COMBUSTION SCIENCE AND TECHNOLOGY

ISSN: 0010-2202 (Print) 1563-521X (Online) Journal homepage: https://www.tandfonline.com/loi/gcst20

Determination of and Fuel Structure Effects on

Laminar Flame Speeds of C1 to C8 Hydrocarbons

S. G. DAVIS & C.K. LAW

To cite this article: S. G. DAVIS & C.K. LAW (1998) Determination of and Fuel Structure
Effects on Laminar Flame Speeds of C1 to C8 Hydrocarbons, COMBUSTION SCIENCE AND
TECHNOLOGY, 140:1-6, 427-449, DOI: 10.1080/00102209808915781

To link to this article: https://doi.org/10.1080/00102209808915781

Published online: 05 Apr 2007.

Submit your article to this journal

Article views: 1660

View related articles

Citing articles: 54 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=gcst20
Combust. Sci. and Tech.• 1998. Vol. 140, pp. 427-449 «;11998 OPA (Overseas Publishers Association) N.V.
Reprints available directly from the publisher Published by license under
Photocopying permitted by license only the Gordon and Breach Science
Publishers imprint.
Printed in Malaysia.

Determination of and Fuel Structure

Effects on Laminar Flame Speeds
of C1 to Ca Hydrocarbons
S. G. DAVIS and C. K. LAW·

Department of Mechanical and Aerospace Engineering, Princeton University,

Princeton, New Jersey 08544 USA

(Received 20 July 1998; In final form 6 October 1998)

Laminar flame speeds determined by using the counterflow twin flame configuration were
compared for various C 1 to C g hydrocarbons, including alkanes, alkenes, alkynes, aromatics,
and alcohols. The data were compared over an extensive range of equivalence ratios at room
temperature and atmospheric pressure. The comparison shows that the laminar flame speeds
of normal alkanes are close throughout the entire range of equivalence ratios studied, except
for methane whose flame speeds are consistently lower. The more unsaturated the molecule
the higher the flame speed for fuels having the same carbon number in the order of
alkanes < alkenes < alkynes. Methyl substitution for hydrogen or branching reduces the flame
speeds for both alkanes and alkenes. The flame speeds of large saturated cyclic species
(cyclohexane and cyclopentane) are close to those of their normal alkane analogs.

Keywords: Flame propagation; burning velocities; hydrocarbon oxidation

INTRODUCTION

Detailed kinetic models have been found to be useful in evaluating

and understanding various fundamental as well as practical combustion
processes. In order to validate their accuracy and test their range of
applicability, it is important that these kinetic models be compared with
experimental data of high fidelity, such as the flame structure and global
response. For the global response, a widely-used parameter is the laminar

• Corresponding author. e-mail: [email protected]

427
428 S. G. DAVIS AND C. K. LAW

flame speed which describes the propagation of a one-dimensional, planar,

adiabatic, premixed flame in the doubly-infinite domain. This response
embodies the fundamental information on the diffusivity, reactivity, and
exothermicity of a given mixture, and has been extensively used to partially
validate proposed kinetic models. The rationale being that if a proposed
model could not even describe such an important response as the laminar
flame speed, its comprehensiveness and utilitywould obviously require
further examination. As such, extensive efforts have been made over the
years to accurately determine the laminar flame speeds of various fuels.
Much of the experimental efforts on hydrocarbon fuels, however, have
concentrated on those in the C, to C 3 range. Comprehensive flame speed
studies on higher hydrocarbons, say, up to Cg and including alkanes,
alkenes, alkynes, aromatics, and alcohols, are scarce and mostly limited to
studies performed in the 1950's or even earlier. The two widely-cited sources
of data are those by Gerstein et al. (1951), who employed the technique of
flame propagation in a tube, and by Gibbs and Calcote (1959), who used
the Bunsen flame. The importance of these studies is two-fold: (I) It gives
valuable insight into the effect of the fuel structure on the laminar flame
speed; (2) it presents data for a wide variety of hydrocarbon fuels, which are
essential for the development of comprehensive chemical kinetic models.
Valuable as these data are, their accuracy is compromised by the inherent
effects of nonadiabaticity as well as flame stretch coupled with mixture non-
equidiffusion. Specifically, recent studies of stretch-corrected flame speeds
of mixtures of air with benzene (Davis et al., 1996), iso-octane and
n-heptane (Davis and Law, 1998) found substantial deviations in the stretch-
affected flame speeds of both Gibbs and Calcote (1959) and Gerstein et al.
(1951), by as much as !Ocm/s. Since these earlier studies constitute the
primary source of flame speed data (Glassman, 1996) that have been used
for the validation of comprehensive mechanisms for several hydrocarbons,
re-determination and re-evaluation of their values are imperative.
In the present study we first used the well-established counterflow twin-
flame technique to experimentally determine the laminar flame speeds of
many of the fuels in the C 3 to C 7 range (propene, n-butane, iso-butane,
l-butene, iso-butene, 1,3-butadiene, n-pentane, cyclopentane, n-hexane,
cyclohexane, benzene, toluene). These data were then compiled together
with those for other fuels (methane, ethane, ethylene, acetylene, propyne,
propane, n-heptane, iso-octane, hydrogen, methanol, ethanol), previously
determined using the same counterflow technique, to form a comprehensive
data base useful for further kinetic modeling studies. The fidelity of the
present database also allowed us to conduct a more quantitatively reliable
 HYDROCARBON FLAME SPEEDS 429

comparison of the effects of fuel structure on the laminar flame speeds of

these fuels.
Furthermore, because of the usefulness of the counterflow technique,
various improvements and enhancements in the experimentation and data
reduction procedure have been implemented since the original work first
appeared (Wu and Law, 1984). Thus an auxiliary objective of this paper is to
consolidate and present in some detail the various aspects of these secondary
features which were either implemented in the present study or reported
previously but were scattered in the literature.

