Effects of The Burner Diameter On The Flame Structure and Extinction Limit of Counterflow Non-Premixed Flames

International journal of spray and combustion dynamics · Volume .2 · Number . 3 .
2010 – pages 199–218 199
Effects of the burner diameter on the

flame structure and extinction limit of
counterflow non-premixed flames
Chang Bo Oh1,*, Eui Ju Lee1 and Jeong Park2
1Division
of Safety Engineering, Pukyong National University, San 100, Yongdang-dong, Nam-gu,
Busan 608-739, South Korea
2Department of Mechanical Engineering, Pukyong National University, San 100, Yongdang-dong,
Nam-gu, Busan 608-739, South Korea

Received September 21, 2009; Revised submission December 20, 2009; Accepted January 05, 2010
ABSTRACT
Experiments and numerical simulations were conducted to investigate the effects of the burner
diameter on the flame structure and extinction limit of counterflow non-premixed methane
flames in normal gravity and microgravity. Experiments were performed for counterflow flames
with a large inner diameter (d ) of 50 mm in normal gravity to compare the extinction limits with
those obtained by previous studies where a small burner (d < 25 mm) was used. Two-dimensional
(2D) simulations were performed to clarify the flame structure and extinction limits of
counterflow non-premixed flame with a three-step global reaction mechanism. One-dimensional
(1D) simulations were also performed with the same three-step global reaction mechanism to
provide reference data for the 2D simulation and experiment. For microgravity, the effect of the
burner diameter on the flame location at the centerline was negligible at both high (ag = 50 s−1)
and low (ag = 10 s−1) strain rates. However, a small burner flame (d = 15 mm) in microgravity
showed large differences in the maximum flame temperature and the flame size in radial
direction compared to a large burner flame (d = 50 mm) at low strain rate. In addition, for normal
gravity, a small burner flame (d = 23.4 mm) showed differences in the flame thickness, flame
location, local strain rate, and maximum heat release rate compared to a large burner flame
(d = 50 mm) at low strain rate. Counterflow non-premixed flames with low and high strain rates
that were established in a large burner were approximated by 1D simulation for normal gravity
and microgravity. However, a counterflow non-premixed flame with a low strain rate in a small
burner could not be approximated by 1D simulation for normal gravity due to buoyancy effects.
The 2D simulations of the extinction limits correlated well with experiments for small and large
burner flames. For microgravity, the extinction limit of a small burner flame (d = 15 mm) was
much lower than that of a large burner flame when ag ≤ 20 s−1. For normal gravity, the extinction
limit of a small burner flame (d = 23.4 mm) was also much lower than that of the large burner
*Corresponding author. E-mail address: cboh@pknu.ac.kr Tel: +82-51-629-6472

Fax: +82-51-629-6463
200 Effects of the burner diameter on the flame structure and extinction limit of
flame when ag ≤ 35 s−1. The effects of the burner diameter on the flame structure and extinction
limit of counterflow non-premixed methane flames were more important in normal gravity than
in microgravity.
Key words: counterflow burner; counterflow flame; burner diameter; extinction limit; non-
premixed flame; simulation; multi-dimensional effects.
1. INTRODUCTION
The counterflow has a very well-defined geometry in fluid dynamics; a flame
established in the flow field can be modeled by a one-dimensional (1D) flame using the
similarity approach. The strain rate of the counterflow flame, derived from the flame
stretch concept, is an important parameter in controlling overall flame behavior. Thus,
it is commonly used to model the local strained flame structure of turbulent premixed
and non-premixed flames perturbed by eddies [1–4]. Experiments and numerical
simulations of counterflow flames also provide useful insights into the fundamentals of
flame physics and chemistry; these include the flame structure, ignition, extinction,
edge flame, and chemical kinetics of laminar flames [1, 2, 5–10]. Since the counterflow
flame is extensively used in combustion research, the counterflow burner configuration
is also well-defined.
To date, two types of counterflow burners have been utilized in combustion research.
Tsuji suggested a cylindrical- or spherical-type counterflow configuration [1, 11, 12]. In
a Tsuji-type counterflow burner, also known as a “Tsuji burner,” a non-premixed flame
is established between the fuel stream issued from the porous media of the cylindrical
or spherical surface, and a uniform stream of ambient air flows to the surface. In the
Tsuji burner, the strain rate is changed simply by varying the uniform air velocity or
cylinder burner diameter. The second type of counterflow burner consists of two ducts
that are vertically and coaxially arranged; the counterflow flame is established between
the two ducts [2, 4–10, 13–15]. In this type of burner, known as the “opposed-jet
burner,” the inner ducts generally have outer annual ducts from which an ambient
curtain fluid such as nitrogen is issued. In an effort to flatten the exit velocity profile,
contraction nozzle–type ducts are usually installed, or glass beads and meshes are
placed inside the ducts of the opposed-jet burner system. In an opposed-jet burner,
the strain rate is easily changed by varying the separation distance of the burner ducts
or the inflowing fuel and air velocities. Of the two types of burner, the opposed-jet
burner, which adopts two ducts or contraction nozzles, has been more widely used in
combustion research in recent years [2, 4–10, 13–15]. The inner diameter (d) of the duct
or nozzle of the burner system is determined according to each research situation when
the opposed-jet burner is used. A lateral flame size that is 10 times the flame thickness
is implicitly known to be sufficient to ensure the one-dimensionality of the flame [12].
The counterflow flame extends radially, and the flame thickness decreases with an
increase in the strain rate; the lateral flame size then becomes sufficiently large
compared to the flame thickness. Thus, there were no obvious problems in previous
studies on counterflow flame since almost all of them focused on cases with relatively
high strain rates [2, 4–8].
International journal of spray and combustion dynamics · Volume .2 · Number . 3 . 2010 201
Recently, it was reported that a low-strain-rate counterflow flame can be affected

