This article was downloaded by: [Ohio State University Libraries]

On: 13 June 2012, At: 13:59

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Combustion Science and Technology

Publication details, including instructions for authors and subscription information:
[Link]

Combustion of Aluminum-Based Slurry Agglomerates

a a a
S. R. TURNS , S. C. WONG & E. RYBA
a
The Pennsylvania State University, 315 Mechanical Engineering Building, University Park,
PA, 16802

Available online: 16 May 2007

To cite this article: S. R. TURNS, S. C. WONG & E. RYBA (1987): Combustion of Aluminum-Based Slurry Agglomerates,
Combustion Science and Technology, 54:1-6, 299-318

To link to this article: [Link]

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: [Link]

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
 Combust. [Link] Tech.• 1987, Vol. 54, pp. 299-318 o 1987 Gordon and Breach Science [Link],
Photocopying permitted by license only Printed in Great Britain

Combustion of Aluminum-Based Slurry Agglomerates

S. R. TURNS, S. C. WONG and E. RYBA 315 Mechanical Engineering Building, The

Pennsylvania State University, University Park, PA 16802

(Received August IS, 1986; in final form May 5,1987)

Abstract-The combustion of aluminum and aluminum/carbon slurry agglomerates was studied in the
post-flame region of a flat-flame burner using cinematography. Diameters of the agglomerates after
both burnout of the carrier fuel (JP'10) and coalescence of the particles ranged from approximately
300 to 800 microns. Burner operating conditions were varied to provide oxygen mole fractions ranging
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

from 0.10 to 0.25 and gas temperatures from 1500 K to 1800 K. Flame environments both with and
without water vapor were considered. Vapor-phase burning with a significant accumulation of oxide
product was observed for aluminum slurry agglomerates. Burning times were approximately four times
those of pure lP-10 droplet at corresponding initial sizes and test conditions. Exercise of an analytical
model which included radiation suggested that gas-phase transport of species from the flame zone and
subsequent condensation at the surface is not likely to be the sole mechanism responsible for surface
oxide accumulation. The combustion characteristics of an aluminum/carbon slurry were found to be
significantly different from those of aluminum-only slurries. Depending upon ambient gas temperature
and oxygen content, appreciable quantities of relatively stable aluminum carbide were formed with no
observable vapor-phase combustion.

INTRODUCTION

This paper is a companion to a recent report on the ignition characteristics of

aluminum-based slurry fuels (Wong and Turns, 1986a). The present work focuses
on the combustion characteristics of slurry droplets after the liquid component
(JP-lO) has burned out and after the remaining aluminum agglomerate has ignited.
IIi the prior work (Wong and Turns, 1986a), we showed that in a hot oxidizing
stream, aluminumlJP-10 slurry droplets allow burning of the liquid JP-10 without
droplet shattering or fragmentation-unlike boron/Jl-LO slurries (Takahashi,
Dryer and Williams, 1986) or aluminum/carbon/Jlvl O slurries {Wong and Turns,
1986a, b)-producing highly porous and friable agglomerates of aluminum
particles. These agglomerates then heat up to essentially the oxide melting point
(ca. 2300 K), presumably destroying the integrity of the oxide coating, and the
individual particles rapidly coalesce into a single molten drop of aluminum.
Concomitantly, a detached envelope flame appears and the aluminum burns. The
processes occurring from particle coalescence to final burnout of the aluminum
are the subject of the present study.
Aluminum combustion has been a subject of considerable study for the past
several decades and has a large body of literature associated with it. Price (1983)
provides an up-to-date review of aluminum combustion. It is generally accepted
that under the proper conditions aluminum can burn in the vapor phase, similar to
a hydrocarbon fuel (Wilson and Williams, 1971 ; Brzustowski and Glassman,
1964a; Kuehl, 1965a, b). Although purely vapor-phase combustion with a
detached envelope flame does occur in dry O 2 / Ar atmospheres, oxide condenses
or is formed on the surface of aluminum droplets burning in air or in
environments containing water vapor, nitrogen, carbon dioxide, carbon monoxide,
etc. (Friedman and Macek, 1963; Prentice and Nelson, 1968; Prentice, 1970;
Prentice, 1974). The accumulation of the surface oxide complicates the
299
 300 S. R. TURNS, S.-c. WONG AND E. RYBA

combustion process and is related to the frequently observed behavior of jetting,

spinning, and fragmentation of the burning aluminum droplets.
Theoretical models have been proposed to describe the combustion of single
metal particles which assume either vapor-phase combustion (Kuehl, 1965b;
Brzustowski and Glassman, 1964b; Kuehl and Zwillenberg, 1968; Klyachko,
1969; Law, 1973), surface combustion (Bartlett et al; 1963), or the simultaneous
existence of surface and vapor-phase combustion (Gremyachkin, Istratov, and
Leipunskii, 1975). Experimental observations are consistent in many respects with
the vapor-phase combustion models, although surface combustion offers a natural
explanation of fragmentation and the formation of oxide bubble products observed
in some experiments. Since oxide forms on the surface of burning aluminum
particles in most environments, a vapor-phase model considering the accumulation
of oxide on the particle surface provides a reasonable simplified treatment. Law
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

(1973) presented a model which allowed oxide accumulation as a result of

diffusion of dissociated oxide products to the surface of the burning aluminum
droplet. This model was adapted and extended for the present study.
The objectives of the present study were to delineate the differences between
single aluminum particle combustion and slurry agglomerate combustion and to
determine the effects of drop diameter, ambient gas temperature, ambient oxygen
mole fraction, and ambient gas water vapor content on the combustion-times of
aluminum-based slurry-fuel agglomerates using a combination of experiment and
analysis. An additional objective was to determine the influence of the same
parameters on the structure and composition of the products remaining after
combustion.

