Combustion

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

MCS 3 Marrakech 8-13 June 2003

Numerical Simulation of Combustion Phenomenon in a


Model Combustion Chamber

F. Bazdidi-Tehrani and M. Rajabi-Zargarabadi
E-mail: bazdid@iust.ac.ir
College of Mechanical Engineering, Iran University of Science and Technology,
Narmak, Tehran 16844, Iran

Abstract
The estimation of turbulent flame characteristics such as temperature and chemical species
are among the important aspects in a gas turbine combustion chamber. In the present work a
two-dimensional combustion simulation is developed for a model combustion chamber such
that the equations of continuity, momentum, energy and gas species mole fractions are
discretized and solved by a Finite Volume method. The k turbulence model and the
discrete transfer radiation model are employed to describe the turbulent flow field and the
radiative heat transfer, respectively. The chemical reaction rates are estimated by both the
eddy dissipation concept model and the Arrhenius rate. The present computational results of
temperature and
4
CH mole fraction distributions using the eddy dissipation concept model
agree well with the experimental results of Yaga et al. [8]. But the predicted results of CO
mole fraction distribution are quite different from the experimental data. The computed
4
CH
and CO mole fraction distributions using the Arrhenius rate are in fairly good agreement with
the available experimental data. However the predicted temperature distribution using the
Arrhenius rate appears to be inconsistent with the experimental data. This could be due to the
inconsideration of the
2
H formation in the two-step reaction scheme.

Introduction
The prediction of flame characteristics including temperature and chemical species are
essential for solving fundamental problems such as the formation of pollutants for instance in
a gas turbine combustion chamber. However, it is difficult to measure these characteristics of
combustion. Hence the need for a suitable combustion simulation model strongly exists. The
numerical simulation of combustion phenomenon is then a very convenient technique.
Furuhata et al. [1] employed the k turbulence model to simulate the turbulent flow
including combustion. However, this model had a limitation on estimating a nonisotropic field
such as the gas turbine combustion chamber [2]. Cook and Bushe [3] used the Large Eddy
Simulation model for estimating the non-premixed combustion. Patankar and Spalding [4]
have developed a three-dimensional combustion simulation in the furnaces using the k
turbulence model. Magnussen and Hjertager [5] developed a mathematical modeling known
as the Eddy Dissipation Concept model for the turbulent combustion. The combustion
characteristics of a pressurized combustor under fuel rich (non-stoichiometric) conditions
were estimated by Yamamoto et al. [6]. Makita et al. [7] employed the three-step and
improved four-step reduced reaction schemes for methane-air combustion. Yaga et al. [8]
have used the LES turbulence model to simulate the turbulent flow along with the eddy
dissipation concept model for predicting reaction rates in a model combustion chamber. The
objective of present paper is to investigate two reaction models, namely the EDC model and
Arrhenius rate, together with the two-step and three-step reaction mechanisms, evaluating
their compatibilities with the k turbulence model.
Combustor flow geometry
The combustor flow geometry is selected according to the experimental setup of Yaga et al.
[8]. The schematic of flow geometry is shown in Figure 1. The size of the combustor is
mm 200 in diameter and mm 800 in length. Methane is supplied through the inner nozzle of
mm 5 inner diameter and air is fed through the outer nozzle of inner diameter of mm 3 2 . In



Figure 1. Schematic of flow geometry [8].

their experiment, the methane and air feed rates (i.e., in stoichiometric ratio, 1 = ) are set as
1 3
h m 0.2

and
1 3
h m 9 . 1

, respectively. The flame temperature is measured by using suction
pyrometer probe equipped with an R-type thermocouple. The tip diameter of gas sampling
probe is mm 32 . 4 . Sampling gases are analyzed by a gas chromatograph [8].

Governing Equations
The general form of the governing equation for steady-state and axisymmetric conditions in
cylindrical coordinates is as follows:

(1)

S )
r
r (
r r
)
x
(
x
) v r (
r r
) u (
x
+

1 1


Where represents the dependent variables including the mass (1), momentum (u, v),
turbulence energy (k), dissipation rate of turbulence energy ) ( , enthalpy ) h ( and mass
fraction ) O H , CO , CO , H , O , N , CH i ; m (
2 2 2 2 2 4 i
= .

is the diffusion coefficient,

S is the
source term and is density. The expressions,

and

S are described in Table 1. The


energy equation contains two source terms related to the chemical reactions and radiation heat
transfer:


(2)

Radiation eaction R
j
eff ij j j j
i
eff
i
i
i
S S ) ( u J h
x
T
k
x
) h u (
x
+ +
|
|
.
|

\
|
+





Where
eff
k is the effective conductivity ) k k k ., e . i (
t eff
+ = and
j
J is diffusion flux of species
j. The first three terms on the right-hand side of equation (2) represent the energy transfer due
to conduction, species diffusion and viscous dissipation consecutively. The last two terms on
the right hand side are explained in the following sections.

