<CN>Chapter 5

<CT>Non-Premixed Micro Combustion

<authors>Yei-Chin Chao and Chih-Peng Chen, Institute of Aeronautics &
Astronautics, National Cheng Kung University, Taiwan
NOMENCLATURE
b ：half-width of the outer wall
c ：normalized half-width of inner wall
cp ：specific heat capacity
d ：inner diameter
D ：mass diffusivity
Di ：concentration-driven diffusion coefficient
Dij ：binary diffusion coefficient
DiT ：thermal-diffusion coefficient
Do ：mean diffusion coefficient evaluated for the oxidizer at the oxidizer stream
temperature
f ：focal length; mixture fraction
Fr ：Froude number
fst ：stoichiometric value of mixture fraction
gx ：gravitational acceleration in x-direction
h ：enthalpy
I ：ratio of the actual initial momentum to that for uniform flow
k ：thermal conductivity
kij ：thermal-diffusion ratio
lD ：radial position of the flame edge
Le ：Lewis number
Lf ：flame length
M ：molecular weight
NDB ：diffusion to buoyancy parameter
NDM ：diffusion to momentum parameter
p ：pressure
Pe ：Peclet number
Pi ：power of incident laser
Qf ：volumetric fuel flow rate
Re ：Reynolds number
Ro ：universal gas constant
S ：molar stoichiometric oxidizer-fuel ratio
Sc ：Schmidt number
T ：temperature
u* ：dimensionless axial velocity component
ud ：molecular diffusion velocity
ue ：fuel exit velocity
w ：species production rate
Wi ：atomic weight of species i
x* ：flame shape
xfl* ：dimensionless flame height
YF, stoic ：stoichiometric fuel mass fraction
Yi ：mass fraction of species i
YO0 ：mass fraction of the oxidizer at the oxidizer stream
Z f* ：normalized flame height
λ ：wave length; thermal conductivity
μ ：viscosity
ν ：(u,v) velocity vector; kinetic viscosity
ζ ：similarity variable
ξ ：mixture fraction
ρ ：density
Φ ：equivalence ratio
5.1. Introduction
5.1.1 Background

The advancement of human civilization is closely related to the ability to use and

control flames. Traditionally, flames have been used for heating, cooking,

illumination, and power generation for industrial and civil applications.. Modern

applications of flames include power generation for electricity and transportation, and

propulsion for both atmospheric aviation and space exploration.

Recently, with increasing demands of micro devices such as micro satellites and

micro aerial vehicles, needs for a micro power source to activate these systems have

significantly increased [1]. These micro systems require high-density power sources

for long periods of operation. The energy density of typical hydrocarbon fuels is about

100 times higher than that of the most advanced batteries. Even with heat losses in the

process of extracting power from burning the fuel, a micro-scale combustion system
has been considered as a viable alternative to batteries. To develop such combustion

systems, an understanding of the physics of laminar micro flames is of vital

importance.

For micro-scale applications, premixing fuel and air is not favored as the process

requires additional volume and weight. The current investigation concentrates

specifically on the flame shape, length, structure, and stabilization mechanism of

microjet diffusion flames. Numerical and advanced non-intrusive diagnostic tools are

used in the study. A comprehensive literature survey including studies of typical

diffusion flames and microjet diffusion flames are is given in the following sections.

5.1.2 Microjet Diffusion Flames

Diffusion flame is the most common flame type of flame in practical combustion

devices. The ability to predict the coupled effects of complex transport phenomena

with detailed chemical kinetics in these systems is critical for modeling reacting

flows, improving engine efficiency, and understanding the phenomena such as flame

extinction and pollutant formation. Laminar jet diffusion flames are fundamental to

combustion and have received much attention as a model system due to their

relatively simple geometry and fluid mechanical characteristics. Efforts have been

devoted to the detailed measurements of thermophysical properties and structures of

laminar flames along with computational efforts [2-–12].

Laminar jet diffusion flames fall into three categories: (1) the Burke--–Schumann

flame [13] controlled by diffusion, (2) the Roper flame [14] controlled by momentum

or buoyancy, and (3) the microflame [15] controlled by diffusion or momentum. The

first two types of flames have long been extensively investigated [16]. The third type

of flame was only recently investigated [15, 17-–23]. However, these studies were

concerned with bulk flame geometrical and thermal characteristics. Due to its small
size and negligible buoyancy force, the microjet flame can be used as a point heat

source with no preference to the orientation in future micro power devices. Therefore,

it is necessary to determine the smallest stable flame that can be maintained and to

investigate its characteristic structure near extinction.

Laminar microjet diffusion flames were theoretically investigated [15] and

numerically studied [17]. The importance of axial diffusion in the theoretical model

and the insignificance of buoyancy effect on the height and shape of the microjet

diffusion flame hashave been identified [15]. More recent investigations of microjet

diffusion flames include those of various fuels: methane [19-–, 20, 23-–, 24], propane

[21, 22], and hydrogen [25] flames. These studies found that the microjet flame hadis

sufficiently smaller in sizes ( few millimeters) and exhibited is nearly spherical in

shape independent of orientation, indicating its buoyancy insensitivity. Because the

spherical shape is similar to that of a microgravity flame, it may serve as a model of

microgravity flames [26]. Also, due to its small size, the heat loss to the burner is

relatively large. Thus, the flame might always operate in a severe, near-extinction

condition. It is also known that different fuels produce different visible flame shapes

near the burner port. For example, the C 2 class of hydrocarbon [15] and hydrogen [25]

flames extend upstream of the burner port, while the propane [21, 22] and methane

yield stand-off flames [19-–, 20, 23-–, 24]. Experimental results of flame shapes of

microjet methane diffusion flames are different using the direct photograph and laser

shadowgraph [20]. This implies that an intensively hot region may exist just ahead of

the flame zone.

5.21.3 Basic Micro Fflame Structure

The laminar microjet methane diffusion flames investigated here are stabilized on

vertical straight stainless-steel tubes with the inner diameter (d) ranging from 186 to
778 µm. The tube wall thickness is 79, 88, 122, 147, 148, and 124 µm for with d =

186, 212, 324, 382, 529, and 778 µm, respectively. Methane fuel is introduced

through the tube into the quiescent atmospheric air. The flame shapes are recorded

using a color CCD (Charge-coupled device) camera with a macro lens. Figures 5.1

and 5.2 respectively show photographs and numerically predicted longitudinal

distributions of OH radicals of the microjet flames. Photographical images and

predicted results clearly indicate the characteristic flame structure of the microjet

diffusion flame, which severely quenches locally at the flame base and the flame

stands off significantly from the tube.

5.32. Methodology
5.32.1 Experimental Setup

Because of the small flame sizes, intrusive techniques are unsuitable for the study of

microjet diffusion flames. Non-intrusive optical methods are used for the present

study. Observations and measurements are described in the following.

5.32.1.1 Burner and Fuel System

The laminar microjet diffusion flames investigated here are stabilized on a vertical

straight stainless-steel (AISI 304) tubes. The fuel flow is fully developed when it exits

the burner. The fuel flow rate is measured with an electrical mass flowmeter

(BROOKS Brooks 5850E), which is well calibrated using a soap-bubble flow meter.

