You are on page 1of 12

Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

CHAPTER 8

LAMINAR DIFFUSION FLAMES


Diffusion flames are different from premixed flames in that the fuel and oxidizer are separated as
reactant and need to diffuse into each other prior to combustion. This difference causes many
consequences, for example,

1. The overall combustion rate of diffusion flames is controlled by the diffusion rate of the fuel and
oxidizer molecules, rather than by the reaction rate as for premixed flames. The chemical reaction
rate in diffusion flames is considered much faster than the diffusion rate.
2. Because of the mixing-controlled process, diffusion flames are generally not characterized by
flame propagation speed or flame thickness as for premixed flames, but rather by flame geometry,
e.g. flame length.
3. Soot is formed in significant quantity in diffusion flames, which cause a different color from
premixed flames. Soot formation can cause various problems for combustion and emissions.
4. Diffusion flames are considered safer than premixed flames because of the segregation of fuel
and oxidizer. The hazards for flashback and detonation are not with diffusion flames.
Diffusion flames find major applications in diesel engines and gas turbines in which liquid fuels are
burned in the form of droplets. In this chapter, we will discuss a classic and (relatively) simple laminar
jet diffusion flame using gaseous fuel, such as a Bunsen flame with closed air port.

1. Physical Description
A primary characteristic of a laminar jet flame is flame geometry, e.g., flame height and shape.

A schematic drawing of a laminar jet diffusion flame is shown in Fig. 8-1, in which two-dimensional
axisymmetry is assumed for the flame geometry as well as for the fields of velocity, temperature, and
species concentrations. As the fuel flows vertically along the flame axis, it diffuses radially outward. At
the same time, the oxidizer diffuses radially inward. The flame surface (or reaction zone) is nominally
defined to exist where the fuel and oxidizer meet in stoichiometric proportions. The products formed at
the flame surface diffuse in the radial direction both inward and outward. For an over-ventilated flame,
where there is more than enough oxidizer in the surroundings to continuously burn the fuel, the flame
length, Lf, is nominally determined by the axial location where
φ (=
r 0,=
x L=
f) 1

Above the nominal flame surface, reactions continue to burn out the incomplete combustion products,
e.g. soot.

In diffusion flames, soot is formed at the fuel side of the reaction zone and is consumed in the upper
part of the flame. Depending on the fuel type and flame residence times, not all of the soot that is
formed may be oxidized in the high-temperature oxidizing regions. In this case, soot “wings” may
appear, with the soot breaking through the flame. This soot that breaks through is generally referred
to as smoke.

In this chapter, we will first discuss a simplified theoretical analysis to derive the flame length of a 2-D,
axisymmetric, laminar jet, diffusion flame. As will be noted, this analysis is significantly more complex
than the analysis for the 1-D planar premixed flame discussed in the last chapter. Our discussion will
focus on the strategies used to simplify the problem with only necessary mathematical derivations.
Following this analysis, the parameters controlling flame size and shape are discussed. Finally factors
controlling soot formation in laminar jet flames are discussed.

1
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Fig. 8-1 Laminar diffusion flame structure (from Turns’s book).

2. Simplified Theories for Laminar Jet Diffusion Flames


The earliest theoretical investigation of the laminar jet diffusion flame was done by Burke and
Schumann in 1928. Their theory assumed constant axial velocity and zero radial velocity, and
provided reasonably well predictions of flame length for axisymmetric (circular port) flames. Among
the following investigators, Roper proposed a new theory in 1977 that retained the simplicity of the
Burke-Schumann analysis but removed the constant-velocity assumption. Roper’s theory provides
reasonable estimates of flame length for both circular and noncircular nozzles.

For the 2-D axisymmetric steady system, we need conservation equations that can handle partial
differentials along the radial and axial directions. These equations are developed in the Chap. 8 of
Turns’s book, and listed below. We will use them directly omitting the due derivations. Following the
governing equations, the strategies used to simplify the problem are discussed. The solutions by
Burke and Schumann and by Roper are briefly reported at the end.

