The effects of multicomponent fuel droplet evaporation on the kinetics

of strained opposed-ow diffusion ames

Chenguang Wang
a
, Anthony M. Dean
a,
, Huayang Zhu
b,c
, Robert J. Kee
b
a
Chemical and Biological Engineering, Colorado School of Mines, Golden, CO 80401, USA
b
Mechanical Engineering, Colorado School of Mines, Golden, CO 80401, USA
c
College of Petroleum Engineering, Xian Shiyou University, Shaanxi 710065, PR China
a r t i c l e i n f o
Article history:
Received 23 July 2012
Received in revised form 9 October 2012
Accepted 13 October 2012
Available online 14 November 2012
Keywords:
Multicomponent fuel droplets
Fuel pyrolysis
Opposed-ow nonpremixed ame
Modeling
a b s t r a c t
With the increasing use of alternative fuels, it becomes important to understand the impacts of their
different chemical and physical properties on combustion processes. The objective of this paper is to
explore the impact of the vaporization of a multicomponent liquid fuel on the combustion kinetics using
an opposed-ow diffusion ame model. The model fuel consisted of a n-heptane, n-dodecane, and
n-hexadecane mixture, selected to represent a FischerTropsch fuel. A computational model is developed
to describe the multicomponent vaporization process. Gas-phase chemical kinetics is modeled using a
reduced mechanism containing 196 species. Results compare pre-vaporized fuel streams with those con-
taining monodispersed initial droplet sizes of 20, 25 and 30 lm. The separation distance between the fuel
and air inlets is either 5 and 10 mm. In all cases the fuel is carried in nitrogen, the pressure is 10 atm, and
the fuel and air inlet velocities are 1 m s
1
. The fuel loading is set to achieve an overall equivalence ratio
of unity. Results show that the nite evaporation rate signicantly impacts the chemical kinetics. In
particular, if the combination of separation length, stream velocity, and fuel volatility is such that fuel
droplets penetrate into the higher temperature region near the ame-front, the rapid increase in evapo-
ration rate signicantly enhances the local vapor phase fuel mole fraction. The high temperature
increases reaction rates, leading to higher peak temperatures as well as increased pyrolysis in the
pre-ame region. For example, the peak temperature predicted for 30 lm droplets is 330 K higher than
that for the pre-vaporized case. This increase occurs in spite of an initial decrease in temperature as a
consequence of fuel vaporization. A similar effect is observed for the pre-ame pyrolysis products;
ethylene, acetylene, and butadiene all increase by about a factor of two for the 30 lm droplet case.
The implications of these ndings regarding the use of alternative fuels is discussed.
2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
1. Introduction
The objective of this paper is to analyze the impact of the vapor-
ization of a multicomponent fuel on the kinetics of non-premixed
strained ames. Such ames occur in diesel engines and turbines,
where the vaporization process is most closely connected to the
combustion event. Analysis of these ames is especially important
with the advent of alternative fuels that might have signicantly
different chemical and physical properties than conventional ones.
Such fuels (e.g., biomass-derived FischerTropsch fuels) might
have substantially different hydrocarbon compositions, such as a
higher concentration of branched alkanes and different boiling
point curves than petroleum-derived diesel fuels [1]. A convenient
framework in which one can examine this coupling of physical and
chemical properties of the fuel is an opposed-ow diffusion ame
(Fig. 1). This provides the advantage of a well-dened ow eld
that can be modeled as a one-dimensional boundary value prob-
lem. As a result, this device is often used for combined experimen-
tal/modeling efforts, such as the analysis of pre-vaporized fuels
[2,3]. Different models have been developed to predict the behav-
ior of fuel evaporation [46]. In this work such efforts are extended
by explicitly coupling the description of the evaporation of a mul-
ticomponent model fuel with the subsequent gas-phase kinetics of
its components. To our knowledge, this is the rst instance of such
an analysis in an opposed ow diffusion ame. Of course, such cou-
pling is frequently accomplished in CFD codes such as KIVA or Flu-
ent to describe the combustion kinetics in engines and turbines
[7,8]. In this work the goal is to develop a more explicit under-
standing of the impact of differential vaporization on the detailed
kinetics in a diffusion ame environment. With a simpler physical
conguration to model, it is possible to work with a more complex
kinetic mechanism.
0010-2180/$ - see front matter 2012 The Combustion Institute. Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.combustame.2012.10.012

Corresponding author. Tel.: +1 303 273 3643; Fax: +1 303 273 3730.
E-mail address: amdean@mines.edu (A.M. Dean).
Combustion and Flame 160 (2013) 265275
Contents lists available at SciVerse ScienceDirect
Combustion and Flame
j our nal homepage: www. el sevi er . com/ l ocat e/ combust ame
This paper is an extension of an earlier effort [9] in which various
single-component fuels, ranging from heptane to diesel fuel, were
modeled with a very simplied kinetic mechanism. The vaporiza-
tion model is extended to account for a multicomponent fuel, and
a much more detailed kinetic mechanism is used to explore the
impact of the various fuel components on the kinetics within the
diffusion ame. A three-component fuel, consisting of n-heptane,
n-dodecane, and n-hexadecane is used to create a simple surrogate
for a FischerTropsch fuel. This combination provides a wide range
of vaporization rates to better dene the impact of this parameter. A
reduced version of the comprehensive n-alkane mechanism devel-
oped by Westbrook and co-workers [10] is used to predict the ame
structure. The results of using monodispersed droplet sizes of 20, 25
and 30 lm are compared to those obtained using pre-vaporized
fuel. The strain rate is varied by changing the separation distance
of the inlet streams. This change has the added advantage of explor-
ing the impact of different residence times of the fuel vapor in the
hot nitrogen stream prior to entering the ame-front region.
Calculations are carried out at a pressure of 10 atm, with room
temperature liquid fuel droplets evaporating into a 950 K nitrogen
stream. These conditions approximate those encountered when
fuel is injected into a diesel engine. The results suggest that the
variation in evaporation rate of the various components can pro-
duce surprisingly large variations in ame behavior, leading to
large changes in the peak ame temperature as well as to substan-
tial differences in the amount of pre-ame chemistry. These results
are most pronounced when the evaporation of the heavier compo-
nents occurs near the ame-front. The increased temperature in
that region leads to a much higher rate of vaporization, producing
an enhancement of the local fuel vapor mole fraction that increases
the reaction rate.
2. Mathematical model
As discussed in previous models, the overall approach is based
upon an iterative algorithm, solving gas-phase conservation equa-
tions in an Eulerian framework and droplet tracking in a Lagrang-
ian framework [9,11]. The coupling is accomplished by source
terms in the gas-phase equations that are derived from the droplet
equations. The droplet equations depend upon the local gas-phase
environment.
2.1. Eulerian gas-phase conservation equations
The conservation equations for gas-phase steady-state strained
laminar axisymmetric opposed-ow ames in similarity form are
well known [9,1217]. A detailed derivation of the stagnation-ow
similarity equations may be found in Kee et al. [17]. After incorpo-
rating the source terms associated with droplet vaporization, the
equations can be summarized as
dqu
dz
2qV S
M
; 1
qu
dV
dz
qV
2
K
r

d
dz
l
dV
dz
_ _
S
V
VS
M
; 2
q

K
k1
u V
k
Y
k
c
p;k
dT
dz

d
dz
k
dT
dz
_ _

K
k1
_ x
k
W
k
h
k
Q
rad
S
T

K
k1
h
k
S
k
; 3
qu
dY
k
dz

dqY
k
V
k

dz

_
x
k
W
k
S
k
Y
k
S
M
: 4
The axial coordinate z is the independent variable. The dependent
variables include the axial velocity u, temperature T, and species
mass fractions Y
k
. The scaled radial velocity V = v/r is also a depen-
dent variable, with v and r being the radial velocity and radial
coordinate, respectively. The mass density q is evaluated using an
ideal-gas equation of state. Thermodynamic parameters include
the species molar weights W
k
, specic heats c
p,k
, and enthalpies
h
k
. The pressure-gradient parameter K
r
= (1/r)(dp/dr) is an
eigenvalue that is determined during the course of the solution.
Transport properties include mixture viscosity l and thermal
conductivity k. The diffusion velocity is represented as
V
k