EXPERIMENTAL CONSIDERATIONS

Apparatus

Wu and Law (1984) first identified the need to account for stretch effects
in the experimental determination of laminar flame speeds. A counterflow
twin flame technique was then proposed which allows the systematic subtrac-
tion of these stretch effects. Figure I shows a schematic of the experimental

FIGURE I Schematic of the flow system and burner assembly.

430 S. G. DAVIS AND C. K. LAW

apparatus suitable for the determination of the laminar flame speeds of both
gaseous and liquid fuels in reduced and elevated pressure environments.
lt consisted of a syringe pump, sonic nozzles, a gas heater monitored by a
temperature and power controller used to vaporize liquid fuels, a nozzle
for atomizing liquid fuels, a vaporization chamber, cooling coils, a mixing
chamber, and a combustion chamber/burner assembly. Nitrogen, which was
used as a diluent in the vaporization process, passed through an electric
resistance heater before vaporizing the liquid fuels in the vaporization
chamber. Oxygen was then added to the flow in the ratio of02/N2 = 1/3.76 to
achieve the desired air composition. Along with oxygen, air was also added to
the system to yield the specified equivalence ratio of the premixture. The flow
then passed through cooling coils so that the premixture entered the burner
at room temperature. A small portion of the air was split from the main line
and passed through a chamber filled with 0.3 urn diameter alumina particles
used for seeding the flow. This seeded flow then merged with the main flow
downstream of the cooling coils. A by-pass valve regulated and ensured
a consistent composition of the premixture flow entering the burner. For
gaseous fuels the vaporization process was not required and the fuel was
added directly to the air stream to obtain the desired mixtures. In the burner,
two symmetrical, planar, nearly-adiabatic flames were generated from the
counterflow nozzles. Each nozzle also had a coaxial flow of nitrogen around
it which helped to isolate the flames from the surrounding environment.
Furthermore, by varying the exit velocity of the nitrogen coflow, flame
stabilization could be more easily achieved. The axial velocity profile along
the centerline was measured using a TSI Argon-ion Laser Doppler
Velocimetry system (496 nm). The measuring volume was 0.3 mm diameter
by 0.9 mm length, and was operated in the forward scattering mode. The
signal from the photomultiplier was analyzed using a TSI processor. The
LDV-determined velocity is estimated to be accurate to within 1-2cm/s.

Liquid Fuel Vaporization

Premixed flames require a homogenous mixture of the fuel and oxidizer.

When working with liquid fuels, it is necessary that all the fuel is vaporized
and well mixed with the oxidizer before entering the burner. In the present
experiments a large fraction of the total nitrogen flow (~70%) was used as
a diluent for vaporization. The diluent nitrogen was first heated to a
temperature which was high enough to vaporize the fuel and bring the
mixture above the saturation temperature of the fuel at the given pressure.
The heated nitrogen flowed coaxially with the fuel, but at the injector head
 HYDROCARBON FLAME SPEEDS 431

was forced through one of the four apertures angled off center, which
sheared the liquid flow both axially and radially and thereby enhanced
atomization. The resulting spray then entered a 300 em? stainless steel
vaporization chamber which served as a cavity and prevented liquid
impingement on the hot surfaces.
Since maintaining the diluent nitrogen at the specified temperature was
crucial for the evaporator design, it was regulated by a temperature-
controlled electrical resistance heater. Similar to the evaporator for the
Princeton Variable Pressure Flow Reactor (Held, 1993), the diluent tem-
perature was set 50°C higher than the calculated temperature to avoid liquid
impingement on the vaporization chamber walls and to account for heat loss
to the injector. This evaporator system showed no flow rate oscillation, as
implied experimentally by the presence of steady, consistent, flames.

Flow Control

Since many of the hydrocarbons under study were liquids, a syringe pump
was used to ensure repeatability and stability, especially in a pressurized
environment. The volumetric flow rate of the syringe pump was calibrated
to the metering controller using water, and density corrected to determine
each fuel's mass flow rate. The uncertainty for the mass flow rate of the fuel
using the syringe pump was estimated to be better than 1%.
Gaseous fuels were metered with a Hasting's mass flow controller while
all other gases (nitrogen, oxygen, and air) used in the experiment were
metered using sonic nozzles. The Hasting's mass flow controller was chosen
over sonic nozzles for gaseous fuels due to its extended range of operating
conditions and because sonic conditions could not be achieved for most
fuels in the C4 family because of their low vapor pressures at room
temperature. Sonic nozzles, however, are convenient for determining the
flow rates of the other gases. The upstream pressure was measured using
a pressure transducer (PSI-Tronix), which had an accuracy of 0.25% full
range. The mass flow controller and sonic nozzles were calibrated using wet
test meters (Precision Scientific) which had accuracies around I %, thus
giving an uncertainty of the gas flow rate on the order of 1%.

Composition Uncertainty

As stated earlier, the mass flow controller and sonic nozzles were calibrated
using wet test meters that had accuracies of ~ 1%. The calibration of the
 432 S. G. DAVIS AND C. K. LAW

controller and nozzles were periodically checked throughout the investiga-

tion, and were consistently within 1% of their originally calibrated value.
The accuracy of the volumetric displacement of syringe pump was better
than 1%. Room temperature did not change by more than ± 1°C
corresponding to a change in density of less than I% for the liquid fuels,
thereby ensuring an accuracy of ~ 1% for desired fuel mass flow rates
using the syringe pump. Therefore the combined uncertainty of the mix-
ture composition was approximately 1- 2% and was verified using gas
chromatography (GC). Different fuel/air mixture samples were also
collected for a variety of fuels and checked against calibration samples
using GC. Results of the flow system were within 1.5% of the calibration
samples, which is well within the range of accuracy using this procedure.

