DYNAMICS OF COMBUSTION SYSTEMS:

FUNDAMENTALS, ACOUSTICS, AND CONTROL

A Short Course of Lectures

V. FUNDAMENTALS OF COMBUSTION
PROCESSES

F.E.C. CULICK
California Institute of Technology
September 2001

 2001 by [Link]. Culick

All Rights Reserved

5-1
 5.1 GOVERNING EQUATIONS FOR ONE-
DIMENSIONAL LAMINAR REACTING FLOWS

Most of the essential ideas on which the equations of motion are

based arise already in analysis of quasi-one-dimensional flows.

However, it is an approximation to construct analyses with the

assumption that properties of the flow vary only in one dimension;
much significant behavior can be treated only in two or three
dimensions.

5.1.1 Global Conservation of Mass

This principle is expressed as the ‘continuity’ equation
with a source of mass:

∂ (ρS c ) ∂
+ (ρuS c ) = wS c
∂t ∂x

wSc=(ρu)sqc, mass addition at the lateral

mass boundary; qc:= circumference
[w]: = e.g.
L3 − s mass converted from condensed phases to
gas within the volume

Rewrite in the form:

∂ρ 1 ∂
+ (ρuSc ) = w
∂t S c ∂x

5-2
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.2 Equations for Conservation of Species

Consider a fixed control volume, a slab perpendicular to the
local average direction of flow (parallel to the axis of a
channel):

∂ρi 1 ∂
+ (ρi ui Sc ) = 1 wi + 1 wci
∂t S c ∂x Sc Sc

Sum these equations over all species to give the

continuity equation and
ρ = ∑ ρi
ρu = ∑ ρi ui
we = ∑ wei
∑ wi = 0

5-3
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.2 Equations for Conservation of Species (cont’d)

Hence the correct bulk velocity is the mass-averaged value:

1
u=
ρ
∑ ρ i ui

ρi
Mass concentration Yi = :
ρ

∂ (ρYi ) 1 ∂
+ (ρui ScYi ) = 1 wi
∂t S c ∂x Sc

Diffusion velocity Vi:

ui = u + Vi

Expand and apply global conservation of mass:

∂Yi ∂Y 1 1 ∂
ρ + ρu i = wi − (ρViYi )
∂t ∂x S c S c ∂x

5-4
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.2 Equations for Conservation of Species (cont’d)

A property of the diffusion velocity:
Definition: ui = u + Vi

⇒ ∑ ρi ui = (∑ ρi ) u + ∑ ρiVi
ρu = ρu + ∑ ρiVi r
ˆ ⋅u
n

∑ ρiVi =0

because Vi is measured with respect to the mass averaged

velocity

Fick’s Law:
∂Yi
ρiVi = −ρ D
∂x Assume: D is same for all species.
In general, D is a matrix
∂Yi
YiVi = − D
∂x

Equations for conservation of species:

∂Y ∂Y 1 ∂  ∂Yi 
ρ i + u i  =  ρ D Sc  + wi
 ∂t ∂x  S c ∂x  ∂x 

5-5
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.3 Momentum Equation

The principle of conservation of momentum must strictly be
applied in the first instance to flow of the mixture:

(ρu )s u|| : axial momentum added at lateral boundary

u|| is speed parallel to u

∂u ∂u ∂p 1 ∂  ∂u  1
ρ + ρu = − +  c  + (u|| − u )(ρu )s qc
µS
∂t ∂x ∂x Sc ∂x  ∂x  Sc

qc:= circumference

Two common assumptions: µ(T) ≈ constant

p(x) ≈ constant

5-6
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.3 Momentum Equation (cont’d)

Example: flame, thickness δ

ρ1 ρ2
u1 u2 ρ 2 < ρ1
δ u2 > u1

Steady flow:

dp du d 2u ∆u ∆u
= −ρu + µ 2 ~ − m& +µ 2
dx dx dx δ δ
δ dp u 2  µ 
~ 1− 
p dx p/ρ  ρuδ 
1
~ M 2 1− 
 Re 

M ~ 0.01 or less in flames

Hence low-speed combustion in narrow regions

takes place at nearly constant pressure

5-7
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.4 Energy Equation

• Literature sometimes confusing to read because the energy

equation can be written in several equivalent forms

• As shown above, the equation should be written for mass-

averaged values:

ρe = Σρi ei ; ρh = Σρi hi etc.

The primitive form of the energy equation must be written

for the stagnation energy (internal and kinetic) of the
mixture, e0 = (1/ρ)Σρie0i :

∂
(Sc Σρi e0i )+ ∂ (Sc Σρi ui e0i )+ ∂ (Sc Σpi ui )
∂t ∂x ∂x
∂ ∂T ∂ ∂u
=  Sc k  +  Sc µu  + h0 s ( pu )s qc
∂x  ∂x  ∂x  ∂x 
dissipation

Note: For simplicity, sources at the lateral boundary

are no longer accounted for in this analysis

5-8
 5.1 ONE-DIMENSIONAL EQUATIONS

5.1.4 Energy Equation (cont’d)

Note: for some systems, the energy of condensed species must be
accounted for by adding ρli, eli, etc.

Rewrite:
Vi 2
∑ρiui e0i = ρue0 + u ∑ρi 2 e0i + ∑ρiVi e0i
∑ piui = pu + ∑ piVi
Vi 2
∑ρi e0i = ρe0 + ∑ ρi 2

∂h ∂h ∂p ∂p 1 ∂ ∂T
ρ + u  =  + u  +  S c k 
 ∂t ∂x   ∂t ∂x  Sc ∂x  ∂x 
1 ∂
− (Sc ∑ piVi ) − ∂ (Sc ∑ρiVi ei )
Sc ∂x ∂x
∂ Vi2  1 ∂  Vi2 
−   ∑ ρi
 
 + 
 Sc ∑ ρiVi  + Φ
 ∂x  2  Sc ∂x  2 
dissipation
K.E. of diffusion (ignore) (ignore)

Dh Dp 1 ∂  ∂T  1 ∂
ρ = +  Sc k  − (Sc ∑ρiVi hi )
Dt Dt Sc ∂x  ∂x  Sc ∂x

p
h = e+
ρ
5-9
 5.2 EQUATIONS FOR MODELING LOW-SPEED
DEFLAGRATIONS (FLAMES)
5.2.1 Muster of Equations
Constant Area

∂ρ ∂
+ ( ρu ) = 0
∂t ∂x
∂Y ∂Y ∂ ∂Y
ρ i + ρu i =  ρ D i  + wi
∂t ∂x ∂x  ∂x 
∂u ∂h ∂p ∂ ∂u
ρ + ρu = − +  µ 
∂t ∂x ∂x ∂x  ∂x 
∂h ∂h ∂p ∂p ∂ ∂T ∂ ∂Y
ρ + ρu = + u +  k  + ∑ ρ D hi i
∂t ∂x ∂x ∂x ∂x  ∂x  ∂x ∂x

Steady Flow

Approximations: (i) p ≈ constant

 equivalent to  (ii) kinetic energy of diffusion ignored
Shvab-Zel’dovich (iii) viscous dissipation ignored
 approximations 
  (iv) binary diffusion
 

Conservation of mass ρu = m

dYi d  dYi 
Species concentrations m =  ρD  + wi
dx dx  dx 

Conservation of energy
dh d  dT  d  dYi 
m = k  +  ∑ ρ D hi 
dx dx  dx  dx  dx 

5-10
 5.2 EQUATIONS FOR MODELING LOW-SPEED
DEFLAGRATIONS (FLAMES)

5.2.2 Alternative Forms of the Steady Energy Equation

(Sc constant)

dh d  dT  d  dYi 
Initial Form: m = k  +  ∑ ρ D hi 
dx dx  dx  dx  dx 

Temperature as Dependent Variable:

dT d  dT  dT dYi
mC p = k 
dx dx  dx 
+ Qw + ρ D
dx
∑ pi dx
C

For reaction ∑ ν i Ai = 0
wi = ν i w ; ∑ hi wi = w∑ ν i hi = −Qw
Q = −∑ ν i hi

If all Cpi are equal:

dT d  dT 
mC p = k  + Qw (a common form)
dx dx  dx 

5-11
 5.2 EQUATIONS FOR MODELING LOW-SPEED
DEFLAGRATIONS (FLAMES)

5.2.2 Alternative Forms of the Steady Energy Equation

(Sc constant) (cont’d)

Enthalpy as Independent Variable:

dh d  k dh  d  k dYi 
m = + ∑ i (Le −1) 
k
dx dx  C p dx  dx  Cp dx 

k energy transfer by conduction

Lewis number : Le = ~
ρ DC p energy transfer by diffusion

Thermal Enthalpy as Independent Variable:

dhT d  dhT  + d k( 1 − Le) dT  + Q w

m =  ρD 
 dx  dx 
0
dx dx  dx

T
1 1
hT = ∑ ρi hiT hiT = ∫ C pi dT ′ Q0 = ∑ h0iρi
ρ T0
ρ

5-12
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION

5.3.1 Steady Flow in a Uniform Duct with Heat Addition

Differential equations for flow with simple heat addition:

Conservation of Mass ρdu + udρ = 0

Conservation of Momentum ρudu + dp = 0

Conservation of Energy dh + udu = dqe − d (∑ hi 0Yi ) = dq

dT dp dT dρ
Entropy Equation ds = C p = R0 = Cv − R0
T p T ρ
dp dρ dT
Equation of State = +
p ρ T

Differential changes of properties with heat addition:

du dρ 1  dq  CHIEF CONCLUSIONS:
=− =
u ρ 1− M 2  C pT  If the area is constant and frictional
losses are ignored:
dp γM 2  dq 
=− (i) Heat addition in subsonic flow
p 1− M 2  C pT 
⇒ acceleration to M = 1
dT 1− γM 2  dq 
=− (ii) Heat addition in supersonic flow
T 1− M 2  C pT 
⇒ deceleration to M = 1
dS  dq 
=  

Cp C T
 p 

5-13
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.2 Changes of State for Steady Flow with Finite Heat Addition

ρ1, u1, p1, T1, s1

q≠0
ρ2, u2, p2, T2, s2

Relations correcting properties in two states of equilibrium:

ρ1u1 = ρ 2u 2
p1 + ρ1u12 = p2 + ρ 2u22
u12 u22
C T1 + = C T2 + − q
p1 p1
(h02 > h01 )
2 2
T2 /T1 C =C 
s2 − s1 = C log  p1
p γ −1
p 2 
( p2 p1 ) γ 
γ −1 2 
Stagnation Temperature : T0 = T 1+ M 
 2 

ρu = constant A unique relation p(ρ) exists, defining the

⇒ “Rayleigh line,” e.g., in the h–s plane as well
p + ρu 2 = constant
as in the p–ρ or p–v plane

5-14
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.2 Changes of State for Steady Flow with Finite Heat Addition
(cont’d)
If Cp is constant, the change of stagnation temperature with heating
is:
1
T02 − T01 = q
Cp

Values of the properties in the end states can be expressed in terms

of Mach number, e.g.:
2 1+
γ −1 2
M2
T02  M 2 (1+ γM 1 ) 2
=
T01  M 1 (1+ γM 2 ) 1+ γ −1 M 2
1
2

The conventional method for solving practical problems is based on

using reference values ( )* for M = 1:
2
T02  M   γ −1 2 
= 2( γ +1)  1+ M 
 1+ γM   
2
T01 2
p γ +1
=
p ∗ 1+ γM 2
u ( γ +1)M 2
=
u ∗ 1+ γM 2
γ +1
∗  
s−s  γ +1  γ
= ln  M 
2
 
Cp   1+ γM  
2

 

5-15
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.2 Changes of State for Steady Flow with Finite Heat Addition
(cont’d)

These formulas give the chart below, taken from A.S. Shapiro,
Dynamics and Thermodynamics of Compressible Flow, Volume 1,
page 169.

q is a parameter
along the curves

5-16
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.2 Changes of State for Steady Flow with Finite Heat Addition
A convenient global view of the general situation is based (cont’d)
on the chart below showing the final Mach number as a
function of the initial Mach number:

M 22 =
1
( 1+ γα2 −β)±(1+ γα) (1−α)2 + γ +1β
 γ-1 β γ 
2γ α − +  
 2γ 2 
γ −1 2 
α = M12 β = 2γQ M12 1+ M
 2 1 
q
Q=
C T01
p

upper upper
CJ CJ

(Taken from
Shapiro, ibid.)

5-17
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION

5.3.3 The Rankine-Hugoniot diagram

The Rankine-Hugoniot diagram is a useful representation of
gasdynamic ‘discontinuities’ showing many general properties
without involving the equation of state. It is based on the
integrated one-dimensional equations for steady flow, evaluated for
the initial and final state of equilibrium upstream and downstream
of a region of ‘discontinuity’:
ρ1u1 = ρ 2u 2 = m u12 u 22
h1 + = h2 +
p1 + ρ1u12 = p2 + ρ 2u 22 2 2
(i) Combine the equations for conservation of mass and momentum
to form the relation defining the Rayleigh Line:
p2 − p1 p2 − p1
= u1u 2 or = m2
ρ 2 − ρ1 v 2 − v1
(i) Combine these three equations of conservation to give:
h2 − h1 = 12 ( p2 − p1 )(v 2 + v1 ) (∆h = ∆p v )
e2 − e1 = 12 ( p2 + p1 )(v1 − v 2 ) (∆e = p∆v )
which define the Rankine-Hugoniot curve, a plot of the final
properties p2 versus v2 for a given initial state:
 
( p )v −  γ +1( p v )+ 2Q 
1

 γ −1
1 2 1 1 0

p2 = 1 
 
(v )−  γ +1  v
2
(C p constant )
 γ 2 −1 
1 2

h = hr + C p (T − Tr ) Q0 = h1r − h2 r

5-18
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION

5.3.3 The Rankine-Hugoniot diagram (cont’d)

Some properties of the Rankine-Hugoniot curve:

γ2 −1 ρ2 γ 2 + 1
(i ) p2 → ∞ : v 2 = v1 i.e., = >0
γ2 +1 ρ1 γ 2 − 1

γ 1 +1 Q
(ii ) p → 0: v = v1 + 2 0 > 0 if Q0 > 0
γ 1 −1
2 2
p1

  γ 2 −1  Q0 
(iii ) v = v : p = p  12 (γ +1)  + (γ −1)  > p
γ −
2 1 2 1 2 2 1
  1  1
p  1

if Q0 > 0 and γ 2 > γ 1

p2 − p1
(iv ) = −m 2 < 0
v 2 − v1
⇒ portion of the curve p2 > p1 AND v 2 > v1 is inaccessible

( v ) Slope of chord from the initial state (p ,v ) to a final state (p ,v )

1 1 2 2

is a measure of the mass flow : − slope = m

( vi ) For a given value of Q , the mass flux m has stationary values

0

at points of tangency with the line ( p1 ,v1 ) − ( p2 ,v 2 )

5-19
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION

5.3.3 The Rankine-Hugoniot diagram (cont’d)

Some properties of the Rankine-Hugoniot curve (cont’d):

dm 2 2T2
(vii) = > 0 : mass flux and entropy increase together
dS 2 (v1 − v 2 )2

dm 2 1-M 22
(viii) = = 0 at M 2 = 1
dp2  m 2  ∂v 2  
(v1 − v 2 )1 − (v1 − v 2 )   
 2T2  ∂p2  s2 

(ix ) For a given value of Q0 , at the tangent points where the

mass flux is maximum (minimum), M 2 = 1 and the entropy
change is minimum where p2 > p1 (maximum where p2 < p1 )
Q0 > 0 : p2 > p1 : detonation
p2 < p1 : deflagration

References: W.D. Hayes, “The Basic Theory of Gasdynamic

Discontinuities”, Section D, in Fundamentals of
Gasdynamics, Princeton University Press, 1958.
L.D. Landau and E.M. Lifshitz, Fluid Mechanics.

5-20
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION

5.3.3 The Rankine-Hugoniot diagram (cont’d)

5-21
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION

5.3.3 The Rankine-Hugoniot diagram (cont’d)

Characteristics of the Rankine-Hugoniot Curves

Flow Segment of M1 M2 Pressure Remarks
Curve Ratio
DETONATIONS
Strong p2
E-F M1 > MCJ M2 < 1 > PCJ Rarely Observed
Detonations p1

Upper C-J p2 Commonly

E M1 = MCJ M2 = 1 = PCJ Observed
Point p1

Weak p2 Rarely Observed

D-E M1 > MCJ M2 > 1 < PCJ (Special F/O
Detonations p1
required)
DEFLAGRATIONS
Weak p2 Commonly
A-B M1 < 1 M2 > 1 < Pcj Observed Flames
Deflagrations p1
(p2/p1 ≈ 1)

Lower C-J p2
B M1 = Mcj M2 = 1 = Pcj Not Observed
Point p1

Strong p2
B-C M1 < Mcj M2 > 1 > Pcj Not Observed
Deflagrations p1

[Adapted from Table 1.7, Liñán and Williams (1993)

Fundamental Aspects of Combustion]

5-22
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.4 Some Notes on General Characteristics and Classifications of
Reacting Flows
• Classifications of flows are usually defined by the relative
significance of the relevant physical processes.
• The physical processes can be divided into two groups
according to their origins in the principles of conservation of
mass, momentum, and energy:
Principle Physical Property or Process
Conservation of Mass Compressibility
Diffusion
Conservation of Momentum Inertial forces (convection)
Pressure
Viscous forces
Body forces
Conservation of Energy Convection of energy
Heat conduction
Diffusion of energy
Energy addition by chemical
reaction
• Superposed on these effects is the distinction between
laminar and turbulent flows.
• The Rankine-Hugoniot diagram provides a convenient clear
classification of simple reacting flows according to the
intensity of heat addition and the Mach number of the flow.
• Problems of combustion system dynamics reside largely in
the range of low Mach number and large rates of energy
addition.

