Twentieth Symposiuln (International) on Combustion/The Combustion Institute, 1984/pp.

505-512

LEWIS N U M B E R E F F E C T S ON T U R B U L E N T B U R N I N G VELOCITY

R. G. ABDEL-GAYED
D. BRADLEY
M. N. HAMID
M. LAWES
Mechanical Engineering Department
University of Leeds
Leeds LS2 9JT

Experimental values of turbulent burning velocities for propane, hydrogen and iso-octane
mixtures with air are reported under conditions of high turbulence and high turbulent Reyn-
olds number. The measurements were made by the double kernel method during explosions
in a fan-stirred bomb, with four fans, capable of speeds of up to 10,000 rpm. The ratio of
turbulent to laminar burning velocity, ut/ue, is correlated primarily with the ratio of r.m.s.
turbulent velocity to laminar burning velocity, u'/ue and a Karlovitz stretch factor given by
the ratio of a strain rate u'/h to a flame gradient given by ud~t, where h is the Taylor
microscale and ~t the laminar flame thickness.
Asymptotic analyses of strained laminar flames, together with the two-eddy theory of tur-
bulent burning, show the additional importance of Lewis number effects. These result in
lean hydrocarbon mixtures being quenched more readily than rich ones, with an opposite
effect for H2 mixtures. This was observed in the experiments. However, full quantitative
agreement between theory and experiment was not achieved, due to the inherent limitations
of the two theories, which are discussed.

1. Introduction tially increases with u'/ut, levels off then, in some

cases, decreases until the flame is quenched.
In both engine performance I and explosion This relationship is examined in the light of re-
hazard z contexts a clear need has emerged for tur- cent theoretical asymptotic analyses of strained lam-
bulent burning data for premixed gases. Two of the inar flames, 5-9 which show a change in burning ve-
present authors have correlateda the ratio of de- locity (most commonly a reduction) with increasing
veloped turbulent to laminar burning velocity mea- strain. In addition to the strain rate, the Lewis
sured on burners, ut/ue, with the ratio of r.m.s. number for the deficient reactant and the activation
turbulent velocity of the cold mixture to its laminar energy for chemical reaction are determining pa-
burning velocity, u'/ue, and the cold turbulent rameters. Such analyses are shown qualitatively to
Reynolds number, RL, (=u'L/v), where L is the contribute towards an explanation of the measured
integral length scale and v the kinematic viscosity. turbulent burning velocity relationships. The mix-
Highly turbulent flames present problems of flame tures investigated were chosen to cover a wide range
stability on burners, which are avoided in explosive of Lewis numbers in order to elucidate the effect
turbulent combustion. Earlier explosion measure- of this parameter on flame straining in turbulent
ments of ut4 were made in a bomb equipped with combustion.
four fans, running at speeds of up to 3,500 rpm.
The fans have been coupled to new high speed
electric motor drives capable of speeds of 10,000 2. Apparatus
rpm. Gaseous mixtures of hydrogen, propane and
iso-octane with air have been exploded in hitherto The explosion bomb comprised a cast steel cyl-
unexplored r6gimes, with values of u' up to 17.3 inder of 305 mm diameter and 305 mm length, with
ms -1 and of RL up to 44,000. Additionally, laser a 150 mm diameter concentric window, of 25.4 mm
doppler velocimetry has been used to characterise thick schlieren quality glass in each end plate. These
the turbulence, in place of hot wire anemometry, windows enabled the whole of the explosion flame
which is less reliable in rapidly reversing flows. to be photographed during the pre-pressure period.
Values of burning velocity so acquired are pre- Four identical eight-bladed fans, machined from a
sented in the present paper. The ratio ut/ut ini- solid block of aluminum were equi-spaced around
505
506 TURBULENT REACTING FLOW

