571

Validation of Coal Combustion Model

by Using Experimental Data of Utility Boilers∗

Kenji YAMAMOTO∗∗ , Masayuki TANIGUCHI∗∗ , Hironobu KOBAYASHI∗∗ ,

Taro SAKATA∗∗∗ and Kazuhiko KUDO∗∗∗∗

Applicability of a coal combustion model was validated by comparing its predictions

with experimental data of utility boilers. The coal combustion model had gasification and
NOx reaction submodels and it was developed by using a small drop-tube-furnace (coal feed
rate 0.6 kg/h). A turbulence combustion simulation program was developed by introducing
the coal combustion model. The program was validated by comparing its predictions with
23 sets of experimental results which contained different plant, load and coal data. The
temperature difference between simulated and experimental results was within 30◦ C at the
furnace exit. The decreasing characteristic of coal burnout with increasing load was well
predicted. The NOx emission difference between simulated and experimental results was less
than 15%. The coal combustion model was judged applicable to utility boilers.

Key Words: Coal Combustion, Gasification, Numerical Simulation, Pulverized Coal, Util-
ity Boiler, NOx , Furnace Exit Gas Temperature, Burnout, Furnace

carbon dioxide or steam, are the main reaction in the fuel-

1. Introduction rich region. The reaction rate parameters of the gasifica-
Decrease of NOx emission is required for pulverized tion reactions have been measured in nitrogen atmosphere.
coal combustion boilers. Studying combustion mecha- The gasification reaction models that have been developed
nisms in fuel-rich regions, where NOx is reduced to N2 , based on the experimental data were sometimes not intro-
is important to develop boilers with lower NOx emission. duced into the simulation programs, because these models
Also, combustion process modeling and development of did not contribute to better prediction accuracy.
a numerical simulation program that predicts NOx con- Yamamoto et al.(4) developed an accurate gasification
centration, coal burnout, and temperature distribution are reaction model based on experimental data measured by
required to develop new boilers in a short time. using a drop-tube-furnace (coal feed rate was 0.6 kg/h).
There were a few pulverized coal combustion simu- The prediction error of coal burnout was less than 3% in
lation programs which predict carbon dioxide and oxygen the SR (stoichiometric ratio) range of 0.4 to 1.5. They
concentrations in furnaces(1), (2) . However, there are none claimed that the gasification reaction rates measured under
which can predict concentration distributions of carbon the coal-air combustion condition were accelerated. Fur-
monoxide and cyan, and NOx emissions. The gasification thermore, about 10% of coal char was found to be con-
reactions, which are the reactions between coal char and sumed by the gasification reactions at SR 0.8 which is a
normal operating condition for utility boilers. Therefore,
∗ the gasification reactions must be introduced into the sim-
Received 7th April, 2004 (No. 02-0738). Japanese Orig-
inal: Trans. Jpn. Soc. Mech. Eng., Vol.69, No.679, B ulation program to predict utility boiler performances.
(2003), pp.730–736 (Received 10th June, 2002) Taniguchi et al.(5) developed a NOx reaction sub-
∗∗
Power & Industrial Systems R & D Laboratory, Hitachi, model based on experimental data measured by using the
Ltd., 832–2 Horiguchi, Hitachinaka-shi, Ibaraki 312–
drop-tube-furnace. The most important reaction to predict
0034, Japan. E-mail: kenji-c [email protected]
∗∗∗ NOx concentration was found to be the reaction between
Kure Division, Babcock-Hitachi, K.K., 6–9 Takara-machi,
Kure-shi, Hiroshima 737–8507, Japan NOx and hydrocarbon radicals formed in the gasification
∗∗∗∗
Department of Mechanical Engineering, Hokkaido Uni- reaction region. Yamamoto et al.(4) developed a turbulent
versity, Kita 13 Nishi 8, Kita-ku, Sapporo, Hokkaido 060– combustion simulation program which adopted the gasifi-
8628, Japan cation reaction submodel and the NOx reaction submodel.