S
Mass 1 0 0
Axial momentum u

eff

x
P
x
v
r
r r x
u
x
eff eff

|
.
|

\
|

+ |
.
|

\
|


1

Radial momentum v

eff

r
P
r
v
r r
v
r
r r r
u
x
eff
eff eff

|
.
|

\
|
|
.
|

\
|

+ |
.
|

\
|


2
1

Mass fraction
i
m

m
eff

i
R
Turbulent kinetic energy k

k
eff



k
G
Eddy dissipation rate

eff

) C G C (
k
k

2 1

Enthalpy h

h
eff

radiation eaction r
S S +

C
1
C
2
C
k


m

h

0.09 1.44 1.92 1.0 1.3 0.85 0.85

Table 1. Source terms and diffusion coefficients for

Reaction Modeling
In this study, the two-step and three-step global reaction mechanisms for methane
combustion are employed so as to estimate the formations of combustible species. The two-
step and three-step global reaction mechanisms are named as Calc. and Calc., respectively.
The two-step reduced reaction scheme for Calc. consists of the following two reactions:


(3)
O H CO O . CH
2 2 4
2 5 0 + +
(4)
2 2
5 0 CO O . CO +

Each reaction rate of this combustion scheme,
i
R , is calculated by the Arrhenius rate [10],
which is presented by equation (5).

(5)
b
o
a
f Arr , i
] m [ ] m [
RT
E
exp A R
(

=

Where A is the pre-exponential factor of rate constant, E is the activation energy, R is the
universal constant of gases and T is the temperature. a and b are empirical constants [10].
The three-step reduced reaction scheme for Calc. consists of the following three reactions:

(6)
2 2 4
2 5 0 H CO O . CH + +
(7)
2 2
5 0 CO O . CO +
(8)
O H O . H
2 2 2
5 0 +


The eddy dissipation concept model [5] is used to estimate the reaction rates of this
combustion scheme. This model relates the rate of reaction to the rate of dissipation of the
reactants and products.

(9)
(
(

=
j j
P
R R
R
i i eddy , i
M v
m
,
M v
m
min
k
B A M v R



Where
i
M is the molecular weight of species i, k is the turbulent kinetic energy and is the
eddy dissipation rate. A and B are the empirical constants [5]. By calculating the species rate
of reaction
i
R , the source term of energy equation due to chemical reactions ) S (
eaction R
may
be estimated as:

(10)
i
i
T
T
i , p
i
o
i
eaction R
R dT c
M
h
S
ref

+ =

Where
o
i
h is the enthalpy of formation of species i in the reference pressure and temperature.
i
R is the volumetric rate of creation of species i.

Radiation Modeling
The radiative transfer equation for an absorbing-emitting (not scattering) medium at
position r, in the direction s, is as follows [11]:

(11)
)) s ( I I (
ds
) s ( dI
b
=

Where ) s ( I is the radiation intensity,
b
I is the black body intensity and is the absorption
coefficient of medium. The weighted sum of gray gases model (WSGGM) is applied to
compute the absorption coefficient [12]. The discrete transfer radiation model (DTRM) is
used to solve the radiative transfer equation [11]. DTRM assumes that all surfaces are diffuse
and the effect of scattering is not included. Also, equation (11) is integrated along a series of
rays emanating from boundary faces. If is constant along the ray, then ) s ( I may be
estimated by equation (12):

(12)
] s exp[ I ]) s exp[ (
T
) s ( I
o

+ = 1
4

Where
o
I is the radiant intensity at the start of the path, which is determined by the
appropriate boundary condition. The energy source term in the fluid, due to the radiation, is
then computed by summing the change in the intensity along the path of each ray that is
traced through the fluid control volume.

Boundary Conditions
Figure 2 shows the boundary conditions employed in the present numerical work.
According to the symmetrical shape of the combustor, the center-line is defined as the axis of
symmetry. Mass flow rates and atmospheric pressure are applied for inlets (i.e., CH
4
and air)
and outlet boundary conditions of the combustor, respectively. Radiative heat transfer is
considered for the outer wall of the combustor.



Figure 2. Computational domain with boundary conditions.