Experimental Specifications

Using the present system, laminar flame speeds were experimentally

determined for mixtures of air and propene, n-butane, iso-butane, I-butene,
iso-butene, 1,3-butadiene, n-pentane, cyclopentane, n-hexane, cyclohexane,
n-heptane, iso-octane, benzene, and toluene. The laminar flame speeds of
propyne were obtained for nitrogen-diluted air mixtures. Results for certain
fuels mentioned above (propene, propyne, n-heptane, iso-octane, benzene,
toluene) were reported in earlier studies, however, the flame speeds were all
determined using the approach and apparatus described herein. Furthermore,
the flame speeds for benzene/air and toluene/air mixtures were re-evaluated in
. the present study due to improvements in flame stability, although the results
did not change much from those previously reported (Davis et al., 1996).
Flame speed data were obtained at atmospheric pressure, room temperature,
and over an extensive range of equivalence ratios for all fuels (1) = 0.7- 1.7)
except for toluene which was limited to a maximum equivalence ratio of
1> = 1.3 due to its partial pressure restriction at room temperature. We realize
that for sufficiently rich burning most of the fuels are considered to be near or
within the sooting regime, depending on the configuration used to determine
the onset of soot formation (Glassman, 1996). Soot, however, was not
observed (1) "5, 1.7) for the present experimental configuration. The fuels
examined along with their vendor-specifications are shown in Table I.
Regarding the specifics of the other flame speed data which were compiled
here, the reader is referred to: methane and propane (Vagelopoulos et al.,
1994); ethane, ethylene, and acetylene (Egolfopoulos et al., 1990); methanol
and ethanol (Egolfopoulos et al., 1992a, I992b); and hydrogen (Egolfo-
poulos and Law, 1990; Vagelopoulos et al., 1994; Aung et al., 1998).
 HYDROCARBON FLAME SPEEDS 433

TABLE [ Fuel specifications

Fuel Purity Vendor

propyne" 98.1% Apogee Technology
propene 99.5% Scotl Specialty Gases
n-butane 99% Matheson Gas
iso-butane 99.5% Matheson Gas
l-butene 99% Matheson Gas
iso-butene 99% Matheson Gas
1,3-butadiene 99% Matheson Gas
n-pentane 99.94% EM Science
cyclopentane 99.48% Aldrich Chemical Co.
n-hexane 99.4% Sigma Chemical Co.
cyclohexane 99.99% EM Science
n-heptane 99.64% EM Science
iso-octane 99.94% EM Science
benzene 99.96% EM Science
toluene 99.5% EM Science
"Major impurities are propane « 0.2%), butane « 0.5%), allene « 0.3), and
butenes « 0.7%).

CONSIDERATION OF STRETCH EFFECTS

The twin flames are created by impinging two symmetrical counterflowing

combustible mixtures generated from two identical aerodynamically-shaped
nozzles. Due to symmetry, conductive heat loss to the downstream is
basically eliminated. It has also been found that radiation minimally affects
the burning intensity of flames which are well within flammability limits
(Egolfopoulos, 1994). Moreover, positive stretch associated with the
counterflow suppresses flamefront cellular instability (Sivashinsky et al.,
1982) which tends to develop for sub-unity Lewis number flames and
renders the wrinkled flame to propagate at a speed very different from that
of the laminar flame speed.
Figure 2 shows that shortly downstream of the nozzle exit, the LOV-
measured axial velocity profile assumes that of the stagnation flow, being
characterized by the negative of the velocity gradient, K. The velocity then
reaches a minimum as it approaches the upstream boundary of the preheat
zone, and then increases due to thermal expansion. After reaching a maxi-
mum, the velocity again decreases due to stagnation. From theory, the local
stretch rate experienced by the flame is given by K (Fig. 2), in units of S-I
(Law, 1988). The minimum of the velocity profile entering the thermal
mixing layer is identified as the reference upstream burning velocity Su.ref
and can be plotted for various K as seen in Figure 3.
It was shown from early theoretical (Matalon and Matkowsky, 1982) and
experimental (Wu and Law, 1984) studies that the strained flame speed,
434 S. G. DAVIS AND C. K. LAW

250

200

..,o
'""'
til

Eo 150
'-'
C
'g 100 cAo
o °00
~ o 00
> o 0
o 0
50 o 0
r---------------..::s~ 00
o
o

-6.0 -5.0 -4.0 -3.0 -2.0 -1.0 0.0

Distance from the Stagnation Surface (mm)

FIGURE 2 Typical experimental axial velocity profile across a stagnation flame, showing
definitions of both K and su,rc:r.