JSME International Journal Series B, Vol. 48, No. 3, 2005

572

The program was validated by comparing its predictions

2. Numerical Simulation Models
with the experimental data measured by using a labora-
tory scale furnace with a coal feed rate of 25 kg/h. The 2. 1 Governing equations
simulation program predicted not only concentration dis- In this paper, flow, combustion and heat transfer equa-
tributions of major gas species but also those of carbon tions were solved to predict the boiler performances. All
monoxide and cyan, which are mainly formed in the fuel- transport processes except radiation were represented by a
rich region. But the applicability of the program to the general convective transport equation which could be writ-
simulation of utility boilers was not examined. Also, the ten for Cartesian coordinates as


prediction accuracy of gas temperature, which is impor- ∂ρφ ∂ ∂φ
+ ρui φ + Γφ =Sφ (1)
tant to predict NOx emission and coal burnout, was not ∂t ∂xi ∂xi
validated. φ = {1,u,v,w,h,Y j ,k,ε,Cchar ,Cash ,Cvol } (2)
The purposes of this paper are (1) to apply the pro- µ + µt
gram to the simulation of utility boilers and (2) to validate Γφ = (3)
σφ
the prediction accuracy of NOx emission, gas temperature, where φ is a dependent variable, Γφ is the effective viscos-
and coal burnout. ity coefficient, and S φ is a source term. The SIMPLE(7)
Nomenclature method was adopted for the calculation of momentum and
pressure. The gas compositions Y j were solved for N2 , O2 ,
Ac : Frequency factor kg/m2 sPa
CO2 , CO, H2 O, H2 , CH4 , NO, and XN. XN represents the
C : Solid concentration in gas kg/kg
sum of HCN and NHi, and it is mainly generated in the
D : Diffusion rate m2 /s
devolatilization and gasification regions. The standard k-ε
Ec : Activation energy J/mol
turbulence model was used. The mass concentrations of
h : Specific enthalpy J/kg
char (Cchar ), volatile matter (Cvol ), ash (Cash ) were solved
k : Turbulence energy m2 /s2
in the Eulerian coordinate system(6) .
k s : Surface reaction rate coefficient kg/m2 sPa
2. 2 Gas combustion model
kd : Diffusion coefficient kg/m2 sPa
There are two important phenomena in gas combus-
Mc : Molecular weight of carbon kg/mol
tion. First, if reaction time is sufficiently long, gas concen-
m p : Weight of one particle kg
trations reach the values of chemical equilibrium. Second,
n : Number density of particles 1/m3
total reaction rate depends on chemical reaction rates and
P j : Partial pressure Pa
a turbulent mixing rate. In particular, the first phenomenon
R : Universal gas constant (=8.31) J/ mol K
is important to predict the gas concentrations in fuel-rich
R j : Reaction rate kg/m3 s
region, but there are only a few simulation programs that
S ext : Surface area of a particle m2
reproduce it(8) . In this paper, total reaction rate was cal-
S char : Char reaction rate kg/m3 s
culated only from the turbulent mixing rate because the
T : Temperature K
chemical reaction rate was high(6) . Eddy dissipation con-
u, v, w : Gas velocity m/s
cept proposed by Magnussen and Hjertager(9) was used to
Wc : Ratio of remaining weight of char to weight of an
calculate the total reaction rate described as follows,
original coal particle kg/kg
Wa : Ratio of remaining weight of ash to weight of an R j = ρ(Y j − Ȳ j )/τt (4)
original coal particle kg/kg where τt is a characteristic time scale of the turbulent reac-
Y j : Mass fraction of gas species j kg/kg tion calculated by the their model. When the gas concen-
ρ : Gas density kg/m3 tration approaches those of chemical equilibrium, the total
ε : Turbulent dissipation rate m2 /s3 reaction rate becomes zero. By using this formulation, the
φ : Conservation variable two important phenomena are reproduced.
Γ : Effective diffusion coefficient kg/ms 2. 3 Coal combustion model
µ : Viscosity kg/ms Lagrangian methods are widely used for the calcu-
Subscripts lation of particle transportation in pulverized coal com-
j : Gas species bustion(10) . However, it was found that convergence of
m : Boundary layer value the solution was not obtained when a few sampling parti-
p : Particle cles were used for the Lagrangian method, so, a simplified
t : Turbulent Eulerian particle transportation method(6) was used in this
0 : Raw coal value study. The transportation equation was solved for three
Superscripts components of the coal (char, volatile, and ash).
∗
: Raw coal value The two-competitive reaction model(11) was adopted
¯ : Equilibrium value to predict the coal devolatilization. The reaction rate was

