COMBUSTION AND FLAME 56: 337-342 (1984)

337

BRIEF COMMUNICATIONS
The Thickness of Laminar Flames
JOZEF JAROSINSKI

Instytut Lomictwa, 02-256 Warszawa,AI. Krakowska110/114.Poland.

T~,~

INTRODUCTION
In addition to the burning velocity, the flame
thickness is one of the most important parameters
characterizing in a synthetic manner the complex
processes in a combustion wave. Unfortunately
there is a small amount of experimental data concerning flame thickness and the rare numerical
values which we can fred in several papers are
frequently based on different and incompatible
definitions and usually are called in question. The
opinion that the flame thickness as such is not defined consolidates this state of affairs.
There are three popular arbitrary definitions of
flame thickness. Zeldovich [1] was the first to
introduce the well known "characteristic thickness" A = ~(cppuuL), where X is the thermal
conductivity, cp the specific heat with constant
pressure, Pu the density of unburned mixture, and
ur. the burning velocity. The dimensional quantity
A is a measure of the thickness within which the
unburned mixture is heated to the ignition temperature by the reaction zone (Fig. 1). Gaydon
and Wolfhard [2] proposed restricting the preheat
zone to the region between the region of ignition
temperature and the point where the temperature
has risen by just 1% with reference to the temperature rise in the zone, from which it follows that
the thickness of the preheat zone is 4.6A. Spalding
[3] defined the flame thickness as a ratio between
the maximum temperature difference and the
maximum temperature gradient which occurs at

t,

/
r,,_

~b

r
4.6~,
~
REACTION ZONE
*Z
-z PREHi:AT ][ONE
Fig. 1. Temperature profile across the flame front.
the point of inflection of the prof'de:
T b - Tu
b=
(dT/dz)max

where T is absolute temperature, z a coordinate

normal to the flame front, and the subscripts b and
u mean burned and unburned, respectively. Simplicity is an advantage of all those definitions, and
the undefined ignition temperature is their weak
point, which leads to substantial scattering of data.

Copyright 1984by The CombustionInstitute

Published by ElsevierSciencePublishingCo., Inc.

52 VanderbiltAvenue, New York, NY 10017

0010-2180/841503.00

338

JOZEF JAROSINSKI

In the present note flame thickness will be

defined on the basis of the energy equation.
THE

METHOD

USED

Following the mathematical analysis of Spalding

[4], let us confider a steady state one-dimensional
laminar flame with the temperature governed by
the energy equation

rig--

-- CpPuUL dz + Q.o

(1)

O.

The new symbols are Q heat of reaction, and

o the rate of chemical reaction (mass of fuel consumed by chemical reaction per unit volume and
time); cp in Eq. (1) wi//be assumed constant.
On introducing
CppuUT.
dx = ~
dz,
X

Tr

Tu
'

Ta--Tu

dr
p =--,
dx

Subscript = meaning adiabatic, we obtain, from

(1),

dp
P d r - p = - ~oF(r)

(2)

under the conditions

r = 01

p =0

(3)

r = rmI

P = Pln

(4)

Equations (3) and (5) imply that x -~ --oo as

r -, 0 (cold boundary problem) and x ~ +oo as
r -* 1 (hot boundary problem)
By integrating Eq. (2) under the boundary
conditions (3) and (5) it can be shown that =
f~ p dT [4], therefore ~ is equal, for AT = 1, to
the area under the p curve on the graph of a function p = f(r). In other words the eigenvalue ~0 is
equivalent to the mean dimensionless temperature
gradient p ~ (where the subscript M is referred to
the mean value of the temperature gradient):
~0 = PM"

(6)

The mean temperature gradient can be expressed by the easily-measured temperature gradient (Fig. 2). In the preheat zone, p = r. In the
reaction zone, the value of the mean temperature
gradient must lie below PM ~ Pm (1-0.5pmX line a
in Fig. 2). If we have the temperature profile from
experiments, we can determine the mean temperature gradient PM very precisely.
Making use of the temperature gradient PM and
the known temperature increases AT = l, it is easy
to determine the corresponding dimensionless
flame thickness Ax = 1/19u = -'- (cp/)t)~ouuL~ M =
S where cp/X i s the mean value over the temperature range of the flame and ~M is the flame thickness resulting from the mean temperature gradient.

for the preheat zone and

to

r = "rm3. P = Pm 3.