the central circumference. The mean diameter of of turbulent energy dissipation caused an apprecia-
each fan was 147 mm and with them all in position ble rate of temperature rise of the gas. Hence the
a central region of 178 mm diameter was created initial temperature for C3H8 and H2 explosions, and
in which there was little mean flow and the tur- for the calibration, was standardised at 328 K. This
bulence was uniform and isotropic. Turbulent ve- was increased to 333 K for CsHI8, conveniently to
locities were varied through changes in fan speed achieve higher equivalence ratios, ~b. All mixtures
between 2400 and 10,000 rpm. Independent speed were made up in the explosion bomb.
control of the Acomel Rl12-M2 induction motor
coupled to each fan ensured equal speeds.
Burning velocities were found by the double ker- 3. Experimental Results
nel method, in which the closing velocity of two
propagating flame fronts of separate kernels was Closure rates of the two contiguous turbulent
measured, l~ This necessitated simultaneous ig- flame fronts were measured along up to three or-
nition at two spark gaps. Each spark plug com- thogonal closure lines, in any one explosion. The
prised two coaxial conductors and these were fitted mixtures employed and their properties are given
horizontally with the gaps 120 mm apart. They were in Table 1. The laminar burning velocities, ue, also
connected in series, in the same capacitative dis- were measured by the double kernel method. With
charge circuit. A 0.05 txf capacity bank was charged point initiation of an explosion, the early stages of
to about 15 kV and discharge through the two gaps flame propagation are not affected by the full spec-
was triggered through the breakdown of an air gap trum of turbulence. It takes time to achieve a fully
by a firing pulse from a synchronisation unit. developed turbulent flame. During the develop-
Flame propagation was recorded with a Hitachi ment phase, the value of turbulent burning velocity
16 HM cin~ camera at a maximum rate of 10,000 might be approximated by multiplying the fully de-
frames see -1. The schlieren technique, with a he- veloped value, ut, by [1 - exp(-{)] ~ where i is
lium-neon laser, defined the flame fronts and re- the elapsed time, from spark to contact of the ker-
vealed something of the structure. Turbulent ve- nels in the present work, divided by the lagrangian
locities and integral length scales were measured, eddy lifetime, 0.44L/u'. 13 It was thus possible to
without any explosion and at different fan speeds, apply a correction to a directly measured burning
by a forward scattering laser doppler counting sys- velocity to give the fully developed value, ut.
tem, with an argon ion laser, rated at 100 mW. A Value of ut for different fan speeds are shown in
Bragg cell gave a maximum frequency shift of 40 Figs. 1-3 for C3H8, H2 and CsHls-air mixtures, re-
MHz. Fuller details of this calibration as well as of spectively, at different equivalence ratios. Associ-
the apparatus and experimental techniques are given ated values of u' also are given on the abscissa. The
in Ref. 12. Values of L ranged from 36 mm at 2400 curves correspond to a least square fit to a second
rpm to 43 mm at 10,000 rpm. In the course of the order polynomial. The spread of values is similar to
present work, calibrations were made for both air that previously observed 12 and some bounds are in-
and a 60% H.z-air mixture at atmospheric pressure, dicated by bars in Fig. 1. The inevitability of ran-
over the full range of fan speeds. Both calibrations dom spread has been discussed and quantified in
proved to be identical. At high fan speeds the rate Ref. 14. When a curve is terminated, it indicates

TABLE 1
Properties of mixtures, all with air

u~ 106v Tb To
Fuel qb m/sec m2/sec K K [3 Le

C3H8 1.3 0.41 16.97 2160 15 000 8.2 0.85

C3Hs 1.1 0.548 17.17 2294 15 400 7.8 0.88
C3H8 0.9 0.466 17.39 2196 16 600 8.9 1.97
C3H8 0.8 0.39 17.50 2062 12 800 7.4 1.98
C3Hs 0.7 0.30 17.61 1900 8 800 5.6 2.03
H2 3.57 2.32 46.37 1666 10 300 7.7 4.71
H2 0.42 1.04 28.46 1224 6 600 7.3 O.70
CsH~s 1.2 0.40 17.93 2283 15 000 7.7 1.16
CsHl8 1.0 0.446 18.10 2355 18 000 8.9 2.48
CsHls 0.92 0.42 18.17 2309 13 800 7.0 3.79
CsHI8 0.85 0.393 18,22 2237 13 400 7.0 3.75
 LEWIS NUMBER EFFECTS ON BURNING VELOCITY 507