EXPERIMENTAL METHODS

Test Apparatus
The ignition of aluminum slurry droplets was studied by photographically
observing probe-supported droplets in the post-flame gases of a flat-flame burner
using the apparatus described in Wong and Turns (1986a). Slurry droplets having
initial diameters ranging from about 500-1100 microns were suspended on 125-
micron diameter silicon carbide coated carbon filaments and were exposed to the
hot gases by rapidly withdrawing a shield. This technique provided molten
aluminum droplets with diameters ranging from 300-800 microns. The
stoichiometry and temperature of the burner product gases were controlled by
adjusting the burner feed gas flow rates.

Instrumentation
Particles were observed with a Redlake Locam 16 rnm motion picture camera
using a magnification of 1.07 : 1. An externally-driven LED marker produced a
timing reference. To obtain silhouette images, the particles were backlit with a
600 W floodlight. The films were analyzed using a Vanguard/Bendix computer-
digitizing motion picture analyzer.
Gas properties were determined as follows: Temperatures were measured with
200 ,urn diameter beryllium oxide/yttrium oxide coated PtlPt-Rh thermocouples.
corrected for radiative heat losses; concentrations were measured by isokinetic
sampling and analysis with a gas chromatograph; and velocities were calculated
 ALUMINUM-BASED SLURRY AGGLOMERATES 301

using the measured burner gas flow rates and the computed density of the post-
flame gases (based on the measured gas temperature and composition).
The surface morphology of agglomerates and combustion residual products
were investigated with a scanning electron microscope (International Scientific
Instruments, model Super ill-A). SEM micrographs were taken using Polaroid type
52 films. The phase composition of the combustion residual products found on the
SiC/C probes was obtained by X-ray diffraction analysis. The small samples were
mounted in Debye-Scherrer cameras, and diffraction patterns were taken using
CuKa radiation. Rough estimates of the relative portions of the phases present
were made visually from the Debye-Scherrer photographs.

Test Conditions
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

Tests were conducted in carbon monoxide/oxygen/nitrogen post-flame gases

having a water mole fraction less than 0.004 and in methane/oxygen/nitrogen
flame gases with water mole fractions ranging from 0.12 to 0.16. For the study of
temperature effects, both dry and wet conditions provided a range of gas
temperature (1510-1809 K) with essentially constant oxygen mole fraction (ca.
0.10). Dry conditions also provided a range of oxygen mole fraction (0.10-0.25)
with essentially constant gas temperature (ca. 1818 K). A more detailed
specification of the flame conditions is presented in Wong and Turns (1986a, b).

SlurrySamples
Three different slurries were studied : (a) A stabilized aluminumlJP-10 slurry with
a solids mass loading of 42.6 percent, (b) the same slurry as above, however,
without stabilizing additives, and (c) a stabilized aluminumlcarbon/JP-10 slurry
with an aluminum mass loading 35.5 percent and a carbon loading of 9.3 percent.
Constituent particle size for the aluminum was nominally four microns, and for the
carbon, submicron.

ALUMINUM COMBUSTION MODEL

Experimental observations for the Al/ JP-10 slurries showed that after coalescence
of the particles constituting the agglomerate, combustion generally proceeded with
a nearly spherical, detached flame surrounding the aluminum droplet. Oxide
smoke was convected out of the flame zone regions in the wake of the burning
droplet. Furthermore, surface accumulation of oxide on the leeward side was
observed. To model this combustion process with a minimum of complexity, the
theoretical vapor-phase combustion analysis of Law (1973) was adapted and
extended . This extended model is schematicaly illustrated in Figure 1a.
In the model, we assume that aluminum vapor and oxidizer react
instantaneously in stoichiometric proportions, v, forming a flame sheet
surrounding the molten aluminum droplet. The relationships among the fuel vapor,
oxidizer and products mass fluxes are illustrated in Figure l b, where MAJ is the
dimensionless mass flow rate of aluminum vapor. The flame temperature is fixed
at the oxide boiling point (3850 K), and the droplet surface temperature is at the
aluminum boiling point (2767 K). The gas-phase conservation equations for
species, mass and energy are solved assuming two spherically symmetric zones, one
 302 S. R. TURNS , s-c. WONG AND E. RYBA

I I \
I \
/
I ~
1\ ~CONDENSED
CONVECTED

/ /
/ \

\
\

\
OXIDE

CONDENSED /
OXIDE I THIN FLAME ZONE
WITH CONOENSED
OXIDE
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

ALUMINUM VAPOR

LIQUID DISSOCIATED OXIDE

ALUMINUM VAPOR

OXYGEN

[Link]