(4)

T= 1,

(5)

.8"

p=0

for the reaction zone

where the subscript m is referred to the maximum
value of the temperature gradient,
~=

MQ~

(T a

.6.
L

.4.

v?

Xco

Tu)Cp2pu2UL 2 '

\\N\

F(T) = ~

.i

" .i

" .i

" i

"

and
coc = 7--

^u

dT.

Fig. 2. Temperature gradient as a function of the temperature in a dimensionless form: a and b-pouible variations of the temperature gradients.

THICKNESS OF LAMINAR FLAMES

339
ltC

The dimensionless number S expressed by the

flame front parameters is the Peclet number based
on ~M- The relation Pr~ = 1IS is equivalent to

~= 1/S.

(7)

THE THICKNESS OF THE PURE

THERMAL FLAME
Pure thermal theory of flame propagation assumes
that the entire heat of reaction is released very near
the Final temperature [1]. If we consider the preheat zone to extend right up to the Final temperature, without any reaction taking place, then we
should modify the boundary condition (4) to
rm = 1 --e,

(8)

where e "~ 1.
In the preheat zone we have F(r) = 0. By making use of (2)-(4) and (8), we have dp= dr and
prowl.
On the other hand it is assumed that, in the
reaction zone bounded by the conditions (4), (5),
and (8), the entire amount of the heat o f reaction
is transferred to the preheat zone; By making use
of (2), (4), (5), and (8) and; on integrating, we
find
-----1/2,

(9)

since it is assumed flrra p dr ~ Pro(1 -- rm)

e ~ ~ and F(r) is defined in such a manner that
fl m F(7") d~" -- I.
By comparing (6), (7), and (9)we have PM ~1/2 and S ~ 2. The relations between the two temperature gradients are shown in Fig. 3.
The maximum temperature gradient Pm corresponds to the line a in Fig. 4. This line cuts off the
"characteristic thickness" A = X/(CppuUi,) [1] on
the abscissa axis.
The mean temperature gradient PM corresponds
to the line b in Fig. 4, which replaces the temperature curve T The area under the line b is equal to
the area under the temperature curve 7. The flame
thickness ~M iS for pure thermal flame twice as
wide as the "characteristic thickness" A.

.2

.4

.6

.8

tO

'1[

Fig. 3. Relation between the two temperature gradients

Pm and PM in p-r coordinates for pure thermal flame.

THE THICKNESS OF A REAL FLAME

It has been known for a very long time that the
eigenvalue ~0 is always less in real flames than 1/2
[4], which is equivalent, according to (7), to S >
2. The estimation of S in real flames (which are,
as regards the flame structure, much different
from flames with a pure thermal mechanism of
propagation) should be based on the analysis of
the temperature curves determined experimentally.
The temperature profiles of lean limit methaneair flames were measured in the normalized flammability tube with a thin wire resistance probe,
similarly to Ref. [5]. The sensor of that probe

11

-Z

.Z

Fig. 4. Relation between the two temperature gradients

Pm and PM in T-z coordinates for pure thermal flame.