Karlovitz Number, Ka
o 0.011 0.022 0.033 0.044 0.055 0.066
50

45

s~, linear

40/
y---
35 ~ SO' non J'mear
0

30 L-..-~~~~~~~_~~_~~~~~-.J

o 100 200 300 400 500

Stretch Rate, K (s-t)
FIGURE 3 Examples or linearly- and nonlinearly-exlrapolated flame speed for n·heptane!air
mixtures, q, = 1.0.
 HYDROCARBON FLAME SPEEDS 435

varies linearly with K for sufficiently small values of K. The data could
Su,rcf,
then be linearly extrapolated to K = 0 to yield the stretch-corrected flame
speed s~, as shown in Figure 3. However, Dixon-Lewis (1990) demon-
strated, using a detailed numerical simulation of the counterflow flame
with a finite separation distance between the nozzle exits, that the linear
extrapolation slightly over-predicts the true flame speed when the local
minimum value of the velocity profile is used for the reference burning
velocity, Su.rcf. Subsequently, an asymptotic analysis using potential flows
(Tien and Matalon, 1991) and allowing for displacement effects showed
that, as K -+ 0, the relation between Su,rcf and K was not linear, and that
the linearly extrapolated value would be slightly higher than its true value.
Realizing that the minimum point in the velocity profile was experimentally
well defined, Tien and Matalon (1991) proposed a nonlinear extrapolation
of Su,rcf as K -+ 0, given by

Su,ref = s~{l - (J.L - I)Ka + Kaln[(a - l)jKa])

where a is the ratio of the densities of the unburned to the burned mixtures,
and J.L is the Markstein number which depends on the diffusive properties of
the mixture, and Ka = K{j°js~ the Karlovitz number with {j0 being the
characteristic flame thickness.
Chao et al. (1997) subsequently performed an asymptotic analysis based
on a finite separation distance (L) of the nozzle exits and plug flow boundary
condition, which approximates more closely the actual axial velocity profile
in the counterflow system. The results again showed a nonlinear variation
as Ka -+ 0 and a nonlinear extrapolation expression was also developed, It
was further shown that, by either decreasing Ka or increasing L, the effect of
nonlinearity is reduced and the accuracy of the linear extrapolation
increases. These results were also substantiated through numerical simula-
tions and experimentation, Specifically, it was shown for various methane/
air flames that there was little difference in the results if the nozzle
separation distances of 14 or 22 mm were used,
Vagelopoulos et al. (1994) further showed computationally and experi-
mentally that in order for the linear extrapolation to be accurate, Ka must
be on the order of 0, I, at least for methane/air, propane/air, and lean
hydrogen/air flames studied. As such, laminar flame speeds obtained from
the linear extrapolation under most of the present conditions would yield
sufficiently accurate results. For very weak flames, though, the accuracy
may decrease because of reduction in the laminar flame speed and increase
in the flame thickness.
436 S. G. DAVIS AND C. K. LAW

For the present study the local minimum in the velocity profile was still
adopted as the unburned reference flame speed due to the unambiguous
identification of that location. The nozzle diameter D was 14 mm, which was
a compromise between the need for a large nozzle separation distance and
hence a large nozzle diameter in order to minimize linear extrapolation
errors, and the limitations on the maximum obtainable liquid fuel flow rates
and problems encountered with buoyancy-induced flame instabilities for
large nozzles. To keep a fixed ratio of nozzle separation distance to nozzle
diameter, an L/ D ratio of 1.15 was utilized.

RESULTS AND DISCUSSION

Both linear and nonlinear extrapolations were attempted, as shown in

Figure 3 for a typical case. Figure 4 shows a comparison between the linearly
and nonlinearly extrapolated results for 1,3-butadiene, propene, and iso-
octane which are fairly representative of all the fuels experimentally meas-
ured using the present apparatus (Tab. I). Results show that the difference
between the two procedures is quite small, with the nonlinear results being
typically 1- 2 cra]« lower for most equivalence ratios. Since Ka for most fuels
at very rich, weakly-burning conditions (</1 = 1.5-1.7) could not be main-

70

,..., 60
--e '"
50
•
• a
• a
. .
0
l,3-butadiene
~ ~ ~
propene
0 •
a I!
'-"
U
ttttttt •a
~a
~ ss iso-octane
::s
o
'"
40 t •a
] •o~ t "'<:;7'\1'\1~
......... ~ t
...."7 ...

"'''' 0• a•
30 \7~
0.
ell at ~'"
t~~ "' .'". •a
... •0 •
<1)
20
§ ~

~ to '" ... ...0

'" '"
0
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio

FIGURE 4 Representative flame speeds obtained using the linear (filled symbols) and
nonlinear (open symbols) extrapolations.
 HYDROCARBON FLAME SPEEDS 437

tained below 0.\, and was generally between 0.2-0.3, the difference between
the linear and nonlinear results increases to about 2- 3 cm/s for such
conditions. It is, however, important to note, say from Figure 3, that the
experimental stretch rates were not sufficiently small, and the experimental
data sufficiently accurate to capture the nonlinear segment of the variation. In
other words, the nonlinear trend and extrapolation are theoretical, instead of
experimental, results. As such, while we do know that the nonlinearly
extrapolated laminar flame speeds are smaller than the linearly extrapolated
ones, the extent of deviation has not been experimentally quantified.
Since the difference between the two techniques is adequately small for
nearly all fuels throughout most equivalence ratios, either technique can be
considered to be sufficiently accurate within the same uncertainty as that for
the velocity measurements. Also, because the linearly-extrapolated values
are well-defined experimentally, and further recognizing that some of the
earlier data were determined using linear extrapolation, only results of the
linear extrapolation will be presented and compared in the following.
Before comparing the various flame speeds, we first note that the flame
temperature has a profound effect on the laminar flame speed due to its
exponential influence. As such, Table II presents the ranked adiabatic flame

TABLE II Ranked flame temperatures of selected fuel/oxidizer

mixtures

Fuel mixture Adiabatic.flame temperature (K), q, = 1.0

acetylene/air 2553
l,3-butadiene/air 2381
ethylene/air 2378
benzene/air 2352
propene/air 2342
toluene/air 2337
l-butene/air 2329
iso-butene/air 2321
cyclopentane-air 2299
propyne/(18%O,-82%N,) 2298
cyclohexane/air 2289
n-heptane/air 2283
n-hexane/air 2280
iso-octane/air 2280
n-pentane/air 2278
n-butane/air 2276
propane/air 2273
iso-butane/air 2272
ethane/air 2266
ethanol/air 2242
methane/air 2231
methanol/air 2227
ethylene/(18%O,-82%N,) 2223
438 S. G. DAVIS AND C. K. LAW

temperatures at ¢ = 1.0 for the various mixtures examined in this study.