DROPLET SURFACE FLAME SHEET FILM RADI(

/
~
ALUMINUM VAPOR I OXIDIZER
I
I
.1. I
I I
I CONDENSED
I PRODUCTS I
I
I I
I
I VAPOR PRODUCTS I
I I
I
I I
I A ISh
1'= I 1'=1'f roo = Sh-2
INN ER REGION OUTER REGION
I
[Link]
FIGURE 1 (a) Aluminum combustion model incorporating oxide accumulation. (b) Mass fluxes of
fuel, oxidizer and combustion products.
 ALUMINUM-BASED SLURRY AGGLOMERATES 303

inside the flame and one outside. The effects of forced convection, enhancing the
combustion rate and producing a slightly asymmetric flame zone, are accounted
for by defining an equivalent spherical system using film theory. Estimates of
Grashoff and Reynold s numbers indicate that forced convection effects are much
more important than natural convection for the present test conditions. Variable
gas phase properties are treated as described in Wong and Turns (1986a) using
properties of aluminum vapor from Svehla (1962). A unity Lewis number also is
assumed.
To maintain the flame temperature at the oxide boiling point, a fraction, 0, of
the oxide is vaporized. A fraction, n, of the vapor products, defined as
Al 20 3(g} == 2 AlO(g) + !02' diffuses inward to, condense at "the surface, .while the
remainder, 1 - rj, moves radially outward. Depending upon the directions of the
mean velocities in the inner region (1 < r< rf ) and the outer region (r> rf), the
condensed products are either trapped in the flame (~I = ~2 = O), or are convected
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

away (~1 = 0, ~2 = I). In general, a third mode can be considered where condensed
products are convected to the surface (~1 = 1, ~2 =O); however, with the
assumptions above, this mode does not occur for aluminum combustion. Law
(1973) has shown that certain inequality constraints must be satisfied for a solution
to exist in each mode .
The oxide which condenses at the droplet surface is assumed to accumulate and
form a spherical segment with height h as shown in Figure 1a. We assume that as
the cap grows, the spherical symmetry embodied in the model is retained over the
solid angle associated with the exposed aluminum surface; thus the combustion
rate of the aluminum is equal to the ratio of exposed area to total surface area
multiplied by the burning rate computed for a sphere having the same diameter as
the combined oxide-aluminum particle . With these assumptions, mass
conservation for the aluminum portion of the droplet can be expressed

dV,\1
- = -4Jlr 2 ( 1 - - m
dt '
h).
2"
/I

AI
/p
AI ,
(1)

where the mass rate of aluminum consumption per unit surface area, rh;l is
obtained from the solution of the gas-phase conservation relations. As the
aluminum is depleted, the oxide volume grows, and its rate of growth can be
related to that of the aluminum as follows:

dJ{,x = _ ~() PAl dv",1 . (2)

dt Pox dt
The overall particle radius , rso can be readily computed from the spherical
geometry as

(3)

and the cap height, h,from the following cubic equation

(4)

To describe the history of the burning aluminum/condensing oxide droplet, Eqs.

(1) and (2) are integrated numerically. To evaluate the burning flux rh;1 at any'
instant in time, it is necessary to solve a system of six non-linear, coupled
 304 S. R. TURNS. s-c. WONG AND E. RYBA

equations related to the gas phase. Details of the formulation of this model are
presented in Wong and Turns (1986b).

RESULTS AND DISCUSSION

General Observations
A typical sequence of events for the stabilized Al/JP-I0 slurry droplet is illustrated
by the backlit photographs and diameter history shown ·in Figure 2 (Wong and
Turns, 1986a). This figure shows that the original slurry droplet swells slightly as
the JP-IO burns out. The transformation from an obviously liquid droplet to a
fuzzy agglomerate (Figure 2, photograph B) occurs at about 0.2-0.3 s where the
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

JP-10 burnout nears completion. The agglomerate of aluminum particles then

heats up, and the individual particles coalesce, indicated by the sharp decline .in
particle diameter between approximately 0.5 and 0.55 s. At the instant when
coalescence appears complete, a vapor-phase diffusion flame surrounding the
molten particle appears, with the sequence of photographs showing both visible
emission from the flame zone and blockage of the backlighting by the oxide
smoke, particularly in the cooler wake regions. Although the overall effect of the
photographic sequence is to suggest a strongly axially symmetrical combustion
process, the actual flame zone (regions of highest luminosity) appears to be more
like a perturbed sphere, or a segment of sphere when the oxide cap blocks a
portion of the surface. Natural light photographs also indicate that the major
deviation from spherical symmetry results from the smoke trail and not the flame
zone.
Interestingly, the backlighting intensity is such that, after ignition, some emission
from the burning droplet can be seen . As a result of the difference in emissivity
between the molten aluminum and the oxide, the liquid aluminum droplet can be
distinguished from the oxide cap which forms on the leeward side of the particle.
Shifting of the position of this cap-like structure results in an asymmetrical smoke
trail as shown in Figure 2, photographs F and G. Weak jetting from the burning
droplets, observed in natural light photographs, occurs occasionally. As burning
proceeds, the overall particle diameter decreases as the relative size of the oxide
cap grows. At about 1.8 s (Figure 2, photograph I), the flame appears to collapse
on the particle surface, and no smoke is visible. For a brief while, the particle
continues to glow with some shifting of the shape of the particle, after which the
particle cools, and no radiation is visible on the film records. For the purposes of
this study, the end of combustion was chosen to occur when the oxide smoke was
no longer discernible. Although the tip portion of the SiC/C support probes
protruding through the droplets appeared to be consumed while the aluminum
droplets burned, only a trace of /3-SiC, and no carbon compounds, were found in
the oxide residuals . Thus, it is not likely that probe burning had a significant
influence on the aluminum combustion process.
Experiments were performed where argon was substituted for the nitrogen
diluent to determine if this influenced the accumulation of oxide on the burning
drop. Work by Prentice and Nelson (1968) and Wilson and Williams (1971)
showed that in oxygen/diluent atmospheres oxide accumulation was decreased or
eliminated by the substitution of argon for nitrogen. However, no discernible
differences were observed on the backlit film records for the combustion of
stabilized AlIJP-IO slurry droplets in CO/0 21 Ar and CO/O/N 2 tlame
 ALUMINUM-BASED SLURRY AGGLOMERATES 305