1
X
k
W

K
jk
W
j
D
kj
rX
k

D
T
k
qY
k
rT
T
; 5
where X
k
are the mole fractions, D
kj
is the matrix of ordinary mul-
ticomponent diffusion coefcients, and D
T
k
are the thermal diffusion
coefcients [17]. The molar production rates of gas-phase species
by chemical reaction are represented as _ x
k
. Radiation heat transfer
between gaseous species and the environment is represented as
Q
rad
[9,18]. Thermodynamic properties and reaction rates are eval-
uated through CHEMKIN software interfaces [17].
As noted above, the purely gas-phase equations are extended to
include source terms associated with droplet vaporization. These
terms include S
M
representing the net gas-phase mass addition,
S
k
representing the source of gas-phase species k, S
V
representing
the source of radial momentum, and S
T
representing the source
of thermal energy. The quantitative evaluation of these terms is
discussed in a subsequent section.
2.2. Lagrangian droplets dynamics
The present model assumes slow-vaporization-limit behavior
[6], neglecting any spatial variations within individual droplets.
Thus, the droplet trajectories, mass, temperature, and composition
can be represented using a system of Lagrangian ordinary differen-
tial equations as
dz
d
dt
u
d
;
dr
d
dt
r
d
V
d
; 6
dm
d
dt
_ m
d
; 7
du
d
dt

F
z
m
d
;
dV
d
dt
V
2
d

F
r
m
d
r
d
; 8
dT
d
dt

_
q
d
m
d
c
p
d
; 9
dY
d;k
dt

_ m
d
m
d

k
Y
d;k
; 10
dF
dt
2FV
d
: 11
Fig. 1. Illustration of the opposed-ow nonpremixed-ame conguration.
266 C. Wang et al. / Combustion and Flame 160 (2013) 265275
In this formulation the independent variable is time t. The depen-
dent variables are the droplet axial position z
d
, radial position r
d
, ax-
ial velocity u
d
, scaled radial velocity V
d
= v
d
/r
d
, total mass m
d
,
temperature T
d
, mass fractions Y
d,k
, and the droplet ux-fraction
function F. The total mass vaporization rate is represented as _ m
d
,
with the mass vaporization rate of fuel species k being represented
as _ m
d;k

k
_ m
d
. Evaluation of the vaporization fraction
k
is dis-
cussed in the subsequent section. The droplet heat capacity is c
p
d
and heat transferred from the gas to the droplet is represented as
_ q
d
. The axial and radial forces, F
z
and F
r
, which represent drag and
thermophoretic forces, affect the droplet trajectories. As originally
discussed by Continillo and Sirignano, the present formulation per-
mits droplets to cross, and re-cross, the stagnation plane [19,9].
2.3. Droplet vaporization
Droplet vaporization is described with a thin-lm model
[12,13,15,20], which provides the needed quantitative relation-
ships between droplets and the surrounding gas-phase environ-
ment. The model is rst applied to evaluate the net vaporization
rate, lumping the multicomponent fuel into a single effective spe-
cies. Vapor-phase mass fractions for the effective fuel are needed at
the droplet surface Y
v,s
and in the surrounding gas outside the
droplet lm Y
v,g
. These mass fractions are dened in terms of the
individual multicomponent fuel species as
Y
v;s

K
k1
Y
v;s;k
; Y
v;g

k
Y
v;g;k
; 12
where Y
v,s,k
and Y
v,g,k
are the mass fractions of the vapor species
mixture at the droplet surface and in the gas far from the droplet,
respectively. The summations consider only the fuel species in the
gas phase.
Using these equivalent fuel-vapor denitions, the net total
droplet vaporization rate (i.e., including all of the component spe-
cies) is evaluated as though the droplets were a single-component
fuel [9]. That is,
_ m
d
pdq
f
D
v;f
Sh

v
ln1 B
v
; 13
where q
f
and D
v,f
are the average mass density and vapor-phase dif-
fusion coefcient within the thin lm surrounding the droplet. A
modied Sherwood number
Sh

v
2
Sh
0
2
FB
v

14
accounts for variations in lm thickness as a result of the Stefan
ow associated with mass transfer at the droplet surface. The
function
FB
v
1 B
v

0:7
ln1 B
v

B
v
15
depends upon the overall Spalding mass-transfer number B
v
as
B
v

Y
v;s
Y
v;g
1 Y
v;s
: 16
In the limiting case of nonvolatile droplets
Sh
0
2 0:552Re
1=2
d
Sc
1=3
; 17
where the Schmidt number is Sc = l
f
/(q
f
D
v,f
) and l
f
is the viscosity
of the lm mixture. The droplet Reynolds number is dened in
terms of the relative velocity as
Re
d
q
f
jv v
d
jd=l
f
; 18
where v and v
d
are the local gas-phase and droplet velocities,
respectively and d is the droplet diameter.
Droplet vaporization is assumed to be sufciently rapid such
that the fuel-vapor concentration at the droplet surfaces are satu-
rated. Thus, the vapor-phase mole fractions at the droplet surface
can be evaluated from the species vapor pressures as
X
v;s;k
X
d;k
p
v;k
p
; 19
where p
v,k
is the saturation vapor pressure of the kth pure species at
the droplet temperature of T
d
, X
d,k
is the mole fraction of kth species
of the liquid-phase mixture of the droplet, and p is the total gas
pressure. The vaporization rate of an individual species _ m
d;k
is eval-
uated using the mass fraction of the vaporizing species
k
as
_ m
d;k
_ m
d

k
, where the mass fraction
k
is evaluated based upon
the denition of the Spalding mass transfer number for individual
species B
k
as

k
Y
v;s;k

Y
v;s;k
Y
v;g;k
B
k
: 20
The Spalding mass transfer number for each individual species B
k
may be evaluated as a function of the mass-transfer number for
the effective single-component fuel B
v
as [21,22]
B
k
1 B
v

g
k
1; 21
where g
k
D
v;f
Sh

v
=D
k;f
Sh

k
, and Sh

k
is the modied Sherwood num-
ber for kth species. Because the relationships between B
k
and B
v
are
nonlinear, an iterative process is needed to evaluate B
k
.
Heat transferred from the gas into the droplet is evaluated as
_
q
d
_ m
d
c
p;v
T T
d