The fuel Reynolds numbers range in the order of 10 to 100.

5.32.1.2 Image Capture System

The traditional way to define qualitative flame properties is usually based on

photographs. A Hhigh-quality recording system is beneficial tofor later digital image

processing. Micro flame images are obtained by a highly sensitive 3-chip color CCD

camera (Sony DXC-9000) with external triggering and recorded in the NTSC format.

The camera shutter time can be adjusted for best images of various flame conditions.
The captured images are then digitized for further analysis. The visible flame heights

have previously been measured using an identical imaging system by Cheng et al.

[25].. Moreover, the flame CH* chemiluminescence image is recorded by a cooled

CCD camera (Cooke SensiCam) with a macro lens and a narrowband interference

filter centered at λ = 431 nm with a full width at half-maximum of 10 ± 2 nm.

5.32.2 Mathematical Formulation

For microjet diffusion flames, detailed measurements are difficult to achieve because

of the tiny flame sizes. The sizes of microjet flame in this study are only several

millimeters and the flame thicknesses are on the order of micrometers. Neither

intrusive nor non-intrusive measurements can obtain sufficient resolution for species

distribution by the existing experimental technologies. Therefore, detailed numerical

simulations [27] with sufficiently small grid resolution are used to comprehend the

flame structures and characteristics. Theoretical models are also applied to examine

the measured flame shapes and flame lengths in this work.

5.32.2.1 Numerical Simulation

To numerically model the laminar microjet diffusion flames, the governing equations

of mass, momentum, energy, and chemical species for a steady axisymmetric reacting

flow can be written in the cylindrical (r, x) coordinate system as

(5.1)

(5.2)

(5.3)

(5.4)

and the state equation

 (5.5)

where ρ, p, T, Y, cp, h, w, R0, M, gx, and v = (u, ) are the density, pressure,

temperature, mass fraction, specific heat capacity of the mixture, enthalpy, species

production rate, universal gas constant, molecular weight, gravitational acceleration in

the x-direction, and velocity vector, respectively. µ, λ, and D are the viscosity, thermal

conductivity, and mass diffusivity, respectively. The subscript i in Equations (5.3)-()–

(5.5) stands for the i-th chemical species. The second term in the bracket of Equations

(5.3) and (5.4) is the thermo-diffusion or Soret diffusion due to the effect of

temperature gradient. The concentration-driven diffusion coefficient is calculated as:

(5.6)

where Dij is the binary diffusion coefficient. The binary mass diffusivity is determined

by the Chapman--–Enskog kinetic theory using Lennard-Jones parameters. The

thermo-diffusion coefficient is calculated as:

(5.7)

where M is the mixture molecular weight and kij is the thermo-diffusion ratio.

A schematic illustration of the computational domain coupled with boundary

conditions is shown in Figure. 5.3. The governing equations are solved using a

commercial flow package. An orthogonal, non-uniform staggered-grid system is used

for solving the discretized equations with a control volume formulation in accordance

with the SIMPLEC (Semi-Implicit Method for Pressure Linked Equations –

Consistent) [28] algorithm. The momentum equations are solved using the second-

order upwind scheme while the central difference method is used for the energy and
species equations. The above equations are solved along the mesh lines in the

computational domain using an iterative ADI (Alternating Direction Implicit) and

TDMA (Tri-Diagonal Matrix Algorithm) techniques. The Iinput of the molecular

transport data is obtained from the CHEMKIN package [289] and then the code

calculates the thermal conductivity and viscosity of the mixture using Wilke’s

formula. Thermal diffusion and buoyancy are included in this analysis but radiation

heat loss is neglected. A skeletal chemical kinetic mechanism [2930] is coupled with

the CFD package computational fluid dynamic program to predict the shape, length,

and structure of the micro methane flame. This skeletal chemical mechanism consists

of ten reversible and fifteen irreversible reactions. It has been demonstrated to be

sufficiently accurate for predicting flame speeds, extinction limits, and the thermo-

chemical properties [2830]. Furthermore, full CH4/Air chemical kinetic mechanism,

GRI-Mech 3.0 [301], is used to investigate the detailed flame structure and

stabilization mechanism.

5.32.2.2 Theoretical Models

It has been shown [15, 21, 312] that simple models derived from similarity analysis

can adequately predict flame height and flame shape for laminar jet diffusion flames.

It is imperative to see the suitability of these models as it applied to microjet methane

diffusion flames. For a steady, axisymmetric, vertical laminar jet at low Mach

numbers, uniform pressure, negligible buoyancy, and negligible mass diffusion, heat

conduction as well as viscous action in the axial direction but fast chemical reaction

rates, the governing equations (5.1)-()–(5.4) can be simplified and represented as the

following differential equation [15]:

(5.8)

The dimensionless axial velocity component (u*) and mixture fraction (f) become
 (5.9)

The flame shape can be determined from Equation (5.9) by assuming that the flame

exists at the stoichiometric value of f:

(5.10)

and the flame length is at the axial position as ζ → 0,

(5.11)

where is the dimensionless flame height, fst is the stoichiometric value of the

mixture fraction (0.055 for CH4--air), Re is the Reynolds number based on the jet

diameter, , and .

Equations (5.10) and (5.11) are derived by assuming negligible axial diffusion in the

flame. The Fflame shapes calculated including the axial diffusion term waswere

shown to better agree with thosethat measured in experiment for low Reynolds

number C2 class hydrocarbon microflames [15]. If the axial diffusion term is taken

into account, then Equation (5.8) becomes,

(5.12)

with the far-field boundary conditions F = 0 and F′= 0 as ζ→ ∞. The flame shape is

then determined by solving Equation (5.12) numerically and substituting F′ into

Equation (5.10).

In the study of microjet propane flames, Matta et al. [21] postulated that flame

behaviors are primarily controlled by diffusion and that buoyancy becomes less

important, at least foer cases near the quenching limit. Therefore, they adopted the jet

diffusion flame model of Turns [323] that does not include buoyancy effects for the
flame length predictions. The flame length depends only upon the volumetric fuel

flow rate, Qf, and not on the Reynolds number:

(5.13)

where D is the mass diffusivity and YF,stoic is the stoichiometric fuel mass fraction. In

addition, Matta et al. [21] hypothesized that the critical fuel flow rate at quenching is

that at which the predicted flame length equals the measured standoff distance.

Equating the flame length in Equation (5.13) to the measured standoff distance yields

a fuel flow rate at quenching. Comparisons of the predictions with experimental

measurements indicated that the behavior of miniature diffusion flames can be

adequately modeled by this simple jet diffusion flame model.