Ue versus u'/ue relationships. An additionally used

dimensionless correlating parameter is the turbu-
lent Reynolds number, Re. For present purposes,
this is conveniently contained within a Karlovitz
flame stretch factor K, which is the ratio of the flow
strain rate to a flame gradient given by ue divided
by the laminar flame thickness, Be. This thickness

-V',
40
~o , ~ FAN SPEEO ,(rpm)
~ ~ ,oo'oo
r - AdR o,}~ 9 9 ~1769

FIG. 1. Propane-air turbulent burning velocities ,o ,%;2. ~ .....

, o , .
at five equivalence ratios. Shading shows quenching
~(o..- ,.o., I.. . . . . 4. . . . . . ~. . . . . . . . .
regions. ~o, o 9149 low ,o" r ,~.3)

~176 o~ K;O-B K;~O

9 9149 = .
14

~1o 1; 2; ~15 310 3n5 40

/
u'm~

FIG. 4. Values of u,/ue for C3Hs-air mixtures at 4)

= 0.7, 0.8, 0.9, 1.1 and 1.3. Lines of constant K
also are shown.

,%0 ,o'oo ~ooo' "oo '

10000 HI - AIR g~'~'*" ""
FANSPEED (rpm) ~,, " , , : , " K=O.25
16 THEORY Oq.~k~I : 9 9 ~), O. 42

FIG. 2. Hydrogen-air turbulent burning velocities ...... ~,

/I"~" o ~.,.-::~--~ ....
at two equivalence ratios. No quenching occurred.
L=O.O3/ ~ K OO2 ~ ' ~ - - U ......................

c~ Am ~ / //Lo.o,?/'.s,'s ,.o.o,

i i , i i i
u'm L
~ 4
FIG. 5. Values of u,/ue for He-air mixtures at 4) =
0.42 and 3.57. Lines of constant K also are shown.

I , ' .6
~oco ,looo 6OO0
30 caH~-AJR ~ 99
FAN SPEED ( r p m )
O7

FIG. 3. Iso-octane-air turbulent burning velocities

t 99 r =l.Z
at four equivalence ratios. Shading shows quench- 20 tHEOrY K.03
ing regions.
. .o" o9 "

that at higher fan speeds, gas phase quenching was ~o K:O.02 0~ 9

observed in the measurement volume. The severity
of this increased with fan speed and in the shaded
regions, for the different values of qb, quenching oc-
curred in every explosion. 0 5 10 15 u'mt 2o 25 30
Values of ut were in the region of 12% greater
than the directly measured values for propane and FIG. 6. Values of ut/ue for ~ mixtures
~ and about twice as much for hydrogen at 4) = 0.85, 0.92, 1.0 and 1.2. Lines of constant
mixtures. Figures 4-6 show the corresponding ut/ K also are shown.
508 TURBULENT REACTING FLOW