Series B, Vol. 48, No. 3, 2005 JSME International Journal

 573

measured by devolatilization experiments(12), (13) which

used a pulsed LASER to heat up particles.
The char reaction model(4) used in this study was de-
veloped based on the experimental data measured by us-
ing a drop-tube-furnace for various conditions of coal-air
combustion. Measured and calculated burnouts were com-
pared for wall temperatures of 1 173 to 1 673 K and the
stoichiometric ratio of 0.4 to 1.5. The prediction error for
(a) Nitrogen species
the burnout was within 3%.
In the char reaction model, char was assumed to re-
act with O2 , CO2 and H2 O as described in Eqs. (5) to (7).
The oxidation reaction was defined as Eq. (5). Gasifica-
tion reactions were defined as Eqs. (6) and (7). The total
char reaction rate depended on gas diffusion rates and an
apparent reaction rate (Eqs. (8) – (10)). The apparent reac-
tion rate means the combination reaction rate of a surface (b) CHi
reaction and pore diffusion. Fig. 1 Reaction schemes
The external surface area was calculated by Eq. (11).
Wc and Wa are the ratios of the remaining weight of char
or ash to the weight of an original coal particle, respec- reaction scheme of nitrogen species. Figure 1 (b) shows
tively. The initial particle diameter of coal was given by the reaction scheme of hydrocarbon radicals. The reaction
the volume-surface mean diameter (d32). The diameter of scheme of volatile N was based on Fenimore’s mechan-
a particle was assumed to be unchanged on devolatiliza- ism(14) . In addition, NOx reduction by hydrocarbon rad-
tion and decreased with combustion of char. The surface icals (CHi)(15) , formation of NOx from char-N, and the
area was calculated as described Eq. (12) from the mass extended Zeldovich mechanism(16) were introduced. The
concentration of char and ash obtained from the transport average difference between calculated and measured NOx
Eq. (1). The number density of particles was assumed to concentration was less than 17% at the furnace exit.
be unchanged in the combustion processes, and was cal- Here, the reaction route of nitrogenous compounds in
culated by Eqs. (13) and (14). low-NOx combustion is described briefly. First of all, XN
1 is generated by devolatilization (R1). Since the oxidation
C + O2 → CO (5) reaction rate (R2) of XN is high in the oxygen-rich region,
2
C + H2 O → CO + H2 (6) XN changes to NOx immediately. Therefore, the increas-
ing point of NOx concentration is detected upstream to that
C + CO2 → 2CO (7)
of XN concentration(17) .
 P j S ext n
S char = (8) Next, the low oxygen concentration region is formed
j 1 1
+ in the fuel-rich region, and is generated by devolatilization
kd j k s j
and char surface reactions. In recent years, with the use of
j = O2 , CO2 , H2 O low volatile coal, NOx emission has been decreased with a


Ec j low-NOx burner(18) . Therefore, the char surface reactions
k s j = Ac j exp − (9)
RT p are important to decrease NOx emission. The main reac-
2Mc D j tions of nitrogen species in the fuel-rich region are NOx
kd j = (10)
d p RT m reduction reactions (R3, R6). In the submodels, hydro-

− 1 carbon radicals were assumed to be generated by the char
∗ Wc + Wa 3 Wc
S ext = S ext (11) surface reactions of gasification.
Wc0 + Wa0 Wc0 Previously, it has been assumed that hydrocarbon rad-

− 13
∗ Cchar +Cash Cchar icals were generated only by devolatilization (R7)(19) . As
= S ext (12) the char in flames had about one mol of hydrogen to eight
Cchar 0 +Cash Cchar 0
Cchar 0 = Cash Wc 0 /Wa 0 (13) mols of carbon(5) , hydrocarbon radicals were assumed to
be formed from the hydrogen (R8). The reaction route R6
n = ρ(Cchar 0 +Cash )/m p 0 (14)
is particularly important for low volatile coal. When R6
2. 4 NOx reaction submodel is excluded, XN concentration may be overestimated, and
The NOx reaction submodel proposed by Taniguchi the NOx reduction reaction (R3) rate may be underesti-
et al.(5) was used in this study. The reaction schemes of mated.
the submodels are shown in Fig. 1. Figure 1 (a) shows the Lastly, in the complete combustion region, XN is ox-