It has been well established that for unburned mixtures containing fuels of
similar structure, the flame speeds are directly correlated to the flame
temperature. There are, however, secondary kinetic effects which cause
deviation of the flame speed for mixtures with similar flame temperatures, as
will be discussed later.
The flame speeds of the normal alkanes from methane through n-heptane
were compared first. Results of methane through n-butane are shown in
Figure 5 and n-butane through n-heptane are shown in Figure 6. It is seen
that most alkanes (ethane and higher) have almost the same flame speeds
throughout the entire range of experimental equivalence ratios, which is
not surprising because their flame temperatures are similar (Tab. II). The
flame speeds of ethane are seen to be slightly higher than the others for
stoichiometric to rich conditions, while those for methane are consistently
lower. This same qualitative trend was also observed in the study of Gerstein
et al. (1951).
Recognizing that the fundamental parameter for flame propagation is the
laminar burning rate per unit area, m = Pus:, where Pu is the density of the
unburned mixture, Figures 7 and 8 show corresponding plots for m, which

50

...... 40 • Oe~o~.o
o· o.
--E
'"
u
........
.. 0
o. o
..0.0
.•
0 .0
•
30 ce· 0 •
o ::
'" q,+
•
ceo 0
<e.
0
] <fo <e.
a. 20 0
Vl
<I)
0 0 0 0 •• +
~
0 0
10 0
•
0
n-butane
propane
0
0
o •
Ii: + ethane
0 methane
0
0.6 0.8 1.0 1.2 1.4 1.6 1.8

Equivalence Ratio

FIGURE 5 Flame speeds of the C,-C. normal alkanes. The methane and propane dala are
taken from Vagelopoulos et at. (1994), and the ethane data are taken from Egolfopoulos et at.
(1990).
 HYDROCARBON FLAME SPEEDS 439

50

-----'"E
40
et
~
rA~~fo
., ..,
C,)
'-'
30
et to
o "
'" ~ to
"0
Il)
Il) ~
~ :
20
••
P.. 0
CIl
Il)
0

§ •
0
n-heptane
n-hexane 0:
10
li: • n-pentane
0 n-butane

0.6 0.8 1.0 1.2 1.4 1.6

Equivalence Ratio
FIGURE 6 Flame speeds of the C.-C, normal alkanes.

0.05

0.04

0.03
~.

0.02 o 0 • n-butane o ••
o 0
o
•o
propane
ethane 0 0 • •
0.01 o methane o

0.6 0.8 1.0 1.2 1.4 1.6 1.8

Equivalence Ratio
FIGURE 7 Mass burning rates of the C,-C, normal alkanes. The methane and propane data
are taken from Vagelopoulos et al. (1994), and the ethane data are taken from Egolfopoulos
et al. (1990).
440 S. G. DAVIS AND C. K. LAW

0.06
--.
,
<IJ
0.05
"'~,,~
M
S
"'~
U
~
'-' 0.04
~
tt
~ ~e
<I)
(;J
0:::
bI)
0.03
.," •
c:
'E::s 0.Q2
o.
lJ:l
<IJ
<IJ
o:s 0.01
•
0
•
n-heptane
n-hexane
n-pentane
0

• 0:

0 n-butane
~
0
0.6 0.8 1.0 1.2 1.4 1.6 1.8

Equivalence Ratio
FIGURE 8 Mass burning rates of the C4-C7 normal alkanes.

removes the dependence of the burning intensity on the mixture density and
hence to a large extent the influence of the molecular weight of the fuel. It
is seen that the same trend in ranking is preserved, indicating that the
similarities and differences exhibited are consequences of the global reaction
rate wand the density-compensated diffusivity >.jcp through m ~ (Ajc p W)I/2,
where A is the thermal conductivity and cp the specific heat. The same
observation holds for all of the following comparisons.
Saturated cyclic C s and C 6 compounds were examined and are compared
to their linear counterparts in Figure 9. The flame speeds of cyclopentane
fall slightly below those of the rest of the fuels, which have approximately
the same value. The difference, however, is within the experimental un-
certainty, thus rendering a definitive interpretation of the results difficult.
Branching in fuel structure, however, has a greater effect on the flame
speed for saturated compounds. Figure 10 compares the flame speeds of
two branched alkanes with those of n-butane, a typical normal alkane. It is
seen that methyl substitution has the tendency to decrease the laminar flame
speed, with the most profound effect occurring at near-stoichiometric
conditions. Since most of the branched and normal alkanes have only very
small differences in adiabatic flame temperature and thermal diffusion
coefficient, these results indicate that the reduction in flame speed due to
branching is mainly caused by differences in their oxidation kinetics, which
will be discussed later.
 HYDROCARBON FLAME SPEEDS 441

';;' 40
'<,

E
~
0","
30

•
o
n-hexane
cylcohexane
e
• n-pentane
o cyclopentane

0.6 0.8 1.0 1.2 1.4 1.6

Equivalence Ratio
FIGURE 9 Flame speeds of cyclic and normal alkanes.

•••••
';;'
--~
40
.. ~ ~ ~ ~ ~~ ..
0"
'"
30

20
.. . .
~~
• 0
o·
~ o.
•~
~
~ .~
•o
• n-butane •• 0
10 o
•
iso-butane
iso-octane
•
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
FIGURE 10 Flame speeds of branched and normal alkanes.