B
T
L
v
_ _
; 22
where B
T
is the Spalding heat-transfer number and T is the gas-
phase temperature outside the lm surrounding the droplet. The
Spalding heat- and mass-transfer numbers are related as
B
T
1 B
v

/
1; 23
where
/
c
p;v
c
p;f
Sh

v
Nu

1
Le
f
: 24
The effective latent heat L
v
of vaporization is represented as
L
v

k
L
v;k
T
d
; 25
where L
v,k
(T
d
) is the latent heat of vaporization of the kth species at
the droplet temperature. The average gas-phase specic heat within
the lm is represented as c
p,f
, and the heat capacity of the fuel-vapor
mixture within the lm c
p,v
is evaluated as
c
p;v

k
c
p;v;k
: 26
The modied Nusselt number Nu

is evaluated in a manner analo-

gous to Sh

v
. The lm Lewis number is dened as Le
f
= a
f
/D
v,f
, where
a
f
is the thermal diffusivity of the gas in the lm and D
v,f
is the mix-
ture-averaged diffusion coefcient of the fuel-vapor species in the
lm. Establishing consistent values of _ m
d
; B
v
and B
T
, and thus eval-
uating _ q
d
, is an iterative process [12].
2.4. Gas-phase source terms
The inuence of droplet vaporization on the gas-phase is repre-
sented in terms of mass, momentum, and energy sources S
M
, S
k
, S
V
and S
T
in Eqs. (1)(4). Sources for individual droplet are repre-
sented as
C. Wang et al. / Combustion and Flame 160 (2013) 265275 267
s
s
M
s
k
s
V
s
T
_
_
_
_
_
_
_
_
_
_

_ m
d
_ m
d

k
_ m
d
V
d
F
r
=r
d
_ m
d

k
h
d;k

_
q
d
_
_
_
_
_
_
_
_
_
_
_
_
_
_
; 27
The liquid-phase specic enthalpy of the droplet species is evalu-
ated as h
d,k
= h
g,k
L
v,k
, where h
g,k
is the specic enthalpy of the va-
por phase. This formulation assures self consistency between the
liquid and gas-phase thermodynamics and the latent heat. If the
thermodynamic properties for the liquid and gas phases are taken
from independent data bases, such self consistency is not inherently
assured.
The source terms S = (S
M
, S
k
, S
V
, S
T
)
T
for the control volume Dz
j
are evaluated as [9,14],
Sz
j

n
d
0
u
d
0
Dz
j

N
j
n1
_
tnDtn
tn
sF dt; 28
where N
j
is the number of times that a droplet enters control vol-
ume Dz
j
, and [t
n
, t
n
+ Dt
n
] is the nth time interval that a droplet
spends in the control volume Dz
j
. The variables n
d
0
and u
d
0
repre-
sent the droplet number density and velocity at the inlet. Although
this formulation enables the droplets to cross the stagnation plane
multiple times, multiple crossings do not occur in the present
results.
2.5. Boundary and initial conditions
The gas-phase problem is a boundary-value problem, with
boundary conditions being specied at the inlet manifolds (cf.
Fig. 1). At the left-hand manifold (i.e., z = 0),
u U
f
; V 0; T T
f
; Y
k
Y
k;f
; 29
where the subscript f represents the fuel manifold. For the studies
here, the gas-phase composition at the fuel inlet is dominantly N
2
,
and sometimes includes low levels of O
2
. Typically, the fuel enters
in the liquid phase, and thus does not directly contribute to the
gas-phase boundary condition. At the right-hand manifold, (i.e.,
z = L),
u U
o
; V 0; T T
o
; Y
k
Y
k;o
; 30
where the subscript o represents the oxidizer. For the studies
here, the gas-phase composition at the oxidizer inlet is air.
The droplets enter through the fuel manifold with the same ax-
ial velocity as the gas. However, the initial droplet temperature is
typically around T
d,0
= 300 K, which is substantially lower than
the surrounding gas phase. The initial droplet composition is spec-
ied in terms of the liquid-phase volume fractions. The inlet drop-
let diameter is specied, which leads directly to an initial droplet
mass. The initial droplet ux fraction is, by denition, F 1 [9].
3. Chemical kinetics model
3.1. Surrogate fuel composition
This study focuses on a three-component fuel mixture that
serves as a relatively simple surrogate for a FischerTropsch fuel.
Specically, the surrogate consists of n-heptane, n-dodecane, and
n-hexadecane in a 1:4:1 mole ratio. This combination approxi-
mates the distribution of alkane species found in some synthetic
aviation fuels [1]. Moreover, detailed validated chemical mecha-
nisms for these alkanes are available [10]. The thermophysical
properties of these alkanes are taken from NIST Standard Reference
Database in the NIST Chemistry WebBook. Figure 2 shows the tem-
perature-dependent vapor pressure and latent heat for the three
fuel components. The markedly different evaporation rates for
these components provide the opportunity to explore the impact
of these differences in volatility on the subsequent chemistry.
The choice of an all-alkane fuel is motivated by both the expected
prevalence of such species in alternative fuels as well as the avail-
ability of reliable alkane kinetic mechanisms.
3.2. Counterow ame conditions
For the studies reported herein, the separation distance be-
tween manifolds is either L = 5 mm or L = 10 mm, and both inlet
axial velocities are xed at U
in
= 1 m s
1
. This variation changes
both the strain rate and the residence time for droplet vaporization
prior to entering the high-temperature ame zone. The gas-phase
temperatures at both inlets is 950 K and the pressure is 10 atm.
These conditions approximate those expected in diesel engines at
the point where the fuel is injected. Three initial droplet diameters
(d
0
= 20 lm, d
0
= 25 lm and d
0
= 30 lm) are considered. In all
cases, the initial droplet temperature is T
d,0
= 300 K, and the initial
droplet velocity is the same as the gas-phase inlet velocity. The ini-
tial droplet number density must also be specied. With the drop-
let diameter specied, the inlet droplet number density is set such
that the overall equivalence ratio would be unity if the fuel and
oxidizer streams would be completely mixed. At the inlet, the fuel
is diluted by nitrogen. For initially 20 lm droplets, these condi-
tions lead to an inlet fuel-droplet loading density of
n
d
0
7:93 10
10
m
3
, with the nominal droplet spacing being
approximately 235 lm. If the fuel were fully vaporized at the inlet,
these conditions would lead to inlet gas-phase fuel mole fractions
of 0.192% for n-heptane and n-hexadecane and 0.767% for n-
hexadecane. Thus the fuel is highly diluted in the inlet carrier
gas. When the initial droplet diameter is varied, the initial loading
density is varied so as to maintain the same fuel mass loading and
the overall equivalence ratio of unity. Simulations are also per-
formed with the pre-vaporized fuel components in the nitrogen
carrier stream. In this case the mass owrate of nitrogen is approx-
imately 1% lower than in the droplet cases.
3.3. Mechanism reduction procedure
The alkane mechanism published by Westbrook et al. [10]
contains 2115 species and 8130 reactions-much too large to use
0
100
200
300
400
500
L
a
t
e
n
t

h
e
a
t

(
k
J

k
g
-
1
)
100 200 300 400 500 600 700
Temperature (K)
C
7
H
16
C
7
H
16
C
16
H
34
C
16
H
34
C
12
H
26
V
a
p
o
r