5-23
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.4 Some Notes on General Characteristics and Classifications of
Reacting Flows (cont’d)
• Writing the conservation equations of motion in dimensionless
variables (x/L, u/ur … etc.) where ( )r are reference values,
produces the following dimensionless groups of properties
characterizing the behavior of a reacting flow:

Dimensionless Group Remarks

u flow speed
Mach Number M= ; M r → 0 for
a speed of sound
incompressible flow (small
disturbances propagate with
infinite speed)

ρuL inertial forces

Reynolds Number Re = ; characterizes
µ viscous forces
transition from laminar to
turbulent flow

C pµ v viscous diffusivit y
Prandtl Number Pr = =
k α thermal diffusivit y

µ v viscous diffusivit y
Schmidt Number Sc = =
ρD D mass diffusivit y

5-24
 5.3 STEADY ONE-DIMENSIONAL FLOWS WITH
FINITE HEAT ADDITION
5.3.4 Some Notes on General Characteristics and Classifications of
Reacting Flows (cont’d)

Sc α thermal diffusivit y
Lewis Number Le = = ; Le = 1
Pr D mass diffusivit y
indicates a condition of
‘similarity’ between the
diffusion of species and
diffusion of temperature

L/u flow 'residence' time

Damkohler Number Da =
τchem reaction time

Ea
Activation Energy βa =
RT

∆h
Heat of Reaction γr =
RT

Zel'dovich Number Ze = γ r β a

ur2 kinetic energy

Froude Number Fr =
gL gravitational potential energy

5-25
 5.4 A GALLERY OF FLAMES

A. Laminar Premixed Flames

From Weinberg, Optics of Flames (1963)

5-26
 5.4 A GALLERY OF FLAMES

A. Laminar Premixed Flames (cont’d)

From Lewis & von Elbe,

Combustion, Flames and Explosions of Gases (1951)

5-27
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames

From Weinberg, Optics of Flames (1963)

From Williams, Combustion Theory (1964)

5-28
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d_

From Williams, Combustion Theory (1964)

5-29
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d)

From Williams, Combustion Theory (1964)

5-30
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d)

From Takagi et al. (1996), 26th Symposium

(International) on Combustion, (pp. 1103-1110)

5-31
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d)

From Shu et al. (1998), 27th Symposium

(International) on Combustion, (pp. 625-632)
5-32
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d)

FIG 1. Schematic of the experimental setup.

The inset is a schematic of a forced, time-
varying, unconfined, axisymmetric laminar
diffusion flame

From Mohammed et al. (1998), 27th Symposium

(International) on Combustion, (pp. 693-702)
5-33
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d)

From Katta et al. (1998), 27th Symposium

(International) on Combustion, (pp. 1121-1129)
5-34
 5.4 A GALLERY OF FLAMES

B. Non-Premixed Flames (cont’d)

From Muller and Scheffes (1998), 27th Symposium

(International) on Combustion, (pp. 1105-1112)

Note: • Flame produced at the interface of a fuel jet in a

rectangular slot and a coflowing air stream (lower speed).

• Vortices produced by acoustic forcing of the fuel stream.

5-35
 5.4 A GALLERY OF FLAMES

C. Turbulent Premixed (with swirl)

OH-PLIF

From Plessing et al. (2000), 28th Combustion Symposium, (pp. 359-366)

Note: • Premixed flame is produced by a nozzle with swirler

(Bedat and Chang) having diameter 50 mm.

5-36
 5.4 A GALLERY OF FLAMES

D. Turbulent Non-Premixed

From Ratner et al. (2000), 28th Combustion Symposium, (pp. 245-252)

5-37
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

Low-speed combustion phenomena fall into two classes defined by the

condition of the reactants: premixed or non-premixed.
Correspondingly, the two fundamental combustion processes are the
laminar premixed flame and the laminar non-premixed or
diffusion flame.

The purpose of this section is to clarify the basic distinguishing

characteristics of these two phenomena.

Possibly the most fundamental difference is that a premixed flame

propagates into the mixture of reactants with a well-defined laminar
flame speed. A diffusion flame does not have such a property.

The laminar flame speed depends chiefly on the particular reactants

involved, the relative proportions of fuel and oxidizer  the
fuel/oxidizer ratio or the equivalence ratio, or more generally the
composition of the mixture  and on the initial temperature of the
reactants.

Related to the existence of a flame speed, premixed flames exhibit lean

and rich “propagation limits”: the flame ceases to propagate if there is
too little or too much fuel.

In contrast, a diffusion flame can exist at the interface of regions

containing separated fuel and oxidizer without propagating. Fuel and
oxidizer are supplied by diffusion to the combustion zone.

5-38
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

Typical Values of Laminar Flame Propagation Speed

F Fuel
: mixture ratio
O Oxidizer

F value of mixture ratio at stoichiometric conditions,

  :
 O s when all fuel and oxidizer burn to products only:
e.g. O2 + 2H2 = 2H2O
F/O
φ: = Equivalence ratio
(F/O)s

From Liñán and Williams (1993) Fundamental Aspects of Combustion

5-39
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

Examples of Simple Premixed and Nonpremixed Laminar Flames

Methane flame on a Bunsen burner

Premixed Nonpremixed

From Liñán and Williams (1993) Fundamental Aspects of Combustion

5-40
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

5.5.1 An Elementary Laminar Premixed Flame

The simplest conceivable example is that described earlier, propagation

of a front in a tube, or a flame above a porous plate. In those and other
cases, it is reasonable to assume that all flow properties vary only in
the direction of propagation, a one-dimensional steady flame.

A second characteristic simplifying analysis is the relatively high rate

of reaction following ignition of fuel and oxidizer. As a direct
consequence, the region of vigorous chemical reaction is relatively thin
compared with the length over which the temperature rises from the
initial value upstream of the flame to the adiabatic flame temperature.
The reason for that behavior is the normally high value of activation
energy and is the basis for a strategy used in combustion theory,
“activation energy asymptotics” (AEA).

Thus the structure of a laminar premixed flame is roughly:

5-41
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES
5.5.1 An Elementary Laminar Premixed Flame (cont’d)
One simple interpretation of steady flame propagation is the following:
(i) energy released in the combustion processes causes high
temperature in the reaction zone;

(ii) the incident flow is heated by thermal conduction (and possibly

radiation) to the temperature necessary for ignition, the
upstream edge of the reaction zone;

(iii) the speed of the incident stream must be such that the flow of
reactants is sufficient to support the rate of combustion, and to
compensate the conduction of heat upstream in order to
maintain a steady distribution of temperature. (This condition
expresses the physical content of the formal result that the
laminar flame speed is found as the consequence of an
eigenvalue problem.
That behavior is essentially the basis for the “thermal theory of flame
propagation” constructed with the following assumptions:
(i) low speed so the pressure is constant;
(ii) all diffusion processes ignored.

In addition, we assume here:

(i) one-step Arrhenius reaction rate;
(ii) constant specific heats.

The basic thermal theory provides results for the laminar flame speed
and the distributions of flow properties, the flame structure.

5-42
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

5.5.1 An Elementary Laminar Premixed Flame (cont’d)

A simple estimate* of the laminar flame speed (Landau and
Lifshitz):

δ thickness of reaction zone

Qt energy/mass released in combustion
w reaction rate
uf flame speed
m m = ρ0uf

(i) rate of energy release by combustion

~ Qtwδ (energy / area-s)

(ii) rate of heat conduction upstream

~ k ( (Tf - T0)/ δ) = k(∆T/ δ)

(iii) rate of mass consumed in the flame ~ w δ

(iv) energy equation for the thermal theory:

dT d  dT 
mC p =  k  + Qt w
dx dx  dx 

*Originally due to Michelson.

5-43
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

5.5.1 An Elementary Laminar Premixed Flame (cont’d)

∆T convection balances
Upstream (w = 0 ) mC p ∆T ~ k conduction
δ
uf ~ k
ρ0C pδ

Q f w δ ~ k ∆T
conduction balances
Combustion Zone
δ rate of energy release

δ ~ k ∆T
Qt w

Overall Energy Balance C p T f −T0  = Qt

 
  Q
 T f −T0  = t
  Cp

k
Flame Thickness δ~
Cpw

n
1 kw
Flame Speed uf ~ ~ p 2 e − E / 2 R0T0
p0 Cp

Estimated Values δ ~ 0.0001m u f ~ 0.5m/s

(100 µ)

5-44
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

5.5.2 Thermal Theory of Laminar Premixed Flames

The purely thermal theory of a planar laminar premixed flame is
governed by the energy equation, here including diffusion:
dT d  dT  dh dY
mC p = k  + Qt w + ρD∑ iT i
dx dx  dx  dx dx
Assume Cpi constant and identical:
dhiT dYi dT dY
∑ dx dx
= C pi
dx
∑ dxi =0

With these assumptions, the simplest thermal theory is based on the

equations:
dT d  dT 
mC p = = k  + Qt w
dx dx  dx 
Qt w = Bp nT α e − E/R0T

• really investigating the propagation of a zone of energy

release through a gas of initially fixed and uniform
composition

• the combustion zone separates reactants and products

• the chief difficulty blocking easy solution is the strong

nonlinear dependence of Qtw on temperature

• a way around this difficulty is to specify Qtw as a function

of space and see what follows — this converts an
‘autonomous’ nonlinear problem to a ‘nonautonomous’
linear problem

5-45
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Infinitesimally Thin Flame Front

• For large E, after ignition at some Tig > T0 (usually Tig is

‘close’ to Tf), combustion occurs so rapidly that it is confined
to a thin region of space, in the limit a sheet:

• On both sides of the flame, there is no combustion:

dT d 2T (assume k constant, not an

mC p =k 2
dx dx essential assumption)

k a characteristic length
T = c1e x/l + c2 l=
mC p for the flame thickness

x < 0: T = T0 + c1e x/l

x > 0: T = Tf

5-46
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Infinitesimally Thin Flame Front (cont’d)

• Integrate the energy equation x1 < x < x2 (Cp constant)

x2 x
 dT  2

mC p [T ] = k   + ∫ Qt wdx
T2
T1
 dx  x1 x1

• Overall energy balance: Qt = Cp (Tf - T0), total heat of

reaction; let x1 → −∞, x2 → +∞:
+∞
mC p (T f − T0 ) = 0 + C p (T f − T0 ) ∫ wdx
−∞

+∞ Total mass reacted in

m = ∫ wdx passage through the
−∞
combustion zone

x
mass of material reacted per unit
∫ wdx : = area time up to station x
−∞

5-47
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Infinitesimally Thin Flame Front (cont’d)

• Integral of the energy equation across the flame sheet:

0+ 0+
 dT 
mC p [T ] = k   + Qt ∫ wdx
0+
0-
 dx  0 - 0-

mC p [T f + − T f- ] = −k   + mQt
 dT 
 dx  0 -

• In the limit, Tf- → Tf:

 dT  mQt
  =
 dx  0 - k

i.e., all energy released in the combustion zone

is conducted upstream

• Hence, c1 = Qt/Cp and the solution is:

Qt x/l
x < 0: T = T0 + e
Cp
x > 0: T = Tf

5-48
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Infinitesimally Thin Flame Front (cont’d)

• For the infinitesimally thin flame:

 x  x
w = Bp nT α e − E/R0T δ  = w0 δ 
l l
∞ ∞
 x  x
⇒ m = ∫ wdx = Bp l ∫ T α e − E/R0T δ dδ 
n

−∞ −∞ l l
 k  α − E/R0T f
m = Bp  n T e
 mC p 
 

1 Bk n/ 2 α/ 2 − E/ 2 R0T f
m = ρ 0u f uf = p Tf e
ρ0 Cp

Laminar flame speed for a planar sheet

(cf. estimate, p. 5-32)
• There are two characteristic lengths in this problem:

(i) the thickness of the reaction zone (0 for a sheet)

(ii) flame thickness, here the characteristic length of

the thermal profile, k/mCp (thinner for increased
speed)

N.B. Distinction

5-49
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Finite Flame Front, Uniform Combustion

• The calculation for an infinitesimally thin front does not show

that the flame speed is an eigenvalue. This example helps to
clarify the matter.
• Assume that combustion is uniform in Ti < T < Tf and absent
elsewhere:

x < 0: T = T0 + c1e x/l

x > 0: T = Tf
d 2T dT
0 < x < xf : k 2 − mC p = −Qt w (Qt w constant)
dx dx

x  
Dimensionless variable ξ =  ξ = ∫ k dx for k variable
l  mC p 
 
Solution: T = Λξ + c2 e ξ + c3 ( 0 < ξ < ξ+ )

Qt wk
and Λ =
(mC p )2

5-50
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Finite Flame Front, Uniform Combustion (cont’d)

• Four matching conditions:

dT
x = 0: T, continuous
dx
dT
x = xf : T, continuous
dx

But there are only three arbitrary constants (c1, c2, c3) so solution
exists only for a particular value of Λ: i.e., Λ is an eigenvalue

Apply the matching conditions to find two formulas for Λ:

 (T f − T0 ) − (Ti − T0 ) e ξ f 
T continuous : Λ= 

 (1 + ξ f ) − e
ξf 

dT
continuous : Λ=
(Ti − T0 ) e
ξ f

ξ
dx e f −1

Hence the thickness of the reaction zone is found as the solution

to:

ξf T − T0 N.B. Ti, ignition

−ξ f
= f temperature, must
1− e Ti − T0 be specified

5-51
 5.5 ELEMENTARY EXAMPLES OF LAMINAR
PREMIXED FLAMES

[Link] Finite Flame Front, Uniform Combustion (cont’d)

And the eigenvalue is:

T f − T0
Λ=
ξf

Λ=
Qt wk
: uf =
1 Qt wk 1
=
(T f − T0 )wk
(mC p )2 ρ 0C p Λ ρ0 C pΛ

1 w0 k
cf. flame sheet: uf =
ρ0 Cp

For a ‘thin’ region of uniform combustion, ξf << 1 (i.e., thickness of

the reaction zone is much smaller than the total flame thickness)

 T − Ti  T −T
 ⇒ Λ = (T f − T0 ) i 0
1
ξ f → 2 f
 Ti − T0  2 T f − Ti

and

1 2(T f − Ti ) wk
uf =
ρ0 (Ti − T0 ) C p

1 kw
cf. P. 5-44, u f ~
ρ0 C p
5-52
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

For analysis of combustion instabilities it is sufficient to consider a

two-phase mixture comprising:

(i) a mixture of reacting perfect gases

(ii) a condensed phase consisting of small particles that may
agglomerate and may vaporize (or condense) and burn

Characteristic (ii) implies that condensed particles interact and that

they possess a distribution of sizes not treated explicitly here.

ρ gi , u gi , e0 gi : local properties of gaseous species i

ρ g , u g , e0 g : mass - averaged properties of the gas phase
ρl , u l , e0l : size - averaged properties
of the condensed (liquid?) phase

5-53
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion

Apply the principles of conservation of mass, momentum, and

energy using Boltzmann’s transport theorem:

In general, VM is the
volume enclosed by the
closed moving control
surface SM.

φ(r,t ):a scalar, vector, or tensor density function defined

over the region in which SM moves.
G : G = ∫ φ(r, t )dV is the amount, in VM, of the property
VM whose density is φ.
Q : velocity of a point in the surface SM (not necessarily
in the direction n̂ normal to the surface).

N.B. Q need not be related to the fluid velocity: it

merely defines the motion of SM moving
through the field.

5-54
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

• Reynolds transport theorem gives the formula for the rate of

change of G within the moving volume VM:

dG d ∂φ ˆ dS
= ∫ φ(r, t )dV = ∫ dV + ∫ φQ ⋅n
dt dt VM VF ∂t VM

VF: volume instantaneously coincident with

VM, and it is legitimate to apply Gauss’
theorem and write:

d ∂φ
∫ φ dV = ∫  + ∇ ⋅ (φQ ) dV
dt VM 
VM  ∂t


Reynolds’ Transport Theorem

References: Reynolds, O. (1903) Collected Scientific Papers

Vol. III, Cambridge University Press.
Culick, F.E.C. (1962) “The Equations of Motion
for Multicomponent Reacting Gas Flows”
MIT Aerophysics Laboratory TR48.

5-55
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

• The required equations of motion are obtained by applying

Boltzmann’s transport theorem with:
φ = ρi mass density, species i
φ = ρ g u g + ρl u l total momentum density, gas
and condensed phases
φ = ∑ ρ gi e0 gi + ρl e0 l total stagnation energy density, gas
and condensed phases

• Q can be ‘any’ velocity, but the conventional choice is the

mass-averaged velocity u

1
u = ∑ ρi u i
ρ

5-56
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Conservation of mass, species i (Q = u )

gi

∂ρ gi
+ ∇ ⋅ (ρ gi u gi ) = wgi + w(gil ) + wegi
∂t
wgi : produced in homogeneous chemical reactions
( l)
wgi : produced from condensed species
wegi : produced from external sources

Conservation of mass, gas phase (sum over i species)

∂ρ g
+ ∇ ⋅ (ρ gi u gi ) = w(gl) + weg
∂t
∑ wgi = 0 (no net production by homogeneous
chemicalreactions)
∑ w(gil ) = w(gl)

∑ wegi = weg
Conservation of mass, condensed phase (Q = u l )
∂ρl
+ ∇ ⋅(ρl u l ) = − w(gl) + wel
∂t

5-57
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Global conservation of mass (sum equations for gaseous and

condensed phases)

∂
(ρ g + ρl )+ ∇ ⋅(ρ g u g + ρlu l )= weg + wel
∂t

Conservation of momentum

(i) construct separate equations for gaseous and condensed

phases, including mutual force of interaction; or

(ii) construct equation for total momentum(Q = u )

ρu = ∑(ρ g u g + ρl u l ) ; ρ = ρ g + ρl

∂
(ρ g u g + ρlu l )+ ∇ ⋅(ρ g u g u g + ρlu lu l )+ ∇p = ∇ ⋅ τt v + m eg + m el
∂t

t
τV : viscous stress tensor
m eg , m el : external sources of momentum

5-58
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Conservation of energy

∂
(∑ρ gi e0 gi + ρl e0l )+ ∇ ⋅(∑ρ giu gi e0 gi + ρlu l e0l )+ ∇ ⋅(∑ piu gi )
∂t
= ∇ ⋅ (τ v ⋅u g )− ∇ ⋅q c + Qe
t

qc internal heat conduction

we , m e , Qe external sources of mass, momentum, and energy,
possibly representing the effects of control

Tg Tl
1 1
egi = ergi + ∫ cvi dTg′ ; e0 g = eg + u g2 ; el = erl + ∫ cdT ′ + ul2
Tr 2 Tr 2

N.B. Ultimately and fundamentally, control can be exerted

only through action involving sources of mass,
momentum, and energy.