The flame speeds of unsaturated hydrocarbons are studied next. Figure II

shows the flame speeds of selected l-alkenes, with a typical normal alkane
(n-butane) shown for comparison. It is seen that the flame speeds for all
the l-alkenes studied are higher than that of n-alkanes, with propene and
442 S. G. DAVIS AND C. K. LAW

80

•
•• •
70
•
-,-..
ell

8u
......
..,
o ::t
60

50
•
• •
•
40
1.
til
<l)
30 •
• ethylene
~ 20 o propene
o
o • •
~ •o l-butene o
10 n-butane

0
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
FIGURE II Flame speeds of the C,-C. normal I-alkenes. The ethylene data are taken from
Egolfopoulos et al. (1990).

I-butene higher by about 5cm/s and ethylene higher by about 30cm/s at

their respective maximums. Furthermore, the peak in the flame speed also
shifts to higher equivalence ratios for the l-alkenes, The higher flame speeds
of ethylene as compared to those of propene and l-butene are caused by two
factors. First, the adiabatic flame temperatures of ethylene/air mixtures are
higher than those of the other two fuels (Tab. II), and second the oxidation
kinetics of ethylene/air mixtures are faster at these conditions. This result is
substantiated by the ethylene results of Egolfopoulos et al. (1990) in an 18%
O 2 in N 2 mixture. With nitrogen dilution the maximum flame speed of
ethylene is still slightly higher than those of propene/air and l-butene/air
mixtures, yet its flame temperature is lower by more than 100 K (Tab. II).
The effect of a triple C-C bond on the flame speed was studied for
acetylene (Egolfopoulos et al., 1990) and propyne (present study). Due to
the cost of propyne, the laminar flame speeds were taken in nitrogen-diluted
conditions (18% O 2 in N 2) , which lowered the flame speed and thus
effectively reduced the required total flow rate. These altered conditions
make quantitative comparisons with fuel/air mixtures inappropriate.
Figure 12 shows the flame speeds of C2 and C 3 species from alkanes to
alkynes. Ethane was not included in the graph for the sake of clarity,
recognizing that its flame speeds are very similar to those of propane. It is
 HYDROCARBON FLAME SPEEDS 443

200
•
0
acetylene
ethylene
"""' • propene
propyne (18%Oz in Nz) •
--a
'"
<..l
'-'
150
0
~ propane
•
• •
o "
'" 100 •
-e
•
ll)
ll)
c,
CIl
ll)
• 0
00 0
0
0
0
a 50
0
Cd
o :.o,.-.9.0H~~.o
[i:
o~ .,<J'. ~ ~
..0
0
~
0
0
0
0.6 0.8 1.0 I.2 1.4 1.6 1.8
Equivalence Ratio

FIGURE 12 Flame speeds from alkanes to alkynes for Cz and C, species. Theacetylene and
ethylene data are taken from Egolfopoulos et al. (1990), and the propane data are taken from
Vagelopoulos et al. (1994).

seen that, for a given number of total carbons, the flame speed increases
with increasing degree of un saturation in the order of alkynes > alkene-
s > alkanes. Also, based on the ethylene results obtained by Egolfopoulos
et al. (1990), the flame speeds of propyne and ethylene in nitrogen-diluted
(18% O 2 in N 2) mixtures are fairly similar (Fig. 13), except for rich equi-
valence ratios. Similar results were also observed by Gibbs and Calcote
(1959) and Gerstein et al. (1951) for propyne- and ethylene-air mixtures. If
this trend can be assumed to apply to the present analysis, then the effect of
unsaturation is greatest for smaller hydrocarbons, and decreases as the
carbon number increases, as observed by Gerstein et al. (1951).
This is further emphasized in the C 4 family where the effect of
unsaturation can be studied for multiple double bonds as well as for linear
and branched molecules. Figure 14 shows the laminar flame speeds of
n-butane, iso-butane, I-butene, iso-butene, and I ,3-butadiene. It is seen that
the more unsaturated the molecule, the higher the flame speed. Further-
more, similar to saturated molecules, branching reduces the flame speeds of
unsaturated molecules as shown by the difference between I-butene and iso-
butene. The effect of branching on the C 4 alkenes is again predominantly
caused by differences in oxidation kinetics, for their adiabatic flame
temperatures and thermal diffusion coefficients are very comparable. It is
444 S. G. DAVIS AND C. K. LAW

80

70 • propyne (18%0, in N,)

0 ethylene (18%0, in N,)
'""'
</)
..... 60
E
o
'-' 50 0
0
o "
'" 40 • • • •
0
0

] ~
•
0.. 30 ~
tI)
• 0

~ 20 ~
•
•
0
~ 0
10

0
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
FIGURE 13 Flame speeds of ethylene and propyne in nitrogen diluted air mixtures (18% 0,
in N,). The ethylene data are taken from Egolfopoulos et al. (1990).

60
• ••• •
50 •• ••
'""'
ell
..... • 0000 •
8
• 00 00
(,)
'-'
40 • o"'~ooooo..,.
"'."'."'."...... 0 0 •
0"," • 0,. o
"'t 0 Ct.