p
r
e
s
s
u
r
e

(
P
a
)
10
3
10
1
10
-1
10
-3
10
5
10
7
10
9
C
7
H
16
C
16
H
34
C
12
H
26
Fig. 2. Vapor pressure and latent heat as functions of temperature for the three
surrogate fuel components.
268 C. Wang et al. / Combustion and Flame 160 (2013) 265275
directly for the counterow ame calculations. In fact, this mech-
anism is so large it is impractical to use it even to generate the tar-
get parameters needed for the mechanism reduction process. Thus,
an alternative approach was used to develop a reduced reaction
mechanism.
First a smaller n-heptane mechanism, consisting of 654 species
and 2827 reactions [23], was considered. An additional simplica-
tion, based on the high volatility of n-heptane, was to assume that
the fuel was pre-vaporized. The counterow diffusion ame code
in CHEMKIN-PRO [24] was used to generate the species and tem-
perature proles for the counterow case, using the conditions
specied earlier with L = 5 mm. An interesting result was that, in
addition to the expected reactions near the ame-front, there
was substantial pyrolysis chemistry in the pre-ame zone, mean-
ing that the reduced mechanism must account for this pre-ame
chemistry for the three component fuel as well.
To focus on the kinetics, two simplied cases were identied
where the original large C
8
C
16
mechanism was used for plug-ow
calculations. This much simpler plug-ow analysis enables the use
of the original large mechanism with the 1:4:1 fuel ratio to gener-
ate targets for the subsequent mechanism reduction. The rst case,
chosen to account for the ame-front kinetics, assumed that both
input streams used in the counterow system were well mixed
at a xed temperature of 2000 K. The calculated n-heptane,
n-dodecane, n-hexadecane, O
2
, CO, CO
2
, H
2
and C
2
H
4
concentra-
tiontime proles over a 2 ms interval were selected as the targets.
(Most of the reaction occurred within the rst 0.1 ms.) The mech-
anism reduction was performed using the DRGEP method within
CHEMKIN-PRO. The absolute error tolerance was set as 10
3
and
the relative error tolerance was set as 5%. These criteria were sat-
ised with a reduced mechanism consisting of 117 species and 708
reactions.
The second case, chosen to describe the kinetics in the pre-
ame zone, was also a plug-ow calculation. Here it was important
to attempt to account for the variation in temperature within the
pre-ame zone. This temperature-distance prole was obtained
from a counterow diffusion ame calculation, using the multi-
component evaporation code described in this work, with an
approximate reduced multicomponent mechanism. This mecha-
nism was only used to generate an approximate temperature pro-
le for the subsequent plug ow calculations. The approach was
needed to account for the drop in temperature accompanying
evaporation and the subsequent temperature changes due to reac-
tion. This calculation predicted substantial pre-ame chemistry in
a narrow zone where there was a rapid rise in temperature (895
1560 K over 0.2 mm). This temperature-distance prole was then
imposed on the plug-ow calculation with the initial composition
corresponding to the fuel/nitrogen mixture. To generate a more ro-
bust mechanism, additional plug-ow calculations considered the
impact of air entrainment. For this case, the equivalence ratio
was set equal to 2 (10% O
2
). This combination of pyrolysis and par-
tial oxidation conditions forced the reduced mechanism to prop-
erly account for the kinetics for both these conditions. The same
target species were used as above. These cases required a more
complex reduction scheme to suitably reduce the mechanism size.
First the DRGEP method with absolute tolerance of 10
3
and a rel-
ative tolerance of 8% was employed to yield a mechanism with 310
species. Then a DRG-sensitivity method with the same tolerances
was then used to produce a 161 species and 722 reactions
mechanism.
The two reduced mechanisms were then merged to generate a
combined mechanism with 196 species and 907 reactions. Predic-
tions using this combined mechanism for a variety of plug-ow
conditions were compared to those obtained with the original large
mechanism to assess the accuracy of the reduction. Such compari-
sons in the high temperature ame-front regime were virtually
identical. Comparisons for the pre-ame pyrolysis conditions were
very similar, as shown in Fig. 3a. Figure 3b compares the effect of
adding 10% O
2
. The observed good agreement for both the pyrolysis
and partial oxidation conditions is encouraging. A more demanding
test of the accuracy of the reduced mechanism involved comparing
the predictions of the counterow ame using this reduced three-
component mechanism considering only heptane as a fuel to those
of the large heptane mechanismthat had served as the basis for the
mechanism reduction. Figure 4 illustrates the predicted results for
temperature and ethylene production. The good agreement be-
tween the temperature proles is particularly encouraging since
temperature was not selected as a target during the reduction pro-
cedure. The ethylene comparisons are also reasonable, especially
considering that the reduced mechanism was based on the C
8
C
16
rather than the C
7
mechanism. A direct comparison of the C
2
H
4
pro-
duction under pre-ame plug-ow conditions using these two
mechanisms showed similar differences, with approximately 10%
less C
2
H
4
produced using the complete C
8
C
16
than when using
the complete C
7
mechanism. These comparisons suggest the valid-
ity of this indirect reduction method; the 196 species reduced
mechanism should be adequate to capture the major features of
the kinetics in the 3-component counterow diffusion ames in
both the pre-ame and ame-front regions.
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
T
e
m
p
e
r
a
t
u
r
e

(
K
)
800
900
1000
1100
1200
1300
1400
1500
1600
Distance (mm)
0.1 0.12 0.14 0.16 0.18 0.2
0.0
0.5
1.0
1.5
2.0
2.5
3.0
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
H
2
H
2
C
2
H
4
C
2
H
4
n-C
12
H
26
Pyrolysis
Oxidation
Original
Reduced
(a)
(b)
Fig. 3. Comparison of the plug ow predictions of the complete and reduced
mechanisms in the pre-ame region. (a) 3-component fuel diluted in nitrogen, (b)
3-component fuel/10% O
2
diluted in nitrogen. Solid lines represent the complete
mechanism; dashed lines represent the reduced 3-component mechanism.
C
2
H
4