environments . This result is probably a consequence of the relatively complex

composition of the ambient gases, i.e., CO, CO 2 , trace H 20 , etc., since Prentice
(1970 , 1974) has shown that oxide accumulation is also related to the presence of
carbon-contairiing species. Whether jetting occurred or not in COIO / Ar
environments could not be ascertained since no natural-light photographs were
taken.

0
1.2
"0
......
"0
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

1.0

0
-l- 0 .8
e::(
0:
0 .6
0:
W
I- 0.4
W
~

-0« 0.2

0
0 0 .4 0.8 1.2 1 .6 2.0

TIME (8)
FIGURE 2 Typical aluminum slurr y d roplet life history.

The sequence of events for a stabilized AVJP-10 slurry droplet in the wet
environments was largely similar to that in dry environments. However, two major
differences were observed : First , relatively violent eruptions or partial
fragmentation of the particles occurred in wet environment s, and jetting from the
particles was stronger, and more frequent, in wet environments than in dry ones.
Fragmentation has been observed by Prentice (1970 , 1974) to be characteristic of
aluminum particles burning in wet environments. Second, the vapor-phase flame
fronts were closer to the burning particles at the early stage after ignition for the
wet environments. Thi s may be mainly due to the higher gas velocities in the wet
test conditions. However, for about half of the combustion time, the molten
aluminum seemed to be nearly encapsulated by the oxide, and the smoke was
confined to a smaller region where aluminum was exposed. As burning neared
completion, the particles vibrated, moving up and down and back and forth on the
SiC probes, producing the last vestiges of an oxide smoke. Encapsulation, which
has also been observed by Prentice (1970 , 1974) may be related to the strong
fragmentation in wet environments.
 306 S. R. TURNS, S.-c. WONG AND E. RYBA

Other than the detailed appearance of the residual combustion products for the
stabilized and unstabilized Al/JP-10 slurries, no major qualitative differences in
the combustion characteristics are obvious. On the other hand, the carbon-
containing slurry behaved quite differently than the AIIJP-lO slurries. For all of the
flame conditions with O 2 levels of about 10.4 percent, no gas-phase flames were
observed for the Al/C slurries, while detached flames were obvious for the AI
slurries except when gas temperatures approached 1500 K. Another major
difference was the shattering and fragmentation of the AIIC slurry agglomerates at
the higher O 2 level (25.2 percent).

COl 021 N2 FLAM E

1.0
dm" 600 11m
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

0 Too" 1809 K
02 " 10.4 "10
Ci
:!:
.....
)(
0.8
0
:!:
.
o.
Ci 0.6
:!:
.....
Ci
:!:

't' 0.4
.....
s:

E
't'
..... 0 .2
-e

0 .5 1.0 1.5 2 .0 2.5 3.0

TIME (5)

FIGURE 3 Predicted life history of aluminum agglomerate after ignition.

Typical Theoretical Results

To assist in interpreting the experimental data, the theoretical model was
exercised. A typical theoretical result is shown in Figure 3. The zero time
represents ignition of the aluminum, and dm is the diameter of the coalesced
aluminum droplet. As the aluminum mass is depleted, oxide vapor produced in the
flame diffuses back to the droplet surface where it condenses and builds the oxide
cap which grows with time. When the aluminum is totally consumed, the oxide
accumulation represents over 40 percent of the original mass of the aluminum,
assuming a solid nonporous product, with the final particle size about 65 percent
of the original droplet size. In the experiments, final particle sizes ranged from 60
to 80 percent of the original molten drop size, depending on the experimental
conditions. Also, the particles contained a number of small voids. Because of the
experimental difficulty in defining the end of combustion and because of the
relatively long theoretical time required to consume the last few percent of the
 ALUMINUM-BASED SLURRY AGGLOMERATES 307

aluminum, theoretical combustion times for 95 percent complete combustion were

used to compare with the experimental data.
It should be pointed out that in applying the model to the present experimental
conditions, radiation effects were necessarily deleted. A parametric study showed
that radiation effects had a minimal effect on burning rates, and hence, burning
times; but more importantly, it also showed that radiation losses could be
sufficiently large to prevent oxide dissociation in the flame and thereby remove the
mechanism for the experimentally observed oxide accumulation on the surface of
the burning droplet. This coupling of radiation and the gas-phase based oxide
accumulation mechanism suggests that surface reaction and/or particle transport,
including thermophoresis, may be responsible for oxide accumulation. An
investigation of these effects was beyond the scope of this work.
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

5.----.,.------.---,---,------,
CO 1 °21 N2 FLAME
T(I)' 1809 K
02' 10.4 "10
4

UJ
:':!: 3
I- THEORY
Cl
(F • 0 l
Z
Z
a:
2
::::>
CD

THEORY
(F· 1.0)

OL-----=-"=-....l.-_ _----l --'-_ _----'-_ _----J

o 200 400 600 800 1000

FIGURE 4 Burning times versus diameter for coalesced aluminum slurry agglomerates.

