Вы находитесь на странице: 1из 5

An Analytical Solution for the Quasi-Steady

Droplet Combustion
FERNANDO FACHINI FILHO

Instituto Nacional de Pesquisas Espaciais (INPE), 12630-000 Cachoeira Paulista, Sao Paulo, Brazil

INTRODUCTION
The relative simplicity in describing the spherical droplet combustion problem is based on the
fact that this process is mainly controlled by the
gas phase in the neighborhood of the droplet,
which has a quasi-steady behavior compared to
the liquid-phase behavior [1, 2]. The quasisteady character of the gas phase processes
explains the success of the steady-state model in
describing the droplet vaporization dynamics
(d 2 law) [3, 4].
Goldsmith and Penner [5] improved the quasi-steady-state model by including the linear
dependence of the transport coefficients and
the specific heats on the temperature, resulting
in a system of nonlinear ordinary differential
equations to describe mathematically the processes in the gas phase. A convenient combination of these equations allowed the nonlinearity
to be removed so that their final form consisted
of a system of linear ordinary differential equations. Kassoy and Williams [6, 7] generalized
that model by including the chemical reaction
and the dependence of the transport properties
on a power of the temperature. They assumed,
however, constant specific heats and Lewis
numbers equal to one. Law [8] extended and
formulated the quasi-steady-state model for the
droplet vaporization adding the dependence of
the gas composition and temperature to the
transport coefficients and specific heat. The
effect of variable transport coefficients on the
combustion problem was studied by Raghunandan and Mukunda [9]. They eliminated such
nonlinearity through a change of variables and
avoided that caused by the chemical reaction by
imposing the knowledge of the flame temperature. Williams [2] solved the same problem
including the chemical reaction. He employed a
function defined by a combination of dependent
variables to remove the nonlinear chemical re0010-2180/99/$19.00
PII S0010-2180(97)00174-0

action term (the SchvabZeldovich formulation).


Puri and Libby [10] treated the problem using
a detailed expression for the transport coefficients, for the reaction CO2 1 H2 3 CO 1 H2O
and the heat transfer by the diffusion of the
species. Due to the complexity of the model the
solution is obtained numerically. Predictions for
vaporization rate constant and flame location
are in good accordance with the experimental
results.
In this paper, an alternative mean to solve
analytically the combustion of the droplet is
presented. The model is not as sophisticated as
that formulated by Puri and Libby [10]. Nevertheless, the dependence of the molecular transport coefficients on the temperature is included.
Moreover, a one-step reaction mechanism is
employed along with constant nonunity Lewis
numbers. The general SchvabZeldovich formulation is used to eliminate the chemical
reaction term. Furthermore, that nonlinearity
from the transport coefficients is removed
through the Goldsmith and Penner procedure.

MODEL
The droplet is assumed to be spherical with
radius a(t) at the instant t (a0 5 a(0) being its
initial value), so that spherical symmetry holds
for the liquid-phase processes. It is assumed
that the density, r l , the specific heat, c l , and the
thermal conductivity, k l , of the liquid phase are
constants. Far from the droplet, the gas has
constant properties, which are: density, r`, thermal conductivity k ` , constant pressure specific
heat, c p . The gas temperature and oxidant mass
fraction are held constants at T ` and Y O` ,
respectively.
The dimensionless variables used in this analysis are defined as
COMBUSTION AND FLAME 116:302306 (1999)
1998 by The Combustion Institute
Published by Elsevier Science Inc.

QUASI-STEADY DROPLET COMBUSTION

u 5 T/T `,

| 5 r / r `,

y O 5 Y O`/Y O`,

y F 5 Y F,

v 5 V a 0/ a `,

x 5 r/a 0, t 5 t/t c.
where t c is the order of magnitude of the
droplet life time, t c 5 a20/a`e, and a ` 5 k ` /
c p r ` , e 5 r ` / r l ,, 1.
For constant liquid density, r l , the dimensionless mass conservation equation for inside
the droplet is given by
3

da /d t 5 23 l ~ t !