Primary assumptions

1. The flow is laminar, steady, axisymmetric, produced by a jet of fuel emerging from a circular
nozzle of radius R, which burns in quiescent, infinite reservoir of oxidizer.
2. Only three “species” are considered: fuel, oxidizer, and product. Insider the flame zone, only fuel
and products exist; beyond the flame, only oxidizer and product exist.
3. Fuel and oxidizer react in stoichiometric proportions at the flame. Chemical kinetics are assumed
to be infinitely fast, resulting in the flame being represented as an infinitesimal thin sheet. This is
commonly referred to as the flame-sheet approximation.
4. Species molecular transport is by simple binary diffusion governed by Fick’s law.
5. Thermal diffusivity and species diffusivity are equal, i.e. Le is unity.
6. Radiation heat transfer is negligible.
7. Only radial diffusion of momentum, thermal energy, and species is important; axial diffusion is
neglected.
8. The flame axis is oriented vertically upward.
Note that additional assumptions are used in the following for further simplifications.

2
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Basic conservation equations

Mass conservation

1∂ ∂
(r ρυr ) + (ρυ x ) =
0
r ∂r ∂x
Mass flow in the Mass flow in the
radial direction axial direction
(kg/m3/s) (kg/m3/s)

Axial momentum conservation

1 ∂ 1∂ 1∂ ∂υ
(r ρυ xυ x ) + (r ρυ xυr ) − (r µ x ) (ρ∞ − ρ )g
=
r ∂x r ∂r r ∂r ∂r
x-Momentum flow x-Momentum flow Viscous force Buoyant force
by axial convection by radial convection (kg/m2/s2) (kg/m2/s2)
(kg/m2/s2) (kg/m2/s2)

Species conservation

∂ ∂ ∂  ∂Y 
(r ρυrYi ) + (r ρυ xYi ) −  r ρD i  =
0
∂r ∂x ∂r  ∂r 
Mass flow of i due Mass flow of i due Mass flow of i due Net mass production rate of i
to radial convection to axial convection to radial diffusion due to chemical reaction
(kg/m3/s) (kg/m3/s) (kg/m3/s) (kg/m3/s)

 i , which is supposed to
Note that with the flame-sheet assumption, the chemical reaction term, m
"'

appear on the RHS of the equation, becomes zero for the fuel inside the flame sheet, and becomes
zero for the oxidizer outside the flame sheet. The flame itself is specified as a boundary condition in
this problem. Since there only three species, the mass fraction of the product can be found as
YPr =1 − YF − YOx
Energy conservation (Shvab-Zeldovich energy equation)

∂  ∂ cP dT 
r ρ D ∫
∂ ∂
(r ρυr ∫ cP dT ) + (r ρυ x ∫ cP dT ) −  =
0
∂r ∂x ∂r  ∂r 
 
Rate of sensible Rate of sensible Rate of sensible Rate of sensible
enthalpy transport enthalpy transport enthalpy transport enthalpy production
by radial convection by axial convection by radial diffusion by chemical reaction
(W/m3) (W/m )
3
(W/m3) (W/m3)

Similar to the species equation, the energy production term, − ∑h 0


f ,i
 i"' , which is supposed to be at
m
the RHS of the equation, is zero except at the flame boundary. This equation applies both inside and
outside the flame but with discontinuity at the flame, which again enters the problem as a boundary
condition.
 We have five conservation equations including two for species conservation. However, the flame,
which is what we are solving for, need be specified outside the equations as boundary conditions;
but we do not know where the flame is located!

 To eliminate the discontinuity and boundary problem, conserved scalars are used to recast the
governing equations so that the flame location can be solved and only simply boundary
conditions are needed.