JSME International Journal Series B, Vol. 48, No. 3, 2005

574

idized to NOx by reaction route R2. NOx is formed by the Table 1 Operating conditions of boilers
combustion of char (R4) because the char contains nitro-
gen. Also, the char reduces the NOx simultaneously (R5).
2. 5 Radiation model
A DT (Discrete Transfer) radiation method(20) , which
is a kind of ray-tracing method, was used to evaluate radia-
tive heat transfer. An energy conservative formulation(21)
was adopted to conserve energy. The number of rays was
36 on every radiative heat absorption element. The DT
method is an economical and efficient method for the cal-
culation of combustion chambers(21) .
A total heat absorption coefficient is the sum of gas
and particulate heat absorption coefficients. The heat ab-
sorption coefficient of combustion gases, which contain
CO2 and H2 O, was calculated by the WSGG (Weighted
Sum of Gray Gases)(22) method, which is a kind of optical
database. On the other hand, heat absorption coefficient
of particles (e.g. raw coal, char, and fly ash) was predicted
by the Mie scattering theory, which involves the particle
diameter and wavelength of thermal radiation. By using
this theory, the change of the particle heat absorption co-
efficient in combustion was calculated.
3. Operating Conditions of Pulverized Coal Combus-
Table 2 Boiler configurations
tion Boilers
Boiler performance values such as NOx emission and
gas temperature are changed under conditions such as
maximum output of boilers, load, and coal type. The nu-
merical simulation program must predict the boiler perfor-
mance variation according to condition changes. In this
study, the simulation program was validated for 23 sets of
conditions described in Tables 1 to 3. Three boilers were
Table 3 Coal analysis
used for the validation. The maximum electric output of
the boilers ranged from 500 to 1 000 MW and their basic
structure is shown in Fig. 2. Boiler furnaces had 24 to 56
HT-NR2 type burners(24) and 8 to 16 OAPs (Over Fire Air
Ports). Burners were used to burn pulverized coal and air
and OAPs were used to burn unburned char or flammable
gases such as CO and H2 . There were several platen su-
perheaters at the top of each furnace.
The measured values were obtained in steady condi-
tions for which boiler loads are ranged from 35 to 100%.
Coal feed rates increased with the load. The properties
of the coals are described in Table 3. The fuel ratio (=
fixed carbon/volatile matter) of the coals ranged from 1.2 tures are compared. The gas temperature in combustion
to 2.4 which covers most coals utilized for pulverized coal fields is important to predict char and NOx reaction rates.
combustion boilers in Japan. Since the reaction rate of However, it is difficult to measure gas temperatures in pul-
the coals with higher fuel ratio is low, the concentration verized coal combustion boiler furnaces for the following
of NOx reduction species such as hydrocarbon radicals is reasons. (1) Inserting a probe into the center of the fur-
low. Therefore, the NOx emission is higher if higher fuel naces is difficult because they are large. (2) measuring
ratio coal is used. temperature continuously is difficult because melting ash
sticks to the probe. (3) measuring temperature for a long
4. Results and Discussion
time is necessary because the combustion fields are un-
First, the calculated and the measured gas tempera- steady.