AlIJP-IO Slurries
Initial diameter effects In Figure 4, experimental and theoretical burning times
versus initial molten droplet size are compared. Two theoretical curves are shown
which essentially bracket the data: The upper curve (F= 0) employs the assumption
of an impervious oxide cap, while the lower curve (F= 1) assumes that there are
sufficient fissures or fumaroles in the cap to completely wet the cap with vaporizing
aluminum. It can be seen that the experimental data lie closer to the totally wet cap
limit (i.e., F= 1). Justification for including some leakage of aluminum through the
oxide can be found in the flame structure seen in Figure 2, photographs E-G,
where a strong concentration of smoke is seen directly above the oxide cap. In fact,
 308 S. R. TURNS, S.-c. WONG AND E. RYBA

it appears that separate flames may exist for the completely exposed aluminum
surface and for the surface covered by the cap . Power-law regression of the data in
Figure 4 shows burning time proportional to diameter raised to the 1.77 ± 0.20
power. This burning time exponent lies between the stagnant environment limit
(n = 2.0) and the forced convection limit (n = 1.5).
In Figure 5, the burning times of the coalesced molten aluminum droplets
measured in the present study are plotted along with the measurements for single
aluminum particles by Wilson and Williams (1971) and Prentice (1974). The
consistency of the data shows that the combustion process for aluminum slurries is
essentially the same as for single aluminum particles, as one might expect since the
agglomerate particles coalesce into a single molten drop.
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

~ o~
o 10 % } PRESENT STUDY cPO
° 25 % 0
•
•
20
21 %
%} PRENTICE qp~
~
"0%%} WILSON 8 WILLIAMS 0V,jj
0
• 10 0 <:lIA.
• 15 :l"l
• 20

• 30 % ffJ

•
•

10° '----_---L_..l--c-L--'-l.--L.l....I..J.._ _ -'----I.----l--L...-LLLJ-'--_ _

10'

FIGURE 5 Comparison of burning times of stabilized slurry droplets with single aluminum particles.
 ALUMINUM-BASED SLURRY AGGLOMERATES 309

Oxygen mole fraction effects The experimental and theoretical results for the
influence of oxygen mole fraction are compared in Figure 6. The data were
obtained through interpolation at dm = 400 and 600 .um on burning-time-versus-
d; plots similar to Figure 4. It can be seen from Figure 6 that both absolute and
trend wise agreement occur between the simplified theory and experiment. The
experimental data again are bracketed by the two theoretical curves (F = 0 and
F= 1) and lie closer to the totally wet cap limit (i.e., F= 1). Both theoretical and
experimental results show that the O 2 level has a strong effect on burning time as
O 2 levels decrease. The sharp increase in burning times at the low 02 levels
indicates that the limit of no dissociation in the flame is being approached.

2 .4
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

CO 1 021 N2 FLAMES

Too" 1818 K
2.0
0 dm " 400 IJm
t>. 600 IJm

-
oil 1.6
_ THEORY

UJ
~
~

Cl
1.2
z
z
Q:
::>
aJ 0 .8

0 .4

OXYGEN MOLE PERCENT

FIGURE 6 Aluminum droplet burning times versus oxygen mole fraction.

Ambient gas temperature effects Results of both experiment and theory showed
very little influence of ambient temperature on burning times over the range of
1400-1800 K for O 2 levels of about 10.5 percent in either wet or dry
environments as shown in Figure 7. Thus the effect of gas temperature is confined
to the agglomerate ignition process (Wong and Turns, 1986a). This result is
reasonable since the flame temperature is essentially fixed by the oxide boiling
point (3850 K) independent of the ambient temperature, unless the ambient
temperature is reduced to such a low value that no vapor products are formed in
the flame.
 310 S. R. TURNS, s.c. WONG AND E. RYBA

2.4 , - - - , - - - - - - , - - - - . - - -_ _----,- -,--_ _--,

CH4/02/N2 FLAMES
• EXPERIMENT
- __ THEORY
2.0
CO/02/N2 FLAMES
o EXPERIMENT

-..
-7-----_~-- THEORY
1.6

w FaO
::I!
I-

Cl
z
1.2
-- ------------- o -..--..
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

Z
a:
::::l
m 0 .8
Fal •
- --------- --------
0.4
dm a 500 11m
O2 • 10.4 -10.7 "Ia

°13'-:0-=-0----'---- '- - ---'---------I.---...J....----I

1500 1600 1700 1800 1900

GA.S TEMPERATURE (K)

FIGURE 7 Aluminum droplet burning times versus ambient gas temperature for wet and dry
environments .