by multi-dimensional flow and the heat loss effects of normal gravity (1 g) [13] or
microgravity (0 g) [14] when an opposed jet–type counterflow burner is utilized. A
comparison of simulations with experiments on a counterflow flame showed that 1D
simulations cannot properly predict the flame structure and extinction limit at low
strain rates in microgravity when a counterflow burner with a finitely small diameter
is used [14]. For normal gravity, the flame structure at the axisymmetric center shows
a multi-dimensional structure, not a 1D one. This multi-dimensional structure affects
the extinction mode and limit of the counterflow flame [13]. These results imply that
the extinction limit and flame structure of a counterflow flame established in a finite-
size opposed-jet burner may be different from those of a 1D counterflow flame obtained
using the similarity approach. In these studies, however, burner size effects on the flame
structure and extinction characteristics of counterflow nonpremixed flames were not
investigated with different diameter burners. Extinction behavior and edge flame
oscillation in which the flame length is less than the burner diameter were investigated
at low strain rates experimentally [15]. The study showed that the flame length is
relevant to lateral heat loss and thus affects the flame extinction and edge flame
oscillation. To the best of our knowledge, the effects of the burner diameter on the
structure and extinction limits of an opposed-jet counterflow flame in normal gravity
and microgravity were not explicitly clarified, although the relation of the lateral flame
size and flame thickness with one-dimensionality has been briefly mentioned in the
literature [12].
In this study, in an effort to clarify the effects of the burner diameter on the flame
structure and extinction limit, we reexamined the results obtained in previous works
with small diameter burners (d < 25 mm) [13, 14]. Two-dimensional (2D) simulations
in normal gravity and microgravity and experiments in normal gravity were performed
for counterflow flames established in a large diameter burner (d = 50 mm). The 2D
simulation and experimental results were compared with those from a 1D simulation.
The detailed effects of the burner diameter on the flame shape and location, local flame
strength, and extinction limit of the counterflow flame are discussed in this paper with
the results from the simulations and experiments.
2. EXPERIMENTS
Experiments on counterflow flames were performed using an opposed-jet burner with
an inner duct diameter (d) of 50 mm and a duct separation distance (L) of 15 mm.
Stainless steel screens were positioned at the duct exits to impose a flat velocity profile.
A water-cooled jacket was installed on the top burner duct to prevent heating of the
metal burner and preheating of the supplied reactants. The fuel, air, and suppression
agent (nitrogen) flows were controlled using mass flowmeters. The details of the burner
system and experimental methods except the burner size are identical to those in
previous studies excluding the burner size [13, 14].
A stable laminar non-premixed methane flame was established under conditions
where the nitrogen mole fraction in the fuel stream (Ca) was 0.79 and then Ca was
adjusted from 0.79 to the critical nitrogen mole fraction at extinction (hereafter known
as the “extinction limit”). Experimental extinction limits as a function of the global
strain rate for methane-air flames in normal gravity were compared with those from
simulations and previous studies [8]. In this study, the global strain rate, hereafter
known as the strain rate ag, was defined as [16]:
2VO  VF ρF  2VrVF  ρF 
ag = 1 + = 1 +  (1)
L  VO ρO  L  Vr ρO 
where the parameters V and ρ denote the velocity and density of the reactant streams at
the duct exit, respectively. L is the duct separation distance, and the subscripts O and F
represent the air and fuel streams, respectively. The velocity ratio Vr is defined as equal
to VO /VF. The velocity ratio was fixed to unity for when d = 50 mm. In addition, the
outflow velocity in lower outer duct was fixed to a very low value of 0.03 m/s to exclude
the outflow velocity effects on the flame structure and extinction limit.
3. NUMERICAL METHODS
A time-dependent axisymmetric configuration was employed for the 2D simulations in
this study to treat the counterflow non-premixed flames that form between two opposed
circular ducts. The governing equations of the simulation were the continuity,
momentum, species, and energy equations. The detailed numerical scheme can be found
elsewhere [3, 13].
The 2D simulations were performed for 1 and 0 g counterflow non-premixed flames
formed between the two ducts with d = 50 mm. In this paper, the term, microgravity,
represents the zero-gravity for the simulations as well as real microgravity for the
experiments. Please be noted that the terms, microgravity and 0 g, are used
interchangeably hereafter. Additional 2D simulations with Vr = 1, as in previous works
[13, 14], were also performed for the counterflow flames formed in the burners; the
inner diameters of the burners were 15 and 23. 4 mm for 0 g and 1 g, respectively. The
simulation method and domain for d = 15 and 23.4 mm were identical to those of
previous works [13, 14]. For 1 and 0 g, the simulation domain for d = 50 mm burner
was –27.6 mm ≤ x ≤ 42.6 mm and 0 mm ≤ r ≤ 140 mm in the axial and radial directions,
respectively. A grid system consisting of 468 × 138 grid points was used for the 2D
simulations. A uniform grid size of 0.15 mm was used in the axial direction. Uniform
grids with a size of 0.3 mm were imposed in the region less than the radius of the burner
duct, and the grid size was expanded radially from the radius of the burner duct. The
grid size was confirmed as conforming to previous works by comparing the computed
extinction limit and maximum temperature with those from experiments [13].
Figure 1 shows the schematic of the burner geometry and simulation domain adopted
in this study. The inflow temperature boundary condition for the reactant and curtain
streams was 298 K. A uniform axial velocity was imposed at the exit of each burner
duct. An axial velocity of 0.03 m/s was imposed at the lower (side A) and upper
boundaries (side C) for the 0 g simulation. The upper boundary of side C, shown in
Figure 1, was treated as an outflow boundary and an axial velocity of 0.03 m/s was
imposed at the lower boundary (side A) for 1 g as in experiments. A no-slip condition
Side C
Product
Air 2D simulation
gas
domain
Flame
L
g d
N2
N2 CH4+N2
Side B
Side A
Figure 1: Schematic of the burner geometry and simulation domain.
was applied on the burner walls, and the wall temperature was taken as 298 K. For all
the species equations, a zero-gradient condition was applied to all boundaries. The air
stream was composed of 21% oxygen and 79% nitrogen by volume, and the fuel stream
was composed of methane diluted by nitrogen. The ambient stream was set to pure
nitrogen as in the experiment to prevent secondary combustion of the fuel.
Taking the computational cost into account, a three-step irreversible reaction
mechanism [17] for methane oxidation was used. The reaction rates of the three-step
reaction mechanism were slightly modified to predict the flame feature more
reasonably. The reaction model was validated by comparing the flame temperature and
extinction limits obtained by simulations with those from experiments in the previous
study. For radiative heat loss, an optically thin radiation model was implemented; the
model considered heat losses from CH4, CO2, H2O, and CO [18]. The 2D simulation
code adopted in this study has already been validated for the counterflow flame
structure [3] and extinction limits [13, 14], and it was further validated for the jet flame
structure [19] by comparing the experimental results.
A previously developed flamelet code, OPPDIF [20], was used in the 1D simulations.
The three-step reaction mechanism and optically-thin radiation model—identical to the
2D simulations—were also adopted for the 1D simulations. The 1D simulation
represented the 0 g flame since the buoyancy effects were essentially not incorporated
into the simulation.
4. RESULTS AND DISCUSSION