can be approximated by a function of a K factor for a laminar flame, the

Lewis number, Le, of the deficient reactant, and a
1) dimensionless activation energy, [3, for the assumed
(1) single step reaction. This energy is the activation
Ue
temperature for the reaction, Ta, divided by the
difference between the mixture adiabatic tempera-
In a laminar flame the strain rate is given by ve-
locity gradients, but in a turbulent flame it is given ture, T~, and the temperature of the reactants, Tu.
by u'/k, where k is the Taylor microscale. This gives These values are given in Table 1. The diffusion
a Karlovitz flame stretch factor, K, where coefficient for the deficient reactant, D, in the Lewis
number expression (=k/pCpD) was found using the
procedure of Ref. 19. Values of specific heat, Cp,
(2) and thermal conductivity, k, were taken from Ref.
\ x I \uU 15, whilst the expression of Ref. 20 was used to
obtain the conductivity of mixtures. Values of Tb
For isotropic turbulence the two length scales are were calculated from JANAF data21 and of To by
related by plotting the logarithm of ue against 1/Tb, as pre-
viously described. 12 Values of Ta for the mixtures
he Av of H2 were taken from fief. 12, whilst those for C3H8
L uI (3) and C8H18 were obtained from the curves in Fig.
7(a) and (b). The figure beside each experimental
where A is a numerical constant, assigned a value point for air mixtures is the equivalence ratio of the
of 40.4 by Abdel-Gayed and Bradley (3). With this mixture and the pressure was 1 atm. throughout.
value, Eqs. (1)-(3)yield In near-stoichiometric flames problems arise in
asigning a value of Le, because the reaction rates
K = 0.157 (u'/ue)eRL -~ (4) are governed not only by the diffusion of the de-
ficient eomponent but also by the more abundant
Values of K, so derived, are shown on the curves
of Figs. 4-6. The viscosity was found from the data
of Svehla 15 and the mixture expression of Wilke.16
Densities were found from the Beattie-Bridgeman 2.0
equation.
I zx~ o I ~.~ 12~
I zx\ I * I 26
1.ol- \" L 9 l, IP,EsE~T
4. Discussion t ~ o o[ ~ J I WORK
~ m 1 - 1 - -
Earlier work3't2 has shown the greatest increase
in ut/u e with u'/ue to occur at the lower values of
the latter. Figures 4-6 show the later levelling off A C3 H8-O2
that occurs in the values of ut/ue, sometimes fol- e C3 H8-O2. N2 |
lowed by a reduction and flame quenching, as also | C3 HS-O2-Ar
| C3 HS-O 2 -He
observed by Karpov and Severin. 17 Quenching is 0.1 ~ I I
associated with higher values of K. 3.5 4 4.5 5
104/Tb
The mixtures were chosen to demonstrate Lewis
number quenching effects. The high values of ut/ 0.5
ue at high RE in Fig. 17 of Ref. 3 rest heavily upon I ; ~ ' ~ N 1.2 (b; C8H18 I I
the data of Ref. 18. However, the present work and 0:4 1 ~ O.8
the previous results reported in Ref. 12 do not sup-
port such high values of UJUe and some explana- "0
tion for this has been tentatively offered in Ref. 2. o ---....~5
0"3
Several theories of turbulent burning postulate
that locally the flame propagation is laminar-like. 3'12
E
0---..4:
It is therefore appropriate to examine the effects of
high strain rates in the present work in the light
of those predicted for laminar flames. We shall use 0.2- I I I I I
4.2 4,4 4.6 4.8 5.0
the asymptotic analysis of Tromans6 for this pur- ~o~/rb
pose. This shows that for Lewis numbers greater
than 0.85 lz straining of the flame reduces the lam- Fie. 7. Determination of activation temperature
inar burning velocity. The ratio of strained to un- from ue. (a) propane mixtures. (b) iso-octane mix-
strained laminar burning velocity, Ue~/Ue becomes tures. The ordinate has a logarithmic scale.
 LEWIS NUMBER EFFECTS ON BURNING VELOCITY 509

one. z7 For the stoichiometric iso-octane mixture and i i ~ i i

1-0
the methane mixture of Fig. 9, the order of reac- ). CH4- AIR
tion for both components was taken to be unity and
Re['. 27 was followed, which suggests the value of
Le is the mean of the values obtained when both
reactants are considered separately. 0.8
Values of UeJUe, with K given by Eq. (4), were
calculated for values of Le and 13 relevant to the \~' \\
present study, from the expressions of Tromans. 6
These are shown as the theoretical curves of Ue~/ 0.6
ue against K, in Fig. 8 (appropriate to C3Hs mix- EXPERf