For the estimation of jet diffusion flame lengths, Roper [14] modified the Burke--

Schumann theory to allow the mass velocity of fuel gas to vary with axial distance as

affected by buoyancy and in accordance with continuity. For the circular burner port,

the following expression can be used to calculate the flame length:

(5.14)

where S is the molar stoichiometric oxidizer--fuel ratio, DO is a mean diffusion

coefficient evaluated for the oxidizer at the oxidizer stream temperature, TO, and Tf is

the mean flame temperature. Equation (5.14) can be applied to estimate the flame

length regardless of whether or not buoyancy is important, and is applicable for fuel

jets emerging into either a quiescent oxidizer or a co-flowing stream, as long as the

flames are over-ventilated.

The solutions of jet diffusion flame theory for the flame length estimations are mainly

dependent upon the volumetric flow rate and hence on the Reynolds number. Chung

and Law [334] extended the Burke--Schumann theory to include the effects of both
streamwise (axial) and preferential diffusion and derived a solution for the flame

length prediction. The solution that is independent of the Reynolds number but

depends on the Peclet number (Pe) can be expressed as

(5.15)

where c is the normalized half-width of the inner wall, YO0 is the mass fraction of the

oxidizer at the oxidizer stream, αn = (Pe2 + 4π2n2)1/2, and Zf* is the normalized flame

height. Equation (5.15) suggests that with the limit of Pe → 0 the flame becomes

independent of Pe, while the streamwise diffusion can be substantially more

significant than streamwise convection. This, however, has not been experimentally

validated.

Equations (5.10), (5.13), (5.14), and (5.15) are all derived by assuming unity Lewis

(Le) and Schmidt (Sc) numbers. If Sc is not equal to unity, 0.704 for methane, then the

flame shape can be represented as [345]:

, (5.16)

where x* is the flame shape normalized by the nozzle diameter and the flame length is

at the axial position as ζ → 0.

In the present study, Equations (5.8) (without axial diffusion) and (5.12) (with axial

diffusion) as well as Equation (5.16) are solved numerically to predict the flame

shapes and flame lengths. Equations (5.13), (5.14), and (5.15) are used only for the

flame length calculations. Comparisons, in the following section, of the measurement

and theoretical results are made to assess the applicability of the simple jet flame

models for microjet methane diffusion flames.

5.43 Characteristics of Microjet Methane Diffusion Flames

5.43.1 Flame Shape
Equation (5.16) as well as Equations (5.8) and (5.12) coupled with Equation (5.10)

are solved numerically to obtain the flame shape. The calculated flame shapes with

the measured and predicted results of Ban et al. [15] for the d = 186 µm flames

operated at several fuel flow rates are compared in Figure. 5.4. The measured flame

shapes are indicated by symbols, the solid and dashed lines denote those from the

calculation with and without axial diffusion, respectively, and the dashed-dot-dashed

lines depict the calculated results of Equation (5.16). Results of Figure. 5.4 indicate

that flame shapes calculated with and without axial diffusion are in poor agreement

with the experimental data and the flame heights are also significantly over-predicted

by nearly an order of magnitude for all ranges of fuel flow rates. With the

consideration of the non-unity Schmidt number effect, the predictions of flame

heights are slightly improved. The comparison shown in Figure. 5.4 clearly

demonstrates that the simple jet flame theory fails completely in the prediction of the

flame shapes of microjet methane flames. For the d = 324 µm flames (not shown

here), similar discrepancies between the predicted and measured flame shapes are also

observed. The failure of the simple jet flame model may be attributed to the failure of

the similarity assumption, which is used in the derivation of the simple model, as

applied to microjet diffusion flames.

5.34.2 Flame Length

Equations (5.11) and (5.16) indicate that the flame length depends only upon the

Reynolds number. It means that different diameter tubes with the same fuel exit

Reynolds number will produce the same flame length. Although the Reynolds number

isdoes not explicitly appeared in Equations (5.13) and (5.14), the flame length can still

be related to the Reynolds number through the dependence on the flow rate (Qf).

Comparison of the measured and calculated nondimensional flame lengths as

functions of Reynolds and Peclet numbers for d = 186 and 324 µm flames is shown in

Figure. 5.5. Figure 5.5(a) shows that the measured nondimensional flame length data

collapse into a straight line and can be scaled with the Reynolds number. Comparison

of the measured and calculated data indicates that the models of Ban et al. [15] , Turns

[323] , and Lee and Chung [345] all result in over-prediction of the flame lengths.

Figure 5.5(a) indicates that the flame length predicted by Roper’s model [14] agrees

very well with the measured data by assuming a mean flame temperature of 1500 K in

Equation (5.14). The slightly difference between the predictions and measurements

could be due to minor effects like buoyancy.

Figure 5.5(b) illustrates the nondimensional flame length as a function of Peclet

number. It should be noted that the model of Chung and Law [334] was derived for a

coflowing Burke--Schumann flame and the definition of Peclet number ( )

is slightly different from that used in the present study ( ), where b is the

half-width of the outer wall and lD is the radial position of the flame edge. The

purpose of this comparison is not to search for a theoretical representation of the

experimental data. Instead, it is to experimentally assess the model prediction when

accounting for the streamwise (axial) diffusion in Burke--Schumann flames [334].

Both measurement and theoretical data show that in the limit of Pe → 0 the flame

length becomes independent of Pe. This suggests that the streamwise diffusion can be

substantially more significant than streamwise convection at low Peclet numbers.

From these comparisons, one may expect that buoyancy acceleration would be more

important than molecular diffusion in the microjet flames. The diffusion--buoyancy

and diffusion--momentum parameters of Baker et al. [22] are examined. The

diffusion-to-buoyancy parameter is defined as

(5.17)
where D is the fuel gas mass diffusivity estimated at the mean flame temperature and

a ≅ 0.6g[(Tf / T0)] is the mean buoyant acceleration [14]. If NDB << 1 then the flame is

buoyancy controlled. If NDB >> 1 then the flame is diffusion controlled. Similarly, NDM

is the diffusion- to- momentum parameter defined as

(5.17)

where , uzo uz0 = Qf I YF,stoic / A is the ratio of the actual initial momentum flux to that

for uniform flow (for parabolic exit velocity I = 1.5 and for uniform flow I = 1.0) and

A is the cross-section area of the burner port. If NDM << 1, then the flame is

momentum controlled. If NDM >> 1, then the flame is diffusion controlled. It is noted

that the Froude number, based on the flame length instead of burner port diameter as

defined by Roper [14], can be recast as Fr = . An examination of the Peclet

number, the Froude number, the diffusion-to-buoyancy parameter, and the diffusion-

to-momentum parameter is sufficient to determine the relative importance of the three

transport mechanisms (buoyancy, momentum, or diffusion) for microjet diffusion

flames. The calculated parameters, based upon experimental data and baseline data (Tf

= 1500 K, T0 = 300 K, I = 1.5, YF,stoic = 0.055, and D = 2.8195 cm2/s) are listed in

Table 5.1 for the d = 186 and 324 µm flames. It is clear that in the analysis

considering only diffusion and buoyancy, the flames are more in the diffusion-

dominated regime for Qf ≤ 5 cc/min because NDB is greater than 1. For Qf ≤ 5 cc/min

the microjet flames are in spherical shape. When Qf > 6 cc/min, the buoyancy

acceleration would be more important than molecular diffusion in the microjet flames

because of NDB < 1. The Froude number for the flames studied here ranges from

O(10-–1) to O(1). This indicates that the flames fall within the transitional to

buoyancy-controlled regime. In addition, the Peclet number ranges from O(1) to

O(10) indicating that the molecular-diffusion velocity is comparable to the convective

velocity only for Qf ≤ 5 cc/min. This is in consistent with the NDM parameter analysis.

