!

CRANNE 05
Proceedings of the !nternational Conference on Recent Advances in Nechanical 8 Naterials Engineering
30-31 Nay 2005, Kuala Lumpur, Nalaysia
Paper No.
APPLICATION OF THE HYBRID METHOD TO NUMERICAL SIMULATION OF DROPLET
EVAPORATION AND COMBUSTION

M. D. Emami* and A. Kamranian


Mechanical Engineering Department
Isfahan University of Technology
Isfahan, Iran

* corresponding author, Email: mohsen@cc.iut.ac.ir


ABSTRACT
This paper presents results of the numerical simulation
of liquid fuels spray evaporation and combustion in
furnaces. The DDM model is used to model the two-
phase liquid-gas flow. The governing equations are
written in Lagrangian and Eulerian coordinates for liquid
and gas-phase, respectively. To couple the governing
equations, a series of source terms due to mass,
momentum and energy transfer of liquid-phase to gas-
phase are added to the conservation equations of each
phase (Two way coupling). The results of the
evaporation case show the droplet heat-up region, and
the droplet evaporation region. The results of
combustion simulation for a fired-heater, also show
general trend of reacting flow in this type of furnaces,
and the lagging in the development of temperature field
due to evaporation process.

Keywords: diffusion flames, spray, combustion

INTRODUCTION
Spray combustion is used extensively in industrial
boilers, ramjet combustors, liquid propellant rocket
engines and aircraft propulsion systems. The
aerodynamic flow pattern, the flame properties and heat
transfer are important characteristics in such
applications. Previous work on spray evaporation and
combustion modeling was reviewed comprehensively
by Law [1] , Sirignano [2] and Faeth [3-5]. In modeling
dilute spray combustion, models can be classified into
three categories according to the treatment of the inter-
phase transport rates. The first is the locally
homogeneous flow model (LHF), in which the inter-
phase transport rates are assumed to be infinitely fast.
The droplet-phase and gas- phase are assumed to be
in kinetic and thermodynamic equilibrium at each point
in the flow. Therefore, there is no velocity slip or
temperature difference between the droplet- phase and
the gas-phase. According to Shearer et al. [6] and Khalil
and Whitelaw[7], this model can be identified as a
single fluid model and can only be used to simulate
small droplet spray combustion. The second model is
known as the Deterministic Separated Flow model
(DSF), in which the droplet temperature and velocity are
solved along the droplet trajectory, independent of the
gas-phase turbulence. The droplet trajectory, in this
model, is a time- averaged trajectory and the effect of
spray droplet-phase on the gas-phase turbulence is
ignored. The third model is the Stochastic Separated
Flow model (SSF), in which both the finite inter-phase
transport rates and the effect of turbulence interactions
between droplet-phase and gas- phase are considered.
This model has been commonly used in various spray
combustion modeling by many researchers, among
them Crowe et al. [8], Soloman et al. [9], Coimbra et al.
[10],. Both the DSF model and the SSF model adopt
Eulerian gas and Lagrangian droplet formulation. The
use of Lagrangian treatment of droplets facilitates the
calculation of the evaporation history and temperature
change of a single droplet. However, in order to obtain
a detailed distribution of droplet velocity and
concentration for comparison with experimental data, a
large amount of droplet trajectories are required.
In this paper, a Hybrid method for simulating dilute
spray combustion is developed, using an Eulerian
treatment for gas-phase and a Lagrangian treatment for
the droplet-phase.
Simulation of spray combustion involves modeling a
number of complex simultaneous inter-dependent
processes. A comprehensive model must account for
turbulent gas-droplet flow, gas-phase combustion,
droplet mass change due to evaporation and radiation
heat transfer. In the present study, a multi-mixture
fraction concept [11] is used to model for the gas phase
combustion. The non-equilibrium radiation model of
Gibbs [12] is used to simulate the radiation exchange in
the gas environment. Turbulence is modeled by the
standard two equations model [13].
The exchange of mass, momentum and energy
between the particles and the gas-phase is accounted
for, via appropriate source terms in the governing
equations.