tures), Fig. 9 (appropriate to previous CH4 \ ~ "-- ..~=1.0

3
mixtures ]2) and Fig. 10 (appropriate to He and iso-
\\ "~,2
octane mixtures). Figures 8 and 10 show striking
differences, due to Le and 13, between lean and rich 0.4
%..,
mixtures. They suggest lean mixtures of C3H 8 and \
CsHls are more easily quenched. On the other hand, \ 1.0
\
with He mixtures, because of the high diffusion (~ Le /8
coefficient of He, the effect is the opposite, and rich 0.2 o 1.0 1.01 12.0
He mixtures are more easily quenched. Shaded re- [~ 0.72 0-973 11-7
gions in these figures are those in which theoretical
solutions are inapplicable. 6
These predictions are in qualitative agreement
with the observed behaviour of ut/ue. Figure 4 0 0.2 oi, K
o16 o~8 1~o
shows lean mixtures of C3H8 to exhibit a greater
decline in ut/ut with an increase in K than do rich FIG. 9. Straining effects on laminar burning ve-
mixtures. Figure 6 shows a more striking decline locities of CH4-air mixtures.
with the weak CsHls mixtures. On the other hand,
as predicted by the asymptotic theory, the rich
mixture of He shows the greater flame straining ef- In the two-eddy theory, the rate o f turbulent
fects in Fig. 5. burning is given by the product of the rate of decay
of eddies and the proportion of them burnt during
I I I I I I their lifetime. 3A2 Figures 4 to 6 show values of ut/
C3H8 - A/R ue obtained from this theory corresponding to one
of the mixtures studied experimentally. One dotted
1-0 curve in each figure does not allow for the effects
of strain on the laminar burning in flamelets. The
other dotted curve allows for straining, through cal-
culation of the change of Ue to ue~ at the known
values of K, Le and 13, as described in Ref. 12.
, ~.. | Comparison with the experimental values shows a
tendency of the theory to overestimate values of ut/
ue for the rich C3H8 mixture and to underestimate
values for the rich CsHls mixture.
The application of another quantitative test can
be seen in Figs. 8 to 10. The points and curves
described there as experimental show, for the dif-
o.~ !'~", ",,~ ~ ~ _ ferent mixtures, at values of K given by Eq. (4),
the ratio of the measured values of ut/ue to the
corresponding values from the unstrained two-eddy
l i ~ 0.7 "" 0.9
theory in which laminar flamelets have a burning
0.2t ~ R ~ ' EXPERIMENTS
velocity, ue. This ratio should represent the effects
of straining on the flamelets and be equal to ues/
ue. Comparisons of these values, for the different
mixtures, with those given by the corresponding
K
curves arising from the asymptotic laminar flame
straining theory, indicate the accuracy of the com-
FI(;. 8. Straining effects on laminar burning ve- bined two-eddy and asymptotic theory. Reference
locities of C3Hs-air mixtures. to Fig. 8 shows that for C3Hs mixtures the asymp-
 510 TURBULENT REACTING FLOW

2.2 i
(a) H2-AIR
I
1.4 t (b) CSHIs-AtR I J
2.0 !

1.8 / 1.2
1.C
1.4I ~
5. 1.2 0.8
THEORY ~ 0 42 ]
~51-0 "I 5"0.~
0.8
O.G EXP 0-4 I.-.-~--(0.85,0.92 ) J
0.4 l ,--1.0 i
~3.57 0.2
0.2 t- ~3,57

0 0.1 0.2 0.3 0.1 0-2 0.3

K K
FIG. 10. Straining effects on laminar burning velocities. (a) Hz-air mixtures, (b) iso-octane-air mixtures.
Values of ~ given on each curve.