(1)

where l 5 m
c p /4 p a0k ` , a 5 a/a0.
The thermal problem inside the droplet may
be described easily assuming either the conduction process is the unique process in transfer of
the heat [11] or this process is so fast that the
temperature is considered uniform [12].
The droplet is thought to be in a quiescent
atmosphere, so that all processes in the gas
phase will have spherical symmetry. Furthermore, there is no fuel in the ambient. Gas
velocities induced by the expansion due to the
phase change, at the liquid gas interface, and
by the thermal expansion across the flame are
much smaller than the sound velocity. Therefore, it is justified to take the pressure as a
constant.
The dependence of the fuel and oxidant
diffusion coefficients, D F and D O , and of the
thermal conductivity, k g, on the temperature is
considered a power equal to 0.5 [13]. The
relations between the transport coefficients of
the gases at the ambient condition (k g` and D i`
[i 5 F, O]), i.e., the oxidant and the fuel Lewis
numbers, L O 5 a ` /D O` and L F 5 a ` /D F`
respectively, are constant.
At the liquid/gas interface, the vapor and
liquid are assumed equilibrium. It is also considered that the fuel molecular weight is equal
to that of the ambient gases mixture, so the
ClausiusClapeyron expression,
y Fs 5 exp@ g ~1 2 u B/ u s!#,

(2)

can be used to relate the temperature, u s , to the


fuel mass fraction, y Fs , at the droplet surface;
where g 5 L/R g T B , T B , R g , and L are the
boiling temperature, the gas constant, and the
latent heat of vaporization, respectively. For g

303
.. 1, the fuel concentraction at the droplet
surface becomes important if u s ; u B 2 O(1/ g )
[14].
It is considered a one-step, irreversible chemical reaction of the BurkeSchumann type. For
each unit of fuel mass consumed, v masses of
oxidant are depleted in the stoichiometric reaction.
The dimensionless conservation equations for
the gas phase [ x . a( t )] can be written as:
x 2|v 5 l ~ t !,

(3)

l y i 1 x 2u n y i
2
5 2s iw i, i 5 O, F,
x2 x x2 x Li x
(4)
1
l u
2 2
2
x x x x

x 2u n

u
5 Qw F,
x

(5)

where s O 5 v/YO` and s F 5 1. The fuel


consumption rate w F is equal to w O /v and Q 5
q/c p T ` .
The closure for the system above is provided
by the dimensionless equation of state of the
gas, written as
1 5 |u,

x . a.

(6)

To integrate the system of Eqs. 4 and 5, the


following boundary condition are used at x 5 a,
2

S
S

D
D

x 2u n y F
5 l ~1 2 y Fs!,
LF x
x 2u n

u
5 l L 1 Q 2 5 l L9,
x

at x 5 x f, u 5 u f,

and

y F 5 y O 5 0,

(7)
(8)
(9)

and
at x 3 `,

u ` 2 1 5 y O` 2 1 5 0,

(10)

where Q 2 is the heat to inside the droplet, L9 5


L 1 Q 2 / l and L9 is obtained by solving the
thermal problem for the liquid phase.

SOLUTION
It is assumed that the combustion process occurs in the BurkeSchumann limit (i.e., the
Damko
hler number, Da, is taken to be infinitely
large). In other words, the reaction characteris-

304

F. F. FACHINI

tic time is infinitely smaller than any mechanical


characteristic time. Therefore, the residence
time of the reactants is infinitely small in the
reaction zone, collapsing the flame into an
infinitely thin zone.
The droplet combustion problem is described
using the general SchvabZeldovich formulation [15, 16], obtained by combining Eqs. 4 and
5. This yields
1
l H
2 2
2
x x
x x

H
l Z
x 2u n
1 N~Z! 2
50
x
x x
(11)

1
l Z
2 2
Le~Z! 2
x x x x

Z
x 2u n
5 0,
x

(12)

where

H
H

Le~Z! 5
N~Z! 5

H~a! 5

x 2u n

and
2l Le~Z!~Z 1 2 Z! 2 x 2u n

Z
5 0,
x

(20)

L F,

Z,1

L O,

Z,1

and Z 1 is

~L F 2 1!/S,

Z,1

~1 2 L O!,

Z,1

Z
x

x5a

x5a

l ~S 1 1! L FL9
2 l L F~1 2 y Fs!
Q
(14)
5 2l L FS~1 2 y Fs!