m
o
l
e

f
r
a
c
t
i
o
n

(
%
)
Distance (mm)
0
0.2
0.4
0.6
0.8
1.0
1.2
T
e
m
p
e
r
a
t
u
r
e

(
K
)
1000
1200
1400
1600
1800
2 2.2 2.4 2.6 2.8 3
Original
Reduced
Counter-
flow
flame
Fig. 4. Comparison of the counterow ame predictions for n-heptane as the fuel
with the complete heptane mechanism to that of the reduced 3-component
mechanism. Solid lines represent the complete heptane mechanism; dashed lines
represent the reduced 3-component mechanism.
C. Wang et al. / Combustion and Flame 160 (2013) 265275 269
4. Results and discussion
4.1. Pre-vaporized fuel results
An initial set of calculations for the counterow ame condi-
tions with the separation distance L = 5 mm was performed with
the three-component fuel pre-vaporized in the hot nitrogen
stream. These simulations served as a basis to compare subsequent
calculations where the fuel is introduced as liquid droplets. To
facilitate comparison with the droplet results where evaporative
cooling occurs, two calculations with inlet temperatures of 850 K
and 950 K served to bracket the temperatures encountered in the
droplet cases. Figure 5 shows predictions near the ame-front
region. The computed temperature proles and vapor phase resi-
dence times (calculated from the axial velocity prole) are shown
in Fig. 5a. The peak temperature of 1690 K for the 850 K inlet tem-
perature is increased to 1769 K for the 950 K case and the ame
front shifts approximately 1 mm toward the fuel side for the higher
temperature. These relatively low peak temperatures are the result
of the nitrogen dilution of the fuel stream. The ames are quite
narrow, approximately 0.3 mm wide at the temperature midpoint,
due to the relatively high pressure of 10 atm. Figure 5b shows that
signicant consumption of all three fuel components begins at
approximately the location where the temperature has increased
to approximately 1000 K. Perhaps the most interesting feature of
these calculations is the signicant extent of reaction in the
pyrolysis zone, as illustrated in Fig. 5c. Ethylene is produced very
early in the nearly uniform inlet temperature region. The ethylene
production is due to the successive b-scission reactions of the alkyl
radicals formed by radical abstraction reactions of the parent fuel
components. There is sufcient reactivity in the pre-ame region
to also produce acetylene and 1,3-butadiene. Vinyl radicals, formed
by radical abstraction from ethylene, can either undergo b-scission
to form acetylene or react with ethylene to form butadiene (C
4
H
6
).
The temperature is increasing rapidly as these species are formed.
Note that even at this point, the O
2
mole fraction is negligible. The
signicant differences in the extent of pre-ame chemistry for the
two cases reveals the importance of the inlet temperatures in driv-
ing the pyrolysis kinetics. One might expect more pre-ame chem-
istry for the pre-vaporized cases than for the droplet cases since
the pre-vaporized fuel components will have spent more time in
the hot nitrogen gas. However, as discussed below, the quantitative
impact of the vaporization process is more complex and more
interesting.
4.2. 20 lm Droplets
Calculations were performed with 20 lm diameter droplets to
explore the impact of a nite vaporization rate. To attain the same
overall equivalence ratio of unity used for the pre-vaporized case,
Distance (mm)
2 2.2 2.4 2.6 2.8 3
0.0
0.2
0.4
0.6
0.8
1.0
1.2
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
C
2
H
4
C
2
H
2
C
4
H
6
Pyrolysis
components
T
in
= 950 K
T
in
= 850 K
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0
5
10
15
20 O
2

m
o
l
e

f
r
a
c
t
i
o
n

(
%
)
n-C
7
H
16
n-C
12
H
26
n-C
16
H
34
O
2
Fuel components
T
e
m
p
e
r
a
t
u
r
e

(
K
)
R
e
s
i
d
e
n
c
e

t
i
m
e

(
m
s
)
1000
1200
1400
1600
1800
0
2
4
6
8
10
Gas-phase flame
(a)
(b)
(c)
Fig. 5. Comparison of the effect on inlet fuel temperature on the counterow
diffusion ame proles for the case where the fuel is all initially in the vapor phase.
The separation distance and computational domain is 5 mm, but only the ame-
zone region is plotted.
0.00
0.05
0.10
0.15
0.20
0.25
(d)
n-C
16
H
34
Droplets
0 0.5 1.0 1.5 2.0 2.5 3.0
Distance (mm)
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
Gas
950K
850 K
(c)
n-C
12
H
26
Droplets
0.00
0.20
0.40
0.60
0.80
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
Gas
950K
850 K
(b)
n-C
7
H
16
Droplets Gas
950K
0.00
0.05
0.10
0.15
0.20
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
850 K
Two-phase (droplets)
Gas-phase alone, 950 K
Gas-phase alone, 850 K
N
2
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(a)
75
80
85
90
95
100
Fig. 6. Comparisons of the counterow diffusion ame proles of vapor mole
fractions of the fuel components for 20 lm droplets (solid lines) with the pre-
vaporized cases (dashed lines).
270 C. Wang et al. / Combustion and Flame 160 (2013) 265275
the initial 3-component droplet number density was set at
n
do
7:93 10
10
m
3
. These liquid droplets, initially at 300 K,
were dispersed in the 950 K nitrogen stream owing at 1 m s
1
,
the same initial conditions used for the high temperature pre-
vaporized case. Figure 6 compares the droplet predictions to those
for the pre-vaporized cases. The slight drop in nitrogen mole frac-
tion (Fig. 6a) between 0 and 2 mm for the droplet case reects the
increase in fuel vapor mole fraction as evaporation occurs. The sub-
sequent drop in nitrogen mole fraction is quite similar for both
cases, reecting the increasing number of moles produced during
the combustion process. Figure 6bd shows vapor-phase mole
fractions for the three fuel components as the liquid components
evaporate. The most volatile n-heptane (Fig. 6b) is fully vaporized
well before entering the ame-front region. Prior to entering the
ame-front region, it achieves a uniform mole fraction that is
slightly larger (3% relative change) than that for the pre-vaporized
cases. This local enrichment, due to the fact that the fuel is now
evaporating over a shorter time interval, is more evident with
the less volatile fuels. The n-dodecane (Fig. 6c) begins to evaporate
later, achieving a uniform mole fraction that is higher than that for
the pre-vaporized cases by about 13%. This trend continues for the
least volatile n-hexadecane (Fig. 6d) where the local enrichment is
approximately 20%. The uniformvapor phase prole is signicantly
shorter than the more volatile components. For all three compo-
nents, the fuel decay proles fall between the those of the two
pre-vaporized cases, consistent with the temperature-distance
proles shown in Fig. 7a.
Figure 7 shows the differences in the extent of pre-ame pyro-
lysis kinetics between the pre-vaporized and the 20 lm droplet
cases. The peak temperature for the droplet case is 1738 K
(Fig. 7a), between that for the pre-vaporized cases (1690
and1769 K). Interestingly, the peak temperature is higher than
the 850 K pre-vaporized case, even though the temperature after
the fuel has evaporated is actually slightly lower (835 K). Although
the overall vapor phase residence time for the droplet case is
slightly longer than the pre-vaporized cases, the time the fuel com-
ponents spend in the vapor phase is shorter. The n-C
7
vapor phase
mole fraction approaches that for the pre-vaporized case at
approximately 1 mm, meaning that it has approximately 1 ms less
time in the heated gas. The time difference is approximately 2 ms
for n-C
12
H
26
and approximately 3 ms for n-C
16
H
34
. Nevertheless,
the amount of pyrolysis is comparable to that predicted in the
two pre-vaporized cases, as shown in Fig. 7bd.
4.3. Effect of initial droplet diameter
A series of calculations were performed wherein the droplet
diameter was changed (with the droplet number density modied
such that the amount of fuel was kept constant). Figure 8a com-
pares the predicted temperatures and droplet-diameter proles
2.0 2.1 2.2 2.3 2.4 2.5
Distance (mm)
0.00
0.01
0.02
0.03
0.04
0.05
0.06
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(d) C
4
H
6
Gas
950 K
Gas
850 K
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(c) C
2
H
2
Gas
950 K
Gas
850 K
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
(b) C
2
H
4
Gas
950 K
Gas
850 K
R
e
s
i
d
e
n
c
e

t
i
m
e

(
m
s
)
T
e
m
p
e
r
a
t
u
r
e

(
K
)
800
1000
1200
1400
1600
1800
0
1
2
3
4
5
6
7
8
(a)
Two-phase (droplets)
Gas-phase alone, 950 K
Gas-phase alone, 850 K
Fig. 7. Comparisons of the counterow diffusion ame proles of temperature,
vapor phase residence time, and mole factions of selected pyrolysis products for
20 lm droplets (solid lines) to the pre-vaporized cases (dashed lines).
2.0 2.1 2.2 2.3 2.4 2.5
Distance (mm)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
M
o
l
e

f
r
a
c
t
i
o
n
800
1000
1200
1400
1600
1800
2000
G
a
s

T
e
m
p
.

