

@I  )i (#) )g (#) ◦ !d (#)


3
(; #) = −I (; #) + I ◦ (Ti ) + I (Tg ) + I(; #) (15)
@# i=1
%(#) %(#) %(#)

with the source term



I(; #) = I (#;  )p̃(i; i ; #) d* : (16)
4

In these equations, the subscripts i and g refer to a class of particle size and to the gas phase,
respectively, I is the radiative intensity averaged over a narrow band, I ◦ (T ) is the equilibrium
intensity at temperature T ,  is the cosine of the angle between the normal to the slab and the
propagation direction, p̃(i; i ; #) is the scattering phase function accounting for all particle classes, #
R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526 521

is the optical thickness de1ned as


 s
#= %(s ) ds ; (17)
0
3

%(s) = (!d i (s) + )i (s)) + )g (s); (18)
i=1
% being the extinction coeOcient, !d i and )i the particle scattering and absorption coeOcients, and
)g the gas absorption coeOcient. RTE may be written in an integral form between optical paths #
and # + P#, assuming small temperature variations inside P#:
 #+P#
I (# + P#; ) = exp(−P#)I (#; ) + ! exp(# − # − P#)I(# ; ) d#
#

3

+(1 − !(#))[1 − exp(−P#)] -i (#)I ◦ (Ti ) + -g (#)I ◦ (Tg ) I; (19)
i=1

where we have introduced the total albedo !(#)= 3i=1 !d i =%(#), and the partial absorption coeOcients

-i (#) = )i (#)=( 3j=1 )j (#) + )g (#)). A numerical diOculty appears in the integration of the source
term in Eq. (19) when the optical path P# is not thin. We introduce here an exponential scheme
and assume that the variation of I in the source term follows
I (# + # ;  ) = exp(−.# )I (#;  ): (20)
The discretization of the zenithal cosines into N values k leads then to the discrete form of the
RTE:
I (# + P#; j ) = exp(−P#)I (#; j )
3

+[1 − !(#)][1 − exp(−P#)] -i (#)I ◦ (Ti ) + -g (#)I ◦ (Tg )
i=1

N
 1
+!(#) I (#; k )Pjk (#)[exp(−.P#) − exp(−P#)]; (21)
1−.
k=1
where the discrete scattering phase function is de1ned as
   2 
1 j +P=2 k +P=2 
Pjk (#) = p̃(i; i ; #) d/ dk dj : (22)
P j −P=2 k −P=2 0

Here, P=2=N is the  increment and / is the azimuthal angle. The integration of Eq. (22) is carried
out at high angular resolution in order to capture the highly forward peaked phase function, especially
for particles of large size parameter. The scattering matrix Pjk is 1rst tabulated vs. wavelength and
temperature and stored before performing radiative transfer calculations.
In order to take into account gas and particle temperature +uctuations in radiative transfer, a Taylor
series expansion of equilibrium intensities is carried out. The intensity I ◦ (Ti ); i being a particle class
or the gas, is replaced by I ◦ (Ti ) with
@2 I ◦
I ◦ (Ti ) = I ◦ (Ti ) + 12 Ti2  2 (Ti ): (23)
@T
522 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

Temperature +uctuations are only accounted for in emission terms. Absorption terms are assumed
to be non-+uctuating since the dominant contribution here is particle radiation and particle radiative
properties vary very slowly with the temperature.
An iterative scheme has been developed to solve the above discrete integral form of the RTE.
The boundary conditions are speci1ed through wall temperatures and isotropic emissivities or per-
fectly re+ecting walls in order to reproduce symmetry conditions. Iterations are undertaken between
ascending and descending directions and the local and directional scattering source terms, as well as
the exponential coeOcient ., are actualized during each iteration step. It was found that the use of
the exponential scheme enhances the convergence of the procedure in terms of elementary optical
thicknesses, although the latter must remain limited to about 0.5 (in normal directions) to obtain a
mesh independent solution.