5-59
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Rewrite equations for conservation of mass: Yi = ρ gi ρ g ; u gi = u gi + v gi

∂Y
ρ g  i + u g ⋅∇Yi  = −∇ ⋅ (ρ g Yi Vgi )+ wgi + wgi( l ) + wei − Yi [wg( l ) + weg ]
 ∂t 

∂ρ g
Gas Phase + ∇ ⋅ (ρ g u g ) = wl + weg
∂t

∂ρ l
Condensed Phase + ∇ ⋅(ρ l u l ) = − wl + wel (defines wl )
∂t

∂ρ
Single - Fluid Model
∂t
(
+ ∇ ⋅ ρu g = W )

ρ = ρl + ρ g
W = we − ∇ ⋅(ρl δ u l )
we = weg + wel
δu l = u l − u g

• density is the total density

• velocity is the velocity of the gas only

5-60
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Rewrite equations for conservation of momentum:

Gas Phase
Du g t
ρg + ∇p = ∇ ⋅ τ v + σ e + δu l wgl + Fl
Dt
σ e = m e − (u g weg + u l wel )
m e = m eg + m el
D ∂
= + u g ⋅∇
Dt ∂t

Condensed Phase
∂u
ρl  l + u l ⋅∇u l  = −Fl (defines Fl )
 ∂t 

Single - Fluid Model

Du g t
ρ + ∇p = ∇ ⋅ τ v + σ e + δu l wgl + δFl
Dt
∂δu l
δFl = −ρl  + δu l ⋅∇δu l + δu l ⋅∇δu g + u g ⋅∇δu l 
 ∂t 
(defines δFl )

5-61
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Rewrite equations for conservation of energy:

(i) Expand and use conservation of mass:

De0 g
+ ∇ ⋅ ( pu g ) = ∇ ⋅ (τ v ⋅u g )− ∇ ⋅q c − ∇ ⋅{∑ Vgi (ρ gi e0 gi + pi )}
t
ρg
Dt
+ δe0 l wl − (e0 g weg + e0 l wel )+ Qe + Q0 l

δe0 l = e0 l − e0 g
∂e
Q0 l = −ρl  0 l + u l ⋅∇e0 l 
 ∂t 

(ii) Remove kinetic energy by using momentum equations:

Deg
+ p∇ ⋅u g = (τ v ⋅∇ )⋅u g − ∇ ⋅q c − ∇ ⋅{∑ Vgi (ρ gi e0 gi + pi )}+ Qe + Ql
t
ρg
Dt
+ δu l ⋅ Fl − u l ⋅ (σ e + δu l wgl )− δe0 l wl − (e0 g weg + e0 l wel )
∂e
Ql = −ρl  l + u l ⋅∇el 
 ∂t 

5-62
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

(iii) Replace energy by enthalpy, hg = eg + p/ρg ; ρghg = Σρgihgi:

 ∂hg  ∂p
+ u g ⋅∇hg  =  + u g ⋅∇p  + (τ v ⋅∇ )⋅u g − ∇ ⋅q c
t
ρg 
 ∂t   ∂t 
 2 
u gi
+ ∇ ⋅ ∑ ρ gi Vgi  hgi + +Q +Q +E
 2  e l l
 
El = δu l ⋅ Fl − u l ⋅ (σ e + δu l + wgl )− δe0l wl
 p 
− (e0 g weg + e0l wel )+  (wl + wel )
 ρg 
 
(iv) Energy equation with gas temperature as independent
variable: T
hgi = hrgi + ∫ C pi dTg′ dhgi = C pi dTi
Tr

ρ gi
hg = ∑ hgi = ∑Yi hgi ⇒ dhg = ∑ hgi dYi + ∑Yi dhgi
ρg
Dhg DYi DTg = ∑ hgi dYi + C p dTg
= ∑ hgi + Cp
Dt Dt Dt

DTg
+ ∇ ⋅ (k∇Tg )+ ∇ ⋅ ∑ ρ g D∇Yi + Qc + Qe + El
Dp
ρgC p =
Dt Dt

Qc = − ∑ ν i hi qc = −k∇Tg ρi Vgi = −ρD∇Yi

5-63
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.1 General Procedure for Deriving the Equations of Motion (cont’d)

Gas Phase
 ∂Tg 
ρgC p  + u g ⋅∇Tg 
 ∂t 
∂p
=  + u g ⋅∇p  + ∇ ⋅ (k∇Tg )+ ∇ ⋅{∑ ρD∇Yi }+ Qc + Qe + Ql + El
 ∂t 

Condensed Phase
∂e
ρl  l + u l ⋅∇el  = −Ql (defines Ql )
 ∂t 

Single - Fluid Model

 ∂Tg 
ρgC p  + u g ⋅∇Tg 
 ∂t 
∂p
=  + u g ⋅∇p  + ∇ ⋅ (k∇Tg )+ ∇ ⋅{∑ ρDhiYi }+ Qc + Qe + δQl + El
 ∂t 
∂δTl
δQl = −ρlC p  + δu l ⋅∇δTl + δu l ⋅∇δTg + u l ⋅∇δTl  (defines δQl )
 ∂t 

5-64
 5.6 EQUATIONS OF MOTION FOR A REACTING
TWO-PHASE MIXTURE

5.6.2 Muster of Equations:

Dρ
Conservation of Mass = − ρ ∇ ⋅u + W
Dt
Du
Conservation of Momentum ρ = − ∇p + F
Dt
DT Dp
Conservation of Energy ρC p = +E
Dt Dt
Equation of State p = ρ RT

Definitions
u := u g W = −∇ ⋅ (ρl δu l )+ wl
t
T : = Tg F = ∇ ⋅ τ v + σ e + δu l wgl + Fl

ρ = ρ g + ρl E = ∇ ⋅(k∇T ) + ∇ ⋅{∑ ρD∇Yi }+ Qc + Qe + δQl + El

D ∂
= + u ⋅∇
Dt ∂t

For wave propagation:

(i) compressibility of the mixture is the compressibility of

the gas
(ii) density of the mixture is the total density of the mixture
(iii) thermodynamic properties are mass-averaged over all
components
(iv) the speed of sound appearing later, a2 = γp/ρ, will
correctly account for the presence of condensed
species, an important result

5-65
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

In practice, the most commonly observed flames are non-premixed;

matches, candles, Bunsen burners, fireplaces, building fires, forest
fires...
In general, non-premixed or diffusion flames pose more difficult
analytical problems than do pre-mixed flames for which asymptotic
methods (e.g., AEA, activation energy asymptotics) are often more
realistically applicable and yield useful results. For example,
Zel’dovich et al. (1984) hardly mention diffusion flames.
As for premixed flames, problems of diffusion flames formulated for
the limit of very fast reaction rates do lend themselves fairly
conveniently to analysis.
The elementary examples discussed here are intended chiefly to
illustrate the two most basic properties of diffusion flames:

(i) unlike premixed flames, a non-premixed flame does not possess a

characteristic (eigenvalue) flame speed;
(ii) in the limit of infinitely fast reaction rates, the fuel and oxidizer
are supplied to the combustion zone in the ratio of their
stoichiometric coefficients in the reaction equations. Hence, even
for finite rates, the maximum temperature in a diffusion flame is
at or near the value for adiabatic combustion under stoichiometric
conditions (φ = 1)

Some references: Burke S.P. and Schumann T.E.W. (1928) “Diffusion Flames,” Industrial and
Engineering Chemistry, Vol. 20, pp 998-1004 [first use of the term ‘diffusion flame’]; Shvab V.A.
(1948) “Relation Between the Temperature and Velocity Fields of the Flame of a Gas Burner,” Gas
Energ. Izd M-L; Zel’dovich Y.B. (1949) “On the Theory of Combustion of Initially Unmixed Gases,”
Zh. Tek. Fiz., Vol. 19, translated as NACA TM 1296 (1951); Friedlander S.K. and Keller K.H. (1963)
“The Structure of the Zone of Diffusion Controlled Reaction,” Chem. Eng. Sci., Vol. 18, pp 365-375);
Findall (1965) “Ignition and Extinction in Combustion of Initially Unmixed Reactants,” J. Fl. Mech.,
Vol. 21, pp 281-303

5-66
 SOME REMARKS ON MODELING
SIMPLE FLAMES

G. G. Brown, “Introduction”, Combustion Symposium, American

Chemical Society, September 1928.
Industrial and Engineering Chemistry, Vol. 20, No. 10, (1928), p.
998

“Combustion is without exaggeration the most important

[chemical] reaction to the human race. All human and animal
existence depends upon combustion as its source of energy. [...]
in many ways our further progress depends upon more intelligent
and efficient control of combustion.”

“These studies of flame propagation are closely allied with the

use of combustion as a direct source of power.”

• The kinetic and dynamical properties of laminar premixed

and non-premixed (diffusion) flames are the fundamental
building blocks required to construct models of global
combustion dynamics in a chamber

• For infinitely fast kinetics, the behavior of both types of

flames is “diffusion limited” and combustion takes place in
infinitesimally thin sheets

• For premixed flames, diffusion of energy

(temperature) is the limiting process, secondarily
influenced by diffusion of species

• For non-premixed flames, diffusion of species is the

limiting process, secondarily influenced by diffusion
of energy (temperature)

5-67
 SOME REMARKS ON MODELING
SIMPLE FLAMES

• In combustion of fossil fuels, homogeneous reaction rates are

normally ‘fast’ due to the dominant Arrhenius factor (large
‘Zel’dovich number ′ )

• Hence, models of combustion occurring in ‘flame

sheets’ are good representations of reality

• The dynamics of elementary flame sheets and ensembles of

flamelets remain matters of current research

• A possible sequence of topics for steady flames:

• Premixed laminar flames

• Diffusion flame sheets

• Strained diffusion flame sheets (counterflow flames)

• Consumption of a fuel strip, strained and unstrained

• Diffusion flame in the field of a vortex

• Results of numerical simulations

• Flamelets, coherent flames, and distributed turbulent

combustion

5-68
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface

We consider here only the case of diffusion-controlled flames,
reaction rate >> diffusion rate

In the limit of infinitely fast reaction rate, chemical reactions are

confined to an infinitesimally thin region, an interface between the
fuel and oxidizer supplies.

For simplicity, assume one fuel species and one oxidizer species
producing one product species. Early reasoning (§4.1) produced
the following analogy between stationary combustion in a laminar
mixing (diffusion) layer and a time-dependent one-dimensional
diffusion flame:

v1 = v + V1

v2 = v + V2

For infinite reaction rate, the concentrations of fuel and oxidizer

cannot be non-zero simultaneously outside the reaction sheet, and,
to be shown, both vanish at the sheet.
Generally, an average speed v is required to keep the
flame sheet stationary in the laboratory reference frame.
5-69
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

To keep the flame at a fixed location (y = 0) in the unsteady
problem, there must be an average speed, varying in time.

The equations of motion are:

 ∂y ∂y  ∂  ∂y 
ρ i + v i  =  ρD i  + wi (∗)
 ∂t ∂y  ∂y  ∂y 
 ∂T ∂T  ∂  ∂T 
ρc p  + v  =  k  + Qw (∗ ∗)
 ∂t ∂y  ∂y  ∂y 

where
∂Yi
(i ) ρYiVi = −ρD (Fick' s Law )
∂y (assumptions)
(ii ) c pi = c p

Hence there is no energy flow by diffusion

(iii ) wi = ν i w (Recall: Single reaction, ∑ vi Ai = 0)

(iv ) Qw = −∑ hi wi = − w∑ ν i hi

In general Q = Q(t )

5-70
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

Problem:

(1) Solve (∗) and (∗∗) on the two sides of the interface
subject to the conditions:

y → +∞ Y1 → 1, Y2 → 0, T → T0
y → −∞ Y1 → 0 , Y2 → 1, T0 → T

(2) Matching conditions (y = 0):

pressure continuous
temperature continuous (unlike premixed flow)
species concentration continuous (otherwise infinite
diffusion rate)
density continuous

The procedure for deriving matching conditions at an

interface is based on integrating the equations of motion
formally over a volume containing the interface:

yu 0+
interface
yl 0−

5-71
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

Conservation of Mass
∂ρ ∂
+ (ρv ) = 0
∂t ∂y
yu y
∂ρ u
∂
∫ ∂t dy + ∫ ∂y (ρv )dy = 0
yl yl

yu → 0 +
(ρv )0+ = (ρv )0− v := mass - averaged velocity
yl → 0 −

Conservation of Species
∂
(ρYi ) + ∂ (ρvYi ) = − ∂ (ρYiVi ) + wi (wi = vi w)
∂t ∂y ∂y
∂ (ρYi ) ∂ (ρvYi ) ∂ (ρYiYi )
yu uy u y u y

∫ ∂t dy + ∫ ∂y dy = − ∫ ∂y dy + ∫ wi dy
yl yl yl yl

0+
yu → 0 +
(ρvYi )0+ − (ρvYi )0− = −{(ρYiVi )0+ − (ρYiYi )0− }+ ∫ wi dy
yl → 0 − 0−

0+

or (ρviYi )0+ − (ρviYi )0− = ∫ wi dy (i =1,2 := F,0 )

0−

vi = v + vi
wi = vi w

5-72
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

Consider the case of a single binary reaction,

fuel, A1
fuel and oxidizer → products:

a1 A1 + a2 A2 → bB
fuel oxidizer oxidizer, A2

The assumption of a flame sheet implies that the reaction rate is

infinitely fast compared with the diffusion rates.
In the limit if infinitely fast reactions, all fuel and oxidizer reaching
the flame sheet are consumed and the concentrations vanish in the
sides opposite from those which the species are supplied:

Not possible because [F] ≠ 0, [0] ≠ 0

in y > 0 implies distributed
reactions, contrary to the assumption
of a flame sheet due to the infinitely
fast reaction rate.
At the sheet, the flows of fuel and
oxidizer are due to diffusion and
convection; hence on their
respective sides, the concentrations
of fuel and oxidizer must be non-
zero, but both concentrations must
vanish at the sheet, to prevent
infinite diffusion rates
which would otherwise be
required at the interface.
5-73
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

Substitute wi = viw and divide the equations written for the fuel,
( )1 and the oxidizer, ( )2: (see p. 5-73)

0+ 0+
1 −1
(ρv1Y1 )0+ = ∫ wdy ; (ρv 2Y2 )0− = ∫ wdy
ν1 0− ν2 0−

Mi : molar mass
ν1 = a1 M1 ; ν 2 = a2 M 2
stoichiometric coefficients expressed
as mass (not moles!) consumed
∂Yi
v1 = v +Vi ; ρViYi = ρ D
∂y

 ∂Y1 
 ρD 
 ∂y  0+ = a1 M1 = f ∗ because Yi → 0
 ∂Y  a2 M 2 at the flame sheet
− ρD 2 
 ∂y  0−

In a diffusion flame sheet the rates of

consumption of fuel and oxidizer are in
mass stoichiometric ratio

5-74
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

Remarks:

(i) In a flow, the motion of the interface (sheet) adjusts so

this condition is satisfied
(ii) It follows from this condition that the maximum
possible adiabatic flame temperature is reached in a
diffusion flame sheet
(iii) For finite reaction rates and finite thickness, the above
condition does not apply; the condition:
yu

(ρv1Y1 )y u
− (ρv1Y1 ) yl = ν i ∫ wdy
yl

implies

(ρv1Y1 )yu − (ρv1Y1 )yl a1 M1 ∗

= =f
(ρv 2Y2 )yl − (ρv 2Y2 )yu a2 M 2

The second terms in the numerator and denominator are

small but non-zero. Hence, as observed and found
numerically, the maximum temperature is somewhat less
than the adiabatic flame temperature, but still high.

5-75
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.1 Matching Conditions at a Reacting Interface (cont’d)

The formal problem to solve for determining the properties
and structure of a diffusion flame sheet is:

∂ρ ∂ (ρv )
+ =0 y→∞
∂t ∂y
ρv1Y1 T → T0
∂Y1 ∂Y 1 ∂  ∂Y1 
+v 1 =  ρD  + w1 Y1 → 1 ; Y2 → 0
∂t ∂y ρ ∂y  ∂y 
∂Y1 ∂T
∂T ∂T 1 ∂  ∂T  , →0
+v = k  + Qw ∂y ∂y
∂t ∂y ρC p ∂y  ∂y 

(ρV1Y1 )0+ a1 M1 ∗ diffusion

= =f
(ρV2Y2 )0− a2 M 2 flame sheet
a1 A1 + a2 A2 = bB

∂ρ ∂ (ρv ) y → −∞
+ =0
∂t ∂y T → T0
∂Y2 ∂Y 1 ∂  ∂Y2  Y2 → 1 ; Y1 → 0
+v 2 =  ρD  + w2
∂t ∂y ρ ∂y  ∂y  ∂Y2 ∂T
ρv 2Y2 , →0
∂T ∂T 1 ∂  ∂T  ∂y ∂y
+v = k  + Qw
∂t ∂y ρC p ∂y  ∂y 

5-76
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates
Ref.: F.E. Marble, Notes on Reacting Flows

To emphasize the action of diffusion, consider an artificial model

problem. Fuel and oxidizer enter through parallel plates and react
at an infinitesimally thin interface. Assume:
(i) uniform temperature and density:
⇒ energy equation is dropped

(ii) that solution for a steady problem exists

(iii) w = kY1Y2δ(y - yf) ≠ 0 only in the flame sheet

k ~ e − E/R0T ; w1 = − a1 M1 w ; w2 = − a2 M 2 w

Species Equations
dY1 d d 2Y1
ρv = − (ρV1Y1 )+ w1 = D 2 − a1 M1 w
dy dy dy
dY2 d d 2Y2
ρv = − (ρV2Y2 )+ w2 = D 2 − a2 M 2 w
dy dy dy
5-77
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
To obtain a steady solution it is necessary that v = 0, i.e., there is no
steady average flow and the problem is entirely one of diffusion
with sources.