3
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Resolving discontinuity with conserved scalars

Mixture fraction for species conservation

The two species conservation equations for the fuel and oxidizer can be replaced by the single
mixture-fraction conservation equation, which for the axisymmetric 2-D system is

∂ ∂ ∂  ∂f 
(r ρυr f ) + (r ρυ x f ) − r ρ D  = 0
∂r ∂x ∂r  ∂r 
Since mixture fraction is a conserved scalar (as discussed in Chap. 6), this equation is sourceless by
definition. It applies to the entire field and has no discontinuity at the flame. The boundary conditions
for mixture fraction can be found as

df
(0, x) = 0 (symmetry)
dr
f (∞ , x) = 0 (no fuel in oxidizer)
f (r ≤ R ,0) = 1 (fuel jet at nozzle exit)
f (r > R ,0) = 0 (no fuel outside nozzle)

The location of the flame can be readily determined when f (r , x) is solved, which is at f = fstoich .

Mixture enthalpy for energy conservation

From the Shvab-Zeldovich energy equation, the standardized enthalpy of the mixture, h, is also a
conserved scalar. So the energy conservation equation for the axisymmetric 2-D system can be
expressed as

∂ ∂ ∂  ∂h 
(r ρυr h) + (r ρυ x h) − r ρ D  = 0
∂r ∂x ∂r  ∂r 
Similar to the above mixture-fraction species conservation equation, this energy equation is
sourceless and applicable to the entire field without discontinuity at the flame. Boundary conditions for
h can be found as

dh
(0, x) = 0
dr
h(∞ , x) =hOx
h(r ≤ R ,0) =
hF
h(r > R ,0) =
hOx
The mass and axial-momentum equations remain as given above, unaffected by the use of conserved
scalar. The boundary conditions for the velocities are

υr (0, x) = 0
dυ x
(0, x) = 0
dr
υ x (∞ , x) =0
υ x (r ≤ R ,0) =
υe
υ x (r > R ,0) =
0

4
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

 By applying the conserved scalars, we now have four conservation equations without the
discontinuity problem, and the boundary conditions for species and energy equations can be
simply defined.

 The number of governing equations can be further reduced by applying non-dimensional


equations.

Further simplification with non-dimensional equations

By defining the following dimensionless variables,


x* ≡ x / R
r* ≡ r / R
υ x* ≡ υ x / υe
υr* ≡ υr / υe
h − hOx ,∞
h* ≡ h* = 1 at nozzle exit ( h = hF ,e ); h* = 0 far away from the nozzle ( h = hOx ,∞ )
hF ,e − hOx ,∞
ρ * ≡ ρ / ρe

plus the mixture fraction f , the dimensionless conservation equations can be written as

Mass conservation
1 ∂ * * * ∂
(r ρ υr ) + * (ρ *υ x* ) =
0 eq. 8-1
r ∂r
* *
∂x
Axial momentum conservation
∂ * * * * ∂ * * * * ∂  µ  * ∂υ x*  gR  ρ∞ * *
(r ρ υ xυ x ) + (r ρ υ xυ r ) −    r  =  − ρ r
∂x * ∂r * ∂r *  ρeυe R  ∂r *  υe2  ρe 
Species conservation
∂ * * * ∂ * * * ∂  ρ D  * ∂f 
(r ρ υ x f ) + (r ρ υ r f ) −  r =0
∂x * ∂r * ∂r *  ρeυe R  ∂r * 

Energy conservation
∂ * * * * ∂ * * * * ∂  ρ D  * ∂h* 
(r ρ υ x h ) + (r ρ υ r h ) −  r =0
∂x * ∂r * ∂r *  ρeυe R  ∂r * 

Also, the dimensionless boundary conditions for the above are

υr* (0, x * ) = 0
υ x* (∞ , x * ) =f (∞ , x * ) =h* (∞ , x * ) =0
∂υ x* ∂f ∂h*
= (0, x *
) = (0, x *
) = (0, x * ) 0
∂r * ∂r * ∂r *
υ x* (r * ≤ 1,0) = f (r * ≤ 1,0) = h* (r * ≤ 1,0) = 1
υ x* (r * > 1,0) = f (r * > 1,0) = h* (r * > 1,0) = 0