Numerical Solution
In the present study, combustion simulation is based on the k turbulence model and the
discrete transfer radiation model (DTRM) for describing the turbulence flow field and the
radiative heat transfer, respectively. The partial differential equation represented by equation
(1), is iteratively solved by the SIMLPE algorithm with the tri-diagonal matrix algorithm
(TDMA) [9]. In this simulation, the equation of continuity, momentum (u,v), enthalpy and gas
species mass fractions are discretized by the Finite Volume numerical method. The wall
function model is applied to estimate the flow near the wall. Second order upwind and second
order central difference schemes are adopted in the convective and diffusive terms,
consecutively [9]. Three structured grid sizes, as presented by Table 2 are investigated to
verify the independence of present numerical solution from the size of grid.

Grid size Case
50*160 1
75*200 2
100*320 3

Table 2. The grids Size.

Figure 3 shows the effect of grid size on the center-line temperature profile, estimated by
Calc.. The result for case 3 demonstrates insignificant changes in the solution and it is hence
considered as the suitable mesh.

200
400
600
800
1000
1200
1400
1600
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
X (m)
T
e
m
p
e
r
a
t
u
r
e

[
K
]
Case 1
Case 2
Case 3

Figure 3. Effect of grid size on center-line temperature profile.

Results and Discussion
Figure 4 illustrates the contours of streamlines in the combustor. A recirculation zone is
appeared in the corner near the outer wall of the combustor, which is created due to sudden
change in the cross section of the flow. In this region, because of the air injection, the
equivalence ratio, , will be reduced and the mixture will become lean.





Figure 4. Contours of streamlines in the combustor.


Figure 5 shows the radial distributions of temperature estimated by Calc. (Arrhenius rate)
and Calc. (EDC), at m 1 . 0 x = downstream of the burner nozzle. The present results using
the EDC model agree well with the experimental results of Yaga et al. [8], except at the
center-line (r/D=0). This is due to the large effect of radiation from the suction pyrometer tip,
causing the thermocouple to indicate excessive temperature [8]. However, the present
predictions using the Arrhenius rate appear to be inconsistent with the available experimental
data, showing somewhat higher trends. This could possibly be due to the inconsideration of
the
2
H formation in the two-step reaction scheme (see equations (3) and (4)).

0
200
400
600
800
1000
1200
1400
1600
1800
2000
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
r/D [-]
T
e
m
p
e
r
a
t
u
r
e

[
K
]
EDC
Arrhenius
Experiment [8]


Figure 5. Radial distributions of temperature at x = 0.1 m.

As for the EDC model, the maximum difference between the predicted temperature and the
experimental data is approximately 23 percent which is displayed at r/D=0.05, and for
r/D>0.05 the difference is less than 4 percent.
Figure 6 demonstrates the radial distributions of
4
CH mole fraction estimated by Calc.
(Arrhenius rate) and Calc. (EDC), for an equivalence ratio of equal to 1, at m 1 . 0 x =
downstream of the burner nozzle. The present predictions employing both the Calc.
(Arrhenius rate) and Calc. (EDC) are in fairly good agreement with the available
experimental data of Yaga et al. [8]. Also it can be seen that the
4
CH concentration in the
recirculation zone (i.e., approximately at r/D>0.05), where the mixture is shown to be lean
(see Figure 4), is very little.


0
0.1
0.2
0.3
0.4
0.5
0.6
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
r/D [-]
C
H
4

M
o
l
e

F
r
a
c
t
i
o
n
EDC
Arrhenius
Experiment[8]


Figure 6. Radial distributions of
4
CH mole fraction at x = 0.1 m.

Figure 7 illustrates the radial distributions of CO mole fraction evaluated by Calc.
(Arrhenius rate) and Calc. (EDC), for an equivalence ratio of equal to 1, at m 1 . 0 x =
downstream of the burner nozzle. The experimental data display a decreasing trend as the
radial distance is increased. The present simulation using the Arrhenius rate shows a fairly
good agreement with the experimental trend. But, the present CO mole fraction values are

0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0.05
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
r/D [-]
C
O

M
o
l
e

F
r
a
c
t
i
o
n
EDC
Arrhenius
Experiment[8]


Figure 7. Radial distributions of COmole fraction at x = 0.1 m.


marginally lower, specially at r/D<0.05. As for the present simulation employing the EDC
model, the trend is quite different from the experimental data, particularly at r/D<0.05. The
reason for this is not known and not enough data is available to clarify it. But, it could
probably be due to the use of a simple CO reaction mechanism [8].