Assertions
(i) v=0

(ii) Yp = 0 , y = ± l (products completely absorbed at the walls)

a2 M 2 kl 2 aM  Stoichiometric 
Define: η= y/ l α= ∗
f = 1 1  
D a2 M 2  ratio of mass 
 consumption rates 
 
Problem to be solved:
Boundary Conditions
d 2Y2
(∗ ) = f ∗ αY1Y2
dη 2 Y1 (η F ) = 0 ; Y1 (1) = 1
d 2Y2 Y2 (ηF ) = 0 ; Y2 (− 1) = 1
(∗∗ ) = αY1Y2
dη2

Procedure:
(i) Subtract f times (∗∗ ) from (∗ ):
d2 n equations → n − 1 equations, 
(Y1 − f ∗Y2 ) = 0  vaguely related to the case Le = 1
dη 2
 

5-78
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
(ii) Define a “composite dependent variable”, an example of
a “Shvab-Zel’dovich transformation”
X = Y1 − f ∗Y2
d2X  X( 1 ) = 1 = c1 + c2
= 0 ⇒ X = c1η + c 2  ∗
dη 2  X( − 1 ) = − f = −c1 + c2
c1 =
1
(1 + f ∗ ) ; c2 = 1 (1 − f ∗ )
2 2
X (η ) = Y1 − f ∗Y2 = (1 + f ∗ )η + (1 − f ∗ )
1 1
2 2

Hence:
∗  1 + f ∗  1 − f ∗
f Y2 = Y1 −   η +
 2  1 + f ∗ 
∗  1 + f ∗  1 − f ∗
Y1 = f Y2 +   η +
 2  1 + f ∗ 

Problem now is to find Y1, Y2 separately

(iii) The equations for Y1 and Y2 are:

d 2Y1
= αf ∗Y1Y2 = αY1 {Y1 − c(η + ψ )}
dη 2

d 2Y2
= αY1Y2 = αY2 { f ∗Y2 + c(η + ψ )}
dη 2

1+ f ∗ 1− f ∗
c= ; ψ=
2 1+ f ∗
5-79
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
(iv) Define a new dependent variable z:
z = Y1 + k (η + η0 ) k , η0 can be set later

⇒ Y1 = z − k (η + η0 )
(use (ii))
f ∗Y2 = z − (k + c ) − η( xη0 + ψ )

dz dY1 d 2 z d 2Y1
= +x ; =
dη dη dη2 dη2

Equation for Y1 gives the equation for z:

d 2z   xη + cψ 
= α[z − x(η + η0 )] z − ( x + c ) η + 0 
dη 2
  x + c 

xη0 + cψ
Choose: η0 = ψ ⇒ =ψ
x+c

d 2z
= α [z − x(η + ψ )][z − c(η + ψ )]
dη2

 Y1 = z − x(η + ψ )
 ∗
 f Y2 = z − c(η + ψ )

5-80
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
(v) Apply boundary conditions
Let z = z 0 @η = 1 , z = zˆ@η = −1
z0 − 1
η =1 Y1 = z − x(η + ψ ) ⇒ 1 = z0 − x(1 + ψ ) ⇒ x =
1+ ψ
z
f ∗Y2 = z − c(η + ψ ) ⇒ 0 = z0 − c(1 + ψ ) ⇒ c = 0
1+ ψ
zˆ
η = −1 Y1 = z − x(η + ψ ) ⇒ 0 = zˆ − x(− 1 + ψ ) ⇒ x =
1+ ψ
zˆ − f
f ∗Y2 = z − c(η + ψ ) ⇒ f = zˆ − c(− 1 + ψ ) ⇒ c =
ψ −1
Assume symmetry, zˆ = zˆ0 , and check to see if the results are
consistent for K and c:
z0 − 1 z 1− ψ
K= = 0 ⇒ z0 =
1+ ψ ψ −1 2
z0 z0 − f ∗ 1+ ψ
c= = ⇒ z0 = f ∗
1+ ψ ψ −1 2

1− ψ ∗ 1+ ψ 1− f ∗
= f ⇒ψ= checks definition of ψ
2 2 1+ f ∗
1− ψ f∗
Hence z0 = =
2 1+ f ∗
z −1 1
K= 0 =−
1+ ψ 2
z0 f∗
and c= =
1+ ψ 2
5-81
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
The equation for z is:
d 2z  1  f∗ 
= α  z + (η + ψ )  z + (η + ψ )
dη2  2  2 
f∗
B.C. : z (1) = z (− 1) =
1+ f ∗
The distributions of reactants are given by the formula
 1
 Y 1 = z + (η + ψ )
2
 f ∗
Y2 = z − (η + ψ )
 2

1− f ∗
where ψ=
1+ f ∗

Thus the original problem of two coupled equations for

Y1 and Y2 has been reduced to a single equation for z.
a2 M 2 kl 2
(vi) Note: α = → ∞ for fast reaction, slow diffusion.
D
1 d 2
z  1   f ∗

Then =  z + (η + ψ )  z + (η + ψ ) → 0 for α → ∞
α dη 144
2
2244 3 144 2
2443
Y1 f ∗Y2
The solutions, the distributions of Y1 and Y2 are almost
linear everywhere except where d 2 z
dη 2
is ‘large’ — a “boundary layer”

5-82
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
The solution for α = ∞ is the “outer” solution in a singular
perturbation problem and ‘singular’ means here that
straightforward solution based on expansion in powers
of (1/2)n fails became when α = ∞, the equation does
not contain the highest derivative and at least one
boundary condition cannot be satisfied by the solution.

(vii) Outer Solution:

On the fuel side of the sheet Y2 = 0
f∗
⇒z= (η + ψ )
2
On the oxidizer side of the sheet Y1 = 0
1
⇒ z = − (η + ψ )
2
The solutions are:
1 f ∗ +1
Y2 = 0 Y1 = z + (η + ψ ) = (η + ψ )
2 2
1 f ∗ +1
Y1 = 0 f Y2 = z − (η + ψ ) = −
∗
(η + ψ )
2 2
Outer solution is

f ∗ +1
Y1 = (η + ψ ) η > −4
2
f ∗ +1 η < −4
Y2 = − ∗
(η + ψ )
2f

5-83
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
1− f ∗
The flame sheet is at η = −ψ = − <0
1+ f ∗
Hence the outer solution looks like:

Remarks:
1− f ∗
(1) Flame sheet exists at η = −ψ = −
1+ f ∗
(2) Direct calculation confirms that reactants
are consumed at rates in stoichiometric ratio:

dY1 1 + f ∗ dY2 1 + f ∗
= ; =
dη 2 dη 2f∗

Fuel Flux 1+ f ∗
= ∗
= f∗
Oxidizer Flux 1 + f
f∗

(viii) What happens in the ‘inner region’ where d2z/dη2 is

not small?

5-84
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
d2 d2
Set ξ = η + ψ ⇒ 2 = 2
dη dξ

1 d 2z  1  f∗ 
=  z − ξ  z − ξ
α dξ 2  2  2 

This equation must be solved in the inner region with

α large and appropriate boundary conditions satisfied.

(ix) Inner Solution:

The idea is to ‘stretch’ the spatial variable and solve
the problem in a thin region including the combustion
zone.
1 d 2z  1  f∗  f∗
=  z − ξ  z + ξ z (± 1) =
α dξ 2  2  2  1+ f ∗

Set ζ = αnz and investigate the problem for α → ∞ but

ξ → 0 so ζ remains finite, the inner region near η=η∗
d 2z 
2 n −1 1 ζ  f∗ ζ 
α =z−  z + 
dζ 2  2 α n  2 α n 
−2n  n 1  n f∗ 
= α  α z − ζ  α z + ζ
 2  2 

Define Z = αn z
3 n −1 d 2Z  1  f∗ 
α =  Z − ζ  Z + ζ
dζ 2  2  2 

5-85
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
Choose n = 1/3 and the problem is independent of α:

d 2Z  1  f∗ 
=  Z − ζ  Z + ζ
dζ 2  2  2 

 1− f ∗ 
Z =α z ; ζ =α z=α
1/ 3 1/ 3 1/ 3
(η + ψ )  ψ = 
∗
 1 + f 
 1 1/ 3  1 
1Y = z + (η + ψ ) = α  Z + ζ
 2  2 

Y = z − f (η + ψ ) = α1/ 3  Z − f ζ 
∗ ∗

 2 2  2 


Boundary Conditions on Z:
1
On fuel side of inner region Z1 → Z
2
f∗
On oxidizer side of inner region Z1 → − ζ
2

5-86
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.2 Diffusion Flames Between Parallel Plates (cont’d)
Complete solution:

Remarks:

(1) This behavior for large but finite reaction rate

is an example of “Activation Energy
Asymptotics” (AEA).

(2) For large Zel’dovich number E/RT where E

is the activation energy in rate ~e–E/RT, the
flame zones are thin in both premixed and
non-premixed flames.

5-87
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.3 The Burke-Schumann Problem

Note on the ‘Burke-Schumann problem (pp § )

Physical situation: Ref VG 5-43

Main assumptions:
(1) Constant diffusion coefficient
(2) Ignore diffusion in streamwise direction
(3) Velocities of fuel and air uniform
(4) Infinitely fast reaction rate.

Result:
A flame sheet develops in a uniform flow due to diffusion
of fuel, oxidizer and products.

5-88
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.3 The Burke-Schumann Problem (cont’d)

Figure 3—Curve I, Underventilated

Cylindrical Flame; Curve II,
Overventilated Cylindrical Flame

AXISYMMETRIC FLOW
“Underventilated”:
Excess fuel
Figure 5—Curve I, Underventilated
Shvab and Zel’dovich Flat Flame; Curve II, Over-
ventilated Flat Flame
extended the analysis,
relaxing assumption (3) and
TWO-DIMENSIONAL FLOW
treating details of the field
Note:
in the region of the flame
“Overventilated”
sheets.
means
excess oxidizer
5-89
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
Zel’dovich number T f − T0 E
β= →∞
T f R0T f
for infinitely fast reaction.

The 1-D time-dependent diffusion flame is the transposition

of the 2-D flame sheet in a shear flow:

The problem treated here is a simplified (Marble) form of

Marble, F.E. and Adamson, T. (1954) [see later, §5.8]

(i) Equations of motion

∂ρ ∂
+ (ρv ) = 0
∂t ∂y
∂Yi ∂Y 1 ∂  ∂Yi 
+v i =  ρD  (reaction rate = 0
∂t ∂y ρ ∂y  ∂y  except in the
flame sheet)
∂T ∂T 1 ∂  ∂T 
+v = k 
∂t ∂y ρc p ∂y  ∂y 
p = ρRT ( p constant )
5-90
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
(ii) Boundary and initial conditions
Assume that flame sheet can be placed at y = 0 by suitable
choice of average velocity v(t).
All t: Y1 (0, t ) = Y2 (0, t ) = 0 All y: T ( y, 0 ) = T0 ; ρ( y, 0 ) = ρ0

t = 0: Y1 ( y, 0 ) = 1 ( y > 0) y → +∞ : Y1 ( y, t ) → 1
Y2 ( y, 0 ) = 1 ( y < 0) y → −∞ : Y2 ( y, t ) → 1

(iii) Matching conditions at the flame sheet

T, ρ continuous at y = 0: T (0+, t ) = T (0−, t )
ρ(0+, t ) = ρ(0−, t )

(a) Integrate energy equation across the flame

 ∂T ∂T ∂  ∂T 
∫ p  ∂t ∂y − ∂y  k ∂y dy = ∫ Qw0δ( y )dy
ρc + v
0−  
0+

[ρc pT ]0− − k ∂∂Ty  = Qw0

0+

  0−

 ∂T   ∂T 
k  −k  = Qw0
 ∂y  0−  ∂y 0+

Net rate of heat loss from rate of energy

flame sheet by conduction
= release by combustion

5-91
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
(b) Integrate continuity equation across the flame sheet
0+
 ∂ρ ∂ 
∫ +
 ∂t ∂y (ρv ) dy = 0
 
ρv = ∑ ρ i v i
0−

(ρv )0+ − (ρv )0− = 0

(ρ1v1 )0+ + (ρ3 v3 )0+ = (ρ 2 v 2 )0− + (ρ3 v3 )0−

ρ3 (0 )[(v 3 )0+ + (− v 3 )0− ] = (− ρ1v1 )0+ + (ρ 2 v 2 )0−

Flow of products away Flow of fuel and flow

from flame zone
= of oxidizer to the flame

(c) Integrate species equations across the flame sheet

(ρ1v1 )0+ = − a1 (ρ2 v 2 )0−

a2

Diffusion rates of fuel and oxidizer in

stoichiometric ratio (ρ1= ρ2= 0 at flame)

5-92
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
(iv) Howarth-Dorodnitsyn transformation

The new difficulty in this problem arises because the

temperature is not uniform: the density therefore varies in both
space and time — a more realistic situation.

In Lagrangian coordinates, constancy of mass for a

fluid element moving vertically implies ρdy = ρ0dη where
η is a new coordinate.

Hence try transforming to the variables

y
ρ
τ=t η = ∫ dy′
0 ρ0
i.e. the independent variables are transformed:
( y,t ) → (η, τ)
 ∂   ∂τ ∂   ∂η ∂   ∂ 
y
∂ ρ  ∂ 
  =  +  =  +∫ dy′ 
 ∂t  y  ∂t ∂τ  y  ∂t ∂η  y  ∂t  η 0  ∂τ ρ0  ∂η  y
 ∂   ∂τ ∂   ∂η ∂  ρ ∂ 
  =  +  =  
 ∂y t  ∂y ∂τ t  ∂y ∂η t ρ0  ∂η t
(a) Transformed continuity equation:

∂ρ ∂ ρ 
 +  v  = 0
∂t  ρ0  ∂y  ρ0 
∂ρ ρ  ρ 
y

∫ ∂τ  ρ   ρ v  −  ρ v  = 0
  dy ′ +
0  0  0  y  0 0
5-93
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
∂ρ ρ 
y

∫ ∂τ  ρ   ρ
  d y ′ + v  = v 0 (t )
0  0  0 y

Definition: ρ 
v 0 (t ) =  v 
 ρ0  y =0

(v) Transformation to similarity variable ζ = η / D0 τ

(η, t ) → ζ = βη r τ s (β, r, s to be determined)

∂ s ∂
= ζ
∂t τ ∂ζ
∂ r ∂ ∂2 r (r − 1) ∂ r 2 2 ∂ 2
= ζ ; = ζ + 2ζ
∂η η ∂ζ ∂η2 η2 ∂ζ η ∂ζ 2

Transformed species equation:

∂Y s r r (r − 1)  r 2 2 ∂ 2Yi
ζ i  τ + v 0 η − η2 D0  = D0 η2 ζ ∂ζ 2
∂ζ  

Choose r =1, v 0 (τ ) = kτ w

∂Yi  sζ w− s  ∂ 2Yi
⇒ + βkτ  = β D0 2
2

∂ζ  τ1+ 2 s  ∂ζ

5-94
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
1 1 k η
Choose s = − , w = − ⇒ v0 = , ζ =β
2 2 τ τ
∂Yi
 1  ∂ 2Yi
− 2 ζ + βk  = β D0 ∂ζ 2
2

∂ζ
1 ∂Yi ∂ 2Yi
− 2 (ζ − 2βk ) =
2β D0 ∂ζ ∂ζ 2
1 1 k
Choose = 2 ⇒ β = & β k = =c
2β 2 D0 4D0 4D0

d 2Yi dY
(∗) + 2(ζ − c ) i = 0
dζ 2
dζ

η c k
where: ζ= v 0 = 2 D0 c=
2 D0 τ τ 2 D0

(vi) Concentration profiles

Solution to (∗) d  (ζ −c )2 dYi 

e =0
dζ  dζ 

ζ ζ
− ( ζ ′ − c )2
Y2 = B ∫ e −(ζ′−c ) dζ′
2
Y1 = A∫ e dζ ;
0 0

ζ>0 ζ<0

ζ → −∞ , Y2 → 1 ζ → ∞ , Y1 → 1

5-95
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
1
Y1 (ζ ) = {erf (ζ − c) + erf (c)} ζ≥0
1 + erf (c)
1
Y2 (ζ ) = {erf (− ζ + c) − erf (c)} ζ≤0
1 − erf (c)
2 z −ζ 2
erf ( z ) =
π ∫0
e dζ ; c unknown

c
(vii) Determining c in v 0 (t ) = 2 D0
τ
Apply the condition that reactants are consumed at rates in
stoichiometric ratio:
(ρ1v1 )0+ = − a1 (ρ2 v 2 )0− = − f ∗ (ρ2 v 2 )0−
a2
 ∂Y   ∂Y 
−  ρD 1  = f ∗  ρD 2 
 ∂y  0+  ∂y  0−

 ∂Y1   ∂Y 
  = − f ∗ 2 
 ∂y  0+  ∂y  0−

∂ 1 ∂  dY   dY 
= ⇒  1  = − f ∗ 2  (∗)
∂y 2 D0 τ ∂ζ  dζ  0+  dζ  0−

d
{erf (ζ − c )} = 2 e −(ζ −c )
2

dζ π
5-96
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
Substitute solutions Y1(ζ) and Y2(ζ) in (∗):

1− f ∗
erf c = a transcendental equation for c
1+ f ∗

(viii) Summary of solution: 1− f ∗

erf c =
1+ f ∗
erf (ζ − c) + erf (c) η
Y1 (ζ ) = (ζ ≥ 0) ζ=
1 + erf (c) 2 D0 τ
y
ρ
η=∫ dy′
erf (− ζ + c) − erf (c) 0 ρ0
Y2 (ζ ) = (ζ ≤ 0)
1 − erf (c) c
v 0 (t ) = 2 D0
τ
(ix) Special Case: f ∗ =1 ⇒ c = 0
Symmetric profile and no average velocity

Gradients of Fuel and

Oxidizer are Equal at
f*=1

5-97
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
(x) Case for f * < 1 (c > 0)
e.g., hydrocarbon fuels: CH 4 + 2O 2 = CO 2 + 2H 2 O
a1 M1 =1.16 a2 M 2 = 2.16 ⇒ f ∗ = 0.5
Oxidizer must be supplied at a higher rate.
c
v 0 (t ) = 2 D0 > 0 : mean flow is upward and
τ the flame sheet propagates
into the oxidizer when f * < 1
Reason:

Shallower oxidizer profile implies lower diffusion rate.