Water vapor effects In the present study, the differences between the theoretical
predictions for wet and dry environments arise principally because of different
convective velocities and minor property differences. In the experiments, however,
the oxide appeared to encapsulate the molten aluminum, apparently causing
relatively violent eruptions or partial fragmentation of the particles in wet
atmospheres, as discussed previously.
Figure 7 shows that the burning times were somewhat shorter in wet than in dry
environments, as was observed by Prentice (1970,1974). However, the shortening
observed in the present study is mainly a result of convective effects, as evidenced
by the shorter times predicted by the theoretical model. Intuitively, encapsulation
would increase the burning times while fragmentation would decrease them. This
suggests that perhaps these two phenomena taken together produce no significant
net effect.
Comparison with JF-IO burning times A comparison between the burning times of
coalesced aluminum agglomerates and those of JP-IO droplets (both with and
without the stabilizing additives) is shown in Figure 8. For corresponding test
conditions, the burning times of aluminum droplets are found to be about 4 times
those of JP-l 0 droplets at diameters of 600-700 microns. Also it can be seen that
the burning times of aluminum droplets are little influenced by ambient
temperatures, with oxygen effects dominating. The burning times of JP-10 droplets,
on the other hand, depend on ambient temperature as well as oxygen level. In our
 ALUMINUM-BASED SLURRY AGGLOMERATES 311

test range, the burning-time exponent in a "d" law" is estimated to be n= 1.8 for the
burning of aluminum particles, while n ranges between 1.8 and 2 for the burning of
JP-10 droplets, depending on test conditions. The experimental uncertainty
associated with the determination of these exponents is such that the confidence
intervals overlap, suggesting that no real difference exists. Since the estimates of
Gr/Re 2 are quite small (of the order of 10- 2 ), and the fact that the computed
burning times are significantly affected by the forced convection conditions of the
experiments, it is more likely that there is indeed no real difference between the
experimental values of burning-time exponents for the AI and JP-I0 droplets, with
both values lying between the forced convection (n= 1.5) and quiescent limits
(n= 2.0).
The effect of stabilizing additives on the burning times of JP-l 0 droplets is not
significant except for the high-oxygen test condition (T", = 1827 K, O 2 = 25.2
percent), where the additives caused increased burning times. This result is
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

consistent with the findings of Polishchuk et al. (1972) which showed that
evaporation rates of kerosene thickened with polyisobutylene depended upon both
the ambient conditions and the concentration of the additive. Moreover,
evaporation times were typically increased by the presence of the additive, but in
some cases, decreases were observed (Polishuk et al; 1972).

I
I
X02_
0
-• 10.4 %} PRESENT STUDY Ql 0 -
0

•
•
25 .2 %
20 .0
21.0 %
%} PRENTICE
000

CL0
'0
0
0
-~
---
co 180
o ~
'"
E
3
10 -
AI
o
)j:IJ
-/l- -
_~
l&J
~
c8j;o o JP-IO

WITHOUT-
--
i= ?; 0 8 ADDITIVES_
o 0• •
o f WITH

.•
t!)
Z
Z
It:
=>
l-
8 i: o
STABILIZING -
ADDITIVES
CD
o
f-
~o o
o
o

I I

do. d m (urn)
FIGURE 8 Comparison of burning times for aluminum droplets with lP-1 0 droplets.
 312 S. R. TURNS, s-c. WONG AND E. RYBA

AIIOJP-10Slurries
Because of the shattering which occurred at high O 2 levels and the physical inability
to measure burning times at low O 2 levels, no quantitative results for combustion of
the Aile slurry were obtained. However, residual combustion products were
analyzed and are discussed in the next section.

Combustion Product Analysis

Aluminum slurries Residual combustion products remammg on the particle
support probes were investigated with a scanning electron microscope (SEM) and
analyzed for composition by X-ray diffraction.
Figure 9 shows SEM micrographs of products obtained for a sequence of tests
with the stabilized AllJP-10 slurry in wet environments at essentially constant
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

oxygen levels. The highest temperature results show a compact mass of a-A120 3
crystals with the individual crystals having a characteristic size of about 10-15
microns. As the gas temperature was lowered, the individual crystal sizes increased ,
reaching about 50-60 microns for T"" = 1417 K. As shown in Table 1, the
composition remained a-Al 203• The last micrograph in the sequence shows an
agglomerate which failed to ignite.

STABILIZED AI/JP-IO SLURRY-WET FLAMES

Too= 1250 K O2=10 .9 %

FIGURE 9 SEM micrographs of residual combustion products for stabilized AI/JP-IO slurry in wet
flames.
 ALUMINUM-BASED SLURRY AGGLOMERATES 313

TABLE I
Residual combustion product analysis-CH/O/N 2 flame environments

Test condition Stabilized Stabilized

T(K) °2(%) HP(%) AI/JP-IO Al/C/JP-IO

1785 10.3 16.2 a-AllO J a-AllO J

1519 10.9 12.4 a-AllO J - 50% c-Al.O,
- 25% 6-AllO.
-35% Al,C J
1417 10.9 12.4 a-AllO J - 50% a-AllOJ
-50% AI.C.
Trace 6-AllO J
1381 10.9 12.4 a-A120 3
Trace 6-Al20 3
Al
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

1250 10.9 12.4

TABLE II
Residual combustion product analysis-CO/0/N 2 flame environments

Test conditions Stabilized Unstabilized Stabilized

T(K) 02(%) AI/JP-10 AI/JP-IO AI/C/JP-10

1827 25.2 a-Al l 0 3

Trace 6-A1 20 3
1809 lOA - 90% c-Al.O, - 50% a-A120 3
- 10% 6-All03 -15% 6-AIP3
Trace Al - 35% Al,C 3
1659 lOA a-AI20J
Trace Al
1548 10.3 - 75% coarse-grained a-
AlP3 a-AI 20 J
- 25% fine-grained Al Trace 6-All03
Trace Al
1510 10.0 Unignited sample- - 67% a-AllOJ
highly oriented Al - 33% o-Al20 J
small amount a-Al l0 3