From the above parameter analysis, we can conclude that the microjet diffusion

flames studied here are not completely buoyancy-free and the molecular- diffusion is

effective only for NDB > 2 and Pe < 2.

5.34.3 Quenching Velocity

To investigate the effect of tube size on extinction behavior, different tube diameters

are used. Figure 5.6 shows photographs of flames operated at fuel exit velocity near

extinction for tube diameters varying from 186 to 778 µm. The most notable feature

of Figure. 5.6 is that the standoff distance is essentially the same, approximately 0.78

mm, for all the tubes. In addition, the flame shapes are remarkably similar over the

range of tube diameters. It has been demonstrated in the previous section and shown

in Figure. 5.5 that the measured and predicted flame length is a linear function of

Reynolds number. Accordingly, a flame can not be sustained if the predicted flame

length for a fuel exit velocity is smaller than the measured standoff distance. Thus, it

is hypothesized that the critical fuel exit velocity at quenching is that at which the

predicted flame length equals the measured standoff distance [21]. Since Roper’s

model [14] is shown above to well predict the flame lengths, the model is further used

for quenching velocity predictions. The quenching velocity for different tube sizes is

determined by equating the flame length in Equation (5.14) to the measured standoff

distance. Comparison of the measured and calculated quenching velocities is depicted

in Figure. 5.7. The predicted quenching velocities (dashed line) are in excellent

agreement with the measured data. The good agreement between the calculated and

measured quenching velocities is due that near the quenching limit,; the microjet

flames are mainly diffusion- controlled and buoyancy has a minor effect on the flame
length. It is also found that the measured quenching velocities follow Re × d =

constant (u × d2 = C) curve. This finding is in agreement with the relationship

proposed [20] but different from Re × d2 = C that was obtained numerically for the

adiabatic wall conditions [19].

5.34.4 Effect of Burner Wall Boundary Condition on the Standoff Distance

In order to clarify the effect of tube materials on the standoff distance, the computed

OH isopleths with different wall thermal conductivity are compared in Figure. 5.8 for

flames of d = 186, 324, and 529 µm. The standoff distance is measured along the jet

centerline from the burner exit to the bottom of blue flame cap. For the same tube

diameter, for example d = 186 µm, the image of Figures. 5.8(a), (d), and (g) are

calculated with thermal conductivity k = 16.2 W/m.K (AISI 304), k = 8 W/m.K

(quartz tube), and k =1 W/m.K, respectively. It can be seen from Figures. 5.8(a)-–(c)

that the computed standoff distance (0.8 mm) is in good agreement with the measured

data (0.78 mm) for the stainless- steel tube. The maximum variation of the calculated

standoff distance is less than 3% for the same tube material with different tube

diameters and that is less than 5% for a fixed tube diameter with different tube

materials. Therefore, it can be concluded that the change of tube materials only has a

negligible effect on the predicted standoff distance, but does influence the quenching

gap between the flame and the tube. It is noted that the detachment of the flame base

from the burner rim is reduced as the thermal conductivity of the tube is reduced. This

is due that reduced thermal conductivity results in less conduction of heat from the

flame edge to the tube, and hence reduces the quenching gap. If an adiabatic wall

condition is imposed, the flame would attach to the burner and no quenching gap is

produced.

5.45. Flame Structure and Stabilization Mechanism

5.45.1 Flame Structure

In order to have a further look into the detailed flame structure of the microjet

methane diffusion flame, the case of a stainless-steel (AISI 304)

tube with an inner diameter (d) of 186 µm and wall thickness of 79

µm at a volumetric flow rate of 3.9 cc/min corresponding to a bulk

velocity of 2.39 m/s is adopted. This flow rate is just slightly above

the extinction limit and the corresponding Reynolds and Froude

numbers are 27 and 3131, respectively.

The computed CH species mass fraction isopleths are compared

with the measured CH* chemiluminescence and flame images in

Figure. 5.9. Note that the CH* image is Abel inverted and it is performed only for the

flame height and flame shape measurements. The CH* intensity can not be correlated

to the CH concentration because the CH* reaction mechanisms are not included in the

present numerical calculation. It can be seen that the comparisons in

terms of flame height, flame shape, and stand-off distance are in

very good agreement. Both experimental and numerical results indicate that the

flame is quenched by the tube wall and creates a gap to allow oxidizer entrainment.

As a result, the flame stands off from the tube. The characteristics of semi-spherical

flame shape, non-buoyancy flame configuration, and large quenching gap between the

flame base and the tube are very similar to those of a microgravity flame, but the heat-

release rate is about two orders of magnitude larger than that of a microgravity flame

[3566]. In view of the success of predicting the overall flame

characteristics of flame heights and shapes of microjet flames by

numerical simulation, computations using GRI-Mech 3.0 are

extended to investigate the flame structure and reaction

characteristics [37].

The computed results of the 2-D temperature, CH, O 2, CH4, CO2, H2O, H2, and CO

species mass fraction contours as compared with mixture fraction contours of lean (ξ

= 0.029), stoichiometric (ξ = 0.055), and rich (ξ = 0.089) limits of the methane flame

are shown in Figure. 5.10. The mixture fraction is defined as [3788]:

(5.18)

where YC, YH, and YO are the mass fraction for carbon, hydrogen, and oxygen atoms,

Wi is the atomic weight of species i, and subscripts f and O refer to fuel jet and

ambient air, respectively. It can be seen that a small amount of O 2 is entrained into the

standoff region from the gap between the burner wall and flame base and a small

amount of CH4 has diffused upstream of the burner port. The entrainment could

results in partial premixing of fuel and oxygen over the standoff distance. The

computed temperature contour shows that the maximum flame temperature locates at

the jet centerline near the stoichiometric mixture fraction contour and the maximum

of the CH isopleths. The unburned mixtures, the burner wall, and the fuel stream are

heated to a temperature higher than 700 K. This suggests that the standoff and the

flame stabilization are strongly related to the characteristic hot zone and heating of the

fuel stream and unburned mixture through the tube wall. The appearance of more

stable species of CO and H2 and products of CO2 and H2O within the standoff region

may further suggest that the high temperature in the standoff region is due to heat-

release from the final-product formation reactions (CO + OH → CO2 + H, OH + H2

→ H2O + H) or due to formation of double reaction zones in a partially premixed

flame. However, the axial distributions of reaction rates (not shown here) and species

mass fractions along the jet centerline indicate that the final-product formation
reactions occur in the downstream region of the maximum temperature location and

no double flame structure exists in this region. Therefore, this high temperature in the

standoff region is most likely due to molecular heat conduction from the flame or

other key radical reactions in the upstream region.