4.1. Heat loss mechanism of counterflow non-premixed flame
In previous studies [13, 14], the extinction limits of CH4 flames as functions of the strain
rate in normal and micro-gravity have been discussed in detail. The extinction limit
Critical mole fraction at extinction,Caext 0.90
0.85
Vr = 3 Vr = 2 Vr = 1
0.80
Vr = 4
0-g, Experiment, d = 20 mm22
0-g, Experiment, d = 15 mm14
1-g, Experiment, d = 23.4 mm23
0.75
0-g, 2D simulation, d = 15 mm14
1-g, 2D simulation, d = 23.4 mm13
0-g, 1D simulation, OPPDIF
0.70
0 20 40 60 80 100
Global strain rate (s−1)
Figure 2: Measured and computed critical nitrogen mole fraction in the fuel stream
for extinction in 0 g and 1 g flames.
increases with decreasing the strain rate until a critical value is obtained. As the strain
rate is further reduced, the extinction limit decreases again. The critical maximum point
of the extinction limit is called the “turning point” [21, 22]. As shown in Figure 2,
experiments and simulations showed that the extinction limit for the 0 g counterflow
flame was higher than that for the 1 g flame at low strain rates due to buoyancy effects
and the multi-dimensional flame structure. The velocity ratios for the 1 g flames were
different for each strain rate, and thus the values of the velocity ratios were included in
the Figure 2. It is noted that the extinction limit for the 0 g counterflow flame obtained
in the experiments was lower than that for the 1D simulation; this was attributed to the
lateral heat loss of flame formed in finite small diameter burners [13, 14].
In an effort to investigate more clearly the lateral heat loss features of a counterflow
non-premixed flame, we present the 2D simulation results of the transport mechanisms
of the heat generated by the combustion reaction of a 0 g counterflow non-premixed
flame; the flame was formed in a d = 15 mm burner at ag = 8 and 40 s−1, as shown in
Figures 3 and 4. In the figures, the solid lines denote the temperature iso-contour lines,
and the arrows denote the heat flux vectors by each transport mechanism. The
temperature difference is 200 K between neighboring iso-contour lines. The high heat-
release region roughly coincides with the high-temperature region above 1600 K,
although the heat release region is not plotted in the figures. The figures show that the
Temp.(K) Temp.(K)
20 1600 20 1600
1400 1400
x (mm)
x (mm)
10 10
0 0
−10 50000(J/m3 s) −10 50000(J/m3 s)

0 10 20 30 0 10 20 30
r (mm) r (mm)
(a) Convection (b) Conduction
Figure 3: Flame temperature and heat transport vectors for the 0 g counterflow non-
premixed flame at ag = 8 s−1 with Vr = 1 and Ca = 0.79. The size of the
arrow shown in the lower part of each figure means the magnitude of the
heat transport vector.
Temp.(K) Temp.(K)
1600 1600
20 20
1400 1400
x (mm)
x (mm)
10 10
0 0
−10 50000(J/m3 s) −10 50000(J/m3 s)