(
K
)
n-C
12
H
26
n-C
16
H
34
n-C
7
H
16
(d) 30 m
0.0
0.2
0.4
0.6
0.8
1.0
1.2
M
o
l
e

f
r
a
c
t
i
o
n
800
1000
1200
1400
1600
1800
2000
G
a
s

T
e
m
p
.

(
K
)
n-C
12
H
26
n-C
16
H
34 n-C
7
H
16
(c) 25 m
0.0
0.2
0.4
0.6
0.8
1.0
1.2
M
o
l
e

f
r
a
c
t
i
o
n
800
1000
1200
1400
1600
1800
2000
G
a
s

T
e
m
p
.

(
K
)
n-C
12
H
26
n-C
16
H
34
n-C
7
H
16
(b) 20 m
0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2
G
a
s

t
e
m
p
e
r
a
t
u
r
e

(
K
)
Distance (mm)
0
5
10
15
20
25
30
D
r
o
p
l
e
t

d
i
a
m
e
t
e
r

(

m
)

800
1000
1200
1400
1600
1800
2000
2200
30 m
25
20
(a)
Fig. 8. Predictions of the impact of changing initial fuel droplet size on the
counterow diffusion ame proles for liquid droplet size, temperatures and vapor
phase mole fractions of the fuel components. Note distance scale in (a) is different
from the other panels.
C. Wang et al. / Combustion and Flame 160 (2013) 265275 271
for 20 lm, 25 lm and 30 lm droplets. It is interesting to note that
during initial droplet vaporization, the droplet diameter actually
increases slightly. This is the result of decreasing liquid-phase
density as the droplet heats, which competes favorably with mass
loss owing to little vaporization during the initial stages of
vaporization. As the initial droplet size increases, the point at
which the droplets are completely vaporized moves well into the
high-temperature region. The peak ame temperature also in-
creases dramatically, from 1738 K for the 20 lm case to 2102 K
for 30 lm droplets, accompanied by a shift away from the fuel-in-
let side. Figure 8bd shows the vapor mole fractions for the fuel
components near the ame-front. For the 20 lm case, all the fuel
is evaporated and nearly uniform vapor mole fractions are estab-
lished before any signicant ame-inuenced temperature in-
creases. For the 25 lm case, only the most volatile n-C
7
reaches a
uniform pre-ame prole; the vapor mole fractions n-C
12
and
n-C
16
are increasing and reaching levels well above those predicted
for the 20 lm case. Again, these higher local vapor phase mole
fractions can be attributed to the increasing vaporization rate as
these droplets enter the high-temperature ame region. This trend
continues for the 30 lm case; here the vapor mole fraction
enhancement is even more pronounced. This increase in the fuel
vapor mole fraction near the ame-front with the larger droplets
results in a greater rate of heat release with the consequent in-
crease in peak temperature. Another potential explanation for
the increased temperature in the ame front region is that perhaps
the delayed evaporation results in a higher O
2
mole fraction in the
region where these droplets evaporate. However, the O
2
proles
for the 20 and 25 lm droplets fall in between those for the two
pre-vaporized cases, while that for the 30 lm droplets are shifted
about 1 mm downstream. The result is that, even for the largest
droplets, the fuel is evaporating into a pyrolysis region and then
swept into the oxidation zone.
Figure 9 shows that the increase in fuel vapor mole fraction also
leads to signicant differences in the amount of pre-ame pyroly-
sis chemistry. Although the temperature is similar for the three
droplet diameters for distances less than approximately 2.5 mm,
the signicant progressive increase in the amount of ethylene
and hydrogen produced with increasing drop size is signicant.
This increase in C
2
H
4
begins near 2 mm, where the temperature
and residence time are similar for all three drop diameters. The
one difference is the increased local vapor phase mole fractions
in this region for the larger drops. Closer to the ame-front, the in-
creased vapor phase residence time in the higher temperature re-
gion for the largest droplets could also contribute to increased
reactivity. Not surprisingly, because of its higher diffusivity, the in-
creases in the hydrogen mole fraction with larger droplets extend
to larger distances.
4.4. Effect of separation distance
The previous results illustrate the substantial impact on both
the peak temperature and the extent of pre-ame pyrolysis chem-
istry if the evaporation rate is sufciently slow that fuel droplets
penetrate into the high temperature region. To explore this effect
further, this sequence of calculations was repeated for cases where
the separation distance is doubled to 10 mm, allowing more time
for the droplet to evaporate before entering the high temperature
region. Figure 10 compares predicted droplet-size proles for the
two separation distances. For both cases, the droplet diameter
initially increases as a result of droplet heating (liquid density
decreases) prior to the onset of evaporation in the hot nitrogen
stream. The effect of increased separation distance gets progres-
sively larger as the droplet size increases. In particular, note the
almost vertical drop in diameter for the 30 lm case at 5 mm
separation. This is due to droplet impingement into the hot ame
region and subsequent very rapid evaporation. In contrast, the
shapes of the proles for the various drop sizes at 10 mm separa-
tion are similar to each other, reecting the slower evaporation
1.5 2.0 2.5 3.0
Distance (mm)
0.0
0.5
1.0
1.5
2.0
2.5
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(d) C
2
H
4
20 m
25 m
30 m
0.0
0.5
1.0
1.5
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(c) H
2
20 m
25 m
30 m
T
e
m
p
e
r
a
t
u
r
e

(
K
)
800
1000
1200
1400
1600
1800
2000
2200
20 m
25 m
30 m
(b)
0
2
4
6
8
10
12
R
e
s
i
d
e
n
c
e

t
i
m
e

(
m
s
)
20 m
25 m
30 m
(a)
Fig. 9. Comparison of the effect of initial fuel droplet size on selected counterow
diffusion ame proles.
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Distance (mm)
D
r
o
p
l
e
t

d
i
a
.

(

m
)
0
5
10
15
20
25
30
35
(a) 5 mm
separation
(b) 10 mm
separation
30 m
25 m
20 m
D
r
o
p
l
e
t

d
i
a
.