The computed axial profiles of temperature and selected species along the centerline

of the jet are shown in Figure. 5.11. The vertical dashed line denotes the location of

stoichiometric mixture fraction. It can be seen that oxygen is entrained and diffused

into the jet center near the burner exit and its concentration decreases with decreasing

CH4 concentration as it flows downstream. The consumption of CH 4 and O2 intersects

at a point slightly upstream of where the maximum temperature (T = 1870 K) is

located. This is usually seen in a normal laminar diffusion flame except that no O 2 is

present upstream of the intersection point in a diffusion flame. The distribution of CO,

H2, H2O, and CO2 is different from a typical diffusion flame structure [323] ,

indicating appreciable amount of reactions in the stand-off region. Moreover, there is

only one CH peak occurring at the maximum temperature position (Figure. 5.11(b)).

This is also consistent with the computed pure CH4 diffusion flame structure [3899] .

To further examine the role of a tribrachial flame in the microjet flame stabilization,

the computed velocity vectors coupled with CH and stoichiometric mixture fraction

contours are depicted in Figure. 5.12 for a small region near the burner exit (0.8 mm ×

1.2 mm). Note that the computational domain is 4 mm × 16 mm in the radial and axial

direction, respectively. The maximum fuel velocity is 9.7 m/s at the centerline of the

burner exit. The velocity vectors show lateral expansion and longitudinal acceleration

as the fuel emerges from the burner exit and approaches the hot zone around the flame

base due to thermal expansion. The flow velocity in the ambient air as well as in the

downstream of flame base remains relatively constant and parallels to the axial
direction, indicating a negligible buoyancy effect. Also, a small amount of air is

entrained into the standoff region from the gap between the burner wall and the flame

base. Figure 5.12 indicates that the flow velocity (0.2 m/s) near the flame edge is less

than the laminar burning velocity (0.4 m/s) and no increase of velocity is observed

after passing the flame edge similar to the observation by Puri et al. [394040]. (2001).

In addition, the flame edge is not located on the stoichiometric mixture fraction point

[4011] . All of these outstanding features suggest that a tribrachial flame may not

exist in a microjet flame near extinction and the flame stabilization must be due to

other unrevealed mechanisms.

5.45.2 Stabilization Mechanism

From the above discussion one might conjecture that the stabilization of a standoff

microjet flame is due to the interaction of flame quenching, preheating of partially

premixed mixture and sustained preflame reaction. The characteristics of flame

quenching are illustrated in Figure. 5.13 by comparing two computed CH contours

under adiabatic and variable wall temperature conditions. With the adiabatic wall, the

flame zone extends upstream of the port and connects to the tube as indicated by CH

contours, while with the variable wall temperature condition (a realistic situation), the

flame is quenched on the tube wall and creates a gap to allow oxidizer entrainment.

As the flame is quenched by the tube wall, heat is transferred through the wall to

accelerate fuel decomposition, initiate further reaction, and produce intermediate

radicals in the vicinity of the exit. As usual, the HO 2 radicals are brisk in the chain-

terminating reactions when the flame is quenched on the wall. Detailed examination

of the intermediates indicates that the HO2 radicals play an important role in

connecting with the stabilization of the flame, as described in the following.

Figure 5.14 shows the calculated results of distribution contours of HO 2, isopleths of

total heat-release rate and CH mass fraction, and mass flux vectors of H, O, OH, and

HO2 radicals in contrast to the stoichiometric mixture fraction location. The CH mass

fraction isopleths and mass flux vectors of O and H radicals are shown on the left

hand side of the figure, while the total heat-release rate isopleths, HO 2 contours, and

mass flux vectors of OH and HO2 on the right hand side. It can be seen that the

maximum HO2 appears in the gap region then decreases to connect with heat-release

rate isopleths. The heat-release rate isopleth depicts a peak reactivity spot (called

reaction kernel [3566] ) located on the lean side of the stoichiometric mixture fraction

with the position of x  0.33 mm and r  0.6 mm. Hydrogen and oxygen atoms as

well as hydroxyl radicals diffuse to both sides of the flame zone and upstream against

the incoming fuel-rich flow, while the HO 2 radicals predominantly diffuse and

displace in the inward direction to the fuel stream near the tube exit.

To further delineate the stabilization process of the microjet flame, the detailed flame

structures across the standoff region (x = -–0.1 mm) and the reaction kernel (x = 0.33

mm) are examined. The radial profiles of the temperature, species mass fractions,

formation rates, net reaction rates, and net heat-release rates of the major elementary

steps at these two characteristic heights are illustrated in Figures. 5.15 and 5.16. In the

figures, the radial location of the tube wall and the stoichiometric mixture fraction (ξ

= 0.055) are indicated by the solid and dashed lines, respectively. For the case in

Figure. 5.15 (x = -–0.1 mm), the distributions of temperature and major species mass

fractions are typical of diffusion flames, except for the appreciable amount of

penetration of O2 to the fuel-side and the efflux of CH4 to the air-side through the gap

(Figure. 5.15(a)). The peak temperature and the production rates of minor species are

primarily formed on the lean-side of stoichiometric mixture fraction. The peak CH 4

consumption rate and the production rates of H 2O and CO2 are also on the air-side
(Figure. 5.15(b)). These facts suggest that in the quenching gap region lean chemical

reactions take place. Figures 5.15(c) and 5.15(d) show the net reaction and heat-

release rates of major elementary steps. It is obvious that the most significant reaction

is the chain-branching reaction H + O 2 ↔ OH + O (R38), followed by the

dehydrogenation of methane primary by OH + CH4 ↔ CH3 + H2O (R98) and OH + H2

↔ H + H2O (R84) reactions to form the product H 2O and to further build up the

radical pool. Although the rate for H production through OH + H 2 ↔ H + H2O (R84),

OH + CO ↔ H + CO2 (R99), and O + CH3 ↔ H + CH2O (R10) reactions are is more

significant than that for OH formation through HO 2 + CH3 ↔ OH + CH3O (R119) and

HO2 + H ↔ OH + OH (R46) reactions , the rate of methyl (CH3) radical production

by OH radical attacking on CH4 (R98) exceeds that by H atoms (R53) in this

relatively low temperature region. In addition, the formation (R168) and destruction

(R119) of HO2 reactions are believed to play an important role for the hot zone and

reaction kernel formation in the quenching gap region. Figure 5.15(d) indicates that

the major contributors to the positive heat release are the reactions, including O + CH 3

↔ H + CH2O (R10), O + CH3 → H + H2 + CO (R284), O2 + HCO ↔ HO2 + CO

(R168), OH + CH4 ↔ CH3 + H2O (R98), OH + CO ↔ H + CO 2 (R99), OH + H2 ↔ H

+ H2O (R84), OH + CH2O ↔ HCO + H2O (R101), O2 + CH2 → 2H + CO2 (R290),

OH + HO2 ↔ O2 + H2O (R87), and HO2 + CH3 ↔ OH + CH3O (R119). The negative

contributors are R38, R166 + R167, R57, and R86.