Series B, Vol. 48, No. 3, 2005 JSME International Journal

 575

Fig. 3 Validation of Furnace Exit Gas Temperature

Fig. 2 Structure of test boiler

FEGT (Furnace Exit Gas Temperature) is measured

indirectly from economizer outlet gas temperature and
heat absorption of heat exchangers. It is defined as the
average temperature at the furnace exit as described in
Fig. 2. A flue gas flow rate and gas concentrations are esti-
mated from operating conditions. It is easy to measure the
economizer outlet gas temperature because it is low (about
400◦ C). The specific enthalpy of the flue gases at the econ-
omizer outlet is calculated from the measured temperature
and the flue gas concentration. The heat absorption of heat
exchangers arranged from the furnace exit to the econo-
mizer outlet is calculated from the inlet and outlet steam
(or water) temperatures. The specific enthalpy of furnace
exit flue gases is calculated from that of economizer outlet
flue gases and the heat absorption. The measurement error
of this method is about ±10◦ C. The FEGT measured by
Fig. 4 Temperature distribution of case 11
this method is the space averaged value at the furnace exit.
The differences of measured and calculated FEGTs
are shown in Fig. 3 and are within ±30◦ C. The operating at the center of the furnace on the burner height level is
conditions used for this comparison are cases 1 to 19. Fur- lower than that near burners. This is because the gasifi-
nace heat absorption can be calculated from boiler thermal cation reactions which occur in the area are endothermic
input and FEGT. The prediction error of the heat absorp- reactions.
tions calculated from FEGTs is within ±5%. The heat ab- Next, the concentration distributions of CO and NOx
sorption depends on reaction rates, which vary according are examined. The CO distribution for case 11 is shown in
to positions of heat sources. Therefore, the highly accu- Fig. 5 (a). The basic mechanism for the low NOx burner is
rate prediction of heat absorption indicates that the reac- improvement of the NOx reduction reaction by providing
tion rates are valid. a wide fuel-rich region due to rapid coal ignition(25) . The
Figure 4 shows the gas temperature distribution for CO concentration near the burners is high because coal is
case 11 on a surface that includes the center of the burn- ignited there by using the low NOx burner. The CO con-
ers. The gas temperature is high near the burners and it centration increases from the burners to the OAP height
decreases towards the exit of the furnace because of heat because the gasification reactions occur in the fuel-rich re-
absorption by the furnace water walls. The temperature gion. Downstream from the OAPs, CO concentration de-

JSME International Journal Series B, Vol. 48, No. 3, 2005

576

Fig. 6 Validation of NOx emissions

(a) CO (b) NOx

Fig. 5 Distributions of gas concentration of case 11

creases to almost zero. The behavior that the CO concen-

tration is high near the burners and decreases downstream
from the OAPs agrees with experimental behavior(26) .
The NOx concentration (0% O2 ) on the same surface
is shown in Fig. 5 (b). The NOx concentration near burners
is high. This is because the XN generated by devolatiliza-
tion (reaction route R1) is oxidized to NOx (R2) rapidly.
In the fuel-rich region from the burners to the height of the
OAPs, the NOx concentration is decreased by NOx reduc- Fig. 7 Validation of burnout
tion reactions R5 and R6. After the mixing of OAP air,
NOx concentration is increased because oxidization reac-
tions through the reaction routes R4, R3, and R9 occur in rate is increased with the boiler load, the residence time
the oxygen rich region. The NOx concentration decreases in the fuel-rich region, which is important to reduce NOx
near the furnace exit. This is because NOx is decreased by emission, is decreased. The calculated results agree with
the char reduction reaction (R5). The behavior mentioned the measured ones except for case 8 (boiler B1, coal C5,
above agrees with the measured behavior(26) . load 35%). The NOx reaction submodel was proper. The
The measured and calculated NOx emissions are combustibility under the operating condition of case 8 is
compared for the cases 8 to 11 and 17 to 23. The re- low because of the high fuel ratio coal and low load op-
sults are shown in Fig. 6. The calculated NOx emission eration. The NOx emission increases due to flame insta-
is the averaged value at the furnace exit. The measured bility which occurs under this condition is not predicted
NOx emission is obtained at the economizer outlet. The accurately. The char and NOx reaction submodel was de-
NOx reaction is assumed to be quenched at the furnace veloped based on the experimental data obtained below
exit. The difference between measured and calculated 1 673 K. The maximum temperature of the calculated re-
NOx emissions is less than ±15%. The NOx prediction sults is higher than 2 000 K. The submodel is found to be
accuracy for the actual utility boilers is almost the same applicable to the high temperature conditions.
as that for the drop-tube-furnace. The NOx model is able Figure 7 shows prediction accuracy of coal burnout
to predict NOx emissions of utility boilers. The measured for cases 8 to 11. The measured and calculated coal
NOx emission increases with fuel ratio, for example C2 to burnouts decrease with increasing boiler load. This is be-
C3 or C3 to C5. This characteristic agrees with the cal- cause the combustion time in the furnace is shortened by
culated one. On the other hand, the measured NOx emis- the increase of coal feed rate. Since the prediction error
sion increases with the boiler load. Since the coal feed of coal burnout is less than 0.3%, char reaction model was