0 10 20 30 0 10 20 30
r (mm) r (mm)
(a) Convection (b) Conduction
Figure 4: Flame temperature and heat transport vectors for 0 g counterflow non-
premixed flame at ag = 40 s−1 with Vr = 1 and Ca = 0.79. The size of the
arrow shown in the lower part of each figure means the magnitude of the
heat transport vector.
heat was transported toward the heat release region by axial convection. The heat was
also transported in the radial direction by convection. Using the heat energy budget for
local points, the amount of heat energy left from the flame region, where the flame
temperature is over 1400 K, by radial convection was not significant. The generated heat
energy left the flame region was mainly due to the conduction process near the
axisymmetric centerline. Near the outer flame edge, however, the heat energy left the
flame region by axial as well as radial (lateral) conduction. The flame size in the radial
direction increased with the strain rate. It has been reported that for non-premixed
flames, the amounts of heat generation and axial conduction loss from the heat release
region are well balanced [24]. These two quantities increase with the strain rate but are
still well balanced. Thus, at high strain rates, the radial conductive heat loss from a heat
release region becomes negligible compared with the axial conduction heat loss. In
addition, the radial conduction heat loss at low strain rates plays a much more
important role in the heat loss mechanism from the overall heat release region; this is
because the flame size is small, as shown in Figures 3(b) and 4(b). The reason that the
extinct limits for the 0 g flame obtained by experiment and 2D simulation were lower
than that for 1D simulation was attributed to the multi-dimensional heat loss featured
in counterflow non-premixed flame with low strain rates, as shown in the previous
study [14].
The flame size in the radial direction became small as the inner diameter of the
counterflow burner ducts decreased while the duct separation distance and strain
rate remained fixed. This implies that a counterflow flame formed in a smaller duct
burner can be more sensitive to lateral heat loss even for a fixed strain rate. To represent
the effects of the burner diameter explicitly, a comparison of the experimental and 2D
simulation results for the extinction limits and flame structures of the non-premixed
flames formed in large and small diameter burners are presented in the following section.
4.2. Extinction characteristics

Figure 5 shows the trend in the extinction limits for 0 g and 1 g counterflow non-
premixed flames with varying strain rates and inner diameters for the burner ducts. The
extinction limits obtained by 2D simulations for 0 g and 1 g counterflow non-premixed
flames agreed very well with experiments in the previous study [13, 14]. As mentioned
earlier, the extinction limit for a 0 g counterflow non-premixed flame formed in a
d = 15 mm burner (case B) was lower than that for a 0 g 1D flame (case A). In addition,
the extinction limit of a 1 g counterflow non-premixed flame with d = 23.4 mm (case C)
was lower than that for a 0 g flame with d = 15 mm (case B). However, the extinction
limit for a 0 g counterflow non-premixed flame formed in a d = 50 mm burner (case D)
was nearly identical to that of a 0 g 1D flame (case A). In addition, it should be noted
that the extinction limit for a 1 g flame formed in a d = 50 mm burner (case E)
approached that of a 0 g 1D flame (case A). The experimental and 2D simulation results
for the extinction limits of a 1 g counterflow non-premixed flame formed in a d = 50
mm burner agreed well with each other. For microgravity, the extinction limit of a small
burner flame (case B) was much lower than that of a large burner flame (case D) when
ag ≤ 20 s−1. For normal gravity, the extinction limit of a small burner flame (case C) was
also much lower than that for a large burner flame (case E) when ag ≤ 35 s−1. The effect
of the burner diameter on the extinction limit of a counterflow non-premixed methane
flame was larger in normal gravity than in microgravity. Figure 5 explicitly shows the
effect of the burner diameter on the extinction limit of counterflow non-premixed
0.90
Critical mole fraction at extinction, Caext
0.85
0.80
0-g, 1D simulation, OPPDIF (case A)

0-g, 2D simulation, d = 15 mm14 (case B)
0.75 1-g, 2D simulation, d = 23.4 mm13 (case C)

0-g, 2D simulation, d = 50 mm (case D)
1-g, 2D simulation, d = 50 mm (case E)
1-g, Experiment, d = 50 mm (case F)
0.70
0 20 40 60 80
Figure 5: Measured and computed critical nitrogen mole fraction in the fuel stream
for the extinction of 0 g and 1 g flames with varying burner inner diameters.
flames. The results imply that the extinction limits of counterflow non-premixed flames
formed in a sufficiently large burner for 0 g and even for 1 g can be approximated well
by 0 g 1D simulations.
Figure 6 shows the simulation results of the maximum flame temperature as a
function of the strain rate at the axisymmetric center of 0 g and 1 g flames when
Ca = 0.79. The velocity ratios, which varied from one to four, correspond to the
experimental conditions shown in Figure 2. The 2D simulation results for the
maximum flame temperature of a 0 g flame in a d = 15 mm burner (case B) was lower
than that for a 0 g 1D flame (case A) at a low strain rate; meanwhile, the maximum
flame temperatures of cases A and B were similar to each other at high strain rates. The
difference in the maximum flame temperatures for cases A and B increased with
decreasing strain rate. These trends were still effective for cases A and C. However, the
maximum flame temperature of a 0 g flame in a d = 50 mm burner (case D) was very
similar to that of a 0 g 1D flame at both high and low strain rates. Even the maximum
flame temperature for a 1 g flame in a d = 50 mm burner (case E) was similar to that for
a 0 g 1D flame. Consequently, figures 5 and 6 show that the effects of lateral heat loss
and buoyancy on the extinction limit and flame temperature became significant with
decreasing strain rates when a small-diameter burner duct was used. However, these
effects were negligible when a large-diameter burner duct was used.
1900
Maximum flame temperature at the centerline (K)
1800
1700
1600
0-g, 1D simulation, OPPDIF (case A)
0-g, 2D simulation, d = 15 mm (case B)
1-g, 2D simulation, d = 23.4 mm (case C)
1500
0-g, 2D simulation, d = 50 mm (case D)
1-g, 2D simulation, d =50 mm (case E)
1400
0 20 40 60 80
Figure 6: Computed 1D and 2D maximum flame temperatures as a function of the