totic theory suggests a greater difference in the ef- are relatively low. The work of Akindele et al2s
fects of Le and 13 on lean and rich mixtures than suggests that when K < 0.28 flame propagation
was observed in practice. Figure 9 shows that, for might occur in a continuous, unfragmented, wrin-
the previously studied CH4 mixtures 12 there is good kled laminar flame front. The two-eddy theory is
agreement between the theoretically predicted and unlikely to apply in this regime.
experimentally measured reductions in burning ve- Clavin and Williams 29'3~have analysed the effect
locity due to straining. However, no measurements of Le on the propagation of wrinkled flames in tur-
were made with rich mixtures and the range of bulent flows of large scale and low intensity, and
Lewis numbers is insignificant. demonstrated the effects of mean stretch and mean
Figure 10 shows that for the ~b = 1.2, iso-octane curvature on ut. In the present work, curvature ef-
mixture, the underprediction of ut/ue by the two- fects can be neglected31 and theoretical values of
eddy theory leads to the erroneous implication that ut/ue from Ref. 29 are shown in Figs. 5 and 6. The
ues/ue might be greater than unity, when the tendency is for ut/ue to be over-estimated.
asymptotic theory shows that, for this mixture, flame The present discussion rests on a compounding
straining must reduce Ues/Ue below unity. As with of two separately developed theories, both with in-
C3Hs mixtures, the asymptotic theory suggests a herent limitations. The two-eddy theory requires
greater difference, due to Le and 13 influences, be- a value to be assigned to fix the intermittency of
tween lean and rich mixtures than, in fact, was ob- dissipative eddies. This value may not remain
served. Probably the rather simplistic approach to unchanged for all conditions. The assumed single
the evaluation of Le, based upon the diffusion solely step reaction in the asymptotic analysis is clearly
of the deficient reactant, over-accentuates Lewis unrealistic, as also may be the requirement that
number effects between weak and rich mixtures. the value of Le should not depart too far from unity.
The theory also suggests big differences between In practice the detailed kinetics and the diffusion
the two H2 mixtures, with the lean one having an of active chain carriers will be influential. Warnatz
increased burning rate as a consequence of flame and Peters 32 employed the detailed chemical ki-
straining. Experimentally, large differences were netics for the two hydrogen mixtures used in the
observed, qualitatively in the right direction, in that present work, in a numerical analysis Of stretched
the rich mixture displayed a greater reduction in laminar flames. The effects of Lewis number on
the laminar burning velocity due to strain. It should burning velocity were no less marked than those
be noted that the values of K for the hydrogen flame suggested by the present experiments. A further
 LEWIS NUMBER EFFECTS ON BURNING VELOCITY 511