Series B, Vol. 48, No. 3, 2005 JSME International Journal

 577

found to be applicable to the simulation of utility boilers. Combustion Science and Technology, Vol.58 (1988),
pp.59–76.
5. Conclusions (9) Magnussen, B.F. and Hjertager, B.W., On Mathemat-
Applicability of a coal combustion model was vali- ical Modeling of Turbulent Combustion with Special
Emphasis on Soot Formation and Combustion, Six-
dated by comparing its predictions with 23 sets of exper-
teenth Symposium (International) on Combustion, The
imental results for utility boilers that had different plant, Combustion Institute, (1976), pp.719–729.
load and coal data. The simulation program predicted (10) Lockwood, F.C., Salooja, A.P. and Syed, S.A., A Pre-
the NOx emission, furnace exit gas temperature and coal diction Method for Coal-Fired Furnaces, Combustion
burnout accurately. and Flame, Vol.38 (1980), pp.1–15.
( 1 ) The difference in measured and calculated NOx (11) Kobayashi, H., Howard, J.B. and Sarofim, A.F., Coal
emissions was less than ±15%. The predicted NOx emis- Devolatilization at High Temperatures, Sixteenth Sym-
sion increased with coal fuel ratio and also with boiler posium (International) on Combustion, The Combus-
tion Institute, (1976), pp.411–425.
load. These characteristics of NOx emission agreed with
(12) Taniguchi, M., Okazaki, H., Kobayashi, H., Azuhata,
the experimental results. S., Miyadera, H., Muto, H. and Tumura, T., Pyrolysis
( 2 ) FEGT (Furnace Exit Gas Temperature) was indi- and Ignition Characteristics of Pulverized Coal Parti-
rectly measured from economizer outlet temperature and cle, Journal of Energy Resources Technology, Vol.123
heat absorption of heat exchangers. The difference in pre- (1995), pp.32–38.
dicted and measured FEGTs was within ±30◦ C. This cor- (13) Okazaki, H., Taniguchi, M., Kobayashi, H., Amano,
responded to 5% error of the furnace heat absorption. K., Azuhata, S. and Hishinuma, Y., Kinetics Rate of
( 3 ) The measured and calculated coal burnouts de- Devolatilization of Pulverized Coal Ignited by Pulsed
Laser, Trans. Jpn. Soc. Mech. Eng., (in Japanese), Ser.
creased with increasing boiler load. The prediction error
B, Vol.62, No.598 (1996), pp.2466–2471.
for the coal burnout was less than 0.3%, and it was the (14) Fenimore, C.P., Studies of Fuel Nitrogen Species
same level for the results of the drop-tube-furnace. in Rich Flame Gases, Seventeenth Symposium (In-
( 4 ) This program predicted the following character- ternational) on Combution, The Combution Institute,
istics: The NOx concentration increased near the burners, (1979), pp.661–670.
decreased in the high temperature fuel-rich region, and in- (15) Bose, A.C. and Wendt, J.O.L., Pulverized Coal
creased downstream from the over fire air port. Combustion: Fuel Nitrogen Mechanisms in the
Rich Post-Flame, Twenty-second Symposium (Inter-
References national) on Combustion, The Combustion Institute,
(1988), pp.1127–1134.
(1) Anagnostopoulos, J.S., Sargianos, N.P. and Bergeles,
(16) Miller, J.A. and Bowman, C.T., Mechanism and Mod-
G., The Prediction of Pulverized Greek Lignite Com-
elling of Nitrogen Chemistry in Combustion, Progress
bustion in Axisymmetric Furnaces, Combustion and
in Energy and Combustion Science, Vol.15 (1989),
Flame, Vol.92 (1993), pp.209–221.
pp.287–338.
(2) Truelove, J.S. and Williams, R.G., Coal Combustion
(17) Abbas, T., Costen, P., Lockwood, F.C. and
Model for Flame Scaling, Twenty-Second Symposium
Romo-Millares, C.A., The Effect of Particle Size
(International) on Combustion, The Combustion Insti-
on NO Formation in a Large-Scale Pulverized
tute, (1998), pp.155–164.
Coal-Fired Laboratory Furnace: Measurements and
(3) Smoot, L.D. (Ed.), Fundamentals of Coal Combustion,
Modeling, Combustion and Flame, Vol.93 (1993),
(1993), p.661, Elsevier Science Publishers B.V.
pp.316–326.
(4) Yamamoto, K., Orita, H., Taniguchi, M. and
(18) Kiyama, K., Tumura, T., Sei, N., Taniguchi, M. and
Kobayashi, H., Char Surface Reaction Model for Pul-
Nomura, S., Combustion Technology for High Fuel
verized Coal Combustion, Trans. Jpn. Soc. Mech. Eng.,
Ratio Coal, Proceedings International Conference on
(in Japanese), Ser. B, Vol.65, No.637 (1999), pp.3196–
Power Engineering-97, Vol.2 (1997), pp.447–481.
3202.
(19) Williams, A., Pourkashanian, M., Bysh, P. and
(5) Taniguchi, M., Yamamoto, K., Kobayashi, H. and
Morman, J., Modeling of Coal Combustion in Low-
Kiyama, K., A Reduced NOx Reaction Model for Pul-
NOx P.F. Flames, Fuel, Vol.73 (1994), pp.1006–1019.
verized Coal Combustion under Fuel-Rich Conditions,
(20) Lockwood, F.C. and Shah, N.G., A New Radiation So-
Fuel, Vol.81 (2001), pp.363–371.
lution Method for Incorporation in General Combus-
(6) Yamamoto, K., Taniguchi, M., Kiyama, K. and
tion Prediction Procedures, Eighteen Symposium (In-
Matsuda, J., A Study of Gasification Reaction in Pul-
ternational) on Combustion, The Combustion Institute,
verized Coal Combustion, 3rd International Sympo-
(1981), pp.1405–1414.
sium on Advanced Energy Conversion Systems and
(21) Coelho, P.J. and Carvalho, M.G., A Conservative For-
Related Technologies, 1-B-5, (2001).
mulation of the Discrete Transfer Method, ASME Jour-
(7) Patankar, S.V., Numerical Heat Transfer and Fluid
nal of Heat Transfer, Vol.119 (1997), pp.118–128.
Flow, (1980), Hemisphere Publishing Corporation.
(22) Smith, T.F., Shen, Z.F. and Friedman, J.N., Evalua-
(8) Smith, P.J. and Fletcher, T.H., A Study of Two Chem-
tion of Coefficients for the Weighted Sum of Gray
ical Reaction Models in Turbulent Coal Combustion,