global strain rate at the center of 0 g and 1 g flames when Ca = 0.79.
4.3. Flame structure

Figures 7 and 8 show the 2D simulation results of the flame temperature and stream
lines of 0 g counterflow non-premixed flames at ag = 10 and 50 s−1, respectively. The
figures represent the overall shapes and flow structures of counterflow non-premixed
flames according to each burner diameter. At ag = 10 s−1, the thickness of the high
temperature region on the centerline of the case when d = 15 mm case (red color) was
slightly narrower than when d = 50 mm. The thickness of the high temperature region
for d = 15 mm decreased radially, while that for d = 50 mm gradually increased radially.
In addition, the high temperature region for d = 15 mm did not go beyond the burner
rim, while that for d = 50 mm extended to the outer region of the burner rim radially.
Figure 7 shows that the counterflow flame structure is multi-dimensional and can be
affected by the size of the burner diameter. On the other hand, at ag = 50 s−1, the flame
shapes for d = 15 and 50 mm were nearly identical to each other at the centerline, as
shown in Figure 8. The flames extended to the outer region of the burner rim radially,
even though the flame size was smaller for d = 15 mm than for d = 50 mm in the radial
direction. The overall flame shape at the centerline showed that the heat loss effect due
to the radial flame size on the centerline flame structure was negligible at a high strain
rate of ag = 50 s−1, even when a d = 15 mm burner was used.
The 2D simulation results for the flame temperature and axial velocity profiles at the
centerline of d = 15 and 50 mm are shown in Figures 9 and 10. For comparison, the
Temperature (K)
503 656 809 962 1115 1268 1421 1574
20
x (mm)
10
−10
−30 −20 −10 0 10 20 30 40 50

r (mm)
Figure 7: Computed flame temperature and stream lines for 0 g counterflow non-
premixed flames at ag = 10 s−1, Vr = 1, and Ca = 0.79. LHS: d = 15 mm,
RHS: d = 50 mm.
Temperature (K)
503 656 809 962 1115 1268 1421 1574
20
x (mm)
10
−10
−30 −20 −10 0 10 20 30 40 50

r (mm)
Figure 8: Computed flame temperature and stream lines of 0 g counterflow non-

premixed flames at ag = 50 s−1, Vr = 1, and Ca = 0.79. LHS: d = 15 mm,
RHS: d = 50 mm.
2100 0.12
0-g, 1D, OPPDIF
1800 0-g, 2D, d = 15 mm
0.08
0-g, 2D, d = 50 mm T
1500
U 0.04
Axial velocity (m/s)

Temperature (K)
1200
0.00
900
−0.04
600
−0.08
300
0 −0.12
0 2 4 6 8 10 12 14
X (mm)
Figure 9: Computed flame temperature and axial velocity profiles at the

axisymmetric centerline of 0 g counterflow non-premixed flames at
ag = 10 s−1, Vr = 1, and Ca = 0.79.
2100 0.30
0-g, 1D, OPPDIF
1800
T 0-g, 2D, d = 15 mm
0.20
0-g, 2D, d = 50 mm
1500 U
0.10
Temperature (K)
1200
0.00
900
−0.10
600
−0.20
300
0 −0.30
0 2 4 6 8 10 12 14
X (mm)
ag = 50 s−1, Vr = 1, and Ca = 0.79.
Temperature (K)
515 679 844 1009 1174 1338 1503 1668
20
x (mm)
10
−10
−30 −20 −10 0 10 20 30 40 50

r (mm)
premixed flames at ag = 17 s−1, Vr = 1, and Ca = 0.79. LHS: d = 23.4 mm,
RHS: d = 50 mm.
results for the 0 g 1D flame are also plotted in the figures. In Figure 9, the flame
temperature and axial velocity profiles for the 0 g 1D flame and d = 50 mm agreed very
well with each other. However, the maximum temperature for when d = 15 mm was
slightly lower than for d = 50 mm or the 0 g 1D flame. This was attributed to the fact
that the amount of radial heat loss when d = 15 mm increased compared to when
d = 50 mm due to the smaller flame size, as shown in Figure 7. The flame temperature
and axial velocity profiles for d = 15 mm, d = 50 mm, and the 1D flame coincided very
well when there was a high strain rate of ag = 50 s–1. This implies that for high strain
rates, the flame structure at the centerline can be approximated well by 1D simulation
even if a small-diameter burner is used as shown in Figure 10.
Figures 11 and 12 show the 2D simulation results of the flame temperature and
stream lines of 1 g counterflow non-premixed flames when ag = 17 and 70 s−1,
respectively. In Figure 11, in contrast to the 0 g flame, the centerline flame structure
when d = 23.4 mm moved discernibly upwards due to buoyancy. If the flame thickness
is defined as the full width at 5% of the maximum flame temperature, the flame
thickness when d = 23.4 mm was narrower than for d = 50 mm at a low strain rate of
ag = 17 s−1. The flame size represented by a high temperature region (red color) for
d = 23.4 mm was smaller than that for d = 50 mm. This implies that the small flame for
d = 23.4 mm was affected by lateral heat loss. When ag = 70 s−1, the centerline flame
location for d = 23.4 mm was moved slightly upward, although the overall flame shape
Temperature (K)
515 679 844 1009 1174 1338 1503 1668
20
x (mm)
10
−10
−30 −20 −10 0 10 20 30 40 50