340
(DISA 55A75) is a 2.2 m m long, 10/am diameter,
Pt-iridium wire. Since the time constant of the
wire is about 5 ms, the temperature of the laminar
flame front could be recorded with no substantial
error. The temperature records transferred into dimensionless coordinates have been compared with
the temperature curve for pure thermal conductive
heating (Fig. 5). The temperature curve as determined by Dixon-Lewis and Williams [6] is also
shown for comparison. The entire flame thickness
determined (in accordance with the beginning and
the end of observable temperature variation) in a
standard flammability tube is approximately twice
the thickness determined for appropriate maximum temperature gradients. The analogous ratio
of thicknesses for the Dixon-Lewis and Williams
curve is approximately 1.5. However, the maximum temperature gradients determined in dimensionless parameters have nearly the same values
for every curve. The curves in Fig. 5 contain all
the information necessary to plot a graph of p =
f(r). The most important fact is the condition
p - r for the preheat zone and the limitation Pra =
const for a substantial part of the reaction zone.
Variation of the temperature gradient o f the curve
1 from Fig. 5 is shown, as an example, in Fig. 6.
By graphical integration of this function we can
Fred, in agreement with (6) and (7), that ~ = PM =
0.197 and S = 5.1 At the same time we can determine the interdependence of the maximum and
mean temperature gradient as Pm/PM = 1.37 (for
curve 2 in Fig. 5, Pm/PM = 1.32).
To find the nondimensional number S =
CppuUSaM/;~ over the entire range of mixture
composition we can use experimental data for
flames of homogeneous methane-air mixtures carried out under conditions approaching those of
adiabatic combustion [7]. The flame thickness
which was used as a basis for comparison was
determined by means of an optical method (the
Gayhart-Prescott schlieren interferometer). In
comparison with probe methods optical methods
have the advantage of not influencing the flame. It
has been proved by temperature profile measurement [5] that the flame thickness as determined in
[7] is, approximately, ~ -- 2(Tb - Tu)/(dT/dz)max,
w h e r e ' ( T b - - Tu) and (dT/dz)raax are the measured values of temperature difference and the
maximum temperature gradient, respectively.

JOZEF JAROSINSKI

to,

/ y/

.6-~
4,

.2'

x
Fig. 5. Experimental temperature profiles of near limit
methane-air flames against the temperature curve for
pure thermal conductive heating in z-x coordinates: 1the curve measured near the limit mixture for upward
propagating flame (approximately 5.3% CH4), h/(cl~ u) =
0.41 cm2/s, u L = 5 cm/s; 2-the curve measured near the
limit mixture for downward propagating flame (approximately 6.1% CH4), k/(cpgu)= 045 cmg/s, uL = 7.5 cm/s;
3-the Dixon-Lewis and Williams curve [6], h/(Cppu) =
041 cm2/s, UL = 5 cm/s; 4-the pure thermal conductive
heating curve.
Thus, the values of the flame thickness found by
Andrews and Bradley were higher than those
corresponding to the mean temperature gradient
Before using them to calculate the number S they
should be corrected by a factor of order 0.66
[ara ~ (1.32/2.00~5]. The results of calculation
o f the nondimensional number S are shown as a
function of the equivalence ratio a in Table 1 and
Fig. 7. These data have been compared with the
1.0'

.a.

.6.
tb
.4.

.2,

.2

,4

.8

1.0

Fig. 6. Relation between the two temperature gradients

Pra and PM in p--r coordinates for the flame corresponding to curve 1 in Fig. 5.

THICKNESS OF LAMINAR FLAMES

341
TABLE 1
Flame Parameters

Equivalence
ratio

uL

(cm/s)

(cm)

Ta or Tllra a

(*C)

h/(c

(cmp,o~/u)
Is)

ui,6CpPu/X

8.7
7.9
7.8
8.0
7.8
8.2
8.8
9.3
10.0
10.0

5.7
5.2
5.1
5.3
5.1
5.4
5.8
6.1
6.6
6.6

0.440
0.455
0.470
0.480
0.492
0.487
0.483
0.473
0.467
0.460

8.9
8.4
8.1
7.8
8.0
8.1
8.5
9.6
10.2
11.4

5.9
5.6
5.3
5.1
5.3
5.3
5.6
6.3
6.7
7.5

0.662

7.4

4.9

0.485

7.6

5.0

CH4-Air Near Adiabatic Co nditions [ 7 ]