problem that arises in turbulent flow is the correct 10. ANDREWS, G. E. AND BRADLEY, D.: Combustion
definition of the strain rate. Under isotropic and and Flame 20, 77 (1973).
isothermal conditions it is appropriate to express it 11. ANDREWS, G. E., BRADLEY, D. AND LWAKA-
by u'/h, but this may not always be valid. BAMRA, S. B.: Fifteenth Symposium (Interna-
tional) on Combustion, p. 285, The Combus-
tion Institute, 1975.
5. Conclusions 12. ABDEL-GAYED, R. G., AL-KHISHALI, K. J. AND
BRADLEY, D . : Proceedings of Ttle Royal Society
(i) Values of ut/ue have been measured in a re- of London A391, 393 (1984).
gime of high turbulence and high RL for 13. ABDEL-GAYED, R. G., BRADLEY, D. AND LWAK-
C3H8, He and CsHls-air mixtures. ABAMRA,S. B.: First Specialists Meeting (Inter-
(ii) As quenching is approached, there is a de- national) of the Combustion Institute, Vol. 1,
crease in values of ut/ue with increase in u'/ p. 95, Bordeaux, 1981.
Us 14. AL-KtnsItALI, K. J., BOSTON, P. M., BRADLEY,
(iii) Quenching effects are greater with the lean- D., LAWES, M. AND PEGG, M. J.: International
er hydrocarbon and the richer H2 mixtures. Conference on Combustion in Engineering, Vol.
This can be ascribed to the effects of Lewis 1, p. 175, The Institution of Mechanical En-
number in strained laminar flames. gineers, 1983.
(iv) Comparisons have been made between the 15. SVEftLA, R. A.: Estimated Viscosities and Ther-
implied reductions in laminar burning veloc- mal Conductivities of Gases at High Temper-
ity due to straining and those predicted by atures, NASA TR R-132, 1962.
asymptotic analysis. There is good qualitative 16. WILKE, C. R.: J. Chem. Phys. 18, 517 (1950).
agreement. 17. KARPOV, V. P. AND SEVERIN, E. S.: Combus-
(v) Quantitative limitations probably arise from tion, Explosion and Shock Waves 16, 41 (1980).
those inherent in the theoretical models. (Translated from Fizika Goreniya i Vzryva 16,
45 (1980).)
18. KOZAEHENKO, L. S. AND KUZNETSOV, I. L.:
Acknowledgments Combustion, Explosion and Shock Waves 1, 22
(1965). (Translated from Fizika Goreniya i Vzryva
Thanks are given to British Gas, British Leyland 1, 31 (1965).)
Technology and the SERC for their support. 19. HIRSCHFELDER,J. O., CURTISS, C. F. AND BIRD,
R. B. : Molecular Theory of Gases and Liquids,
p. 539, John Wiley and Sons, 1965.
REFERENCES 20. MASON, E. A. AND SAXENA, W. H.: Physics
Fluids 1, 361 (1958).
1. TABACZYNSKI,R. J.: Prog. Energy Combust. Sci. 21. JANAF Thermochemical Tables, Second Edi-
2, 143, 1976. tion, NSRDS-NBS 37, United States Depart-
2. ARDEL-GAYED, R. G. AND BRADLEY, D.: Fuel- ment of Commerce, 1971.
Air Explosions, (J. Lee, C. M. Guirao and D. 22. DAMKOHLER,G.: Z. Elektrochemie Angewandte
E. Grierson, Ed.), Studies Series, Study No. Phys. Chem. 46, 601 (1940). (English Transla-
16, p. 51, University of Waterloo Press, 1982. tion, NACA TM 1112 (1947).)
3. ABDEL-GAYED, R. G. AND BRADLEY, D.: Phil. 23. WAGNER, P.: Burning Velocities of Various Pre-
Trans. Royal Society of London, 301, 1, 1981. mixed Turbulent Propane Flames on Open
4. ARDEL-GAYED, R. G. AND BRADLEY, D.: Six- Burners, NACA TN 3575 (1955).
teenth Symposium (International) oll Combus- 24. GIBBS, G. J. AND CALCOTE, H. F.: J. Chem. and
tion, p. 1725, The Combustion Institute, 1977. Engineering Data 4, 226 (1959).
5. BUCKMASTER,J.: Seventeenth Symposium (In- 25. SOKOLIK, A. S., KARPOV, V. P. AND SEMENOV,
ternational) on Combustion, p. 835, The Com- E. S.: Combustion, Explosion and Shock Waves
bustion Institute, 1979. 3, 36 (1967). (Translated from Fizika Goreniya
6. TROMANS,P. S.: The ASME Symposium on Fluid i Vzryva 3, 61 (1967).)
Mechanics of Combustion Systems, p. 201, 26. METGHALCttI, M. AND KECK, J. C.: Combustion
Boulder, Colorado, 1981. and Flame 48, 191 (1982).
7. BUCKMASTER, J. AND MIKOLAITIS, D.: Combus- 27. JOUL1N, G. AND M|TANI, T.: Combustion and
tion and Flame 47, 191 (1982). Flame 40, 235 (1981).
8. DURRIN, P. A.: J. Fluid Mech. 121, 141 (1982). 28. AKINDELE, O. O., BRADLEY, D., MAK, P. W.
9. DANESHYAR, H., MENDES-LOPES, M. C. AND AND McMAHON, M.: Combustion and Flame 47,
LUDFORD, G. S. S.: Nineteenth Symposium 129 (1982).
(International) on Combustion, p. 413, The 29. CLAVIN, P. AND WILLIAMS, F. A.: Prog. Aero-
Combustion Institute, 1983. nautics Astronautics 76, 403 (1981).
512 TURBULENT REACTING FLOW

30. CLAVIN, P. AND WILLIAMS, F. A.: J. Fluid Mech. Plane Premixed Hydrogen-Air Flames, Paper
116, 251 (1982). presented at the Ninth International Collo-
31. KLIMOV, A. M.: Zh. Prikl. Mekh. tekh. Fiz. 3, quium on Dynamics of Explosions and Reactive
49 (1963). Systems, Universitd de Poitiers, July, 1983.
32. WABNAI'Z, J. AND PETEBS, N.: Stretch Effects in