5
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Note that the conservation equations and boundary conditions for h and
*
f are of identical form.
ν µ
Also, if buoyancy is neglected, and assume Sc= = is unity, i.e. µ = ρ D , the momentum
D ρD
equation and boundary conditions for υ x* are also of identical form to h and f . Therefore the
*

following single conservation equation replaces the individual axial momentum, species, and energy
equations:

∂ * * * ∂ ∂  1 ∂ζ 
(r ρ υ x ζ ) + * (r * ρ *υr*ζ ) − *  r * *  =
0 eq. 8-2
∂x *
∂r ∂r  Re ∂r 

where the generic variable ζ= υ x*= f= h* and Re = ρeυeR / µ .

 With non-dimensional equations, we now have only two conservation equations to solve, eq. 8-1
and eq. 8-2. Note, however, that ρ* appears in every term of both equations and needs to be
solved.

 The relation between ρ* and f (or h*) can be established via state relationships.

Expressing density and temperature with mixture fraction using state relationships

Density can be related to mass fraction and temperature via ideal gas law
PMWmix
ρ= eq. 8-3
RuT
where
1
MWmix = eq. 8-4
∑ (Yi / MWi )
Our target is therefore to find Yi = Yi ( f ) , T = T ( f ) , so that ρ = ρ ( f ) can be obtained.

With the flame-sheet approximation, the definition of the mixture fraction can be used to relate YF, YOx,
and YPr to f across the field.

Inside the flame ( fstoich < f ≤ 1 )


f − fstoich 1− f
YF = ; YOx = 0 ; YPr =
1 − fstoich 1 − fstoich

At the flame ( f = fstoich )


YF = 0 ; YOx = 0 ; YPr = 1

Outside the flame ( 0 ≤ f < fstoich )


f f
YF = 0 ; YOx = 1 − ; YPr =
fstoich fstoich

1
where fstoich = for the reaction 1 kg fuel + ν kg oxidizer → (ν + 1) kg products . Figure 8-2
1 +ν
shows YF, YOx, and YPr as a function of mixture fraction, where linear relations are seen for all three
species.

6
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Fig. 8-2 Mixture fraction in the simplified jet diffusion flame model with flame-sheet approximation (left).
Species mass fraction and temperature as a function of mixture fraction (right) (from Turns’s
book).

To determine temperature as a function of mixture fraction, further assumptions are made: 1) the
specific heat is constant and equal for all species; 2) enthalpies of formation for the oxidizer and
product are zero, i.e., ∆hC =
hF0 .

With these assumptions, the calorific equation of state for the mixture becomes

h = ∑Yi hi =YF ∆hC + cP (T − Tref )


Using the definition of h* above mentioned, an expression for the dimensionless enthalpy can be
obtained in the form of T. Since f = h from eq. 8-2, the relation between mixture fraction and
*

temperature can thus be obtained, where the temperature is expressed explicitly as a function of the
mixture fraction, T = T ( f ) . Consequently, the temperature inside the flame sheet, at the flame, and
outside the flame sheet can be obtained in terms of the mixture fraction (details available in Turns’s
book).

 In summary, we have obtained Yi = Yi ( f ) and T = T ( f ) , which allow us to express ρ * = ρ * ( f )


(using eq. 8-3 and 8-4) and subsequently solve the two conservation equations, eq. 8-1 and 8-2.

Solutions for flame length


Burke and Schumann made the earliest attempt to analyze circular two-dimensional laminar jet
diffusion flames in a way similar to the above. In order to solve the equations, they assumed a single
velocity across the field ( υ x = constant , υr = 0 ). The constant axial velocity eliminated the need to
solve the axial momentum equation and by default, neglected buoyancy. The zero radial velocity
results in ρυ x = constant from the mass conservation equation. Therefore, the problem is reduced
to solve only one conservation equation, although the solution of which is still complicated
mathematically. The flame lengths predicted by Burke-Schumann theory are in reasonable agreement

7
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

with the experiment. This somewhat surprising agreement is, recognized by themselves and
confirmed later by others, primarily due to offsetting assumptions of neglecting buoyancy and
neglecting radial diffusion. In reality, the effect of buoyancy is to cause the flame to be narrower,
which in turn, increases diffusion rates that tend to cause the flame to be wider.