(

m
)
0
5
10
15
20
25
30
35
30 m
25 m
20 m
Fig. 10. Comparison of the effect of changing the separation distance from 5 mm to
10 mm on the prole of liquid fuel diameter as the diameter varies from 20 lm to
30 lm.
272 C. Wang et al. / Combustion and Flame 160 (2013) 265275
of the larger droplets. For this case the droplets completely evapo-
rate prior to entering the hot region. Figure 11 shows analogous
droplet-temperature trends.
As a consequence of the larger separation distance with the
associated lower droplet temperatures at a given distance, the
evaporation rate is slower, resulting in less enhancement of the va-
por phase mole fractions of the three fuel components. This results
in vapor phase fuel mole fractions closer to those predicted for the
pre-vaporized cases, as seen in Figs. 1214. With the most volatile
component (n-C
7
), Fig. 12 shows that, for the 20 lm case, the vapor
phase mole fractions closely approach those for the pre-vaporized
cases at both separation distances. However, even for this volatile
component, the evaporation rate is sufciently slow for larger
droplet diameters to observe differences. For the shorter separa-
tion, signicant enhancement is observed, more so for the 30 lm
case. Such enhancements are much diminished at the larger sepa-
ration. For the mid-volatile n-C
12
, Fig. 13 shows a more pronounced
effect of both droplet size and separation distance, with substantial
deviations from the pre-vaporized cases for all droplet sizes at
5 mm separation. The deviations are smaller at 10 mm separation,
but still signicant. For the 5 mm case, neither the 25 lm or 30 lm
droplets have fully vaporized before reaching the ame-front. In
contrast, all of the components for the larger separation have
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Distance (mm)
300
350
400
450
500
550
600
650
700
D
r
o
p

T
e
m
p
.

(
K
)
300
350
400
450
500
550
600
650
700
D
r
o
p

T
e
m
p
.

(
K
)
30 m
25 m
20 m
30 m
25 m
20 m
(a) 5 mm separation
(b) 10 mm separation
Fig. 11. Comparison of the effect of changing the separation distance from 5 mm to
10 mm on the droplet temperature as the diameter varies from 20 lm to 30 lm.
C
1
2
H
2
6

m
o
l
e

f
r
a
c
.

(
%
)
(b) 10 mm separation
0.00
0.20
0.40
0.60
0.80
1.00
1.20
Distance (mm)
30 m
25 m
20 m
Gas
C
1
2
H
2
6

m
o
l
e

f
r
a
c
.

(
%
)
(a) 5 mm separation
30 m
25 m
20 m
Gas
0.20
0.40
0.60
0.80
1.00
1.20
0
0 1.0 2.0 3.0 4.0 5.0
0.5 1.0 1.5 2.0 2.5
Distance (mm)
0.00
950 K
850 K
Fig. 13. Comparison of the effect of changing the separation distance from 5 mm to
10 mm on vapor phase mole fraction of n-C
12
as the diameter varies from 20 lm to
30 lm.
0.05
0.10
0.15
0.20
0.25
0.30
0.35
C
1
6
H
3
4

m
o
l
e

f
r
a
c
.

(
%
)
Distance (mm)
0.00
(b) 10 mm separation
30 m
25 m
20 m
Gas
950 K
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0
0 1.0 2.0 3.0 4.0 5.0
0.5 1.0 1.5 2.0 2.5
Distance (mm)
0.00
C
1
6
H
3
4

m
o
l
e

f
r
a
c
.

(
%
)
(a) 5 mm separation
30 m
25 m
20 m
Gas
Fig. 14. Comparison of the effect of changing the separation distance from 5 mm to
10 mm on vapor phase mole fraction of n-C
16
as the diameter varies from 20 lm to
30 lm. Long dashed line shows prediction for pre-vaporized fuel entering at 950 K;
short dashed line for 850 K.
C
7
H
1
6

m
o
l
e

f
r
a
c
.

(
%
)
0.00
0.05
0.10
0.15
0.20
0.25
(b) 10 mm separation
30 m
25 m
20 m
Gas
0 1.0 2.0 3.0 4.0 5.0
Distance (mm)
Two-phase
Gas, 950 K
Gas, 850 K
C
7
H
1
6

m
o
l
e

f
r
a
c
.

(
%
)
0.00
0.05
0.10
0.15
0.20
0.25
0 0.5 1.0 1.5 2.0 2.5
Distance (mm)
(a) 5 mm separation
30 m
25 m
20 m
Gas
Fig. 12. Comparison of the effect of changing the separation distance from 5 mm to
10 mm on vapor phase mole fraction of n-C
7
as the diameter varies from 20 lm to
30 lm.
C. Wang et al. / Combustion and Flame 160 (2013) 265275 273
reached uniform pre-ame proles. As illustrated in Fig. 14, the
trend of increasing vapor mole fraction enhancement with increas-
ing droplet size continues for the least volatile component n-C
16
.
Figure 15 illustrates the impact of different evaporation rates on
the peak ame temperature, due to different separation distances.
The 10 mm separation distance leads to peak temperatures much
closer to those predicted for the pre-vaporized cases. The peak
temperatures were 1796 K, 1749 K, and 1724 K for the 30 lm,
25 lm, and 20 lm droplet sizes respectively. These can be
compared to peak temperatures of 1804 K and 1729 K for the
two pre-vaporized cases. Although the magnitude of the differ-
ences relative to the gas phase are much smaller than for the
5 mm separation, the trend is the same, with the largest droplets
having the highest peak temperature. The variation in the cooling
rate during evaporation, shown in the lower panels of Fig. 15, is
directly related to the evaporation rate. The increased surface-to-
volume ratio for the smaller droplets leads to increased evapora-
tion rates and faster cooling. In the lower panel for the 10 mm case,
note that even the 30 lm droplet has completely evaporated
800
820
840
860
880
900
920
940
960
T
e
m
p
e
r
a
t
r
u
e

(
K
)
(d)
30 m
25 m
20 m
Gas, 950 K
Distance (mm)
T
e
m
p
e
r
a
t
r
u
e

(
K
)
800
1000
1200
1400
1600
1800
Gas
(c) 10 mm separation
0
Distance (mm)
Two-phase
Gas, 950 K
Gas, 850 K
800
820
840
860
880
900
920
940
960
T
e
m
p
e
r
a
t
u
r
e

(
K
)
Distance (mm)
(b)
30 m
25 m
20 m
Gas, 950 K
T
e
m
p
e
r
a
t
u
r
e

(
K
)
800
1000
1200
1400
1600
1800
2000
2200
0 1.0 2.0 3.0 4.0
1.0 2.0 3.0 4.0 5.0 6.0
0 0.5 1.0 1.5 2.0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Distance (mm)
(a) 5 mm separation
30 m
25 m
20 m
Gas
Two-phase
Gas, 950 K
Gas, 850 K
5 mm separation 10 mm separation
Fig. 15. Comparison of the effect of changing the separation distance from 5 mm to 10 mm on the temperature prole. The lower panels expand the temperature range to
show the temperature drop during the evaporation process.
0.00
0.01
0.02
0.03
0.04
0.05
0.06
4.0 4.2 4.4 4.6 4.8 5.0
Distance (mm)
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(h) C
4
H
6
30 m
25 m
20 m
Gas, 950 K
0.0
0.1
0.2
0.3
0.4
0.5
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(g) C
2
H
2
30 m
25 m
20 m
Gas, 950 K
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(f) C
2
H
4
30 m
25 m
20 m
Gas, 950 K
850 K
6
8
10
12
14
R
e
s
i
d
e
n
c
e

t
i
m
e

(
m
s
)
(e) Residence time
30 m
Gas, 850 K
10 mm separation
Two-phase
Gas, 950 K
Gas, 850 K
0.00
0.02
0.04
0.06
0.08
0.10
0.12
2.0 2.1 2.2 2.3 2.4 2.5
Distance (mm)
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(d) C
4
H
6
30 m
25 m
20 m
Gas
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(c) C
2
H
2 30 m
25 m
20 Gas
0.0
0.5
1.0
1.5
2.0
2.5
M
o
l
e