r (mm)
premixed flames at ag = 70 s−1, Vr = 1, and Ca = 0.79. LHS: d = 23.4 mm,
RHS: d = 50 mm.
for d = 23.4 mm was different from the shape for d = 50 mm, as shown in Figure 12.
These flame structures can be easily identified by the flame temperature and axial
velocity profiles at the centerline shown in Figures 13 and 14. At the low strain rate of
ag = 17 s−1 shown in Figure 13, the temperature and axial velocity profiles when d = 50
mm shifted slightly to the oxidizer side (upper duct side) compared to the 0 g 1D flame.
On the other side, the temperature and axial velocity profiles for d = 23.4 mm shifted
greatly to the oxidizer side; the shift was greater compared to the other two cases due
to buoyancy. This means that the 1 g flame formed in a small burner with d = 23.4 mm
is greatly affected by buoyancy at low strain rates. The buoyancy effect on the flame
location in a d = 50 mm burner is not important. For a high strain rate of ag = 70 s−1,
shown in Figure 14, the buoyancy effects on the flame location decreased for both
d = 23.4 and 50 mm. The results show that the temperature and axial velocity profiles
for d = 50 mm were very close to those for a 0 g 1D flame, although the shift in the
temperature and axial velocity profiles for d = 23.4 mm is still discernible. This
implies that for both low and high strain rates, the buoyancy effects on the flame
structure including location are negligible; thus, the 1 g counterflow non-premixed
flame structure at the centerline in a d = 50 mm burner can be approximated by a 0 g
1D simulation.
Table 1 shows the important calculated parameters that characterize the flame
structure as a function of the burner inner diameter and strain rate. The local strain rate
2100 0.15
0-g, 1D, OPPDIF T
1800 1-g, 2D, d = 23.4 mm
1-g, 2D, d = 50 mm 0.10
U
1500
0.05

Temperature (K)
1200
0.00
900
−0.05
600
−0.10
300
0 −0.15
0 2 4 6 8 10 12 14
X (mm)
ag = 17 s−1, Vr = 1, and Ca = 0.79.
2100 0.40
0-g, 1D, OPPDIF
1800
T 1-g, 2D, d = 15 mm 0.30
1-g, 2D, d = 50 mm
U
0.20
1500
Temperature (K)
0.10
1200
0.00
900
−0.10
600
−0.20
300 −0.30
0 −0.40
0 2 4 6 8 10 12 14
X (mm)
ag = 70 s−1, Vr = 1, and Ca = 0.79.
Table 1: Calculated parameters at the axisymmetric centerline as a function of

the inner burner diameter and global strain rate with Vr = 1 and Ca = 0.79.
Global strain rate 10 s−1 50 s−1