Figure 5.16 shows the radial distributions of calculated variables across the reaction

kernel (x = 0.33 mm). In general, the distributions of variables are similar to those at x

= -–0.1 mm, except for the increased levels of species production rates, net reaction

rates, and heat-release rates resulting in higher species concentrations and peak

temperature (T = 1746 K) in the form of a flame. In addition, the production rates

(Figure. 5.16(b)) show that unlike those at x = -–0.1 mm, the H2, CO, CH3, and CH2O

species are primarily formed on the fuel-side and consumed on the air-side of the

stoichiometric mixture fraction; and the H, O, and OH radicals are produced on the

air-side and consumed on the fuel-side. The peak consumption rate of CH 4 is on the

fuel-side and that of O2 and the production rates of the H 2O and CO2 are on the air-

side. Figures 5.16(c) and (d) show that the chain-branching reaction (R38) is still the

most significant reaction and the methyl oxidation reaction (R10) is the major

contributor to the heat-release rate. It should be noted that the methyl (CH 3) radical is

produced more by the H atom through (R53) than by the OH radical through (R98) as

compared to those reactions at x = -–0.1 mm. Also, the formation (R168) and

destruction (R119) of HO2 reactions, that play an important role for the hot zone

formation in the gap region (x = -–0.1 mm), become insignificant at this location. The

dominant exothermic reactions, including R10, R284, R58, R290, R98, R99, R84,

R101, R125, and R168, contribute up to 81.5% of the total heat-release rate at the

reaction kernel (1.15  109 W/m3), thereby stabilizing a standoff microjet flame.

From the analysis of chemical kinetic structures, the sequence of flame stabilization

can be summarized schematically in Figure. 5.17 and explained as follows. As the

flame is quenched on the tube wall, heat is transferred through the tube wall to

accelerate fuel pyrolysis near the tube exit, produce CH3 and H intermediate radicals

in the standoff region, and initiate further reaction in the vicinity of the flame base.

Consequently, the chain-branching reaction (R38), the methane consumption

reactions (R98) and (R53), and the final product formation (R84) and (R99) are

enhanced to further build up the radical pool of H, O, OH, and CH 3 in the gap region.

Then the methyl (CH3) radical oxidations (R10) and (R284) not only release

significant heat to enhance further reactions but also produce the formaldehyde
(CH2O) radical. The CH2O radical is attacked by the OH radical (R101) and H atom

(R58) to form the formyl (HCO) radical and followed by the HCO oxidation (R168)

to produce the HO2 radical. Finally the HO2 radical predominantly diffuses and moves

inward to the fuel stream near the tube exit and reacts with CH 3 (R119) to release heat

near the tube wall region. These key reactions are believed to form the hot zone (T =

800 ~ --1450 K) that connects the visible flame base and provides heat for sustaining

and enhancing further H2---–O2 chain reactions and CH3 formation (R53 and R98) and

oxidation (R10 and R284) at the flame base, which in turn result in the formation of

the reaction kernel responsible for flame stabilization. It should be noted that although

the reaction rate and heat-release rate of the HO 2 formation (R168) and consumption

(R119) reactions are not as significant as those of the other reactions;, they indeed

play an important role in the formation of the hot zone and high heat-release reaction

kernel [4135].

5.5 6 Summary
Flame characteristics, in terms of flame shape, flame length, flame structure, reaction

rate profiles, quenching limit, and stabilization mechanism, of microjet methane

diffusion flames operated at various fuel exit velocityies ranging from just above

quenching to below blowoff are investigated. Comparisons of the measured primary

flame parameters, such as flame heights, flame shapes, and quenching velocities with

theoretical predictions indicate that Roper’s model [14] can satisfactorily predict the

characteristics of microjet methane flames. Comparisons of the predicted quenching

velocity with measured results indicate that quenching occurs when the flame length

equals the standoff distance. It is also found that the quenching velocity for different

tube diameters collapses to the curve Re × d = const. An order of magnitude analysis

of the dimensionless parameters suggests that the microjet diffusion flames studied
here are not buoyant buoyancy-free and the molecular- diffusion is effective only for

NDB > 2 and Pe < 2. Numerical simulations of the flames stabilized at the tip of a 186

µm tube indicate that the computed flame shape and flame length are in excellent

agreement with experimental results. Furthermore, the calculated flame structures

show that the diffusion process dominates over the premixing process in the standoff

region, suggesting that the flame burns in a diffusion mode. Besides, the calculated

OH mass fraction isopleths indicate that the change of tube materials has a minor

effect on the standoff distance near extinction, but does influence the quenching gap

between the flame and the tube.

The computed flame heights as a function of fuel exit velocities indicate that

buoyancy may play a role for flames with larger tube diameters and higher exit

velocities. However, for the case of d = 186 µm flame with ue < 3.68 m/s, the

buoyancy effect becomes minor and the flame is diffusion controlled. Although

partial premixing may occur in the standoff region, its mixing intensity is not strong

enough to generate reaction and hence no double flame structure is observed. Further

examination of the computed flame structure reveals that the flame is stabilized by a

consequence of flame quenching on the tube wall, formation of HO 2 layer, forming

the hot zone through reaction of HO2 with CH3, yielding subsequent key radical

reactions, and finally forming the reaction kernel. The major reaction path near the

flame base is also summarized schematically.