0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5

16
23
30
37
45
45
41
34
26
19

0.260
0.170
0.135
0.115
0.100
0.100
0.115
0.145
0.200
0.270

1390
1550
1710
1830
1940
1920
1850
1770
1690
1610

0.478
0.496
0.520
0.535
0.550
0.548
0.535
0.525
0.515
0.505

CH4-Air Limit Conditions [8 ]

0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5

8.5
18.3
26.2
34.0
37.5
36.0
30.5
24.0
16.5
9.5

0.460
0.210
0.145
0.110
0.105
0.110
0.135
0.190
0.290
0.550

1060
1180
1310
1410
1490
1480
1420
1350
1290
1230
Hz-Air [9, 10]

1.0

195

0.025

1550
C3Hs-Air [9, 10]

1.0

40

0.092

1500

a It is assumed that TXim = 0.8Ta(K).

results o f measurements carried out in narrow
channels near the quenching conditions [8]. The
quenching distance was used as a measure o f flame
thickness. It was found [8] that the quenching
distance was twice as wide as the flame thickness
determined in [7]. Two points corresponding to
the quenching conditions o f hydrogen-air and
propane-air flames were considered in addition.
The analysis o f data for fuels other than methane
was bounded by those two points, because o f the
lack o f reliable data for flame p a r a m e t e r s - f l a m e
thickness, in particular. F r o m [8] it follows that

the difference between the thickness of the

adiabatic flame and the limit flame is minimum for
the stoichiometric mixture. The two points for
the stoichiometric mixture have been calculated
assuming that the flame thickness is half the
quenching distance [8] and using the data o f
Potter and Berlad [9] and Lewis and Elbe [10].
The constancy o f the number S in Fig. 7 for any
mixture composition and the amount o f heat
losses mean, from the physical point o f view, that
the fundamental geometrical proportions o f flame
in these different conditions are constant.

342

JOZEF JAROSINSKI

6,
o~
@
4
CH 4- a i r , odiabotic

|lame [ 7 ]

x C H 4 - o l r , limit florae [ 8 ]
H 2. air [9,10]
*

.,

'

CsHe-air [9,10]

'

1'.o

1:2

EQUIVALENCE RATIO

O(

Fig. 7. The nondimensional number S ---cppuULGM/~ for different equivalence ratios ~.

CONCLUSIONS
A simple method for determining the thickness of
a laminar flame on the basis of the energy equation has been developed. The physical significance
of the eigenvalue in the energy equation is highlighted. The method was used to determine the
thickness of the pure thermal flame and that of a
real flame, characterized by the measured temperature profdes.
The results presented in Fig. 7 should be
treated as a first approximation. They need further,
more accurate investigation. The knowledge of the
temperature profile is a key to the determination
of the flame thickness. If the temperature profde
is measured, Pro, PM, S, and GM and the like can
be easily determined.

2. Gaydon, A. G., and Wolfhard, H. G., Flames-Their

Structure, Radiation and Temperature, Chapman
and Hall, London, 1953.
3. Spalding, D. B., Some Fundamentals of Combustion, London, 1955.
4. Spalding, D. B., Combust. Flame 1:296 (1957).
5. Jarosifiski, J., Strehlow, R. A., and Azarbarzin, A.,
Nineteenth Symposium (lnternationa 0 on Combustion, 1982 (in press).
6. Dixon-Lewis, G., and Williams, A., Eleventh Symposium (International) on Combustion, Pittsburgh,

1967, p. 951.
7. Andrews, G. E., and Bradley, D., Combust. Flame
19:275-288 (1972) Figs. 4, 7.
8. Jaxosinski, J., Combust. Flame 50 (1983).
9. Potter, A. E., and Berlad, A. D., Sixth Symposium
(International) on Combustion, Reinhold, New York,
1957, p. 26.
10. Lewis, B., and yon Elbe, G.,Combustion, Flamesand
Explosions o f Gases, Academic Press, New York,
1961, Tables 5, 6.

REFERENCES
1. Zeldovich, Ya.B., The Theory of Combustion and
Detonation, Publ. Academy of Sciences, USSR,
1944 (in Russian).

Received 24 January 1983; revised July 1983