Hence to obtain the higher total rate of supply, the
mean flow into the sheet from the oxidizer side is
required. Note: although v0 ≠ 0 means that the flame
moves with respect to the reactants, the propagation
speed v0 is not strictly an eigenvalue.
5-98
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
(xi) Temperature Profile
Energy equation
∂T ∂T 1 1 ∂  1 ∂T 
+v =  ρk 
∂t ∂y c p ρ ∂y  ρ ∂y 
1
Assume k~: ρk = ρ 0 k 0
T
∂T ∂T k  ρ ∂  ρ ∂T 
+v = 0  0  0 
∂t ∂y ρ0 c p  ρ ∂y  ρ ∂y 

Transform variables ( y, t ) → (η, T )

∂T ∂T D0 ∂ 2T Pr
+ v0 = Le =
∂τ ∂η Le ∂η2 Sc

where
ρ0 c p D0 Pr cµ v
Le = ; Pr = p 0 , Sc = 0
k Sc k0 D0

Assume Le = 1

∂T ∂T ∂ 2T
+ v0 = D0 2
∂τ ∂η ∂η

Same as species equations:

∂Yi ∂Yi ∂ 2Yi

+ v0 = D0 2
∂τ ∂η ∂η
5-99
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES
5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )
(cont’d)
Hence solution is:
 A{erf (ζ − c ) + erf c} + B ζ≥0
T =
C{erf (− ζ + c ) − erf c} + D ζ≤0

η c
ζ= ; υ0 = 2 D0
2 D0 τ τ

(xii) Apply Boundary and Matching Conditions

η = 0 T = Tf Flame temperature is independent
of time because reaction rate is
infinitely fast

erf (− c ) = −erf (− c ) ⇒ B = D = T f

ζ → +∞ : T → T0 and T0 − T f = A{1 + erf c}

ζ → −∞ : T → T0 and T0 − T f = C{1 − erf c}

Hence the solution is:

 erf (ζ − c ) + erf c
= Y1 ζ≥0
T f − T  1 + erf c
=
T f − T0  erf (− ζ + c ) − erf c = Y ζ≤0
 1 − erf c
2

5-100
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )

(xiii) Determine Tf (cont’d)
Conservation of energy (power) at the flame sheet:
 ∂T   ∂T 
k  −k  = −Qw
 ∂y 0+  ∂y  0−
∂ 1 ρ d
=
∂y 2 D0 τ ρ0 dζ
Qw := total rate of energy release in combustion
of mixture of mf + m0
m& f mass flow rate of fuel
energy mass of fuel 
= ⋅ = q f m& f q f energy released per unit
mf time mass of fuel

 ∂Y  ρD ρ  dY1 
At flame sheet, m& f = −ρD 1  = −  
 ∂y  0+ 2 D0 τ ρ 0  dζ 0+
ρD ρ  dY1 
⇒ Qw = q f m& f = −q f  
2 D0 τ ρ0  dζ  0+
Substitute in power balance at the flame sheet:
k ρ  dT   dT    ρD ρ  dY1  
  −    = − q f −   
2 D0 τ ρ0 d ζ dζ
  0+   0−   2 D0 τ ρ 0  dζ  0+ 
 dT   dT   ρD  dY  k
  −   = q f   1  Le = =1
 dζ  0+  dζ  0−  k  dζ  0+ ρc p D

 dT   dT  q f  dY1 
  −  =  
dζ d ζ
  0+   0− c p  dζ 0+
5-101
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )

From solutions for (Tf – T), Y1, and Y2 : (cont’d)

 dY   dY   q  dY1 
(T f − T0 ) 1  −  2   = f  
 dζ  0+  dζ  0−  c p  dζ  0+
d  k
 ds erf ( ± ζ m c ) = ± e −c 2
; c=
4D0
0+

 − e −c 2 − e −c  q f ∗ − e −c
2 2

T f − T0  − =
1 − erf c 1 + erf c  c p 1 + erf c
1− f ∗ 1 + erf c 1
erf c = ⇒ =
1+ f ∗ 1 − erf c f ∗

qf f ∗
T f − T0 =
cp 1+ f ∗

This is exactly the result obtained from the global power balance
constructed directly:
(m& f + m& 0 )c pT0 + m& f q f = (m& f + m& 0 )c pT f
m& f (c pT0 + h fr ) qf f ∗ m& f
(m& f + m& 0 )c pT f T f = T0 +
cp 1+ f ∗
f =
m& 0

f ∗ h fr + h0 r
q f = c pT0 +
1+ f ∗

m& 0 (c pT0 + h0 r )
5-102
 5.7 LAMINAR DIFFUSION–CONTROLLED
NON–PREMIXED FLAMES

5.7.4 Time-dependent One-Dimensional Diffusion Flame (β → ∞ )

(cont’d)
(xiv) Flame Propagation and Concentration Profiles

Fuel and oxidizer are supplied at rates:

2
 ∂Y1   ρ  1  ∂Y1  D0 e −c
m& f =  ρD  = (ρD )    = ρ0
 ∂y 0+ ρ
 0 2 D τ
0  ∂y 0+ πτ 1 + erf c
2
D0 e −c m& f 1 − erf c & ∗
m& 0 = ρ0 = = f
πτ 1 − erf c m& 0 1 + erf c

For a typical flame at 1 atmosphere,

D0 gm
ρ0 ~ 10 − 2 for τ ~ 10 −3 s
τ s
2
1 e −c
f* c π 1 + erf c f*<1
hydrocarbon
0.1 0.994 0.127 steeper
flames ∇(oxidizer)
0.5 0.305 0.386
in air
1 0 0.564
solid steeper
2 -0.305 0.771 ∇(fuel) f*>1
propellant
10 -0.994 1.272
General behavior
> ∇Y2 Flame
f* c ∇Y1 <
Propagation
<1 >0 < into oxidizer

>1 <0 > into fuel

5-103
 5.8 STRAINED LAMINAR FLAMES

In real flows — notably in turbulent flows — flames are subject to

translation, rotation, straining (stretching) and bending (curvature).

Locally, translation and rigid body rotation have no effect on the

burning properties.

However, both straining and curvature influence the burning rate (flame
speed), ignition and extinction. Those effects must be accounted for in
proper modeling of turbulent flows — they are part of the “fundamental
processes”.

In this section only straining is investigated. Several idealized situations

illustrate the presence of strain for premixed and diffusion flames
imagined here to be sheets in the limit of infinitely fast reaction rates:

Premixed flames with strain:

reactants reactants

products

products reactants

5-104
 5.8 STRAINED LAMINAR FLAMES

Stagnation point flow reactants

products

As the distance from the stagnation point increases to

planes parallel to the plate, the vertical velocity decreases,
but the increment of flame speed due to the strain also
decreases so a flat flame sheet may exist.
Diffusion flames with strain:
counterflow flame droplet combustion
fuel (contains strain and curvature)
oxidizer

oxidizer

fuel
vortex
combustion
oxidizer

Note that even for premixed flames, diffusion is significant.

5-105
 5.8 STRAINED LAMINAR FLAMES

5.8.1 The Kinematics of Strained Flows

The general motion of a fluid element consists of translation, rotation
and shear. If translational motion is removed—i.e. in a reference frame
moving with the local mass-averaged velocity, due to rotation and
shearing strains, a fluid element at two successive instants of time,
undergoes a change like:

t t + δt

Note that for purely rigid body rotation at the angular rate Ω, a point at
the distance r from the origin has the velocity Ωr perpendicular to the
radial line.
To analyze the rotational and deformational motion, consider the
relative motion of the two nearby points P and O referred to an origin
Os in an inertial reference frame: r
dδr r r
= δr ⋅ ∇u
dt
or

δx&   u x u y  δx  δx 

 =   = [U ] 
δy&   v x v y  δy  δy 

5-106
 5.8 STRAINED LAMINAR FLAMES

5.8.1 The Kinematics of Strained Flows (cont’d)

The matrix [U] can be split into antisymmetric and

symmetric parts:
[U ] = [U ] + [U ]
(a) (s)

and “it can be shown that”:

 
0 −
1
(v − u )
[U ( a ) ] =  1 
x y
2 (pure rotation)


2 x
(v − u y ) 0 

 
u
1
(u + v )
[U ( s ) ] =  1 
x y x
2
 (straining)
 (u y + v x ) vy 
2 

Remarks: (1) for pure rotation: u x = v y = 0 ; − u y = v x = θ&

0 − θ& 
⇒ [U ( s ) ] = 0 ; [U ] =  &
(a)

θ 0
r
(2) for potential flow, ∇ × u = 0 ⇒ vx − uy = 0

⇒ θ& = (v x − u y ) = ω = 0
1 1
2 2
(3) in potential flow there is no rotation of a fluid
element, but there is straining.

5-107
 5.8 STRAINED LAMINAR FLAMES

5.8.1 The Kinematics of Strained Flows (cont’d)

e.g. in a potential vortex, the diagonals of a fluid element always
remain parallel to their positions in some chosen initial state:

c
vθ =
r

t t + δt

(4) absence of vorticity or shear does not necessarily

imply absence of straining; if vorticity is present,
a fluid element possesses rotation having the
angular velocity ω:
1
ω = ( vorticity)
2
(the ½ arises because of the conventional
r
definition of vorticity = ∇ × u )

(5) here we consider flows having pure straining and

no rotation, [U(a)] = 0, so

 
u
1
(u + v )
[U ] = [U ( s ) ] =  1 
x y x
2

 (u y + v x ) vy 
2 

5-108
 5.8 STRAINED LAMINAR FLAMES

5.8.1 The Kinematics of Strained Flows (cont’d)

Examples of pure straining flows in incompressible flows:
(i) One-dimensional shear:
r  ∂v ∂u 
u = ky v = 0 ⇒ vorticity = ∇ × u = kˆ −  = k
 ∂x ∂y 

shear with no strain

(vorticity ≠ 0)

(ii) Stagnation point flow: strain with no shear

u = εx r
∇ ⋅u = ε − ε = 0
v = − εy
∇ × u = kˆ(v x − u y ) = kˆ(0 − 0 ) = 0
r


 ux
1
(u x + v y ) ε 0 
[U ] =  1
(s) 2
 = 0 − ε  strain ≠ 0
 (u y + v x ) 0   
2 

 0
1
(v x − u y )
[U ] =  1
(a) 2
 = [0] rotation = 0
 (v x − u y ) 0 
2 

This is a potential flow that causes straining

but possesses no shear (vorticity = 0)
5-109
 5.8 STRAINED LAMINAR FLAMES

5.8.1 The Kinematics of Strained Flows (cont’d)

Streamlines for incompressible stagnation point flow:

strain but no shear.

u = εx ; v = − εy
x = y = 1: u = ε
v = −ε
x = y = −1 : u = − ε
v=ε
x = − y = 1: u = ε
v=ε
− x = y = 1 : u = −ε
v = −ε

x = y = 0: u = v = 0
5-110
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Consider now the problem of a planar flame sheet in a flow containing
an internal stagnation point; this is usually called a “counterflow
diffusion flame”:
fuel

oxidizer
Assume
(i) Irrotational two-dimensional flow:
r
∇ × u = 0 ⇒ vx = uy
(ii) Variable density:
∂ρ ∂ ∂
+ (ρu ) + (ρv ) = 0
∂t ∂x ∂y
(iii) Uniform and constant pressure (hence
no momentum equation required)
(iv) Stretching (straining) parallel to the x–z
plane and independent of y:
u = ε(t ) x
By continuity, the stretching u(x) induces
a vertical velocity unknown at this point
(v = –εy if ρ = constant)
(v) Thermodynamic properties vary only in the
y-direction.
ρ = ρ( y, t ) etc.
5-111
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Remarks:

(i) It is not obvious a priori that solution to such a

problem exists. Recall that for the plane
diffusion flame with no strain, an average
vertical velocity v0 of the flow is required to
keep the flat flame sheet stationary in the
laboratory system.

(ii) Here, straining induces a vertical velocity,

varying in x, which in the plane of the sheet
must have just the right value locally to
compensate the influences of the parallel strain
on the properties of the flame. In particular, the
tendency of the flame to diffuse and thicken
must be precisely compensated by the
kinematics of the strain field, producing a
steady globally flat flame sheet.

(iii) The main purpose of the following analysis is to

determine the effect of strain on fuel
consumption by a steady flat diffusion flame.

(iv) The procedure follows generally the sequence

of steps summarized in Section 5.6.3.

5-112
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Procedure:
(i) Conservation of mass with u = ε(t)x ; ρ=ρ(y, t):
∂ρ ∂
+ ε(t )ρ + (ρv ) = 0
∂t ∂x
(ii) Conservation of species
∂Yi ∂Y 1 ∂  ∂Yi   ∂Yi by 
+v i =  ρD   = 0 
∂t ∂y ρ ∂y  ∂y   ∂x assumption 
(iii) Conservation of energy
∂T ∂T 1 ∂  ∂T   ∂T 
+v = k   = 0
∂t ∂y ρ c p ∂y  ∂y   ∂x 
(iv) Integrate the continuity equation:
y
∂ρ ρ 
y y
ρ
∫ ∂t  ρ 
  dy '+ ε (t )∫ρ dy '+ ρ v = 0
0  0 0 0  0 0
(ρv)0 = 0 in the stagnation point flow

∂ρ
y
ρ
∫ ∂t  ρ 
  dy ′ + v = −ε(t )η
o  0 ρ 0

(v) Apply the Howarth-Dorodnitsyn transformation

y
ρ
( y, t ) → τ = t ; η=∫ dy '
ρ
0 0

∂Yi  ρ  ∂Yi 1 ∂  ρ 2 D ∂Yi 

y
ρ
+  v + ∫ dy ' =  
∂τ  ρ0 ρ
0 0  ∂η ρ 0 ∂η ρ
 0 ∂η 

5-113
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Assume:
ρ 2 D = ρ02 D0 , ρk = ρ0 k0 and
ρ0 c p D0
Le = =1
k0

i.e. again, the “effective” transport properties for temperature

are the same as those for species.

Unstrained (§5.6.3) Strained

∂Yi ∂Yi ∂ 2Yi ∂Yi ∂Yi ∂ 2Yi

+ v0 = D0 2 + (v 0 − εη) = D0 2
∂τ ∂η ∂η ∂t ∂η ∂η
y
ρ
y
ρ
τ = t ; η = ∫ dy ' τ=t ; η= ∫ dy ′
0 ρ0 ρ
0 0

(vi) The species and energy equations become

∂Yi ∂Yi ∂ 2Yi

− ε ( τ) η = D0 2
∂τ ∂η ∂η
∂T ∂T ∂ 2T
− ε(τ)η = D0 2
∂τ ∂η ∂η

5-114
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

(vii) Introduce the composite variables (Shvab-Zel’dovich)

T q
χ = Y1 − f ∗Y2 ψ = − f Y1 Y1 : fuel
T0 c pT0 Y : oxidizer
2

and χ , ψ satisfy

∂χ ∂χ ∂ 2χ
− ε(τ)η = D0 2
∂τ ∂η ∂η (χ , ψ are continuous
throughout the field)
∂ψ ∂ψ ∂ 2ψ
− ε ( τ) η = D0 2
∂τ ∂η ∂η

with boundary conditions:

qf
y = +∞ : χ (∞, t ) = 1 ψ(∞,t ) = 1 −
c pT0
y = −∞ : χ (− ∞ , t ) = − f ∗ ψ(− ∞,t ) = 1

(viii) Transformation from Eulerian to Lagrangian variables

Consider a fluid element at τ = 0 , η = ηi ; the element suffers
a change of position in the vertical direction for τ > 0.

• in the equivalent incompressible flow, the velocity in the

vertical direction is (if ui = εx , vi = – εy):
dη (η is the position of
v i = −ε(τ )η =
dτ the fluid element)

5-115
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

τ
− ∫ ε ( τ′ )dτ′
• integrate: η = η0 e 0

where η = η0 at t = 0. Hence from the Lagrangian point

of view, η0 , the initial position of a fluid element, can
be taken as a new independent variable

η = η0 : τ=0
η(η0 , τ) : τ>0
τ
− ∫ εdτ ' η0
• let g = e 0
: η= and g& = εg
g

(ix) Transform independent variables η , τ → η0 (τ ) , τ∗ = τ

∂ ∂η0 ∂ ∂
= = g (t )
∂η ∂η ∂η0 ∂η0
∂2 2 ∂
2
=g
∂η2 ∂η02
∂ ∂ ∂η ∂ ∂ ∂
= + 0 = ∗ + εgη
∂τ ∂τ * ∂τ ∂η0 ∂τ ∂η0
because
∂η ∂
= εg ⇒ 0 = (ηg ) = g&η = εg (τ )η = εg (τ∗ )η
dg
dt ∂τ ∂τ

5-116
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Apply this transformation to the equations for χ

and ψ:
∂χ ∂χ ∂ 2ϕ
− ε(t )η = D0 2
∂τ ∂η ∂η
 ∂ ∂  ∂χ 2 ∂ χ
2
 + εgη χ − εηg = D0 g
 ∂τ * ∂η 0  ∂η 0 ∂η02

Note:
∂χ ∂χ 2
The velocity v =-εy→-εη
− D0 g 2 2 = 0
∂τ * ∂η0 doesn’t appear because η0 is a
∂ψ Lagrangian variable and the
2 ∂ ψ
2
and − D0 g =0 equations describe changes in a
∂τ * ∂η02
frame moving with a fluid
element.