GAS PHASE MODEL
The canonical form of the governing equations
describing the steady state, subsonic flow of a gaseous
mixture in the absence of body forces, may be written
as:
( )

S =
~
u (1)
where is the considered quantity for which the
transport equation has been derived, is the density of
the gaseous mixture, u
~
is the mixture velocity,

is
the effective diffusion coefficient, and

S is the source
term, which may contain terms due to the second (drop)
phase. The above formulation assumes turbulent flow
for the gaseous mixture. It is also assumed that the
effect of turbulence could be taken into account by a
turbulent diffusion coefficient, which may be written as:

t
t
=
,

(2)
where

2
k
C
t
= (3)
is the turbulent viscosity and

is the turbulent
Prandtl/Schmidt number.
The effective diffusion coefficient in eq.(1) may be
written as:
, , t l
+ =
(4)
where
, l
is the laminar diffusion coefficient of .
The set of governing equations for the gas phase
comprises the continuity equation, the momentum
equations, the energy(enthalpy) equation, the k
equations, the mixture fraction equation, and the
variance of mixture fraction. Table(1) shows the
relevant quantities associated with eq.(1) for the
aforementioned governing equations.

Table 1: The diffusion coefficients and the source terms
for the transport equations

SPRAY DROPLET PHASE MODEL
In the following, a Lagrangian reference frame is
adopted, following individual drops as a function of time
as they pass through the flow field. The drop position is
then given by [14]:

+ =
t
p p p
dt u x x
0
0

(7)
where
0 p
x

is the initial location of the drop. The

governing equations of droplet-phase can be written in
the following forms:

Conservation of mass
p
p
r
dt
dm
= (8)
where,
p
m is the droplet mass and
p
r is the rate of mass
transfer to the gaseous phase due to evaporation
/surface reaction.

Conservation of momentum in the i
th
direction :
i
i
x
p
Dx
p
ip p ip
g
m
F
m
u
dt
dm
dt
du
+ =
(9)
where
( )
ig
u
ip
u
ig
u
ip
u
D
C
p
d
g D
F

=
2
8

is the drag force

acting on the droplet,
i
g is the gravity force in i
direction,
D
C is the drag coefficient,
p
d is the drop
diameter and subscripts g and p refer to gas and
droplet phase, respectively.

It should be noted that for the droplet momentum
equation, only the drag force is considered as the
dominant source term.

Conservation of energy:
( ) ( )
mass evp comb conv rad ref p p p
Q Q Q Q Q T T C m
dt
d
+ =
. . . .


(10)
where, is the percentage of heat due to surface
combustion that goes into the particle and ( )

1 is the
percentage of heat released into the gaseous phase. As
the present study is concerned with the evaporation of
light hydrocarbon fuels, surface combustion is not
considered, and =0 is used for the current
computations. Up on substitution of appropriate
expressions for the terms on the right hand side of the
above equation, the following equation for the droplet
temperature is derived [14]:
( ) ( )
( )
p p
evp p
p p
c p
g p
p p
p g g
g p
p p
p p p
C m
h r
C m
H r
T T
C m
d k Nu
T T
C m
d
dt
dT
. 4 4
2
1 .

+ =

(11)
where
evp
h
c
H
g
k
g
Nu p C
p
, , , , , , are the droplet emissivity,
the Stephan-Boltzman constant, droplet mean heat
capacity, the Nusselt number, the gas conduction
coefficient, heat of combustion and the latent heat of
droplet phase, respectively. Parameter is a binary
switch, which is set equal to one in the case of droplet
evaporation, and is set to zero for droplet surface
combustion.

The rate of mass transfer to the gas-phase,
p
r , may be
related to the mass transfer coefficient between the
droplet phase and the gaseous environment as:
( )

=
, . ,
*
i surf i mass p
A h r (12)
where
. , surf i
=
m v
i sat
T R
P
,
,
, i
=
m v
i
T R
P
,
,
( )
2
. surf
m
T T
T
+
=

and
*
mass
h is
a corrected mass transfer coefficient to account for
higher evaporation rates[14]. The corrected Sherwood
number,
*
Sh , and its associated mass transfer
coefficient,
*
mass
h , may be written as:
Equation

S
Continuity 1 0 Smp
Momentum u
i

eff
up
S +

i i
j
eff
j
x
p
x
u
x


Turb.kinetic energy k
k

*
G
Turb. dissip.