1 1 S,

Z.1

2 q zj

Z,1

and where

~S 1 1! L Fu s
1 y Fs, Z~a! 5 Sy Fs 1 1
Q
(13)

H
x

Z.1

qH` 2 @~S 1 1! LF/Q 1 1# 2 ~1 2 LO!qZ`, Z , 1

where s 5 s O .
Notice that u n is not written as a function of
H and Z because this term will be further
eliminated through an additional expression
relating H and Z.
The boundary conditions for Eqs. 11 and 12
are

H
2 l N~Z!~Z 1 2 Z! 5 0
x
(19)

~S 1 1! L F~L9 2 u s!/Q 2 1,

Z 5 Sy F 2 y O 1 1, S 5 sL O/L F,

x 2u n

l ~H 1 H 1! 2 x 2u n

where H 1 is

H 5 ~S 1 1! L Fu /Q 1 y O 1 y F,

where the subscript f indicates the flame condition. The functions H and Z and their first
derivative are continuous everywhere [15].
As the second derivatives of the functions H
and Z are discontinuous through the flame, Z 5
1, according to Eqs. 11 and 12, the integration is
performed separately in the regions a # x # x f
and x f # x # `, respectively. This yields

(15)

H~ x f! 5 ~S 1 1! L Fu f /Q, Z~ x f! 5 1

(16)

H~ x 3 `! 5 ~S 1 1! L F /Q 1 1,

(17)

Z~ x 3 `! 5 0

(18)

q H` 5

x 2u n H
l x

x3`

and q Z` 5

x 2u n Z
lLO x

.
x3`

The flux q H` is determined by applying conditions (Eq. 16) and the continuity of the functions H and Z and their first derivative to Eq.
19, thus eliminating the conditions at the flame.
This yields
q H` 5

~S 1 1! L F
~1 1 L9 2 u s! 2 L F 1 L O
Q
(21)

Using the same procedure with Eq. 20 it is


possible to find the flux q Z` ,
q Z` 5 2~s 1 1!.

(22)

Hence, Eqs. 19 and 20 can be written as [5]

l
1
H
5
x 2u n ~H 1 H 1! 2 N~Z!~Z 1 2 Z! x
52

1
Z
.
Le~Z!~Z 1 2 Z! x

(23)

QUASI-STEADY DROPLET COMBUSTION


Integrating the last two terms of Eq. 23 one
finds
H 1 H1 5 2

(24)

where the constant C 1 is an integration constant


given by:
~S 1 1! LFL9/$Q@S~1 2 YFs!#%1/LF,

Z,1
1/LO

~S 1 1! LF~1 1 L9 2 us 2 Q!/ @Q~1 1 s!#

, Z,1

The flame temperature can be found from


Eq. 24 by applying conditions (Eq. 16) to it.
Choosing the solution corresponding to x f # x,
one may write

u f 5 1 1 ~ u s 1 Q 2 1 2 L9!

F S D G
12

s
s11

b na/x f 5

@ x n/~ x 1 q ` 2 1!# dx

(28)

uf

N~Z!
~Z 2 Z!
Le~Z! 2 1 1

1 C 1~Z 1 2 Z! 1/Le~Z!,

305

1/LO

(25)

Notice that if s .. 1, then u f 5 1, which is the


case for the flame located far from the droplet.
If this condition is met, the quasi-steady model
will not describe correctly the flame behavior
and the gas-phase temperature [1719].
The droplet surface temperature is found by
imposing conditions (Eq. 13) on Eq. 24 and
eliminating the flame temperature, u f . The resulting expression gives u s as a function of b n ,
through the effective latent heat L9,

where q ` 5 x 2 u n ( u / x)/ l u x3` is the heat


conducted to the ambient, which is determined
by adding q H` to q Z` . This yields
q ` 2 1 5 L9 2 u s 2 Q

(29)

It is seen from Eqs. 25 and 26 that the


influence of the fuel Lewis number on the
droplet combustion takes place during the droplet heating period and the maximum temperature reached at the surface. After the droplet
reaches its final temperature, u s ; u B , the value
of L9 approaches L, and the flame temperature
and its location and the vaporization constant,
b n , will depend on the oxidant Lewis number
strongly.
The flame temperature increases when the
oxidant Lewis number decreases. The increase
in the flame temperature is due to the fact that
the oxidant diffusion rate to the flame is faster
than the heat loss rate to the ambient. The
decrease in the oxidant Lewis number leads to
an increase in the oxidant mass flux to the flame,
thus the flame approaches the droplet surface,
for the chemical reaction to occur stoichoimetrically. The presence of the flame near the
droplet increases the heat flux to the droplet
which in turn increases the vaporization rate.
Next we present the results for u s 5 u B of the
droplet combustion of methyl and ethyl alcohol,
benzene, and heptane.