(x) Introduce the new time variable t* to eliminate g2 :

t∗
∗
∫ εdτ
∗
g dτ = dt
2 ∗
; g (τ ) = e 0
τ ∗2
τ∗ 2 ∗ ∗
∫ ε ( τ1 ) dτ1
t = ∫e
∗ 0
dτ∗2
0

Then:
∂χ ∂ 2χ
− D0 2 = 0
∂t ∗ ∂η0
∂ψ ∂ 2ψ
− D0 2 = 0
∂t ∗ ∂η0

5-117
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Solutions are error functions (again)

η0
Let ζ = ∗
and
2 D0t

χ = Y1 − f ∗Y2 = A1 + A2 erf ζ
T q
ψ= − f Y1 = B1 + B2 erf ζ
T0 c pT0

Apply the boundary conditions (vii)

χ : η0 → ∞ : Y1 = 1, Y2 = 0 ⇒ A1 + A2 = 1; erf (∞ ) = 1
η0 → −∞ : Y1 = 0 , Y2 = 1 ⇒ A1 − A2 = − f ∗ ; erf (− ∞ ) = −1

1− f ∗
A1 =
2
1+ f ∗
A2 =
2
T q
ψ : η0 → ∞ : = 1, Y1 = 1 ⇒ B1 + B2 = 1 − f
T0 c pT0
T
η0 → −∞ : = 1, Y1 = 0 ⇒ B1 − B2 = 1
T0
1 qf
B1 = 1 −
2 c pT0

1 qf
B2 = −
2 c pT0

5-118
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Solutions:

∗ 1− f ∗ 1+ f ∗
Y1 − f Y2 = + erf ζ
2 2
T qf  1 qf  1 qf
− Y1 = 1 − −
 2 c T erf ζ
T0 c pT0  2 c T
p 0  p 0

(xi) Flame position:

As for the diffusion flame with no strain, the flame position
is defined by Y1 = Y2 = 0:

1− f ∗ 1 + erf ζ f
erf ζ f = − ⇒ f∗=
1+ f ∗ 1 − erf ζ f

If f ∗ < 1, erf (ζ f < 0 ) ⇒ ζ f < 0

Interpretation: ηof
ζf =
2 D0t ∗
yf
ρ
ηof = gη f = g ∫ dy
ρ
0 0

For a lean flame ( f * < 1), the flame sheet lies in y < 0 and
moves downward in time because
1
ζf ~
t∗

5-119
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

(xii) Rate of Fuel Consumption

m& f = −(ρv iYi ) f + = −(ρvYi + ρViYi ) f +

At the flame sheet, Yi = 0 so

(ρvYi ) f + = 0
and
 ∂Y 
m& f = −(ρViYi ) f + =  ρD i  (ρ D = ρ D )
2 2
0 0
 ∂y  f +
ρ 2  ∂Yi 
= D 
ρ0  ∂η  f +

 ∂Y ∂ζ ∂ξ 
m& f = ρ 0 D0  i  (1)
 ∂ζ ∂ξ ∂η  f +

∗ 1 − f ∗ (1 + f ∗ )
Y1 − f Y2 = + erf ζ
2 2
− ζ 2f
 ∂Y   ∂Y 
⇒ 1 =
1
(1 + f ∗ )e − ζ 2f
⇒  1 =
2 e
(2)
 ∂ζ  f + π  ∂η  f + π (1 − erf ζ f )

η0 ∂ζ 1
ζ= ⇒ =
2 D0t ∗ ∂η0 2 D0t ∗

5-120
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Consider the special case ε(t) = constant:

t
− ∫ εdt ′
η = η0 e 0
= η 0 e − εt ; g (τ) = e ετ
τ ∗2

( )
τ∗ ∗ ∗
∫ ε ( τ1 ) dτ1 τ∗
2
∗ 1 2 ετ∗
t∗ = ∫ e 0
dτ∗2 = ∫ e 2 ετ2 dτ∗2 = e −1
0 0 2ε

dζ 1 2ε
⇒ = ∗ (3)
dη0 2 D0 (e 2 ετ − 1)

dη 0
= g (τ∗ ) = e 2ετ
∗
η0 = g ( τ ∗ )η ⇒ (4)
dη

Substitute (2), (3), (4) into (1):

∗ −ζ 2
D0 e ετ e f
m& f = ρ 0 (2ε)
π e 2 ετ − 1 1 − erf ζ f
∗

Short times ετ∗ → 0

−ζ 2 −ζ 2
D0 1 + ετ∗ e f D0 e f
m& f ≈ ρ0 (2ε) ≈ ρ0
π 2ετ∗ 1 − erf ζ f πτ∗ 1 − erf ζ f
where 2
1 − erf ζ f =
1+ f ∗
This is identical to the result found for the
unstrained flame, §5.6.4 (xiv)
5-121
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Remarks:
(i) Results for species concentrations Y1, Y2 and the
temperature T/To are formally the same as those for
the time-dependent diffusion flame sheet without
strain.

(ii) The differences in the time variable (a non-uniform

‘scaling’) and in the variable η (to account for the
vertical motion induced by the strain field) cause
changes in the similarity transformation variable
η
ζ=
2 D0 τ∗

(iii) For large times, ετ∗ → ∞,

ετ∗ −ζ 2f
D0 2ε e e
m& f ετ→

→∞ ρ 0 ∗
π e ετ 1− erf ζ f
 
2
−ζ
2D e f
m& f ≈ ρ0  ε
 π 1− erf ζ f 

Hence the strain rate causes increased fuel

consumption — fairly evident if the flame is
to remain in a fixed position.

5-122
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Unstrained Strained

(iv)  1 ηf
 β = arise in similarity ζf =
2 D0 4D0 τ∗
 K
transformation
c = pp. (5-71, 5-72) (similarity variable)
 2 D0

c K
v 0 = 2 D0 =
τ τ

∂Yi ∂Yi ∂ 2Yi ∂Yi ∂Yi ∂ 2Yi

+ v 0 (τ) = D0 2 + (v 0 − εη) = D0 2
∂τ ∂η ∂η ∂τ ∂η ∂η

m& f e ετ 1
 ετ → 0
= 2ετ →  2ετ
2D0 ε  1 + f ∗  −ζ 2f e −1  1 ετ → ∞
ρ0  e 
π  2 

m& f m& f ~
1
Σ
2D0 ε  1 + f ∗  −ζ 2f
ρ0  e m& f ~ Σ
π  2 

5-123
 5.8 STRAINED LAMINAR FLAMES

5.8.2 Strained Diffusion Flame in Stagnation Point Flow (cont’d)

Recall (§5.5.3 xi) the result for an unstrained diffusion flame sheet:
2
D0 e −c 1− f ∗ k
m& f = ρ 0 ; erf c = ∗
; c :=
πτ 1 + erf c 1+ f 2 D0

D0  1 + f ∗  −c 2 2
εt = 0: m& f = ρ 0 ∗  e ⇒ 1 + erf c =
πt  2  1+ f ∗

Here, for εt → 0:
D0  1 + f ∗  −ζ 2f ηof 1− f ∗
m& f = ρ 0  e ; ζf = ; erf ζ f = −
πτ  2  2 D0t ∗ 1+ f ∗

1− f ∗ 
− erf c = erf (−c) = ⇒ erf (−c) = erf ζ f = − 
∗
 1 + f 

Hence numerically, c2 = ζ f2 and the two results agree so far

as the fuel consumption rate is concerned: i.e. ε → 0 in the
counterflow flame gives the same number for m& f as the
case ε = 0 gives.

However, the flow for ε → 0 here is quiescent and is

therefore identical to the earlier case (ε = 0) only for f * = 1,
when v0 = 0.

5-124
 5.8 STRAINED LAMINAR FLAMES
5.8.3 Compare Results for Strained and Unstrained Diffusion
Flame Sheets
UNSTRAINED (§ 5.6.4) STRAINED (§ 5.7.2)
2 −ζ 2f
D0 e −c D0 2Στ*e Στ*
e
m& f = ρ0 (5 − ) m& f = ρ0 (5 − )
πτ 1− erf c πτ* e 2 Στ* −1 1+ erf ζ f

(4 − 72 ) (c ↔ −ζ f )
k η0 f
c :=
4 D0
ζ f := (5 − )
τ := τ : * = t 4 D0 τ*
1− f * 1− f *
erf c = erf ζ f = −
1+ f * 1+ f *
qf f* f *h fr + hor
Flame Temperature: T f2 = T0 + ; q f = c pT0 (5 − )
c p 1+ f * 1+ f *

mass consumption rate of fuel

f * :=
mass consumption rate of oxidizer
Note: Different definitions of c, ζf due to different methods of solution
used here:
Unstrained flame: similarity solution
Strained flame: Shvab-Zel’dovich variable
Limit Behavior for the Strained Flame:
ετ* → 0: (short times or vanishingly small strain rate)
2
D0 e −ζt
m& f → ρ 0 result for strain = 0
πτ* 1− erf ζ f
ετ* → ∞: (long times or large strain rate)
 D0 e −ζt 
2

m& f → ρ0  2ε independent of time

 π 1 − erf ζ f 
 m& f ~ ε
Nonzero strain rate increases the fuel consumption rate.
5-125
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

• In Section 4.6 reasoning was given to show that a flame sheet

generated at the interface of oxidizer and fuel streams, a steady two-
dimensional problem could be interpreted as the time evolution of a
one-dimensional diffusion flame:

• However, in the initial portion of the flow adjacent to the edge of the
splitter plate, where the flame is ignited, chemical reaction is
distributed and there is no flame sheet. The approximation as a
flame sheet is not valid: an entirely different sort of analysis is
required. This problem was treated in a well-known paper by
Marble and Adamson (1954), Jet Propulsion, pp. 85-94 for the case
of a flame developing between a stream of reactants and a hot stream
of combustion products.
• In the case covered here, the flame is a pre-mixed flame
ignited by the hot products and diffusion is a significant
process: a purely thermal theory is not applicable

5-126
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

Reactant
Stream

Hot Combustion
Products

• In the early stage, the heat generated by combustion is small and a

perturbation method may be used: the velocity, temperature and
concentration fields do not differ much from their forms in the
absence of chemical reaction.

• Further downstream an integral method is applied.

5-127
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition

• These physical processes dominate in the early stage:

• development of a laminar mixing zone
• development of the temperature field due to heat conduction
• interdiffusion of the streams, and development of a zone of
chemical reaction.
• The pressure is practically uniform throughout the field and in
steady flow the governing equations are those given in ' 4.2.
Development of the region may be approximated as a boundary layer
flow and the governing equations are:

∂
(ρu ) + ∂ (ρv ) = 0
∂x ∂y
∂u ∂u ∂  ∂u 
ρu + ρv =  µ 
∂x ∂y ∂y  ∂y 
∂T ∂T ∂  ∂T  −E
ρuC p + ρvC p =  λ  + qAe RoT
∂x ∂y ∂y  ∂y 
∂Y ∂Y ∂  ∂Y  −E
ρu + ρv =  ρD  − Ae RoT
∂x ∂y ∂y  ∂y 

where
ρ r mass reactants/volume
Y= =
ρ total mass/volume

5-128
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

(i) Howarth-Dorodnitsyn Transformation

As in Section 5.6.3, we convert the problem defined
above to an equivalent incompressible flow problem with
x& → ζ = x , y → η
the transformation
y
ρ
η=
∫
0
ρ1
dy

where ρ1 is the density of reactions far from the combustion

zone.

The velocity components in the transformed flow are:

u =u
η
ρ ∂ ρ
v = v + u ∫  dy
ρ1 0 ∂x  ρ1 

and the equations for the equivalent incompressible flow are:

∂u ∂v µ1
+ =0 where: ν=
∂ζ ∂η ρ1
∂u ∂v ∂ 2u C p µ1
u +v =ν 2 Pr =
∂ζ ∂η ∂y k1
∂T ∂T ν ∂ 2T q −E
µ1
u +v = + Ye RοT Sc =
∂ζ ∂η Pr ∂η C pT
2
ρD1
∂Y ∂Y ν ∂ 2Y Y − E RοT
u +v = − e
∂ζ ∂η S c ∂η2 T
5-129
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

• Solution is to be obtained subject to the boundary conditions applied
far from the region of combustion.
η → +∞ : u = u1 , T → T1
η → −∞ : u = u 2 , T → T2
(ii) Definition of the Extent of the Ignition Stage
The main idea for solution in the early stage is that the
influences of chemical reaction and heat release are small
and that, because of the Arrhenius factor, what reaction does
occur is chiefly at temperatures near T2 (i.e., in the vicinity
of the interface).
• The reaction is most vigorous, that is, the rate has
maximum value, where
d
dη
( −E
Ye R T = 0
ο
) (assume T constant)

dY E
⇒ = −Y
dT RοT

For Pr = Sc = 1, and for the boundary conditions imposed

here, the temperature and concentration profiles are similar:

T2 − T 0 T = T2
Y (y) = →
T2 − T1 1 T = T1

5-130
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

Hence dY −1
=
dT T2 − T1

dY
and the two expressions for imply (T ≈ T2) that the
dT
concentration where the reaction rate is maximum is:
RοT2 1
Y=
E 1 − T1 T2

so the rate of combustion is

RοT 1 1 −E R T
e ο 2

E 1 − T1 T2 τ

Because u ≈ u 2 where the reaction takes place, the heat

released in the distance x, i.e. during the time x u2 , is
−E
x e RT ο 2
x
:=
u 2 τE  T1  l
1 − 
RοT2  T2 

where the characteristic length l is

E  T1  E R T
l = τu 2 1 − e ο 2

RοT2  T2 

5-131
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

The idea now is to develop the solution for small ζ = x l , the

region near the beginning of the joining of the two streams.

(iii) Solution for small ζ = x l

(a) Momentum Equation

The equation for conservation of mass is satisfied by

defining a stream function ψ:
∂ψ ∂ψ
u= ; v=−
∂η ∂ζ
This is a boundary layer problem; guided by Blasius’
solution; assume that the stream function can be written
in the ‘similarity’ form,
ψ = u1νx f (σ )

with
η
σ=
u1 νx

Substitution in the momentum equation leads to the

result that f (σ) must satisfy
 f ′(∞ ) = 1

 u
2 f ′′′ + ff ′′ = 0  f ′(− ∞ ) = 2 = Λ
 u1
 f ′′(∞ ) = f ′′(− ∞ ) = 0

5-132
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

(b) Energy and Concentration Equations

T T E
Let θ = , θ1 = and θ0 =
T2 T2 RοT
Energy Equation:
∂θ ∂θ ν ∂ 2 θ q Y − θ0 θ
u +v = + e
∂ζ ∂η Pr ∂η2 C pT2 τ

Concentration Equation:

∂Y ∂Y ν ∂ 2Y Y −θ0 θ
u +v = = − e
∂ζ ∂η Sc ∂η 2
τ

1 
(1) No reaction  → 0  :
τ 

Assume similarity solutions θ(0)(σ), Y(0)(σ):

2 ( 0 )″ ′
θ + fθ ( 0 ) = 0
Pr
2 ( 0 )″ ′
Y + fY ( 0 ) = 0
Sc

B.C. θ( 0 ) (∞ ) = θ1 Y ( 0 ) (∞ ) = 1
θ( 0 ) (− ∞ ) = 1 Y ( 0 ) (− ∞ ) = 0
′ ′
θ′(∞ ) = θ′(− ∞ ) = Y ( 0 ) (∞ ) = Y ( 0 ) (− ∞ ) = 0

5-133
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

(2) Small Heat Addition (χ = σ l small)

Now θ = θ(σ, ζ ); Y = Y (σ, ζ ) :

1 ∂ 2 θ f ∂θ ∂θ q 1 
− θ  −1 
+ ′
− fχ =− ΛYχθ a (1 − θ1 )e  θ 
a

Pr ∂σ 2
2 ∂σ ∂χ C pT2

For small χ (χ << l ) , write

θ = θ ( 0 ) (σ ) + χθ (1) (σ ) + χ 2 θ ( 2 ) (σ ) + L

Similarly for Y (σ, χ )

u
For Λ = 2 = 1 , f = σ :
u1
 θ( 0 ) (σ ) − θ1 1 1 
 1− θ =  1 − erf  Pr σ 
 1 2   2 

 Y ( 0 ) (σ ) − 1 = 1 1 − erf  1 Scσ 
 2  2 
2 z −t 2 dt
erf ( z ) =
π ∫0
e

Substitute θ = θ(0), Y = Y(0) in the right-hand sides of energy and

concentration equations to obtain inhomogeneous equations for
θ(1) and Y(1), e.g.
 1 
1 ∂ 2 θ(1) f ∂θ(1) q − θ  −1 
′ =− ΛY θ0 (1 − θ1 )e  θ 
0

+ − fθ (1) (0) 0

Pr ∂σ 2
2 ∂σ C pT2
Solutions can be obtained by introducing a Green’s function.
5-134
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

Results (Marble and Adamson):
Pr = 0.91; Sc = 1.00 ; θ1 = 0.286 ;
q
θ0 = 23.96 ; = 1.29 ; T1 = 30
C pT1

5-135
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

Remarks:
(i) ‘Ignition’ is indicated by development of a bulge in the
temperature profile at some value of χ i = xi l and σ2.
A vertical slope in θ(v1) occurs at
 1 
q − θ  −1 
θ (σ1 ) = Y ( 0 ) (σ1 )θ a (1 − θ1 )e  θ ( σ1 )
a (1 )
(1) 

C pT2
and the axial location of the bulge is found from
∂θ ∂θ dθ ( 0 ) dθ (1)
=0⇒ ≈ +χ
∂σ ∂σ dσ dσ
 dθ ( 0 ) dσ 
⇒ χ1 = −  (1 ) 
 dθ dσ  σ c

(ii) Numerical solution ⇒ σi and χi is found by applying

asymptotic expansion for θ(0) and θ(1).

(iii) The distance from the origin where the two streams
first make contact to the point where a bulge develops
in the temperature profile may be called the
‘detachment length.’ Its value is most sensitive to the
initial temperature, as the figure shows.

(iv) The analysis does not predict true “blow-off” but the
detachment length does rapidly become enormous for
T2/T1 < 3+.

5-136
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.1 Early Stage : Ignition (cont’d)

xi , cm
DETACHMENT LENGTH,

RATIO OF STREAM TEMPERATURES, T2 / T1

Variation of Detachment Length with

Temperature of the Hot Gas Stream
5-137
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame

• Following ignition, the flame develops large temperature changes
and gradients, placing the problem outside the range of small
perturbation theory.

• Marble and Adamson apply a form of Karman’s integral method.

The chief difference from previous applications if that the equation
for species conservation must be included in the analysis. This
approximate analysis covers both the region 0 < x < xi to ignition and
x > xi where the flame develops. Hence the results overlap those
given in the preceding section.