4. Results and discussion

Calculations have been 1rst carried out for a fully developed gas velocity 1eld and an x-independent
temperature pro1le with or without turbulent dispersion of particles. Gas velocity and temperature
pro1les, typical of the pro1les near a wall point in the inner region of the MPS-P230 motor of the
Ariane5 launcher, are shown in Fig. 5. The wall temperature is assumed equal to 1500 K and its
emissivity is assumed grey and equal to 0.8. The turbulence intensity is relatively small since the
velocities are small in the internal parts of the motor. A bi-modal particle size distribution is assumed
in this region with 70% (in mass) small particles of dp = 1 m and 30% particles of dp = 80 m.
The intermediate particles are mostly produced through fragmentation and coalescence phenomena
near the nozzle.
Fig. 6 shows the distribution of the radiative power per unit volume computed without parti-
cle turbulent dispersion, i.e., particle concentrations are simply assumed proportional to gas density
at the local temperature. The results from three assumptions are illustrated in this 1gure: (i) a
uniform temperature pro1le with gas and particle temperatures equal to the temperature outside the

3500 15

3000
Temperature (K)

10 mean axial velocity


-1

2500
ms

5
2000 1/2
k

1500 0
0.00 0.01 0.02 0.03 0.04 0.05 0.00 0.01 0.02 0.03 0.04 0.05
distance to wall (m) distance to wall (m)

Fig. 5. Gas temperature, velocity and k 0:5 pro1les used for radiative transfer prediction in an internal part of the motor.
R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526 523

0.5 Isotropic scattering (0.63)


Anisotropic scattering (1)

nondimensional radiative power


0.3
isothermal profile (3.63)
0.1

-0.1

-0.3

-0.5

-0.7

-0.9

-1.1
0.000 0.010 0.020
distance to the wall (m)

Fig. 6. Non-dimensional radiative power per unit volume vs. the distance to the wall, computed with data representative
of the inner regions of the MPS-P230 motor of Ariane5 launcher. The number in parentheses indicates the resulting wall
radiative +ux divided by the +ux obtained from the full calculations.

boundary layer (3380 K); (ii) the temperature pro1le shown in Fig. 5 but with isotropic particle
scattering; and (iii) the full calculations with the non-uniform temperature pro1le and anisotropic
scattering properties. It is shown in this 1gure that the cold boundary layer attenuates signi1cantly
the radiative +ux coming from the outer hot regions. This layer is neither optically thin nor thick.
Thus the precise knowledge of the two-phase aerothermal 1elds inside the boundary layer is crucial
for radiative transfer predictions. The results from calculations assuming isotropic scattering show
that the highly forward scattering character of the considered particles makes them less eOcient
in blocking the radiation coming from the outer regions. Assuming isotropic scattering leads to
underestimation of the radiative wall +ux of 37%.
Fig. 7 shows the evolution of wall radiative +uxes when particle turbulent dispersion is accounted
for. For this simulation, the particles are injected in one-dimensional gas velocity and temperature
1elds (shown in Fig. 5) but their concentrations and mean temperatures are allowed to vary along
the axial coordinate according to the turbulent dispersion. The 1gure shows that this turbulence
dispersion results in a signi1cant increase in the total radiative wall +ux. In fact, the turbulent
dispersion leads to an increase of particle concentration in the immediate viscinity of the wall but
to a signi1cant decrease of this concentration in the region of maximum turbulence intensity, as
shown in Fig. 8. This e6ect, is a result of centrifugal ejection of particles by energy containing
eddies, as observed also in [19,20], and their accumulation in less turbulent regions. It is more
pronounced for the small particles characterized by small inertia. The result for radiative transfer
is a less eOcient extinction of radiation coming from the outer region by the cold boundary layer.
Fig. 7 also shows the contributions of the di6erent particle classes and the gaseous phase to radiative
transfer. The small particles are by far the most eOciet. particle classes and the gaseous phase to
radiative transfer.
524 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

nondimensional wall radiative flux

Gas contribution
Particles of dp=1µm
Particles of dp=80µm
0.5
Total with turbulent dispersion
without turbulent dispersion

0
0.0 0.2 0.4 0.6 0.8
distance from the entrance (m)

Fig. 7. Non-dimensional radiative wall +uxes with and without particle turbulent dispersion.

3e+15 0.20

0.15
2e+15 mechanical turbulence intensity
N (m )
-3

Im

0.10

1e+15
0.05

with turbulent dispersion


withou turbulent dispersion
0 0.00
1 10 100 1000
+
y

Fig. 8. Number densities of small size particles with and without turbulent dispersion at section x = 100 cm from the
entrance. y+ is the non-dimensional distance to the wall (y+ = 10 corresponds here to y  0:03 cm).

The turbulent dispersion of particles introduces also a thermal non-equilibrium e6ect between
condensed and gaseous phases. The particles ejected from the turbulent regions towards the wall
have a certain thermal relaxation time and are, in general, at higher temperature than the gas. Fig. 9
R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526 525

300

with turbulent dispersion, without Pr


with turbulent dispersion, with Pr
200
without turbulent dispersion
Tp-T g (K)

100

-100
1 10 100 1000
+
y

Fig. 9. Di6erences between particle and gas average temperatures at x = 100 cm. Particles of 80 m diameter.

shows this e6ect for the particles of 80 m diameter. Temperature di6erences reach about 200 K,
but the e6ect for small particles is very weak since their thermal relaxation time is much more
smaller. The e6ect of this thermal non-equilibrium on radiative +uxes remains limited to about 2%
since radiative transfer is mostly due to the small particles. Similarly, the direct e6ect of temperature
+uctuations on radiative +uxes was found very small in this application due to the small levels of
turbulence.