( )
k
C G C

2 1


Mixture frac. f
f
Sf
**
Mix. Frac. var. g
g

k
g C
x
f
x
f
g
m m f t
t

,
2

Enthalpy h
h
hp
S
**
* ij ij t
S S G 2 =
, and

=
i
j
j
i
ij
x
u
x
u
S
2
1

** S
f
is the mixture fraction source term due to second
phase
*** S
hp
is the enthalpy source term due to second phase
3 . 1 , 1 , 92 . 1 , 44 . 1 , 09 . 0
2 1
= = = = =


k
C C C

= =

P
P P
P P
P P
h
h
Sh
Sh
i i sat
i sat
i
mass
mass
, ,
,
,
* *
ln
(13)
where P refers to static pressure and subscripts
, sat stand for the saturation and ambient conditions,
respectively. Solutions of eqs. (7),(8),(9) and (11) by the
RK4 solution algorithm specifies the trajectory of the
particle, and its velocity along the points of this
trajectory [14]. The particle diameter may be calculated
from the following formula:
3
6
p p
d m

= (14)

NUMERICAL PROCEDURE
The governing equations for the gas phase in Eulerian
coordinates are integrated over the computational cell,
using the FV approach on collocated grid. In the gas-
phase the two-equation turbulence model ( k ) and a
multi-mixture fraction model[11] is used for turbulence
and combustion, respectively. The coupled Lagrangian
equations are solved by RK4 algorithm. The
convergence criterion for the gas-phase is set such that
maximum norm L
2
of the normalized residuals of
f k p w v u , , , , , , and g is less than 0.001.

RESULTS AND DISCUSSION
Two cases are considered in the present study. The
first case is concerned with the evaporation of drops of
equal sizes in a stagnant fluid. The second case is the
simulation of evaporation and combustion of a liquid
fuel spray in an industrial furnace.

Evaporation case
This case considers evaporation of liquid fuel drops in a
stagnant fluid. The input data for the computer
simulation is shown in table (2). The computational
domain is a 2D-rectangular region. Liquid fuel is
injected into this region at the centroid of the domain.
The gas phase temperature and pressure are 673 K
and 1 bar, respectively[15]. This case has been studied
experimentally by Blardini et al[16 ]. The present study
compares the predictions of the current computational
model with the experimental results of [16] and the
numerical predictions based on the Spalding model[17],
the finite conductivity model[15], and the numerical
model of Blardini et al[16].

Table 2 : Simulation data for the first case(evaporation)
Dimensions (mm) 20*40
Injection velocity(m/s) 6
Orifice dia.(mm) 0.28
Fuel Tetra-decane
Fuel Temp.(K) 300

Figure (1) shows the variation with time of D
2
/D
0
2
, the
squared of the ratio of the drop diameter to its initial
diameter. The experimental data and other numerical
predictions from reference [16] are also shown on this
figure for comparison. As shown on fig.(1), the ratio is
increased initially during a small time interval, where the
drop absorbs heat from surroundings and its diameter
is increased due to volume expansion. Evaporation
starts near the end of this time interval, resulting in a
decrease in the diameter ratio. Both the experimental
and numerical results follow the well-known d
2
-law
beyond this time interval. The present prediction shows
better agreement with experimental data than the other
numerical results extracted from reference [16]. The
present results show the increase in the drop diameter
due to initial heating of the drop. The evaporation time
for the drop is 36 msec according to present results,
which is in close agreement with the extrapolated value
based on the experimental data. Overall, the present
simulation is superior to those of reference [16]
Fig. 1: Variation of D
2
/D
0
2
with time (evaporation case).
Time (sec)
D
r
o
p
l
e
t
T
e
m
p
.
(
K
)
0 0.01 0.02 0.03
300
350
400
450
500
T= 526 K= BoilingPoint

Fig. 2: Variation of drop temperature with time
(evaporation case).