• Procedure:
(i) The chief assumptions are:
• uniform and constant pressure
• molecular weights, specific heats and transport
coefficients equal for the two components
⇒ ρ ~ 1/T and Pr, Sc constant.
• Arrhenius reaction rate
• adopt approximations for a boundary layer

(ii) Integrate the partial differential equations (given above) across

the reacting shear layer. Different thickness must be assigned
to the velocity, temperature and concentration profiles because
Pr ≠ Sc ≠ 1 is assumed.

(iii) Choose ‘reasonable’ representations of the profiles

across the layer; based on previous work by Lock
(1951) sections of sinusoids are used.

5-138
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

(iv) Substitution of (iii) into (ii) produces sets of coupled
differential equations for the thickness of the layers.

(v) The equations found in (iv) are integrated numerically. Use of

the results in (ii) produces approximations for the streamwise
development of the profiles.

• Remarks
(i) Pr < Sc assumed ⇒ diffusion of reactant into the hot stream is
slower than the diffusion of thermal
energy.
Hence the thermal layer is thicker than the concentration layer
initially—before a combustion wave or flame develops.
(ii) Subsequently, the abrupt rise of temperature from the cold
(reactant) stream to the hot (product) stream moves into the
region y > 0.

(iii) Results have been obtained only for the case u1 = u2 so there is
no shear layer. Hence the case, for example, of flow past an
edge with a recirculation zone is not included. See later work:
Marble and Candel (1978) 17th Symposium on
Combustion (p. 761)
Subbaiah (1983) AIAA J., Vol. 21 (p. 11)
Yang and Culick (1986) Combustion Science
and Technology, Vol. 45 (p. 1)

5-139
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

• Representations of the profiles:

∆ 1 , ∆ 2 : thickness of concentration profiles

β1 , β 2 : thickness of temperature profiles
β 0 : y > β 0 , chemical reaction neglected

5-140
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

Velocity: (u 0 constant on y = 0, x > 0)

 u 0  π y

 1 − 
 sin ( y > 0)
u u 0  u1  2 δ1
= +
u1 u1  u 2 u 0  π y
 −  sin ( y < 0)
 u1 u1 
 2 (− δ 2 )

[u1
= u 2 in the analysis]

Concentration:

 π  y − β0 
(1 − Y (β )) sin   (y > β )
∆ − β
0 0
 2  1 0 
Y = Y (β 0 ) + 
Y (β ) sin π  y − β 0  (y < β )
 0 2  − ∆ 2 − β 0 
0


5-141
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

Temperature:
 π  y − β0 
(θ
 1 − θ ) sin   (y > β )
β − β
0 0
 2  1 0 
x < xi : θ = θ0 + 
(1 − θ ) sin π  y − β 0  (y < β )
 2  − β 2 − β 0 
0 0

 π  y − β0 
(θ
 1 − θ ) sin   (y > β )
β − β
0 0
 2  1 0 
θ = θ0 + 
(θ(− ∆ ) − θ ) sin π  y − β 0  (− ∆ < y < β0 )
 2  − ∆ 2 − β 0 
2 0 2


 
θ=
1
(θ(− ∆ 2 ) + 1) − 1 (θ(− ∆ 2 ) − 1) cos π  − y − β 0  (− β < y < −∆ 2 )
2  − ∆ 2 − β0 
2
2 2

1
• For u1 = u 2 = u 0 : β 0 = (β1 − β 2 ) ⇒ β 2 = β1 − 2β 0
2
β + ∆2
Y (β 0 ) = 0
∆1 + ∆ 2

5-142
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

• Equations for the thickness (β 0 , β1 ; ∆1 , ∆ 2 )

Four coupled equations for x < xi and five for x > xi

d
x < xi : (∆1 + ∆ 2 ) = f1 (∆1 + ∆ 2 ; Y (β 0 ))
dx
d
Y (β 0 ) = f 2 (∆1 + ∆ 2 ; Y (β 0 ))
dx
d
(β1 − β 0 ) = f 3 (β1 − β 0 ; ∆1 + ∆ 2 )
dx
dβ 0
= f 4 (∆1 + ∆ 2 ; Y (β 0 ))
dx

d
x > xi : (∆1 + ∆ 2 ) = g1 (∆1 + ∆ 2 ; Y (β 0 ))
dx
d
Y (β 0 ) = g 2 (∆1 + ∆ 2 , Y (β 0 ); β1 − β 0 )
dx
d
(β1 − β 0 ) = g 3 (∆1 + ∆ 2 , Y (β 0 ); β1 − β 0 )
dx
dβ 0
= g 4 (β1 − β 0 )
dx

d   ∆1 + ∆ 2 
(
 2β − ∆ )
2 
Y (β ) − 1  = g 5 (∆ 1 + ∆ 2 ; β 1 − β 2 )
β − β
0
dx   1 0 

The fifth equation arises from the initial conditions at x = xi

5-143
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

Results:
• Initially, for Pr > Sc, the concentration thicknesses in both the
reactant and product streams are thinner than the temperature
profiles.
• As reaction proceeds, development of a combustion wave
requires the reaction must occur mainly at high temperature
and the thermal profile becomes thinner on the hot side.
• The concentration and temperature profiles have the same
thickness at x = xi where a ‘bulge’ first appears: the local
temperature then exceeds T2.
• Downstream of xi, the wave propagates into the reactant
stream: hence the temperature profile requires different
representations up and downstream of xi.

Variation of Maximum Temperature on

Later Stage of Flame Development
5-144
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

Temperature Profiles Leading to Laminar Flame

5-145
 5.9 IGNITION AND COMBUSTION IN A
LAMINAR MIXING REGION

5.9.2 Development of the Laminar Flame (cont’d)

Thicknesses of Concentration and Temperature Profiles

5-146
 5.10 COMBUSTION OF FUEL STRIPS

Ref.: F.E. Marble, Notes on Reacting Flows

Building on the results obtained for strained and unstrained

diffusion flame sheets, the sequence of topics to be covered is:
• combustion of a fuel strip
• combustion of a fuel strip in a straining field
• combustion in the field of a vortex
These ideas form part of the foundation for constructing a theory of
a turbulent combustion field composed of a distribution of coherent
flames or flamelets. The theory is incomplete and even in that
form has been worked out only for steady combustion. An
interesting problem is determining the dynamical response of such
a model of turbulent combustion to an impressed sinusoidal
acoustics field.

5.10.1 Combustion of an Unstrained Fuel Strip

y = y0
(t = 0 )
y = − y0

5-147
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

This is a step towards the situation existing during combustion in a
vortex field:

Analysis of the fuel strip burning in flame sheets is based on the

same equations used in Sections 5.4 - 5.6, and treated with the same
approximations. The boundary conditions differ.

Procedure of Analysis:
i) Apply the transformations of Howarth-Dorodnitsyn and then
to the similarity variable ζ :
y
ρ η
y →η= ∫ dy′ → ζ =
0 ρ0 2 D0t

Species equations are again, for Le = 1

∂Y1 ∂ 2Y1 ∂Y2 ∂ 2Y2 ∂T ∂ 2T

= D0 2 ; = D0 2 ; = D0 2
∂t ∂ζ ∂t ∂ζ ∂t ∂ζ

5-148
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

ii) Boundary conditions for flame sheets (Yi only):

Fuel : Y1 (ζ → +∞ ) =1 Y1 (ζ → ζ f ) = 0
Oxidizer : Y2 (ζ → −∞ ) =1 Y2 (ζ → −ζ f ) = 0

± ζ f (t ): time-dependent locations of the flame sheets

iii) Solutions can be expressed in terms of error functions:

2 ζ −z2 erf (∞ ) = 1 ; erf (− ∞ ) = −1

erf (ζ ) =
π ∫0
e dz
erf (0) = 0

iv) Construct solutions in terms of composite or Shvab-

Zel’dovich variables:

• Recall for case of fuel and oxidizers in half-planes:

5-149
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

• Here superpose such solutions to obtain a representation of
the composite concentration in a strip:

Initially

t>0

• These remarks suggest the form of solution:

Y1 − f ∗Y2 = A1erf (ζ + ζ 0 ) + A2 erf (ζ − ζ 0 ) + A3

flame sheet flame sheet Constant required

initially initially to give three
constants for
at ζ = −ζ 0 at ζ = +ζ 0 satisfying three
( y = − y0 ) ( y = + y0 ) initial conditions
5-150
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

Note: 1) Positions ± ζ f (t) of the flame sheets in general vary

with time;
2) Owing to symmetry, and global construction of
mass, there can be no average motion for y → ± ∞ ,
unlike the case of a single flame sheet.

v) Apply the initial and boundary conditions:

+ ∞ ( y > 0) erf (+ ∞ ) = 1
Initial conditions: t = 0+, ζ → 
− ∞ ( y < 0) erf (− ∞ ) = −1

y
Y1 − f ∗Y2 = A1erf (ζ + ζ 0 ) + A2 erf (ζ − ζ 0 ) + A3 ζ~
t

+ ∞ y + y0 > 0 (− y0 < y < +∞ )

ζ + ζ0 → 
− ∞ y + y0 < 0 ( y < −∞ )

+ ∞ y − y0 > 0 ( y0 < y < +∞ )

ζ −ζ0 → 
− ∞ y − y0 < 0 ( y < y0 )

5-151
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

Schematically, for t → 0+ :

∗ ∗
erf (ζ + ζ 0 ) → 1
Y1 − f Y2 = − f
erf (ζ − ζ 0 ) → 1
y = y0 ; ζ = ζ 0
∗
erf (ζ + ζ 0 ) → 1
Y1 − f Y2 = 1
erf (ζ − ζ 0 ) → −1
y = − y0 ; ζ = −ζ 0

erf (ζ + ζ 0 ) → −1
Y1 − f ∗Y2 = − f ∗
erf (ζ − ζ 0 ) → −1

Hence, the initial conditions can be applied in the three regions to

give 3 equations for the 3 constants:

y > y0 : − f ∗ = A1 + A2 + A3
− y0 < y < y0 : 1 = A1 − A2 + A3
y < y0 : − f ∗ = − A1 − A2 + A3

Solutions:

A1 =
1
(1 + f ∗ ); A2 = −
1
(1 + f ∗ ); A3 = − f ∗
2 2

5-152
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

vi) The solution for the Shvab-Zel’dovich composite variable is
therefore:
 
Y1 − f ∗Y2 =
1
(1 + f ∗ )erf  η + η0  − erf  η − η0  − f ∗
2   2 D0t   2 D0t 

vii) The formula for the time-dependent positions of the flame

sheets is found by applying the boundary condition that
Y1 = Y2 = 0 at y = ± y f (t ) , i.e. η = ±η f (t ):

 
0=
1
(1 + f ∗ )erf  η f + η0  − erf  η f − η0  − f ∗
2   2 D0t   2 D0t 

y0
ρ
Note: η0 = ∫ dy = y0 if ρ0 = density of fuel
ρ
0 0
≈ constant
Solution to the above transcendental equation gives the two
values ± η f (t ).
The two flame sheets propagate into the fuel and annihilate
each other at y = η f = 0 after the time t B computed as the
solution to
 η0  f∗
erf  =
 ∗
 2 D0t B  1 + f

5-153
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

Example: η0 = y0 = 0.5 cm; f ∗ =1

4D0 t

t B ≈ 0.85

viii) Fuel consumption rate is calculated from diffusion rate of fuel

to the flame sheet:
 ∂Y 
m& f = −2ρD 1 
 ∂y  y f

 − η f −η0  
2 2
 η f + η0 
m& f ∗ − 
1 + f   2 D0t 
 
 2 D0t 
 
= e − e 
D0 2  
2ρ0 
πt

5-154
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

Example:
f ∗ =1 ηf
η0
m& f
D0
2ρ0
πt

D0 t
η0
1
Remarks: 1) Initially, m& f ~ and the consumption rate
t
varies with time roughly like:

m& f 1
m& f ~ for 2 semi-infinite planes
t of fuel and oxidizer, an
isolated flame sheet with
no strain (p. 5-80)

tB time, t

5-155
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

Remarks: 2) Burnout time for the strip:

η0 −1  f∗ 
From (vii): = erf  
∗
2 D0t B  1 + f 
D0t B 1 1
=  → ~1
η0 −1  f∗  f ∗ =1
2erf −1 ( 12 )
2erf  
∗
 1 + f 

1 ( thickness)2 an estimate
η0 = ( thickness ) : tB ~
2 4 D0 ( f ∗ = 1)

Interpretation
For some interval 0 < t = ti < t B , the flame sheets propagate nearly
as isolated flame sheets until they interact, i.e. the distance between
them is roughly twice the ‘thickness’ of their profiles

Hence the thickness of an

unstrained flame sheet is
 ∂Y1  1 roughly
  ≈
 ∂y  0+ π D0 t π D0 ti = ∆ f
(p. 5- )

5-156
 5.10 COMBUSTION OF FUEL STRIPS

5.10.1 Combustion of an Unstrained Fuel Strip (cont’d)

Estimate time of independence by setting 2∆ f = thickness of the

fuel strip
2 π D0 ti ≈ thickness
( thickness)2
⇒ ti ≈
4π D0

Hence
Burnout time
~ π (f ∗
= 1)
Time of independence

That is, the flames are ‘independent’ for roughly 1/3 of their lifetime.

5-157
 5.10 COMBUSTION OF FUEL STRIPS

5.10.2 Strained Fuel Strip

As in Section 5.6, assume a condition of uniform strain, u = εx:

Procedure:

i) Apply the usual transformations to find the same equations

given in Section 5.6 (ε = constant)
y
ρ
Howarth-Dorodnitsyn: η= ∫ dy ′
ρ
0 0

t
∫ εdt′
Strained Scaling: ξ = ηe 0
= η e εt = η g ( t )
t′
t 2 ∫ εdt ′′
1 2 εt
Time Scaling: τ = ∫e 0
dt ′ = (e −1)
0 ε

5-158
 5.10 COMBUSTION OF FUEL STRIPS

5.10.2 Strained Fuel Strip (cont’d)

For Le = 1, the equations are ∂Yi ∂ 2Yi

= D0 2
∂t ∂ξ
∂T ∂ 2T
= D0 2
∂τ ∂ξ

Hence the equations are formally the same as those for a fuel strip
with no strain, except for the change of variables:

η → ξ = e εt η and t 
→ τ = (e − 1) ε
1 2 εt

→0
→t
2ε
dη
⇒ = −εη
dξ
ii) The solution for the Shvab-Zel’dovich variable is therefore
that produced in Section 5.7.1, (vi), but in the new variables
accounting for strain:

 
Y1 − f ∗Y2 =
1
(1 + f ∗ )erf  ξ + ξ0  − erf  ξ − ξ0  − f ∗
2   2 D0 τ   2 D0 τ 

Note: Take ρ = ρ 0 @ t = τ = 0 ⇒ ξ 0 = η0 = y0

iii) Positions of the two flame fronts:

Two roots of:
 ξ f + ξ0   ξ f − ξ0  2 f ∗
erf   − erf 


 2 D τ  = 1+ f ∗
 2 D τ
0   0 

5-159
 5.10 COMBUSTION OF FUEL STRIPS

5.10.2 Strained Fuel Strip (cont’d)

A numerical result for the position of the flame fronts:

ζf
ζ0
f ∗ = 0.2

f ∗ =1

2 D0t
ζ0
Remark:

a) 2 D0 τ ~ diffusion thickness. Hence, when 2 D0 τ ~ ξ0 , the

flames influence one another.

b) For f ∗ = m& f m& 0 = 0.2 , the required mass flow rate of oxidizer
is greater (5x) than that of the fuel. Hence the flames move
apart during some initial period, until the concentration
gradient in the fuel is too small to provide the necessary fuel
flow rate and then move together, eventually annihilating each
other when ξ f = 0 at t ~ 4ξ 0 D0

5-160
 5.10 COMBUSTION OF FUEL STRIPS

5.10.2 Strained Fuel Strip (cont’d)

b) cont’d
For example, if the combustion of
(x ~ t )
fuel issuing from a slot is
modeled in this way, the shape of
the flame is approximated by the
above result (modified by the
effects of vertical convection) of f
a candle flame or Bunsen burner.

iv) Burnout time: Y1 = Y2 = 0 @ ξ = 0 , by symmetry, so as in

Section 5.7.1 (viii):

D0 τ B 1 D0 τ B
= ⇒ ~ 1 for f ~ 1
ξ0  2 f  ξ0
2erf −1  
 1 + f 

1
or τB ~ ( thickness )2
4 D0

Hence the real time of burnout, tB, is found as follows:

τB = (e − 1) ~
1 2 εt B 1
2ε 4D0

1  ε  1
tB ~ log e 1 +  ε →
D0 <<1
2ε  2D0  4D0

5-161
 5.10 COMBUSTION OF FUEL STRIPS

5.10.2 Strained Fuel Strip (cont’d)

v) Fuel consumption rate, as in Section 5.7.1 (viii), but
1 1 1
→ = 2ε 2 εt and
t τ e −1

  ξ f + ξ0  
2 2

(1 + f ) 2  − 2 D0 τ 
 ξ f −ξ0 
∗ −  
m& f  2 D0 τ 
 
= e − e 
2D0 ε e 2 ετ − 1  
2ρ0 
π

± ξ f (t ) := positions of the flame fronts

For interpretation, recall results for the sequence of problems.

5-162
 5.10 COMBUSTION OF FUEL STRIPS

5.10.3 Interpretation of Results for Combustion of Fuel Strips

• Compare the various results for the consumption rate of fuel

(N.B. t now written everywhere for dimensional time)

(i) Unstrained Diffusion Flame (p. 5-80)

1
erf c =
2
f∗
D0 e −c m& f a1M 1
m& f = ρ0 f ∗ := = (p.5 - 51)
πt 1 + erf c m& 0 a2 M 2
f ∗ := stoichiometric ratio
of mass flow rates
1
m& f ~ decreases as t → ∞ because the concentration
t gradients decrease with time

(ii) Strained Diffusion Flame (pp. 5-98,99,100)

D0 εe εt  1 + f ∗  −ξ 2f  2 
m& f = ρ 0  e  ξ f = −c; 1 − erf ξ f = ∗

πt e 2 εt − 1  2   1 + f 

 D0  1 + f ∗  −ξ 2f Identical to an unstrained
ρ0 πt  2 e (t → 0) flame for short times
  


→
 Fuel consumption is constant
2 D  1 + f ∗
 −ξ 2f
ρ0 0
ε  e (t → ∞ ) for long times and increases
 πt  2  with the strain rate

5-163
 5.10 COMBUSTION OF FUEL STRIPS

5.10.3 Interpretation of Results for Combustion of Fuel Strips

(iii) Unstrained Fuel Strip (p. 5-131)

 − η f −η0  
2 2
 η f + η0 
D0  1 + f   2 D0t 
∗
  − 
 2 D0t 

 
m& f = 2ρ0  e −e 
πt  2  


OXIDIZER

m& f 1 FUEL
m& f ~
t
OXIDIZER

tB time, t

The burnout time is roughly three times the interval during which
the flames burn independently after ignition.