Ignited sample-
not available

For dry environments, the individual crystals were much less apparent with a
generally more smooth appearance. The phase compositions, however, were
comparable in both wet and dry environments for high-temperature conditions
(T", ~ 1659 K), as can be seen in Tables I and n. For low-temperature conditions
(T", ~ 1548 K), however, some unburned aluminum was found in the residual
combustion products.
Figure 10 shows SEM micrographs of products obtained for the unstabilized All
JP-10 slurry in dry environments with essentially constant oxygen levels. Except for
the highest temperature result, the samples appear relatively smooth, without the
obvious apprearance of individual crystals. The unstabiJized Al/JP-10 slurry
products showed the presence of both o-Al20 3 and Al along with a-Al 20 3 (see
Table Il). Thus it appears that the influence of the stabilizing additives extends all
the way to the final combustion products for low-temperature conditions. For high-
temperature conditions, however, only ignition times appear to be affected.
 314 S. R. TURNS, S.-C. WONG AND E. RYBA

UNSTABILIZED AI/JP-IO SLURRY -DRY FLAMES

Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

FIGURE 10 SEM micrographs of residual combustion products for unstabilized AI(JP-IO slurry in
dry flames.

Aluminum/carbon 'slurries An interesting finding for the aluminum/carbon slurry

was the presence of aluminum carbide (AI4C3 ) in the product residue (Tables I and
Il), Apparently heterogeneous reactions occurred for the AI/C agglomerates,
producing both AI20 3 and A1 4C3 , as suggested by the brief period of incandescence
observed for the AI/C slurry agglomerates following the JP-lO burnout. Results for
tests where AI/C slurry droplets were burned at a low-temperature condition
(T", = 1519 K, O 2 = 10.9 percent) and subsequently exposed to a high-temperature
condition (T", = 1785 K, O 2 = 10.3 percent) showed slow conversion of the AI4C3 as
illustrated in Table m.
TABLE 1II
Conversion of AI4C) to Al20 ) at T., = 1785 K, O 2 = 10.3 percent

Time expo sed to high-temperature

(s)

o 50 50
15 50 50
30 75 25
60 80 20
 ALUMINUM-BASED SLURRY AGGLOMERATES 315

The equilibrium mole fraction of Al 4C 3 at both the high and low temperature
conditions was less than 5'10- 6 • Consequently, once AJ4C3 is formed it is difficult
to produce the final oxidation product, A1 2 0 3 , within a short time at O 2 levels of
about 10 percent. Diffusional resistance of the oxidant through the agglomerate
matrix probably contributes to the kinetic limitation.
At high O 2 levels, however, a-A12 0 3 was the predominant species present.
Moreover, it was only at this high O 2 test condition that coalescence was observed
for the A1/C agglomerates. As ambient temperatures were decreased, a greater
proportion of A1 4C3 was found. Vapor-phase combustion was not obvious at any of
the test conditions. SEM micrographs of the Al/C slurry products formed in wet
environments are shown in Figure 11. Two distinct phases are readily obvious with
small globules dispersed over a matrix. Higher magnification study of the globules
showed a crystal structure quite similar to the Al 2 0 3 crystals apparent in Figure 9.
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

STABILIZED AI/C/JP-10 SLURRY-WET FLAMES

Too=1677K 02=10.9%

Too = 15 19K 02 = 10.9% Too= 1417 K 02 = 10.9 %

FIGURE II SEM micrographs of residual combustion products for stabilized AJ/C/JP-I 0 slurry in
wet flames.

Dispersed products Smoke samples were collected on quartz slides for the
stabilized AllJP-l 0 slurry at the Too = 1809 K, O 2 = 10.4 percent test condition.
SEM analysis showed most particles were roughly spherical. A smoke-particle size
distribution determined for a 200 particle sample showed that the number of
particles decreased continuously above the 0.1 !Lm size. Because of the limited
 316 S. R. TURNS, S.-C. WONG AND E. RYBA

resolution of the SEM, particle sizes less than approximately 0.1 ,urn could not be
clearly resolved. Thus, it is possible that the modal diameter may be smaller than
0.1 ,um.

CONCLUSIONS

Upon particle coalescence and ignition, vapor-phase burning occurred for AJ/JP-
10 slurry droplets, similar to that observed by others for single aluminum particles.
As the droplet bums, oxide accumulates usually forming a massive product which
has a final diameter of 60 to 80 percent of the initial aluminum droplet diameter.
Burning times for the aluminum agglomerates were approximately four times those
for JP-10 at corresponding diameters and test conditions.
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

A relatively simple aluminum combustion model provided reasonable agreeemnt

with qualitative features of aluminum burning, as well as quantitative agreement
with observed burning times, as long as particle fragmentation did not occur.
Inclusion of radiation in the model suggests that dissociation of the oxide products
in the flame and subsequent product diffusion to the surface are not solely
responsible for surface oxide accumulation for particles in the size range studied,
thus alternative mechanisms such as convection, thermophoresis and surface
reactions may also contribute.
Complete oxidation of an aluminum/carbon slurry occurred only at the highest-
temperature (ca. 1800 K) flame conditions. For combustion at lower temperatures,
appreciable quantities of aluminum carbide were found along with alumina in the
condensed products. Subsequent oxidation of the aluminum carbide in a high-
temperature flame was quite slow. Since equilibrium calculations indicated that
AJ4C3 should be essentially non-existent at the experimental conditions, it is likely
that AJ4C3 conversion to AJ20 3 is limited by either chemical kinetics or diffusional
resistance within the agglomerate, or both.