References
[1]. Kovacs, G. T. A., 1998, Micromachined Transducers-Source Book.
McGraw-Hill, New York.
[2]. Mitchell, R. E., Sarofim, A. F., and Clomburg, L. A., 1980,
“Experimental and Numerical Investigation of Confined Laminar
Diffusion Flames,” Combustion and. Flame, vol. 37, pp. 227-–244. doi:
http://dx.doi.org/10.1016/0010-2180(80)90092-9.
[3]. Smooke, M. D., Mitchell, R. E., and Grcar, J. F., 1984, in: Elliptic
Problem Solvers II, edited by Birkhoff, G. and Schoenstadt, A.,
Academic, New York.
[4]. Smooke, M. D., Lin, P., Lam, J., and Long, M. B., 1990,
“Computational and Experimental Study of a Laminar Axisymmetric
Methane/Air Diffusion Flame,” Proceeding of Combustion Institute, vol. 23,
pp. 575-–582. doi: http://dx.doi.org/10.1016/S0082-0784(06)80305-X.
[5]. Smooke, M. D., Xu, Y., Zurn, R., Lin, P., Frank, J., and Long, M. B.,
1992, “Computational and Experimental Study of OH and CH
Radicals in Axisymmetric Laminar Diffusion Flames,” Proceedings of
Combustion Institute, vol. 24, pp. 813-–821. doi:
http://dx.doi.org/10.1016/S0082-0784(06)80099-8.
[6]. Smooke, M. D., Ern, A., Tanoff, M. A., Valdati, B. A., Mohammed,
R. K., Marran, D. F., and Long, M. B., 1996, “Computational and
Experimental Study of NO in an Axisymmetric Laminar Diffusion
Flame,” Proceedings of Combustion Institute, vol. 26, pp. 2161-–2170. doi:
http://dx.doi.org/10.1016/S0082-0784(96)80042-7.
[7]. Norton, T. S., Smyth, K. C., Miller, J. H., and Smooke, M. D., 1993,
“Comparison of Experimental and Computed Species Concentration
and Temperature Profiles in Laminar, Two-dimensional Methane/Air
Diffusion Flames,” Combustion Science and Technology, vol. 90, pp. 1-–34.
doi: http://dx.doi.org/10.1080/00102209308907601.
[8]. Takagi, T., and Xu, Z., 1994, “Numerical Analysis of Laminar
Diffusion Flames—Effects of Preferential Diffusion of Heat and
Species,” Combustion and Flame, vol.96, pp. 50-–59. doi:
http://dx.doi.org/10.1016/0010-2180(94)90157-0.
[9]. Walsh, K. T., Long, M. B., Tanoff, M. A., and Smooke, M. D., 1998,
“Experimental and Computational Study of CH, CH*, and OH* in an
Axisymmetric Laminar Diffusion Flame,” Proceedings of Combustion
Institute, vol. 27, pp. 615-–623. doi: http://dx.doi.org/10.1016/S0082-
0784(98)80453-0.
[10]. McEnally, C. S., Pfefferle, L. D., Schaffer, A. M., Long, M. B.,
Mohammed, R. K., Smooke, M. D., and Colket, M. B., 2000,
“Characterization of a Coflowing Methane/Air Non-premixed Flame
with Computer Modeling, Rayleigh-Raman Imaging, and On-line
Mass Spectrometry,” Proceedings of Combustion Institute, vol. 28, pp.
2063-–2070. doi: http://dx.doi.org/10.1016/S0082-0784(00)80614-1.
[11]. Nguyen, Q. V., Dibble, R. W., Carter, C. D., Fiechtner, G. J., and
Barlow, R. S., 1996, “Raman-LIF Measurements of Temperature,
Major Species, OH, and NO in a Methane-Air Bunsen Flame,”
Combustion and Flame, vol.105, pp. 499-–510. doi:
http://dx.doi.org/10.1016/0010-2180(96)00226-X.
[12]. Chou, C.-P., Chen, J.-Y., Yam, C. G., and Marx, K. D., 1998,
“Numerical Modeling of NO Formation in Laminar Bunsen Flames –
—A Flamelet Approach,” Combustion and Flame, vol. 114, pp. 420-–435.
doi: http://dx.doi.org/10.1016/S0010-2180(97)00317-9.
[13]. Burke, S. P. and Schumann, T. E. W., 1928 “Diffusion Flames”,
Industrial Engineering Chemistry, vol. 20, pp. 998-–104. doi:
http://dx.doi.org/10.1021/ie50226a005.
[14]. Roper, F. G., 1977, “The Prediction of Laminar Jet Diffusion
Flame Sizes: Part I. Theoretical Model and Part II. Experimental
Verification,” Combustion and Flame, vol. 29, pp. 219-–234. doi:
http://dx.doi.org/10.1016/0010-2180(77)90112-2.
[15]. Ban, H. Venkatesh, S. and Saito, K., 1994, “Convection-
Diffusion Controlled Laminar Micro7 Flames,” Transactions of the ASME,
Journal of Heat Transfer, vol. 116, pp. 954-–959.
[16]. Williams, F. A., 1985, Combustion Theory. Addison-Wesley, New
York.
[17]. Nakamura, Y., Ban, H., Saito, K., and Takeno, T., 1997, “Micro
Diffusion Flames in a Cold Boundary,” Proceeding of the Central State
Section Meeting, pp. 160-–163.
[18]. Nakamura, Y. and Saito, K. 2001, “Thermal and Fluid Dynamic
Structures Of Micro-Diffusion Flames,” Nagare (in Japanese), vol. 20,
pp. 74-–82.
[19]. Nakamura, Y., Kubota, A., Yamashita, H., and Saito, K.,
November 30, 2003, “Near Extinction Flame Structure of Micro-
Diffusion Flames,” The International Symposium on Micro-Mechanical
Engineering, Paper No. ISMME2003-111, pp. 163–170.
[20]. Ida, T., Fuchihata, M., and Mizutani, Y., 2000, “Microscopic
Diffusion Structures With Micro Flames,” Proceedings of the Third
International Symposium on Scale Modeling, ISSM3-E7, Nagoya, Japan.
[21]. Matta, L. M., Neumeier, Y., Lemon, B., and Zinn, B. T., 2002,
“Characteristics of Microscale Diffusion Flames,” Proceedings of
Combustion Institute, vol. 29, pp. 933-–939. doi:
http://dx.doi.org/10.1016/S1540-7489(02)80118-3.
[22]. Baker, J., Calvert, M. E. and Murphy, D. W., 2002, “Structure
and Dynamics of Laminar Jet Micro-Slot Diffusion Flames,”
Transactions of the ASME, Journal of Heat Transfer, vol. 124, pp. 783-–790.
[23]. Cheng, T. S., Chao, Y.-C., Wu, C.-Y., Li, Y.-H., Nakamura, Y., Lee,
K.-Y., Yuan, T., and Leu, T. S., 2005, “Experimental and Numerical
Investigation of Microscale Hydrogen Diffusion Flames,” Proceedings of
Combustion Institute, vol. 30, pp. 2489-–2497. doi:
http://dx.doi.org/10.1016/j.proci.2004.07.025.
[24]. Nakamura, Y., Ban, H., Saito, K., and Takeno, T., 2000,
“Structures of Micro (Millimeter Size) Diffusion Flames,” Proceedings of
the Third International Symposium on Scale Modeling, ISSM3-E7, Nagoya, Japan.
[25]. Cheng, T. S., Wu, C.-Y., Chen, C.-P., Li, Y.-H., Chao, Y.-C., Yuan, T.,
and Leu, T. S., 2005, “Detailed Measurement and Assessment of
Laminar Hydrogen Jet Diffusion Flames,” Combustion and Flame, vol.
146, pp. 268-–282. doi: http://dx.doi.org/10.1016/j.combustflame.2006.03.005.
[26]. Yoo, S. W., Law, C. K., and Tse, S. D., 2002, “Chemiluminesent
OH* and CH* Flame Structure and Aerodynamic Scaling of Weakly
Buoyant, Nearly Spherical Diffusion Flames,” Proceedings of Combustion
Institute, vol. 29, pp. 1663-–1670. doi: http://dx.doi.org/10.1016/S1540-
7489(02)80204-8.
[27]. Cheng, T. S., Chen, C.-P., Chen, C.-S., Li, Y.-H., Wu, C.-Y., and
Chao, Y.-C., 2006, “Characteristics of microjet methane diffusion
flames,” Combustion Theory and Modeling, vol. 10(5), pp. 861-–881. doi:
http://dx.doi.org/10.1080/13647830600551917a.
[28]. Van Doormal, J. P. and Raithby, G. D., 1984, "Enhancements of the
SIMPLE Method for Predicting Incompressible Fluid Flows", Numerical Heat
Transfer Vol. 7, Iss. 2, p. 147-163.
[289]. Kee, R. J., Rupley, F., Miller, J., Coltrin, M., Grcar, J., Meeks, E.,
Moffat, H., Lutz, A., Dixon-Lewis, G., Smooke, M. D., Warnatz, J.,
Evans, G., Larson, R., Mitchell, R., Petzold, L., Reynolds, L.,
Caracotsios, M., Stewart, W. and Glarborg, P., 1999, User Manual, The
CHEMKIN Collection Release 3.5, Reaction Design, Inc., San Diego, CA.
[2930]. Smooke, M. D., 1991, “Reduced Kinetic Mechanisms and
Asymptotic Approximations for Methane-Air Flames,” Lecture Notes in
Physics, Springer-Verlag, Berlin, vol. 384, pp. 1-–28. doi:
http://dx.doi.org/10.1007/BFb0035362.
[301]. Smith, G. P., Golden, D. M., Frenklach, M., Moriarty, N. W.,
Eiteneer, B., Goldenberg, M., Bowman, C. T., Hanson, R. K., Song, S.,
Gardiner, W. C., Lissianski, V. V., Qin, Z., 1999, GRI-Mech Homepage,
Gas Research Institute, Chicago, www.me.berkeley.edu/gri_mech/.
[312]. Spalding, D. B., 1979, Combustion and Mass Transfer. Pergamon
Press, New York.
[323]. Turns, S. R., 2000, An Introduction to Combustion: Concepts and
Applications, McGraw-Hill, New York, pp. 352.
[334]. Chung, S. H. and Law, C. K., 1984, “Burke-Schumann Flame
With Streamwise and Preferential Diffusion,” Combustion Science and
Technology, vol. 37, pp. 21-–46. doi:
http://dx.doi.org/10.1080/00102208408923744.
[345]. Lee, B. J. and Chung, S. H., 1997, “Stabilization of Lifted
Tribrachial Flames In A Laminar Nonpremixed Jet,” Combustion and
Flame, vol. 109, pp. 163-–172. doi: http://dx.doi.org/10.1016/S0010-
2180(96)00145-9.
[35]. Cheng, T. S., Chao, Y.-C., Chen, C.-P. and Wu, C.-Y., 2008,
“Further Analysis of Chemical Kinetic Structure of a Standoff Microjet
Methane Diffusion Flame Near Extinction”, Combustion and Flame, vol.
152, pp. 461-467.
[3566]. Takahashi, F., and Katta, V. R., 2002, “Reaction Kernel
Structure and Stabilization Mechanisms of Jet Diffusion Flames in
Microgravity,” Proceedings of Combustion Institute, vol. 29, pp. 2509-–
2518. doi: http://dx.doi.org/10.1016/S1540-7489(02)80306-6.
[3677]. Chen, C.-P., Chao, Y.-C., Cheng, T. S., Chen, G.-B., and Wu, C.-
Y., 2007, “Structure and Stabilization Mechanism of a Microjet
Methane Diffusion Flame Near Extinction,” Proceedings of the Combustion
Institute, vol. 31, pp. 3301-–3308. doi:
http://dx.doi.org/10.1016/j.proci.2006.08.069.
[3788]. Bilger, R. W., Stårner, S. H., Kee, R. J., 1990, “On Reduced
Mechanisms for Methane – Air Combustion in Nonpremixed Flames,”
Combustion and Flame, vol. 80, pp. 135-–149. doi:
http://dx.doi.org/10.1016/0010-2180(90)90122-8.
[3899]. Blevins, L. G., and Gore, J. P., 1999, “Computed Structure of
Low Strain Rate Partially Premixed CH /Air Counterflow Flames:
Implications for NO Formation ,” Combustion and Flame, vol. 116, pp.
546-–566. doi: http://dx.doi.org/10.1016/S0010-2180(98)00059-5.
[394040]. Puri, I. K., Aggarwal, S. K., Ratti, S., and Azzoni R., 2001,
“On the Similitude Between Lifted and Burner-Stabilized Triple
Flames: A Numerical and Experimental Investigation,” Combustion and
Flame, vol. 124, pp. 311-–325. doi: http://dx.doi.org/10.1016/S0010-
2180(00)00201-7.
[4011]. Echekki, T., and Chen, J. H., 1998, “Structure and
Propagation of Methanol–Air Triple Flames,” Combustion and Flame, vol.
114, pp. 231-–245. doi: http://dx.doi.org/10.1016/S0010-2180(97)00287-3.
[412]. Cheng, T. S., Chao, Y.-C., Chen, C.-P. and Wu, C.-Y., 2008,
“Further Analysis of Chemical Kinetic Structure of a Standoff Microjet
Methane Diffusion Flame Near Extinction,” Combustion and Flame, vol.
152, pp. 461–467. doi: http://dx.doi.org/10.1016/j.combustflame.2007.10.007.
Van Doormal, J. P. and Raithby, G. D. (1984), "Enhancements of the
SIMPLE Method for Predicting Incompressible Fluid Flows", Numerical Heat
Transfer Vol. 7, Iss. 2, p. 147-163.
Nomenclature
b ：half-width of the outer wall
c ：normalized half-width of the inner wall
cp ：specific heat capacity
d ：inner diameter
D ：mass diffusivity
Di ：concentration-driven diffusion coefficient
Dij ：binary diffusion coefficient
DiT ：thermal-diffusion coefficient