5-164
 5.10 COMBUSTION OF FUEL STRIPS

5.10.3 Interpretation of Results for Combustion of Fuel Strips

(cont’d)
(iv) Strained Fuel Strip (p. 5-135 ff)

  ξ f + ξ0  
2 2

(1 + f ) 2  − 2 D0t 

 ξ f −ξ0 
∗ −  
m& f  2 D0 t 
 
= e − e 
2D0 ε e 2 εt − 1  
2ρ0 
π

m& f  1 
 
D0 (1 + f )  εe −ξ f
∗  2

2ρ0 
π 2

Roughly three periods in the burnup of a strained fuel strip:

{ Initially transient εt → 0, so τ → t (p. 5-136)

The two flames burn independently like simple
unstrained diffusion flames

D0 1
m& f ~ ρ 0 ~
πt t

5-165
 5.10 COMBUSTION OF FUEL STRIPS

5.10.3 Interpretation of Results for Combustion of Fuel Strips

(cont’d)

| After t ~ 1/ε, the influence of the straining rate

becomes significant and the two flames burn
individually as strained diffusion flames; an
approximately steady state is revealed:

D0
m& f ~ ρ 0 ε
πt

The ‘steady’ burn interval is a consequence of the

stretching supply fuel and oxidizer to the flames,
tending to compensate the reduced rates due to the
reduced diffusion gradients.

5-166
 5.10 COMBUSTION OF FUEL STRIPS

5.10.3 Interpretation of Results for Combustion of Fuel Strips

(cont’d)
y
ρ η  similarity
Unstrained fuel strip: y → η = ∫ dy′ → ξ =  
0 ρ0 2 D0t  variable 

For a fuel strip subject to constant strain, η is replaced by

ξ = e εt η and t by (e − 1)
1 2 εt
2ε
Hence the position of the flame in the equivalent or
corresponding incompressible flow is
η f = ξ f e − εt

i.e. straining causes the flames to move together, accompanied

by increased diffusion gradients.

Initially the flames are separated by 2η0 2η0

The flame structures begin to interact when the flame sheets are
separated by twice the diffusion thickness, 2∆ ~ 2 D0t

Numerical results suggest that the interactions begin when

εt ~ 1 , so ∆ ~ D0 ε
2η0 η0
~ = e − εt B
2∆ D0 ε

5-167
 5.10 COMBUSTION OF FUEL STRIPS

5.10.3 Interpretation of Results for Combustion of Fuel Strips

(cont’d)
2η0 η0  ε 
~ = e −εt B log e −εt B η0 =0
2∆ D0 ε  D0 

1  1 ε 
⇒ interval of ' steady burning' , t B ~ log 

ε  η0 D0 

} After a period of nearly constant burn rate, the remaining fuel

in a strained fuel strip is quickly consumed, in a time ~ 0(1/ε),
with an associated ‘pulse’ of energy release.

Recall: If Q& is the rate of energy released by combustion, the

∂Q&
source for pressure waves is because the
∂t
associated density change appears as a monopole
source of mass to the field.

5-168
 5.11 SOME BASIC MECHANISMS FOR
INSTABILITIES IN COMBUSTORS OPERATING
WITH GASEOUS REACTANTS

5.11.1 General Remarks on the Context

• The instabilities in question here are observed as unstable

fluctuations of pressure.
• Hence the mechanisms must be realized as sources of pressure
(acoustic) waves which in turn may be identified as unsteady
sources of mass, momentum or energy.
• The relevant terms in the wave equation are
1 ∂ 2 p′ R ∂Q ′ R ∂
∇ p′ − 2 2 =
2
+ (CvTW )′ − a 2∇ ⋅ F′ + L
a ∂t Cv ∂t Cv ∂t
′
 ∂u  r
nˆ ⋅ ∇p′ = − ρ  ⋅ nˆ + nˆ ⋅ F′ + L
 ∂t 
r
Q : energy ; W : mass ; F : momentum
• Other contributions on the right-hand sides are related chiefly to
nonlinear interactions among the fluctuations; and to linear and
nonlinear interactions between the fluctuations and the mean flow
field.
• Subsequent to the modal expansion and spatial averaging, the
surface and volumetric source terms appear in the forcing
functions for the oscillator equations:
η
&& n + ωn2 ηn = Fn
Fn2
pa 2
Fn = ∫



ψ n
C
R ∂
v ∂t
Q{′ + (C vT W )′
}
− a 2
r
F ′ ⋅ ∇ ψ n

 dV

r ′
 ∂u 
+ ∫∫ ψ n  ρ  ⋅ nˆ dS
 ∂t 
5-169
 5.11 SOME BASIC MECHANISMS FOR
INSTABILITIES IN COMBUSTORS OPERATING
WITH GASEOUS REACTANTS

5.11.2 Some Classes of Possible Mechanisms

a) Elementary Properties of Flame Sheets

• Fluctuations of area (stretching or straining)

• Fluctuations of reaction consumption rate/area (stretching)
• Fluctuations of energy release (stretching; sensitivity to
pressure and temperature)

b) Macroscopic Combustion Processes

• Flame sheets, including segments (flamelets)

• Combustion in vortices: winding-up flame sheets
• Distributed combustion:
¾ Flamelet or coherent flame theories of turbulent
combustion fields
¾ pdf (probability distribution function) theories of
turbulent combustion fields
• Recirculation Zones and Hysteresis
• Fluctuations of Mixture Ratio: pressure fluctuations can
influence the injection system, including the pre-mixer, and
cause uniformities of mixture ratio injected by the stream to
the combustion zone. Some calculations have been reported
by Lieuwen and Zinn, 36th AIAA Aerospace Sciences
Meeting, 1998

5-170
 5.11 SOME BASIC MECHANISMS FOR
INSTABILITIES IN COMBUSTORS OPERATING
WITH GASEOUS REACTANTS

5.11.3 Present Status of the Dynamics of these Processes

(a) Flame sheets: a commonly invoked mechanism to-date for
investigations of active control (Cambridge, MIT, Politecnico
Milan, …); the dynamical response has been associated usually
with fluctuations of the area, first proposed as a mechanism by
Blackshear (1952) NACA TN 2772. Kasken and Noreen
(1956), Trans. ASME, Vol. 77, pp. 885-895, applied the idea to
interpret excitation of oscillations by vortex shedding.

(b) Combustion in Vortices: vortex shedding from a bluff body

flameholder, with delayed pulses of combustion rate and energy
release was first proposed by Rogers and Marble (1952), Jet
Propulsion, V. 26, (pp. 456-462), as a mechanism for transverse
instabilities (‘screech’).

5-171
 5.11 SOME BASIC MECHANISMS FOR
INSTABILITIES IN COMBUSTORS OPERATING
WITH GASEOUS REACTANTS

5.11.3 Present Status of the Dynamics of these Processes (cont’d)

(c) Distributed Combustion: has received practically no attention.
An interesting problem: determine the dynamical response of a
turbulent combustion field represented by flamelets (which may
also involve statistical distributions).

(d) Recirculation Zones: Russian researchers (Natanzon et al;

Beliy, Doulatov et al) have suggested the mechanism:

(i) depending on its history, a recirculation zone may give rise

to two states of steady combustion:

Ignition by Recirculation Zone

Self-Ignition in the Bulk Flow

5-172
 5.11 SOME BASIC MECHANISMS FOR
INSTABILITIES IN COMBUSTORS OPERATING
WITH GASEOUS REACTANTS

5.11.3 Present Status of the Dynamics of these Processes (cont’d)

(ii) if one of these states is stable and one unstable, the
possibility exists for hysteresis in the dynamics, as well as
in steady-state, and hence

5-173
 5.12 COMBUSTION IN A VORTEX

References:

Marble, F.E. (1984) “Growth of a Diffusion Flame in the Field of a

Vortex”, Advances in Aerospace Science, C. Casci, Ed. (pp. 395-
413).

Karagozan, A. R. (1982) “An Analytical Study of Diffusion Flames in

Vortex Structures”, Ph.D. Thesis, California Institute of
Technology.

Norton, O.P. (1983) “The Effects of Vortex Flames with Finite Reaction
Rates”, Ph.D. Thesis, California Institute of Technology.

Karagozan, A. R. and Marble, F.E. (1986) “Study of a Diffusion Flame

in a Stretched Vortex” Combustion Science and Technology, Vol. 45
(pp. 65-84).

Laverdant, A.M. and Cardel, S.M. (1989) “Computation of Diffusion

and Premixed Flames Rolled up in Vortex Structures”, J.
Propulsion, Vol. 5, No. 2 (pp. 134-143).

Macaraeg, M.C., Jackson, T.L. and Hussaini, M.Y. (1992) “Ignition and
Structures of a Laminar Diffusion Flame in the Field of a Vortex”,
Combustion Science and Technology, Vol. 87 (pp. 363-387).

Fiechtner, G.J., Carter, C.D., Katta, V.R., Gord, J.R., Donbar, J.M.,
and Rolon, J.C. (1999) “Regimes of Interaction Between a
Nonpremixed Hydrogen-Air Flame and an Isolated Vortex”, 37th
Aerospace Sciences Meeting, Paper AIAA No. 99-0320.

Katta, V.P. and Roquemore, W.M. (1999) “Studies on Unsteady

Opposing-Jet Diffusion Flames”, 37th AIAA Aerospace Sciences
Meeting, AIAA Paper No. 99-0322.

5-174
 5.12 COMBUSTION IN A VORTEX

5.12.1 Kinematics of the Vortex – Flame Interaction

• The basic problem: at t = 0 superimpose a point vortex on a flat
diffusion flame and follow the history of the flame sheet.

• Azimuthal velocity induced by an ideal potential vortex,

dθ Γ
v θ := r =−
dt 2πr
a more realistic model of the vortex contains a viscous core
(Bachelor, 1967, An Introduction to Fluid Mechanics).

vθ = −
Γ
2πr
(
1 − e −ξ
2
) ξ=
r
2 νt
 µ
ν = 
ρ


FUEL

OXIDIZER

• Straining causes
(i) reduced thickness of the diffusion layers,
increasing the consumption rate.
(ii) increased flame area, amplifying the integrated
consumption rate/volume.
(iii) eventual extinction due to rapid flow of cool
reactants to the flame and/or local depletion of
one of the reactants.
5-175
 5.12 COMBUSTION IN A VORTEX

5.12.1 Kinematics of the Vortex – Flame Interaction (cont’d)

• General character of the field:
r << 2 νt , near the center of the vortex :
−Γ r
vθ → (solid body rotation )
2πν 4t

r >> 2 νt , in the far field :

−Γ
vθ → (potential vortex )
2πr
• Close to the center, the strain rate is very small, far away the
movement of the sheet from its original location is small,
giving the distribution of strain rate shown for several
Reynolds numbers, Γ
R= :
2πν

γ
Near the center: Intermediate distance, ~ 0(1)
2 νt
arcs of
circles

5-176
 5.12 COMBUSTION IN A VORTEX

5.12.1 Kinematics of the Vortex – Flame Interaction (cont’d)

• The strain rate is given by the formula
  R 
4ζ 2 (1 − e − R 4 ζ )1 − 1 + e − R 4 ζ 
εt =   4ζ   ; ζ=
Γt
1 + 4ζ (1 − e − ζ ) 2πr 2
2 R 4 2

and for R/4ζ → 0, the limit of a potential vortex,

2
 Γt 
 2
4ζ 2 πr 
εt = = 
1 + 4ζ 2
 Γt 
2

1+  2 
 πr 
In this limit, the flame surface density (flame area/volume) is
2
1  Γt  Γt t
Σ= = 1+  2  → 2 3 for 2 → ∞
πr  πr  πr r
Near the origin, the flame surface density becomes indefinitely
large due to the ‘winding-up’.

• The kinematical structure of the spiral flame sheet depends on the

dimensionless similarity variable r Γt , meaning that the shape
of the spiral is independent of time, but the size increases as time
passes.

• Within the spiral, the spacing between sheets is approximately

1 π2 r 3
~
Σ Γt

5-177
 5.12 COMBUSTION IN A VORTEX

5.12.1 Kinematics of the Vortex – Flame Interaction (cont’d)

• Because most of the diffusion and combustion processes take
place outside the viscous (‘solid body’) core and can be estimated
as those within a potential vortex.

• Further analysis (Marble, …) gives the result for the consumption

rate increased by the velocity field induced by the vortex:
∆m& f
m& f
2
3
1
[ −1
]
~ Γ D03 1.2327 − 1.0694(Sc R ) 6 + L

where m& f is the consumption rate associated with combustion of a

flat sheet having the same area (length) as the flame in the spiral;
Sc is the Schmidt number ν/D0.

5-178
 5.12 COMBUSTION IN A VORTEX

5.12.1 Kinematics of the Vortex – Flame Interaction (cont’d)

• Norton finds that the consumption rate exhibits a weak pulse
during combustion in a vortex.

• That the part of combustion occurs in a pulse seems

not to have been found in other works (numerical
simulations).
5-179
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

References:

Wiliams, F.A. (1975) “A Review of Some Theoretical

Combustions of Turbulent Flame Structure”, AGARD CP
164.

Marble, F.E. and Broadwell, J.E. (1977) “The Coherent Flame

Model for Turbulent Chemical Reactions”, Project
SQUID, TRW Report TRW-9-PV.

Damköhler, G. (1940) Z. Electrochem, Vol. 46 (pp. 601- )

Schelkin, K.I. (1943) J. Technical Physics, USSR, Vol. 18 (pp.

520- )

Hawthorne, W.R., Weddell, D.S. and Hottel, H.C. (1948)

Third Symposium on Combustion, Flame and Explosion,
(pp 266- )

Hottel, H.C. (1952) Fourth Symposium (International) on

Combustion, (pp. 97- )

Karlovitz, B. Denniston, D.W.J. and Wells, F.E. (1951) J.

Chemical Physics, Vol. 19 (pp 451- )

Wohl, K., Shore, L. von Rosenberg, H. and Weil, C.W.

(1952) Fourth Symposium (International) on Combustion
(pp. 626- )

5-180
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

References (cont’d):

Darabika, N., Giovangilgi, V., Trouvé, A., Candel, S.M. and

Esposito, E. (1987) “Flamelet Modeling of Turbulent
Premixed Flames”, AGARD CP 422.

Al-Abbadi, N. and Chen, L.-D. (1994) “A Flamelet Modeling

for Pollutant Reduction in Diffusion Flames”, AIAA Parer
No, 94-2732.

Peters, N. (1986) “Laminar Diffusion Flamelet Concepts in

Turbulent Combustion”, 21st Symposium (International) on
Combustion (pp. - ).

Libby, P.A. and Williams, F.A. (1981) “Structure of Laminar

Flamelets in Premixed Turbulent Flames”, Combustion and
Flame, Vol. 44 (pp. 287- ).

5-181
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

• The idea is that combustion in a turbulent field takes place in

elements of thin flame fronts (flamelets), i.e. in zones of
concentrated combustion. Reaction rates (per unit volume and
time) are then computed as averages over a distribution of
flamelets.

• An alternative approach, for example, is based on relating local

reaction rates to mean values of products of fluctuations.

• The term ‘coherent’ means that while subject to convection,

distortion, straining, … an element of flame retains its identity.

• An attractive feature of this representation is that to a large

extent the details of the chemistry can be separated from the
details of the turbulence.

• The dynamical behavior of a reacting turbulent flow represented

as a collection of flamelets has not been analyzed.

5-182
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

5.13.1 Summary of the Procedure

• The figure, taken from Veynante et al (1991) shows the
speculated connection between the global situation, a ‘turbulent
diffusion flame’ and the elementary problem, a laminar
diffusion flame in a strain field.

• A similar picture holds for a turbulent premixed flame, but the

elementary flame problem is the premixed case (p. )

5-183
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

5.13.1 Summary of the Procedure (cont’d)

• The analysis is conveniently partitioned into four parts:

(i) the elementary laminar combustion problem local

properties and structure of the ‘flamelets’ in a strain field;

(ii) a description of the turbulent flow field based on

equations and closure rules for the flow variables;

(iii) an equation or equations connecting the flamelet models

to the turbulent flow (a ‘flame area balance equation’);

(iv) models and equations for chemical activity outside the

regions covered by (i) and (iii).

• The chief results are the consumption/production rates, mass per

unit volume and time, ρVDiΣ
M
ρ : density of mixture,
L3
VDi : volume rate of consumption/production of
specis i, per unit area of flame surface
L3
[VDi ] := 2 = L
L −s s
L2 1
∑ : flame surface area density, 3 =
L L
[ρVDi ∑] := M3 ⋅ L ⋅ 1 = 3M
L s L L −s
5-184
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

5.13.1 Summary of the Procedure (cont’d)

• A form of the equation governing the flame surface density is:

 Rate of Change 
     
 of Flame   Turbulent   Production of Σ 
  =   +  
 Surface   Diffusion of Σ   by Strain 
 Density, Σ     

   Dissipation of Σ 
 Dissipation of Σ   
–   –  by Mutual 
 by Extinction   Annihilation 
   

• A representation of the turbulent flow field is required to give the

velocity field, strain rate, diffusion,...

5-185
 5.13 MODELING A TURBULENT COMBUSTION
FIELD AS A DISTRIBUTION OF FLAMELETS
(COHERENT FLAMES)

5.13.1 Summary of the Procedure (cont’d)

The scheme of this approach is illustrated with the flow chart
(Veynante, et al., 1991)

5-186
 End of Section V

5-187