L9/$1 2 exp@ g ~1 2 u B/ u s!#% 1/LF}


5 Q 1 ~1 1 L9 2 us 2 Q!@s/~s 1 1!#1/LO

(26)

If s .. 1, the equation for the vaporization


process without combustion is recovered, as
expected.
Integration of Eq. 5 in the interval a $ x $ x f
yields a relation between b n and a/x f :

b n~1 2 a/x f! 5

uf

us

@ x n/~ x 2 u s 1 L9! dx,


(27)

recalling that l 5 b n a. Repeating that integration in the interval x $ x f , another relation


between b n and x f /a is found:

RESULTS
In order to compare the results from this model
to that presented by Puri and Libby [10], they
are written (Eq. 1) in terms of the droplet
diameter instead of the droplet radius. With this
modification, the dimensional droplet vaporization rate is R 5 8 a ` r ` b n / r l . The results of this
model along with those from Puri and Libby
[10] are shown in Table 1.
The droplet vaporization rate determined in
this analysis is about 50% higher than that
found by Puri and Libby [10] and that by
experimental means. Agreement is found only
for the benzene. This difference in the results

306

F. F. FACHINI
2.

TABLE 1
Comparisons of results

T f (K) model

Fuel

3.
R (mm2/s)

x f /a

model
model
[10]
model
[10]

Methanol 2363.1
Ethanol
2375.0

9.27
15

4.9
4.3

0.83
0.84

0.56
0.57

Benzene
Heptane

30.6
33.2

17.9
9.8

0.85
1.44

0.65
0.79

2163.5
2631.3

expt
[10]
0.54
0.65
0.80
0.78

T ` 5 298 K, Y O` 5 0.23.

can be explained by the assumption of one-step


chemical reaction that overestimated the flame
temperature.

4.

5.
6.
7.
8.
9.
10.
11.

CONCLUSION
Although the model admits the transport coefficients dependence on the temperature and
the fuel and oxidant Lewis numbers different
from unity, the results do not have good concordance with the experimental results. The
principal point is that the chemical reaction is to
be considered one-step, which leads to an overestimation of the flame temperature, and, consequently, the vaporization rate and the flame
position. To correct this overestimation an effective heat of combustion could be applied in
the form Q e 5 k f Q [18 20].
This work was supported in part by Fundaca
o
ao Amparo a
` Pesquisa do Estado de Sa
o PauloFAPESP under Grant 93/2597-9.

12.
13.

14.
15.
16.

17.
18.
19.
20.

Williams, F. A., Combustion Theory, 2nd ed., AddisonWesley, Menlo Park, CA, 1985, pp. 52 69.
Godsave, G. A. E., Fourth Symposium (International)
on Combustion, The Combustion Institute, Pittsburgh,
1953, pp. 818 830.
Spalding, D. B., Fourth Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh,
1953, pp. 847 864.
Goldsmith M., and Penner S. S., Jet Propulsion 24:245
251 (1954).
Kassoy, D. R., and Williams, F. A., AIAA J. 6:1961
1965 (1968).
Kassoy, D. R., and Williams F. A., Phys. Fluids 11:
13431351 (1968).
Law, C. K., Phys. Fluids 18:1426 1432 (1975).
Raghunandan, B. N., and Mukunda, H. S., Combust.
Flame 30:71 (1977).
Puri, I. K., and Libby, P. A., Combust. Flame 76:67 80
(1991).
Park, J. M., Ablow, C. M., and Wise, H., AIAA J.
4:10321035 (1966).
Williams, F. A., J. Chem. Phys. 33:133144 (1960).
Reid, R. C., Prausnitz, J. M., and Poling, B. E., The
Properties of Gases and Liquids, 4th ed., McGraw-Hill,
New York, 1986.
Lin
an, A. (1990). Vaporizacio
n de y Condensacio
n en
Gotas, private notes.
Lin
an, A., Discurso en la Real Academia de Ciencias
Exactas, Fsicas y Naturales, Madrid, 1991.
Lin
an, A., and Williams, F. A., Fundamental Aspects of
Combustion, Oxford University Press, UK, 1993, pp.
111151.
Crespo, A., and Lin
an, A., Combust. Sci. Technol.
11:9 18 (1975).
Waldman, C. H., Fifteenth Symposium (International)
on Combustion, 1975, pp. 429 442.
Fachini, F. F. (1992) Ph.D. thesis, Universidade Politecnica de Madrid, Spain.
Saastamoine, J. J., Aho, M. J., and Linna, V. L., Fuel
72:599 609 (1993).

REFERENCES
1.

Boghi, R., Clavin, P., Lin


an, A., Pelce
, P., and Sivashinsky, G. I., Modelisation des Phenome`nes de Combustion, Editions Eyrolles, Paris, 1985, pp. 76 102.

Received 22 March 1996; accepted 24 April 1997.

Вам также может понравиться