]
30
•
o..,~ 0 ~. o •
0.. o~ 0
",. •
tI)
0
20 ~ •
o
1,3-butadiene
0 ",.
0

~
l-butene ~
~ 10 •
0
iso-butene
iso-butane
0
"'0

n-butane
'"
0
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
FIGURE 14 Flame speeds of various alkanes and alkenes in the C. family.

difficult, however, to determine if the higher flame speeds of the normal

diene (1,3-butadiene) compared to the normal alkene (I-butene) are based
solely on their higher adiabatic flame temperatures or are coupled to their
oxidation kinetics as well.
 HYDROCARBON FLAME SPEEDS 445

Flame speeds of some simple aromatics were also compared in the pre-
sent study. The results of benzene and toluene are plotted along with the
saturated ring compound cyclohexane (Fig. 15). It is seen that the un-
saturated aromatic ring has higher flame speeds than the saturated ring
compound. The reason for the higher flame speeds is not trivial and is
possibly caused by differences in their oxidation kinetics and flame
temperatures, which are inherently coupled. It is also noted that the
substitution of a hydrogen by the methyl group on the aromatic ring
decreases the flame speed quite considerably, similar to methyl substitu-
tion on normal alkanes. This effect is largely caused by differences in the
oxidation kinetics because the two fuel/air mixtures have similar adiabatic
flame temperatures and insignificant differences in thermal diffusivities.
The laminar flame speeds of methanol and ethanol were previously
determined using the counterflow twin-flame technique (Egolfopoulos et al.,
1992a, I992b) and are compared with the corresponding alkanes in
Figure 16. The data for the alcohols were experimentally obtained at higher
temperatures and were extrapolated to their respective values at 298 K
shown in the figure. The comparison shows that the effect of hydroxyl
substitution of hydrogen is to enhance the laminar flame speeds for
methanol (as compared to methane) which is consistent with the result of
Gibbs and Calcote (1959), while having minor differences for ethanol as
compared to ethane.

........
til 40
• •••
••••••
...... •• 0 0 0 0•0•· ·
•
8u • 0 0.
'-" •• 0 o·
o ..,"
30
•• 0 ••
• ~o •
'"~c.. 20 ~
CIl •
••
<L>

~ • benzene
ti:
10 0 toluene •
• cyclohexane

0
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
FIGURE 15 Flame speeds of simple aromatics.
446 S. G. DAVIS AND C. K. LAW

60

r-. 50 0 0

--....-8
til

.. -.
00
~o. 0 0
·0
(.) 40 0
•
.~ •
0","
0 0

• • 0
.0·•
30
"2Il) • 0
0.
en 20 ••
~ •• 0
11)

~
0
O. • methane • •
u:: 0 methanol •
10 • • ethane •
0 ethanol
,
0
0.6 0.8 1.0 1.2 1.4 1.6 1.8
Equivalence Ratio
FIGURE 16 Flame speeds of simple alcohols and their respective aliphatic parent molecules.
The methanol. ethanol. methane. and ethane data are taken from Egolfopoulos et al. (l992a.
1992b). Vagelopoulos er al. (1994), and Egolfopoulos er al. (1990). respectively.

For completeness in presentation and compilation, Figure 17 shows the

laminar flame speeds of hydrogen/air mixtures measured by Egolfopoulos
and Law (1990) and Vagelopoulos et al. (1994) who employed the coun-
terflow twin flame technique, and by Aung et al. (1998) who used outwardly-
propagating spherical flames. Due to stretch compensation, both techniques
yield very similar results and inclusion of the latter study extends the experi-
mental range of equivalence ratios to ¢ = 4.0.
Finally, the influence of the oxidation kinetics on the flame speed is
discussed. Specifically, the flame speed may be influenced by differences in
the reactivity of the intermediates formed during the fuel oxidation. For
example, the flame speeds for ethylene/diluted-air (Tad = 2223 K) and
propene/air (Tad = 2342 K) are similar, despite the 100K difference in flame
temperature. Thus at the same flame temperature, ethylene has a higher
flame speed than propene. The difference may be explained by the higher
reactivity of the vinyl radical (major radical formed in ethylene oxidation),
as compared to the allyl radical (major radical formed in propene
oxidation). Allyl, being a resonantly stabilized radical, is more stable than
vinyl against reaction with molecular oxygen or thermal decomposition
leading to the production of the H-atom. It should be noted that the
 HYDROCARBON FLAME SPEEDS 447

300
• Hz/Air Mixture
0
250 0
~
Vl • 0
...... •
ao 200
~ 0
0
a
'-"
...,::s •
• 0
"0 150 • 0

"
Q)
Q)
0.. •
••
til 100
Q)

§ •
u: 50 <! • counterflow flame

0
,•
• 0 propagating spherical flame

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5

Equivalence Ratio
FIGURE 17 Hydrogen flame speeds. The counterflow data are taken from Egolfopoulos and
Law (1990) and Vagelopoulos et al. (1994), and the spherical flame data are taken from Aung
et al. (1998).

reaction between vinyl and oxygen is a secondary chain-branching reaction,

which promotes the rate of flame propagation.
Differences in the flame speed of propene and ethylene may also ex-
plain why linear alkanes have higher flame speeds than branched alkanes.
Specifically, linear alkanes produce a larger quantity of ethylene during
combustion than branched-chain alkanes. The latter produce mainly
branched-chain or large alkenes (e.g., propene and iso-butene).
Comparing toluene and benzene, the primary radicals during oxidation
are benzyl and phenyl, respectively. Benzyl is more resonantly stable than
phenyl. Its reactivity towards molecular oxygen is lower than that of phenyl,
and may be the cause of lower flame speeds for toluene. Comparing
methanol and methane, methoxy is much less stable than methyl, because
methoxy can quickly decompose to formaldehyde and the H atom, while a
similar decomposition of methyl is insignificant in flames. For this reason,
methanol has a higher flame speed than methane.
The above discussion points to the fact that the kinetic effects on flame
speeds can be mostly attributed to the stability/reactivity of the initial
radicals generated in fuel oxidation. Specifically, if this initial radical is
resonantly stabilized or is unable to further decompose rapidly to generate
the most reactive H-atoms, one should expect a slowdown in the flame
propagation rate.
448 S. G. DAVIS AND C. K. LAW

SUMMARY

Laminar flame speeds, determined by the counterflow twin flame configura-

tion, were compiled from the present study and those taken from previous
works. The compilation includes species from the C\ to C, compounds of
alkanes, alkenes, alkynes, aromatics, and alcohols. It is shown that the laminar
fla'me speeds of the normal alkanes are very comparable throughout the entire
range of equivalence ratios studied, except those for methane which are
consistently lower. Structural branching lowers the flame speeds for both
alkanes and alkenes, and it was determined that this effect was a result of
differences in the oxidation kinetics of normal and branched hydrocarbons.
For a given carbon number, the further from saturation the higher the flame
speed such that alkynes > alkenes > alkanes.