DoO ：mean diffusion coefficient evaluated for the oxidizer at the oxidizer stream
temperature
f ：focal length; mixture fraction
Fr ：Froude number
fst ：stoichiometric value of mixture fraction
gx ：gravitational acceleration in the x-direction
h ：enthalpy
I ：ratio of the actual initial momentum to that for uniform flow
k ：thermal conductivity
kij ：thermal-diffusion ratio
lD ：radial position of the flame edge
Le ：Lewis number
Lf ：flame length
M ：molecular weight
NDB ：diffusion to buoyancy parameter
NDM ：diffusion to momentum parameter
p ：pressure
Pe ：Peclet number
Pi ：power of incident laser
Qf ：volumetric fuel flow rate
Re ：Reynolds number
Ro0 ：universal gas constant
S ：molar stoichiometric oxidizer---fuel ratio
Sc ：Schmidt number
T ：temperature
u* ：dimensionless axial velocity component
ud ：molecular- diffusion velocity
ue ：fuel exit velocity
w ：species production rate
Wi ：atomic weight of species i
x* ：flame shape
xfl* ：dimensionless flame height

YF, stoic ：stoichiometric fuel mass fraction

Yi ：mass fraction of species i
YO0 ：mass fraction of the oxidizer at the oxidizer stream
Zf* ：normalized flame height

λ ：wave length; thermal conductivity

μ ：viscosity
ν ：(u, v) velocity vector; kinetic viscosity
ζ ：similarity variable
ξ ：mixture fraction
ρ ：density
Φ ：equivalence ratio
γ