Roper published a new theory in 1977 that retained the simplicity of the Burke-Schumann analysis but
removed the constant-velocity assumption. Roper’s theory provides reasonable estimates of flame
length for various burner geometries (circular, square, slot etc.) and flow regimes (momentum-
controlled, buoyancy-controlled, and transitional). Roper’s theoretical results for circular-port burner is
shown below, which applies regardless of whether or not buoyancy is important, and are applicable
for fuel jets emerging into either a quiescent oxidizer or a coflowing stream, as long as the oxidizer is
in excess, i.e. the flames are over-ventilated.
0.67
QF (T∞ / TF )  T∞ 
Lf =  
4π D∞ ln(1 + 1 / S)  Tf 
where QF is the initial volumetric flow rate, S is the molar stoichiometric oxidizer-fuel ratio, D∞ is a
mean diffusion coefficient evaluated for the oxidizer at the oxidizer stream temperature, T∞ , and TF
and Tf are the fuel stream and mean flame temperature, respectively.
It is important to note that for circular-port flames, the flame length does not depend on initial velocity
or port diameter but rather on the initial volumetric flow rate, QF.

Numerical solutions using computers and finite-difference techniques allow laminar jet flames to be
modeled in much greater detail than the analytical approach discussed above. Shown in Fig. 8-3 is
the calculated CH4-air flame at the same conditions as the simplified theories discussed above but
with detailed velocity, temperature, and species-concentration fields resolved. A chemistry set of 79
reactions and 29 species was used, and both radial diffusion and axial diffusion were considered.
From these experimentally validated calculations, we can see that the flame sheet (reaction zone), as
indicated by the peak OH concentrations, is well below the flame tip which is at ~ 6 cm height. Also
the reaction zone is very thin near the lower edges of the flame, but is considerably broadened near
the top due to buoyancy effect.

Fig. 8-3 Numerical solutions of temperature (left) and OH molar fraction (right) in CH4-air laminar jet
diffusion flame. (Source: Smooke et al. 23rd International Symposium on Combustion, 1990)

8
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

3. Factors Affecting Flame Length through Stoichiometry


The molar stoichiometric ratio, S, is important for determining flame length. The definition of S is in
terms of the nozzle fluid (fuel) and the surrounding reservoir fluid (oxidizer), i.e.

 moles of ambient fluid 


S = 
 moles of nozzle fluid  stoich

Thus, S depends on the chemical compositions of both the nozzle-fluid stream and the surrounding
fluid. The following parameters, all of which affect S, are of interests in many applications.

Fuel type

The molar stoichiometric air-fuel ratio for a pure fuel with the formula CxHyOz can be expressed as

x +y /4−z /2
S= eq. 8-5
xO2

where xO2 is the molar fraction of oxygen in the air.


The flame lengths, relative to methane, of H2, CO, and C1-C4 paraffins are shown in Fig. 8-4, which
were calculated for circular-port burner using Roper’s equation. It is seen that flame length increases
as H/C ratio of the fuel decreases. This trend is strong for small hydrocarbons but becomes less
significant for larger hydrocarbons.

Fig. 8-4 Dependence of flame length on fuel stoichiometry. (from Turns’s book).

Primary Aeration

Gas-burning appliances typically premix 40-60% of the stoichiometric air with the fuel gas before it
burns as a laminar jet diffusion flame. This primary aeration is to make the flame short and prevent
soot formation. In this case, the stoichiometric ratio can be evaluated by treating the nozzle fluid as a
mixture of the fuel and primary air:

1 −ψ pri
S=
ψ pri + (1 / Spure )

9
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

where ψ pri is the fraction of the stoichiometric requirement met by the primary air, and Spure is the
molar stoichiometric ratio associated with the pure fuel.