Inner burner diameter (d) 15 mm 50 mm 15 mm 50 mm
Local strain rate (s−1) 19.2 18.7 83.1 90.8
Maximum HRR (J/cm3-s) 58.7 73.6 197.5 202.5
0-g Maximum temperature (K) 1651 1709 1757 1758
Flame thickness* (mm) 12.1 12.3 5.9 5.9
SMHRR (J/cm3) 3.1 3.9 2.4 2.2
Global strain rate 17 s−1 70 s−1
Inner burner diameter (d) 23.4 mm 50 mm 23.4 mm 50 mm
Local strain rate (s−1) 62.3 15.2 114.5 114.8
Maximum HRR (J/cm3-s) 146.2 111.8 235.6 243.1
1-g Maximum temperature (K) 1831 1748 1740 1746
Flame thickness* (mm) 7.9 9.4 5.0 5.0
SMHRR (J/cm3) 2.3 7.4 2.1 2.1
*The flame thickness is defined as the full width at 5% of the maximum flame temperature.
was evaluated at the location of the maximum heat release rate (HRR) of the flame at
the centerline. The specific maximum heat release rate (SMHRR) was defined as the
maximum heat release rate divided by the local strain rate [2]. It has been confirmed in
a previous study that the parameter SMHRR is a good indicator for the local flame
strength [13, 14]. A small SMHRR implies that the flame is weak and easy to
extinguish. For 0 g flames, the parameters shown in Table 1 for when d = 15 and 50 mm
are relatively similar to each other at high strain rate (ag = 50 s−1). However, the
parameters, especially the maximum temperature and heat release rate, are different
for flames in d = 15 and 50 mm burners at low strain rate (ag = 10 s−1). The SMHRR
value showed that a flame in a d = 15 mm burner can be extinguished more easily
than a d = 50 mm burner at low strain rate (ag = 10 s−1). On the other side, at high
strain rate (ag = 50 s−1), the flame in a d = 15 mm burner is not extinguished more easily
comparing to d = 50 mm flames. The extinction limits according the burner diameter
are consistent with the results shown in figure 5; this is because the flame strength is
directly related the extinction limit. For 1 g flames, the parameters for when d = 23.4
and 50 mm are very similar to each other when there is a high strain rate of ag = 70 s−1.
This means that the effect of the burner diameter on the centerline flame structures and
extinction limits are negligible at high strain rate. However, the parameters when
d = 23.4 mm were different from when d = 50 mm case at low strain rates (ag = 17 s−1).
The SMHRR value when d = 23.4 mm case was much lower than when d = 50 mm,
even though the maximum temperature when d = 23.4 mm was higher than when
d = 50 mm. This is because the local strain rate increment is much larger than the
increment of the maximum heat release rate as the burner diameter decreases. This is
also due to buoyancy effects—which are closely related to the burner size—on the
flame structure, such as the local heat release rate and strain rate mentioned earlier.
Thus, the extinction limit when d = 23.4 mm was much lower than when d = 50 mm at
low strain rates; this extinction feature also correlates well with the results shown in
Figure 5. In addition, the result implies that the maximum flame temperature is not a
good indicator for the flame strength even in steady situations.
5. CONCLUDING REMARKS
A fundamental study was conducted to investigate the effects of the burner diameter on
the flame structure and extinction limit of counterflow non-premixed methane flames in
normal gravity and microgravity. One- and two-dimensional simulations were performed
to clarify the multi-dimensional flame structure and extinction limit of counterflow non-
premixed flames with a three-step global reaction mechanism.
The flame structure of counterflow non-premixed flames with respect to the flame
location and flame thickness was investigated through simulations. For microgravity, the
effect of the burner diameter on the flame location at the centerline was negligible at
both high (ag = 50 s−1) and low (ag = 10 s−1) strain rates. However, a small burner flame
(d = 15 mm) in microgravity showed large differences in the maximum flame
temperature and the radial flame size compared to a large burner flame (d = 50 mm) at
low strain rate (ag = 10 s−1). At normal gravity, the flame for a small burner (d = 23.4 mm)
showed differences in the flame thickness, flame location, local strain rate, and
maximum heat release rate compared to a large burner (d = 50 mm) at low strain rates
(ag = 17 s−1). However, the difference between the small and large burner flames became
negligible at high strain rates (ag = 50 s−1) excluding the flame location. For normal
gravity and microgravity, the flame structure of a counterflow non-premixed flame
established in a large burner (d = 50 mm) can be approximated fairly well by a 1D
simulation of the 0 g flame. For normal gravity, however, the structure of a low-strain-
rate counterflow non-premixed flame in a small burner could not be approximated by
the 1D simulation of a 0 g flame due to buoyancy effects.
The two-dimensional simulation results for the extinction limits agreed well with the
experimental results for small and large burner flames. The radial heat loss mechanism
was also confirmed by investigating the heat flux vectors of convection and conduction.
For microgravity, the extinction limit of a small burner flame (d = 15 mm) was much
lower than that of a large burner flame (d = 50 mm) when ag ≤ 20 s−1. For normal
gravity, the extinction limit of a small burner flame (d = 23.4 mm) was also much lower
than that of a large burner flame (d = 50 mm) when ag ≤ 35 s−1. At low strain rate, the
effect of the burner diameter on the extinction limit of a counterflow non-premixed
methane flame was larger in normal gravity than in microgravity. The extinction feature
of the microgravity counterflow flame was correlated to the flame size, indicating some
radial (lateral) heat loss. However, the extinction feature of the normal gravity
counterflow non-premixed flame was related mainly to the buoyancy effects, which are
also related to the burner diameter size. These extinction features can be identified by
investigating the specific maximum heat release rate.
ACKNOWLEDGEMENT
This work was supported by the Pukyong National University Research Fund for 2007
(PKS-2007-011).
REFERENCES
[1] Tsuji, H., Counterflow diffusion flames. Progress in Energy and Combustion
Science, 1982, 8, 93–119.
[2] Sung, C. J., Liu, J. B. and Law, C. K., Structural response of counterflow diffusion
flames to strain rate variations, Combustion and Flame, 1995, 102, 481–492.
[3] Oh, C. B., Lee, C. E. and Park, J., Numerical investigation of extinction in a
counterflow nonpremixed flame perturbed by a vortex, Combustion and Flame,
2004, 138, 225–241.
[4] Smooke, M. D., Puri, I. K. and Seshadri, K., A comparison between numerical
calculations and experimental measurements of the structure of a counterflow
diffusion flame burning diluted methane in diluted air, Proceedings of the
Combustion Institute, 1988, 21, 1783–1792.
[5] Balakrishnan, G., Trees, D. and Williams, F. A., An experimental investigation of
strain-induced extinction of diluted hydrogen-air counterflow diffusion flames,
Combustion and Flame, 1994, 98, 123–126.
[6] Fotache, C. G., Wang, H. and Law, C. K., Ignition of ethane, propane, and butane
in counterflow jets of cold fuel versus hot air under variable pressures,
[7] Chelliah, H. K., Law, C. K., Ueda, T., Smooke, M. D. and Williams, F. A., An
Experimental and theoretical investigation of the dilution, pressure and flow-field
effects on the extinction condition of methane-air-nitrogen diffusion flames,
Proceedings of the Combustion Institute, 1990, 23, 503–511.
[8] Puri, I. K. and Seshadri, K., Extinction of diffusion flames burning diluted
methane and diluted propane in diluted air, Combustion and Flame, 1986, 65,
137–150.
[9] Santoro, V. S. and Lin ~án, A. and Gomez, A., Propagation of edge flames in
counterflow mixing layers: Experiments and theory, Proceedings of the
[10] Gutheil, E. and Sirignano, W. A., Counterflow spray combustion modeling with
detailed transport and detailed chemistry, Combustion and Flame, 1998, 113,
92–105.
[11] Tsuji, H. and Yamaoka, I., The counterflow diffusion flame in the forward
stagnation region of a porous cylinder, Proceedings of the Combustion Institute,
1967, 11, 979–984.
[12] Han B., Ibarreta, A. F., Sung, C.-J. and T’ien, J. S., Experimental low-stretch
gaseous diffusion flames in buoyancy-induced flowfields, Proceedings of the
[13] Oh, C. B., Hamins, A., Bundy, M. and Park, J., The two-dimensional structure of
low strain rate counterflow nonpremixed-methane flames in normal and
microgravity, Combustion Theory and Modelling, 2008, 12(2), 283–302.
[14] Hamins, A., Bundy, M., Oh, C. B. and Kim, S. C., Effect of buoyancy on the
radiative extinction limit of low-strain-rate nonpremixed methane-air flames,
[15] Park, J. S., Hwang, D. J., Park, J., Kim, J. S., Kim, S., Keel, S. I., Kim, T. W. and
Noh, D. S., Edge flame instability in low-strain-rate counterflow diffusion flames,
[16] Seshadri, K. and Williams, F. A., Laminar flow between parallel plates with
injection of a reactant at high Reynolds number, International Journal of Heat
and Mass Transfer, 1978, 21, 1783–1792.
[17] Dryer, F. L. and Glassman, I., High-temperature oxidation of CO and CH4,
Proceedings of the Combustion Institute, 1973, 14, 987–1003.
[18] Ju, Y., Guo, H., Maruta, K. and Liu, F., On the extinction limit and flammability
limit of non-adiabatic stretched methane-air premixed flames, Journal of Fluid
Mechanics, 1997, 342, 315–334.
[19] Hwang, C.-H., Oh, C. B. and Lee, C.-E., Effects of CO2 dilution on the
interactions of a CH4-air nonpremixed jet flame with a single vortex,
International Journal of Thermal Science, 2009, 48, 1423–1431.
[20] Jutz, A. E., Kee, R. J., Grcar, J. F. and Rupley, F. M., Oppdif: a fortran program
for computing opposed-flow diffusion flames, Sandia Report SAND96-8243,
1987.
[21] T’ien, J. S., Diffusion flame extinction at small stretch rates: the mechanism of
radiative loss, Combustion and Flame, 1986, 65, 31–34.
[22] Maruta. K., Yoshida, M., Guo, H., Ju. Y. and Niioka, T., Extinction of low-
stretched diffusion flame in microgravity, Combustion and Flame, 1998, 112,
181–187.
[23] Bundy, M., Hamins, A. and Lee, K. Y., Suppression limits of low strain rate non-
premixed methane flames, Combustion and Flame, 2003, 133, 299–310.
[24] Oh, C. B., Hwang, C.-H., Lee, C.-E. and Park, J., Unsteady extinction mechanism
of nonpremixed flame interacting with a vortex, Chemical Engineering Science,
2006, 61, 3643–3652.