Figure (2) shows the variation of drop temperature with
time according to present simulation. In the heating
period, the drop temperature increases with time due to
the absorption of heat from surroundings. As the drop
temperature reaches the fuel boiling point, evaporation
starts and the drop temperature remains constant
during the evaporation period.
Figure (3) presents the evaporation ratio as a function
of time. This ratio is defined as the ratio of the mass of
drop evaporated up to the current time, to the initial
mass of the drop. It is evident from this figure that a
negligible amount of fuel is evaporated in the heat up
Time (sec)
0 0.01 0.02 0.03 0.04
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
Curreent Model
Blardini et al.Model
Finite Cond. Model
Spalding Model
Experiment
D
2
/D
0
2

period, but the evaporation rate grows exponentially
afterwards.

Evaporation and combustion case
The second case in this study considers the
evaporation and combustion of liquid fuel spray in an
industrial furnace. The furnace geometry is shown in
Fig.(4), where only 3 burners out of 32 burners of this
furnace are considered to assess the evaporation
/combustion model.
Time (sec)
X
V
0 0.01 0.02 0.03
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1

Fig. 3: Variation of the evaporation ratio with time
(evaporation case).

Fig. 4: Furnace geometry (second case).

As the combustor is symmetric with respect to the
vertical plane, only half of the geometry is simulated.
Table (3) shows the flow conditions for this simulation.

Table 3 : Simulation data for the second case
Injection velocity(m/s) 8.2
Fuel Tetra-decane
Fuel Temp.(K) 300
Air Temp.(K) 300
Air swirl Number 0.5
Air velocity (m/s) 14.5

Due to space limitations, only the temperature contours
at different cross sections of the furnace are presented
here. Figure (5-a,b,c) shows temperature contours at 3
cross sections in the y-z plane. These are one section
adjacent to the burner planes (x/h=0.03), and two
sections further downstream at x/h=0.1 and x/h=0.5.
Fig. 5(a,b,c): Temperature contours at 3 cross sections
in the y-z plane (second case).

As it is seen from Fig.(5), the temperature profiles are
not symmetric. This asymmetry is due to the existence
of swirl in the combustion air stream. The extension of
Fuel injection
Air

hot region toward the right part of the cross sections is
also justified based on this reason. It is inferred from
this figure that the three main vortices at the first cross
section plane interact with each other and form a single
large vortex further downstream.



Fig. 6(a,b,c): Temperature contours at 3 cross sections
in the x-z plane (second case).

Figure (6-a,b,c) shows temperature contours at 3 cross
sections in the x-z plane. These are one section
between wall and the first burner planes (y/w=0.085), a
section at the first burner plane (y/w=0.223), and a
section between the first and the second burner planes
(y/w=0.36).

Figure (6) shows that the high temperature region is
formed near the upper part of the furnace (x>3). The
main reason for this pattern is the delay time which is
associated with evaporation of the liquid fuel. As the
evaporation temperature is about 526 K, the drops are
not evaporated instantaneously as they are injected into
the furnace. In fact, both the heat up period and the
evaporation period present in the process. After
evaporation of fuel droplets, the fuel vapor mixes with
the combustion air, and a diffusion flame is formed
where the mixture fraction is close to the stoichiometric
mixture fraction of tetra-decane. The maximum
predicted temperature is about 2600 K, which is high for
this type of fuel. This high value is attributed to the
assumption of a single step reaction, which is an
inherent assumption of the flame sheet model of
combustion.

CONCLUSIONS
The current computer code with the flame sheet model of
combustion and a particle evaporation and combustion
model shows good capability in simulating two phase
reacting flows. The predicted evaporation characteristics
for the first studied case represent close agreement with
the experimental data.