Acknowledgments

The authors gratefully acknowledge the support of this research by AFOSR

under the technical monitoring of Dr. Julian M. Tishkoff. We would also
like to thank Dr. C. J. Sung of Princeton University and Professor Hai
Wang of the University of Delaware for helpful discussions.

References

Aung, K. T., Hassan, M. I. and Faeth, G. M. (1998) Effects of Pressure and Nitrogen Dilution
on Flame/Stretch Interaction of Laminar Premixed H2/O,jN2 Flames. Combust. and
Flame, 112, I.
Chao, B. H., Egolfopoulos, F. N. and Law, C. K. (1997) Structure and Propagation of
Premixed Flame in Nozzle-Generated Counterflow. Combust. and Flame, 109,620.
Davis, S. G., Wang, H., Brezinsky, K. and Law, C. K. (1996) Laminar Flame Speeds and
Oxidation Kinetics of Benzene-Air and Toulene-Air Flames. Twenty-Sixth Symposium
(International) on Combustion, The Combustion Institute, Pittsburgh, p. 1025.
Davis, S. G. and Law, C. K. (1998) Laminar Flame Speeds and Oxidation Kinetics of iso-
Octane/Air and n-Heptane/Air Flames. Twenty-Seventh Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, in press.
Davis, S. G., Law, C. K. and Wang. H. (1998) An Experimental and Kinetic Modeling Study
of Propyne Oxidation. Twenty-Seventh Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, in press.
Dixon-Lewis, G. (1990) Structure of Laminar Flames. Twenty-Third Symposium (International)
on Combustion, The Combustion Institute, Pittsburgh, p. 305.
Egolfopoulos, F. N., Zhu, D. L. and Law, C. K. (1990) Experimental and Numerical
Determination of Laminar Flame Speeds: Mixtures of CrHydrocarbons with Oxygen
and Nitrogen. Twenty-Third Symposium (In/emotional) on Combustion, The Combustion
Institute, Pittsburgh, p. 471.
Egolfopoulos, F. N. and Law, C. K. (1990) An Experimental and Computational Study of the
Burning Rates of Ultra-Lean to Moderately-Rich H2 /0 2/N2 Laminar Flames with Pressure
Variations. Twenty- Third Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, p. 333.
 HYDROCARBON FLAME SPEEDS 449

Egolfopoulos, F. N. (1992) Geometric and Radiation Effects on Steady and Unsteady Strained
Laminar Flames. Twenty-Fifth Symposium (International) on Combustion, The Combus-
tion Institute, Pittsburgh, p. 1375.
Egolfopoulos, F. N., Du, D. X. and Law, C. K. (I 992a) A Comprehensive Study of Methanol
Kinetics in Freely-Propagating and Burner-Stabilized Flames, Flow and Static Reactors,
and Shock Tubes. Combust. Sci. and Tech., 83, 33.
Egolfopoulos, F. N., Du, D. X. and Law, C. K. (l992b) A Study on Ethanol Oxidation Kinetics
in Laminar Premixed Flames, Flow Reactors, and Shock Tubes. Twenty-Fourth
Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, p. 833.
Gerstein, M., Levine, O. and Wong, E. L. (1951) The Determination of Fundamental Burning
Velocities of Hydrocarbons by a Revised Tube Method. J. Am. Chern. Soc., 73, 418.
Gibbs, G. J. and Calcote. H. F. (1959) Effect of Molecular Structure on Burning Velocity.
J. Chern. Eng. Data, 4, 226.
Glassman, I. (1996) Combustion, 3rd ed., Academic Press, San Diego, California.
Held, T. J. (1993) The oxidation of methanol, isobutene and methyl tertiary-butyl ether, Ph. D.
dissertation, Princeton University.
Law, C. K. (1988) Dynamics of Stretched Flames. Twenty-Second Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, p. 1381.
Matalon, M. and Matkowsky, B. J. (1982) Flames as Gasdynamic Discontinuities. J. Fluid
Mech., 124,239.
Sivashinsky, G. I., Law, C. K. and Joulin, G. (1982) On Stability of Premixed Flames in
Stagnation-Point Flow. Combust. Sci. Technol., 28, 155.
Sun, C. J., Sung, C. J., Wang, H. and Law, C. K. (1996) On the Structure of Nonsooting
Counterflow Ethylene and Acetylene Diffusion Flames. Combust. and Flame, 107,321.
Tien, J. H. and Matalon, M. (1991) On the Burning Velocity of Stretched Flames. Combust. and
Flame, 84, 238.
Vagelopoulos, C. M., Egolfopoulos, F. N. and Law, C. K. (1994) Further Considerations on
the Determination of Laminar Flame Speeds With the Counterflow Twin Flame
Technique. Twenty-Fifth Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, p. 1341.
Wu, C. K. and Law, C. K. (1984) On the Determination of Laminar Flame Speeds from
Stretched Flames. Twentieth Symposium (International) on Combustion, The Combustion
Institute, Pittsburgh, p. 1941.