Effects of The Burner Diameter On The Flame Structure and Extinction Limit of Counterflow Non-Premixed Flames

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effects of The Burner Diameter On The Flame Structure and Extinction Limit of Counterflow Non-Premixed Flames

Uploaded by

Copyright:

Available Formats

International journal of spray and combustion dynamics · Volume .2 · Number . 3 .

2010 – pages 199–218 199

Effects of the burner diameter on the

Nam-gu, Busan 608-739, South Korea

*Corresponding author. E-mail address: cboh@pknu.ac.kr Tel: +82-51-629-6472

Recently, it was reported that a low-strain-rate counterflow flame can be affected

Figure 1: Schematic of the burner geometry and simulation domain.

4. RESULTS AND DISCUSSION

Critical mole fraction at extinction,Caext 0.90

−10 50000(J/m3 s) −10 50000(J/m3 s)

−10 50000(J/m3 s) −10 50000(J/m3 s)

4.2. Extinction characteristics

Critical mole fraction at extinction, Caext

0-g, 1D simulation, OPPDIF (case A)

0.75 1-g, 2D simulation, d = 23.4 mm13 (case C)

Figure 6: Computed 1D and 2D maximum flame temperatures as a function of the

4.3. Flame structure

503 656 809 962 1115 1268 1421 1574

−30 −20 −10 0 10 20 30 40 50

503 656 809 962 1115 1268 1421 1574

−30 −20 −10 0 10 20 30 40 50

Figure 8: Computed flame temperature and stream lines of 0 g counterflow non-

Axial velocity (m/s)

Figure 9: Computed flame temperature and axial velocity profiles at the

515 679 844 1009 1174 1338 1503 1668

−30 −20 −10 0 10 20 30 40 50

515 679 844 1009 1174 1338 1503 1668

−30 −20 −10 0 10 20 30 40 50

Axial velocity (m/s)

Table 1: Calculated parameters at the axisymmetric centerline as a function of

Global strain rate 10 s−1 50 s−1

You might also like