Acknowledgements

The authors acknowledge the helpful co-operation with engineers from Snecma Propulsion Solide
(F. Laturelle, J.M. DSeoclSezian, and Ph. Le Helley) and the partial 1nancial support from this company
and from CNES during the Ph. D. preparation of one of the authors (R.D.).

References

[1] Sabnis JS. Modeling of particulate +ow e6ects in solid rocket motors. AIAA Lecture Series, 35th AIAA Aerospace
Sciences Meeting, Reno, 1997.
[2] Brewster MQ, Parry DL. Radiative heat feedback in aluminized solid propellant combustion. J Thermophy 1988;2:
123–30.
[3] Ishihara A, Brewster MQ, Sheridan TA, Krier H. The in+uence of radiative heat feedback on burning rate in
aluminized propellants. Combust Flame 1991;84:141–53.
S
[4] Mignon Ph. Etude thSeorique du couplage convection turbulente-rayonnement dans un eS coulement de gaz dans un
S
canal. ThermodSegradation de la paroi. ThUese de Doctorat, Ecole Centrale Paris, 1992.
526 R. Duval et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 84 (2004) 513 – 526

[5] Zeman O. Dilatation dissipation: the concept and application in modeling compressible mixing layers. Phys Fluids
1990;A2:178–88.
[6] Sou1ani A, Mignon P, Taine J. Radiation e6ects on turbulent heat transfer in channel +ows of infrared active gases.
ASME-HTD 1990;137:141–9.
[7] Duval R. Transferts radiatifs dans des chambres de combustion de propulseurs aU propergols solides aluminisSes. ThUese
S
de Doctorat, Ecole Centrale Paris 2002– 05, 2002.
[8] Shirolkar J, Coimbra C, McQuay MQ. Fundamental aspects of modeling turbulent particle dispersion in dilute +ows.
Prog Energy Combust Sci 1996;22:363–99.
[9] Dombrovsky LA. Radiation heat transfer in disperse systems. New York: Begell House; 1996.
[10] Reed RA, Calia VS. Review of aluminum oxide rocket exhaust particles. AIAA Paper no. AIAA-93-2819, 1993.
[11] Parry DL, Brewster MQ. Optical constants of Al2 O3 smoke in propellant +ames. J Thermophy 1990;5:142–9.
[12] Plastinin YA, Sipatchev HP, Karabadzhak GF, Khmelinin, BA, Szhenov EY, Khlebnikov AG, Shishkin YN.
Experimental investigation of alumina particles’ phase transition and radiation. AIAA Paper no. AIAA-98-0862,
1998.
[13] Pluchino AB, Masturzo DE. Emissivity of Al2 O3 particles in a rocket plume. AIAA J 1981;19:1234–7.
[14] Mularz EJ, Yuen MC. An experimental investigation of radiative properties of aluminum oxide particles. JQSRT
1972;12:1553–68.
[15] Konopka WL, Reed RA, Calia VS. Measurements of infrared optical properties of Al2 O3 rocket particles. AIAA
Paper no. AIAA-83-1568, Montreal, 1983.
[16] An1mov NA, Karabadzhak GF, Khmelinin BA, Plastinin YA, Rodionov AV. Analysis of mechanisms and nature
of radiation from aluminum oxide in di6erent phase states in solid rocket exhaust plumes. AIAA Paper no
AIAA-93-2818, 1993.
[17] Sou1ani A, Taine J. High temperature gas radiative property parameters of statistical narrow-band model for
H2 O; CO2 and CO, and correlated-k (ck) model for H2 O; CO2 . Int J Heat Mass Transfer 1997;40:987–91.
[18] Ludwig CB, Malkmus W, Reardon JE, Thomson J. Handbook of infrared radiation from combustion gases. NASA
SP-3080, 1973.
[19] Brooke J, Hanratty T. Free-+ight mixing and deposition of aerosols. Phys Fluids 1994;6:3404–15.
[20] Eaton JK, Fessler JR. Preferential concentration of particles by turbulence. Int J Multiphase +ow, 1994; 20(Suppl):
169 –209.

You might also like