ACKNOWLEDGEMENTS

Thi s work was performed under Prime Contract No. F3361 S-84-C-2410 with the U.S. Air Force Wright
Aeronautical Laboratories and administered under Subcontract No. C8197(8827}-394 by
Battelle-Columbus Laboratories. Technical monitors for this study were C. R. Martel (AFWAL) and A.
Levy (Battelle). Slurry samples were provided by Sun Refining and marketing Company. E. Ryba is a
member of the Department of Materials Science and Engineering .

NOMENCLATURE

d diameter
D effective mass diffusivity
Gr Grasshoff number
h oxide cap height
m mass flow, kg/s
rhO mass flux, kg/s-m?
M dimensionless mass flow, ri1I(41lpDrs )
n burning time exponent
Re Reynolds number
 ALUMINUM-BASED SLURRY AGGLOMERATES 317

r radial coordinate
f dimensionless radial coordinate, r/r,
r"',71 r"' ,m film radii for heat and mass transfer, respectively
Sh Sherwood number
t time
V volume

Greek Symbols
SI ,2 parameters defining aluminum combustion mode
fraction of ()diffusing inward
"o
() fraction of oxide vaporized at the flame
8(1 + v)
v stoichiometric mass ratio of oxidizer to fuel
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

p density

Subscripts
1,2 inner and outer zones, respectively
00 ambient
AI aluminum
m molten
o initial
ox oxide
s surface

REFERENCES

Bartlett, R. w., Ong , Jr., J. N., Fassell, Jr., W. M., and Papp, C. A. (1963). Estimating aluminum particle
combustion kinetics. Comb . and Flame 7, 227.
Brzustowski, T. A., and Glassman, I. ( I 964a). Vapor-phase diffusion flames in the combustion of
magnesium and aluminum. II: Experimental observations in oxygen atmospheres. Progress in
Astronautics and Aeronautics, 15, AIAA, Academic Pres s, NY, 117.
Brzusrowski , T. A, and Glassman, I. (1964b), Vapor-phase diffusion flames in the combustion of
magnesium and aluminum. 1: Analytical developments. Progress in Astronautics and Aeronautics,
IS, AIAA, Academic Press, NY, 75.
Friedman, R., and Macek, A (1963). Combustion studies of single aluminum particles. Ninth
Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, p. 703.
Gremyachkin, V. M., Istratov, A. G., and Leipunskii, O. I. (1975). Model for the combustion of metal
droplets. Comb . Expl. and Shockwaves 11, 313.
Klyachko, L. A. (1969). Combustion of a stationary particle of low-boiling metal. Comb. Expl: and
Shockwaves5,279.
Kuehl , D. K. (1965a). The ignition and combustion of small diameter aluminum wires. Pyrodynamics 3,
65.
Kuehl, D. K. (1965b). Ignition and combustion of aluminum and beryllium. AIAAi. 3,2239.
Kuehl, D. K., and ZwiUenberg, M. L. (1968). Predictions of burning times of metal particles. AIAA
paper No. 68-494.
Law, C. K. (1973). A simplified theoretical model for the vapor-phase combustion of metal particles.
Combust. Sci. and Tech. 7, 197.
Polishchuk, D. l., Yankevich, V. L., and Nechitailo, I. N. (1972). Evaporation and combustion kinetics of
thickened hydrocarbon fuels. Advances in Aerosol Physics 2, 119.
Prentice, J. L. (1970). Combustion of pulsed-heated single particles of aluminum and beryllium. Comb.
Sci and Tech. 1,385.
 318 S. R. TURNS, s-c. WONG AND E. RYBA

Prentice, L L., (1974). Aluminum droplet combustion: Rates and mechanisms in wet and dry oxidizers.
Naval Weapons Center, Rept, TP 5569.
Prentice, J. L., and Nelson , L. S. (1968). Differences between the combustion of aluminum droplets in air
and in an oxygen-argon mixture. J. Electrochem . Soc. 115,809.
Price, E. W. (1983). Combustion of metalized propellants. Progress in Astronautics and Aeronautics 90,
AIAA, NY, 497.
Svehla, R. .A. (1962). Estimated viscocities and thermal conductivities of gases at high temperatures.
NASA Technical Report R-I32.
Takahashi, E, Dryer, E L., and Williams , E A (1986). Combustion behavior of free boron slurry
. droplets. Twenty-FirstSymposium (International) on Combustion, Pittsburgh (in press) .
Wilson, Jr. R. P., and Williams, F. A. (1971). Experimental study of the combustion of single aluminum
particles in 0/Ar. Thirteenth Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, p. 833 .
Wong, S.•C; and Turns, S. R. (1986a). Ignition of aluminum slurry droplets. Comb . Sci. and Tech. 52,
22 I.
Wong, S. -C, and Turns, S. R. (I986b). Ignition and combustion of aluminum and aluminum/carbon
Downloaded by [Ohio State University Libraries] at 13:59 13 June 2012

slurry agglomerates. AFWAL-TR-86-2005.