NOMENCLATURE
CD Drag coefficient
Cp specific heat, J/kgC
C Turb. Viscosity coeff.
d Diameter, m
D Diameter, m
f Mixture fraction
g Mixture fraction variance
hevp Evaporation enthalpy, j/kg
Hc Heat of combustion, j/kg
k Turb. Kinetic energy, (m/s)
2
p Pressure, pa
Q Energy (heat), j
T Temperature, K
S General source term
Smp Mass source due to second phase, kg/m
3
Sup momentum source due to second phase, N/m
3
Sf Mixture fraction source due to second phase, kg/m
3

Shp energy source due to second phase, j/m
3

u x-direction velocity, m/s
v y-direction velocity, m/s
w z-direction velocity, m/s
x Spatial coordinate, m
y Spatial coordinate, m
z Spatial coordinate, m

Greek Letters
density, kg/m
3

Viscosity, pa-s
General variable
Diffusion coefficient, variable dependent
Prandtl/Schmidt number
Turbulence dissipation, m
2
/s
3

p
Particle/drop emissivity

0
Binary switch in eq.(10)

Stephan-Boltzman constant

Subscripts
t turbulent
p Particle/droplet
i Spatial direction
g Gas phase
D drag
m mass
ref reference
comb combustion
rad radiation
conv convection
evp evaporation
mass Mass transfer

REFERENCES
[1] Law, C. K., Int. J. Heat and Mass Transfer, Vol. 18,
pp. 1285-1292, 1975.
[2] Sirignano, W.A., "Fuel Droplet Vaporization and
Spray Combustion Theory", Prog. Energy Combust. Sci.,
Vol.9, pp.291-322, 1983
[3] Faeth, G. M., Evaporation and Combustion of
Sprays, Prog. Energy Combust. Sci., Vol. 9, pp. 1-79,
1983.
[4] Faeth, G. M., "Spray Combustion Models-A Review",
AIAA Paper No. 79-0293, 1979.

[5] Faeth, G. M.,Current Status of Droplet and Liquid
Combustion, Prog. Energy Combust. Sci., Vol. 3,pp.
191-224, 1977.
[6] Shearer, A. J., Tamura, H. and Faeth, G. M., J. of
Energy, Vol. 3, p. 271, 1979.

[7] Khalil, E. E., and Whitelaw, J. H., Aerodynamic and
Thermodynamic Characteristics of Kereson- Spray
Flame, Sixteenth Symposium (International) on
Combustion ,pp. 569-576, The Combustion Institute,
Pittsburgh, 1976.
[8] Crowe, C. T., Sharma, M. P. and Stock, D. E., J. of
Fluid Engr., Vol. 99, p. 325, 1977.
[9] Solomon, A. S. P., "A Theoretical and Experimental
Investigation of Turbulent Sprays", PhD thesis, The
Pennsylvania State University, 1984
[10] Coimbra, C.F.M., Coelho ,P.J, McQuay, Q.M.,
Carvalho, M.G. Proc Third Int. Conf Combust. Technol.
Clean Environ., Lisbon, Portugal 1995

[11] M.D. Emami S.Ziaei-Rad and H. Afshin, Application
of the Multi-mixture Fraction Model in Numerical
Simulation of Turbulent Reacting Flows, Proceedings of
the 10
th
Asian Congress of Fluid Mechanics (Int.) , 17-21
May 2004, Peradeniya, Sri Lanka

[12] Emami, M. D., Prediction of Finite Rate Chemistry
Turbulent Combustion, Phd Thesis, University of London,
1999.

[13] B.E. Launder, D.B. Spalding, The numerical
computation of turbulent flow, Comp. Meth. Appl. Mech.
Eng. 3 (1974) 269.

[14] Kamranian, A., "Numerical Simulation of
Evaporation and Combustion of Oil Fuel Sprays in
Industrial Furnaces", M.Sc. thesis, Isfahan University of
Tech., 2003
[15] Bertoli, C., and Migliaccio, M.na, A Finite
Conductivity Model For Diesel Spray Evaporation
Computations, Int. J. Heat and Fluid Flow, Vol. 20, pp.
552-561, 1999.
[16] Belardini, P., Bertoli, C., Lazzaro, M., and Massoli,
P., Single Droplet Evaporation Rate: Experimental and
Numerical Investigations, Proceeding of Second
International Conference on Fluid-Mechanics,
Combustion, Emission and Reliability in Reciprocating
Engin, Capri, Italy, pp. 265-270, 1992.
[17] Spalding, D. B., The Combustion of Liquid Fuels
Proceding of the Fourth International Symposium on
Combustion , The Combustion Institute, Pittsburgh, PA.,
1953.