JSME International Journal Series B, Vol. 48, No. 3, 2005

578

Gases Model, ASME Journal of Heat Transfer, Vol.104 (25) Azuhata, S., Narato, K., Kobayashi, H., Arashi, N.,
(1982), pp.602–608. Morita, S. and Masai, T., A Study of Gas Composition
(23) Kim, C. and Lior, N., Easily Computable Good Ap- Profiles for Low NOx Pulverized Coal Combustion,
proximations for Spectral Radiative Properties of the Twenty-First Symposium (International) on Combus-
Particulate-Gas Components and Mixture in Pulverized tion, The Combustion Institute, (1986), pp.1199–1206.
Coal Combustors, Fuel, Vol.74 (1995), pp.1891–1902. (26) Narato, K., Azuhata, S., Arashi, N. and Morita, S., De-
(24) Ishida, T., Yamanouchi, H., Morita, S. and Kaneda, H., tailed Mapping of the Structure of a Pulverized Coal
Advanced Technology for Coal Fired Boilers, Hitachi Flame in a Laboratory Scale Combustor, Trans. Jpn.
Hyoron, (in Japanese), Vol.74, No.11 (1992), pp.803– Soc. Mech. Eng., (in Japanese), Ser. B, Vol 55, No.517
808. (1989), pp.2787–2794.

Series B, Vol. 48, No. 3, 2005 JSME International Journal