The effect of primary aeration on the length of methane flames on the circular-port burner is shown in
Fig. 8-5. Note that in the range of 40-60% primary aeration, flame lengths are reduced by 85-90%
from their original no-air-added lengths. The maximum amount of air that can be added, the rich limit,
is controlled by safety considerations, e.g. flashback.

Fig. 8-5 Effect of primary aeration on laminar jet flame lengths. (from Turns’s book).

Oxygen content of oxidizer

The amount of oxygen in the oxidizer has a strong influence on the flame lengths, as seen in Fig. 8-6.
Small reductions in O2 content from the normal 21% value for air result in greatly lengthened flames.
With pure O2 oxidizer, flame lengths for methane are ~ 1/4 their value in air. The same equation for
evaluating the fuel type effect, eq. 8-5, can be used to calculate the effect of O2 content.

Fig. 8-6 Effect of oxygen content in the oxidizing stream on flame length. (from Turns’s book).

10
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Fuel dilution with inert gas

Diluting the fuel with an inert gas also has the effect of reducing flame length via its influence on the
stoichiometric ratio. For a fuel of CxHyOz formula,

x +y /4−z /2
S=
 1 
  xO2
 1 − xdil 
where xdil is the diluent molar fraction in the fuel stream.

4. Soot Formation and Destruction


Soot formation is an essential feature of non-premixed hydrocarbon flames. As discussed previously,
the yellow color of diffusion flames is caused by the blackbody radiation of soot particles inside flame.
The radiant heat transfer to the combustor liner could be detrimental to the durability of the materials
in gas turbine engines. Soot is also a major pollutant from diesel engines which utilize non-premixed
combustion for propulsion.

With some knowledge about hydrocarbon combustion chemistry, it is no wonder that the mechanism
of soot formation is exceedingly complex, which in fact remains as the one of the most challenging
problems for combustion researchers. It is generally agreed that soot is formed over a limited range of
temperatures, say 1300 K < T < 1600 K, in diffusion flames and the formation proceeds in four steps.

1. Formation of precursor species


2. Particle inception
3. Surface growth and particle agglomeration
4. Particle oxidation

The process of forming precursor species involves first breaking down the fuel molecules to small
species. The small species, mainly C2H2 and C3H3, form the first aromatic ring which then grows into
polycyclic aromatic hydrocarbons (PAHs). Through chemical reactions as well as coagulation, the
PAHs grow to a critical size (3000 – 10,000 atomic mass units) and transform from large molecules
into small particles. The small particles experience surface growth and agglomeration and grow
further in size. Soot gets oxidized as the particles pass the reaction zone (φ = 1) and enter the
oxygen-rich region (φ < 1). For a vertical jet diffusion flame, the soot is always formed interior to the
reaction zone lower in the flame (φ >> 1), and is oxidized primarily in the upper part of the flame
exterior to the reaction zone. Figure 8-7 shows the temperature and soot formation at an axial location
for an ethene (C2H4) jet flame, from which the location of soot formation relative to the reaction zone
can be clearly identified. Diffusion flames may emit soot (or smoke) if the formed soot is not
completely oxidized before the flame contour closes at the flame tip.

Fuel molecular structures have strong impact on soot formation. The tendency of a fuel to soot can be
experimentally determined by the smoke point test. Smoke point tests measure the maximum flame
length or fuel flow rate, which are linearly related for circular-port burner, without soot (smoke) emitted
at the flame tip. The larger the flame length or fuel flow rate, the lower the sooting propensity of the
fuel. In the order of increasing sooting propensity, the rank of fuel type is generally, alkanes < alkenes
< alkynes < aromatics.

11
Advanced Thermodynamics - Semester 2 2016 - Lecture Notes

Fig. 8-7 Radial profiles of temperature and soot formation (indicated by scattered light) for a laminar
ethene jet diffusion flame. (from Turns’s book)

12

You might also like