Conclusions
1- The estimated results of temperature using the EDC model agree well with the experimental
results of Yaga et al. [8]. However, the present predictions using the Arrhenius rate appear to
be inconsistent with the available experimental data. This could possibly be due to the
inconsideration of the
2
H formation in the two-step reaction scheme.
2- The predicted results of CH
4
mole fraction distributions employing both Calc. (Arrhenius
rate) and Calc. (EDC) are in fairly good agreement with the available experimental data.
3- The present estimation of CO mole fraction distributions using the Arrhenius rate shows a
fairly good agreement with the experimental trend. As for the present simulation employing
the EDC model, the trend is quite different from the experimental data, particularly at
r/D<0.05, and it could probably be due to the use of a simple CO reaction mechanism.
4- For engineering applications such as the gas turbine combustion chamber, the choice of a
suitable reaction model together with a reaction mechanism, is of considerable importance for
predicting accurately the turbulent combustion.

Nomenclature
axial distance (m) x specific heat (j/kg K )
p
c
stoichiometric coefficient for reactant x

generation of turbulent kinetic energy
k
G
stoichiometric coefficient for product x
enthalpy (j/kg) h
eddy dissipation rate (m
2
/s
3
) Intensity of radiation I
absorption coefficient (1/m)
effective thermal conductivity(w/m.K)
eff
k
viscosity (Pa s)
turbulent kinetic energy (m
2
/s
2
)
k
equivalence ratio{
.
) / (
) / (
Stoich
Actual
a f
a f
} mass fraction for species i (-) m
density (kg/m
3
)

molecular weight (kg/kmol) M
Prandtl number (-)
h

pressure (pa) P
Schmidt number (-)
m

radiative heat flux (w/m2) r


q
shear stress tensor
ij
radial distance (m) r
diffusion coefficient


universal constant of gases

(j mol
-1
K
-1
)
R
Subscripts
reaction rate of species i (mol m
-3
s
-1
) i
R
Arrhenius Arr
volumetric source term of quantity

S
air a
temperature (K) T
fuel f axial velocity (m/s) u
oxidant o radial velocity (m/s)
v

References
[1] Furuhata,T., Nogami, T., Tanno, S., Miura, T., Miyabuch, Y., Abe, T., Sugimoto, T. and
Ureshi, K., Simulation of Combustion Characteristics in the Combustor for Gas Turbine
Generator , Experimental Heat Transfer, Fluid Mechanics and Thermodynamics, 538-
588, (1991).
[2] Aoki, H., Tanno, S., Miura, T. and Ohnishi, S. Three dimensional Spray Combustion
Simulation in a practical boiler, JSME International Journal. Vol. 35, 428-434, (1992).
[3] Cook, A. W. and Bushe, W. K., A Subgrid-Scale Model for the Scalar Dissipation Rate in
Non-premixed Combustion, Physics of Fluids, Vol. 11, 746-748, (1999).
[4] Patankar, S.V. and Spalding, D.B., A Computer Model for Three-Dimensional Flow in
Furnaces, 14th International Symposium on Combustion. The Combustion Institute,
Pittsburgh, PA, 1973, pp. 185-207.
[5] Magnussen, B.F., and Hjertager, B.H., On Mathematical Modeling of Turbulent
combustion with Special Emphasis on Soot Formation and Combustion, Sixteenth
International Symposium on Combustion, The Combustion Institute, Pittsburgh, PA,
1976, pp. 719-729.
[6] Yamamoto, T., Miazaki, T., Furuhata,T., Arai, N., Kobayashi, N. & Miura, T.,
Temperature Profile in the pressurized Methane-Air Combustor, Journal of Flow
Visualization and Image Processing, Vol. 5, 51-62, (1998).
[7] Makita, T., Miazaki, T., Furuhata,T. and Arai, N., Turbulent Combustion Characteristics
of a Pressurized Methane-Air Combustor under Fuel-Rich Condition , International
Joint Power Generation Conference, Miami Beach, Florida, 2000, Num. 15048.
[8] Yaga, M., Endo, H. Yamamoto T., Aoki H. and Miura T. Modeling of Eddy Characteristic
Time in LES for Calculating Turbulent Diffusion Flame, International Journal of Heat
and Mass Transfer, Vol. 45, 2343-2349, (2002).
[9] Patankar, S.V., Numerical Heat Transfer and Fluid Flow, Hemisphere Publishing Co.,
1980.
[10] Turns, S.R., An Introduction to Combustion, McGraw-Hill, 1996.
[11] Modest, F.M. Radiative Heat Transfer, McGraw-Hill, 1993.
[12] Modest, F.M.,The Weighted Sum of Gray Gases Model for Arbitrary Solution Method in
Radiative Transfer , International Journal of Heat Transfer, Vol. 113, NO. 3, 650-656,
(1991).

You might also like