f
r
a
c
t
i
o
n

(
%
)
(b) C
2
H
4
30 m
25 m
20 m
Gas
R
e
s
i
d
e
n
c
e

t
i
m
e

(
m
s
)
2
3
4
5
6
7
8
9
10
(a) Residence time
30 m
Gas
5 mm separation
Two-phase
Gas, 950 K
Gas, 850 K
Fig. 16. Comparison of the effect of changing the separation distance from 5 mm to 10 mm on the vapor phase residence time and selected pyrolysis products as the diameter
varies from 20 lm to 30 lm.
274 C. Wang et al. / Combustion and Flame 160 (2013) 265275
before entering the high-temperature region, whereas only the
20 lm droplet vaporized completely with 5 mm separation.
Figure 16 shows the impact of the variations in vapor-phase
mole fractions on the pre-ame pyrolysis chemistry at different
separation distances. Note that the residence times are signi-
cantly longer at the larger separation (Fig. 16 a and e). As expected,
the extent of reaction for the pre-vaporized cases is slightly larger
at the larger separation distance due to the increased time in the
vapor phase. In contrast, there is signicantly less pre-ame reac-
tion for the droplet cases at the larger separation, and the variation
of the extent of reaction with droplet size is much smaller than it is
for the smaller separation. The larger enhancements of local fuel
vapor mole fractions with the droplets for the smaller separation
cases more than offset the decreased residence time for this case.
The effect of varying droplet size is much less at the larger separa-
tion distance where even the heaviest component in the largest
droplet has time to vaporize before entering the ame-front. The
greatly enhanced vaporization rates that result if the droplets pen-
etrate into the high temperature region near the ame-front thus
may have profound effects upon the extent of both the pre-ame
and ame-front chemistry.
5. Summary and conclusions
An opposed-ow diffusion ame model was used to explore the
impact of the evaporation of a multicomponent fuel on the kinetics
in the pre-ame and ame-front regions. The fuel consisted of
three alkanes with much different volatilities, selected to cover
the range expected in a typical aviation fuel. The advantage of
using the alkanes is that validated kinetic mechanisms are avail-
able. A reduced kinetic mechanism containing 196 species was de-
rived from an initial mechanism containing 2115 species. Results
were obtained comparing pre-vaporized fuel streams to those con-
taining monodispersed droplet sizes of 20, 25 and 30 lm carried in
nitrogen at separation distances of 5 and 10 mm with both inlet
velocities at 1 m s
1
. In all cases the overall equivalence ratio
was set to be unity and the pressure was 10 atm.
It was observed that nite evaporation rates signicantly im-
pacted the kinetics. In particular, if the combination of separation
length, stream velocity, and fuel volatility was such that the fuel
droplets entered the higher temperature region near the ame-
front, the rapid increase in evaporation rate in that region signi-
cantly enhanced the local vapor phase fuel mole fraction. This in
turn increased the reaction rates, leading to higher peak tempera-
tures as well as increased pyrolysis in the pre-ame region. For
example, the peak temperature predicted for the 30 lm droplet
was 330 K higher than a pre-vaporized case at the same inlet tem-
perature. This increase occurred in spite of an initial decrease in
temperature as the droplet evaporates. A similar effect is observed
for the pre-ame pyrolysis products; ethylene, acetylene, and
butadiene all increased about a factor of two for the 30 lm droplet
case.
These variations in temperatures and species concentrations
resulting from droplet evaporation could have signicant implica-
tions for various combustion applications. The importance of the
local fuel mole fraction in terms of dramatically changing the reac-
tion rate requires careful attention to details of the fuel evapora-
tion rate. Even when combustion kinetics is a primary focus, it is
clearly important to consider fuel volatility characteristics when
designing surrogate fuel blends [25].
Acknowledgment
This effort was supported by the Ofce of Naval Research via
Grant N00014-08-1-0539.
References
[1] B.L. Smith, T. Bruno, J. Propul. 24 (2008) 618623.
[2] J.A. Cooke, M. Bellucci, M.D. Smooke, A.A. Gomez, A. Violi, T. Faravelli, E. Ranzi,
Proc. Combust. Inst. 30 (2005) 439446.
[3] A. Holley, Y. Dong, F. Egolfopoulous, Combust. Flame 144 (2006) 448460.
[4] M. Arias-Zugasti, D.E. Rosner, Combust. Flame 135 (2003) 271284.
[5] S.S. Sazhin, A.E. Elwardany, E.M. Sazhina, M.R. Heikal, Int. J. Heat Mass Transfer
54 (2011) 43254332.
[6] W.A. Sirignano, G. Wu, Int. J. Heat Mass Transfer 51 (2008) 47594774.
[7] Y. Ra, R.D. Reitz, Combust. Flame 155 (2008) 713738.
[8] Y. Ra, R.D. Reitz, Combust. Flame 158 (2011) 6990.
[9] R.J. Kee, K. Yamashita, H. Zhu, A.M. Dean, Combust. Flame 158 (2011) 1129
1139.
[10] C.K. Westbrook, W.J. Pitz, O. Herbinet, H.J. Curran, E.J. Silke, Combust. Flame
156 (2009) 181199.
[11] H. Zhu, R.J. Kee, L. Chen, J. Cao, M. Xu, Y. Zhang, Combust. Theory Model 16
(2012) 715735.
[12] B. Abramzon, W.A. Sirignano, Int. J. Heat Mass Transfer 32 (1989) 16051618.
[13] E. Gutheil, W.A. Sirignano, Combust. Flame 113 (1998) 92105.
[14] A.M. Lentati, H.K. Chelliah, Combust. Flame 115 (1998) 158179.
[15] W.A. Sirignano, Fluid Dynamics and Transport of Droplets and Sprays, second
ed., Cambridge University Press, 2010.
[16] A.U. Modak, A. Abbud-Madrid, J. Delplanque, R.J. Kee, Combust. Flame 144
(2006) 103111.
[17] R.J. Kee, M.E. Coltrin, P. Glarborg, Chemically Reacting Flow: Theory and
Practice, Wiley, Hoboken, NJ, 2003.
[18] R.S. Barlow, A.N. Karpetis, J.H. Frank, J.Y. Chen, Combust. Flame 127 (2001)
21022118.
[19] G. Continillo, W.A. Sirignano, Combust. Flame 81 (1990) 325340.
[20] S.S. Sazhin, Prog. Energy Combust. Sci. 32 (2006) 162214.
[21] H. Watanabe, Y. Matsushita, H. Aoki, T. Miura, Combust. Flame 157 (2010)
839852.
[22] E. Rivard, D. Brggemann, Chem. Eng. Sci. 65 (2010) 51375145.
[23] M. Mehl, W.J. Pitz, C.K. Westbrook, H.J. Curran, Proc. Combust. Inst. 33 (2011)
193200.
[24] CHEMKIN-PRO Release 15101. Reaction Design, Inc., San Diego, CA, 2010.
[25] C.J. Mueller, W.J. Cannella, T.J. Bruno, B. Bunting, H.D. Dettman, J.A. Franz, M.L.
Huber, M. Natarajan, W.J. Pitz, M.A. Ratcliff, K. Wright, Energy Fuels 26 (2012)
32843303.
C. Wang et al. / Combustion and Flame 160 (2013) 265275 275