Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
1874396X/10 2010 Bentham Open
Open Access
Numerical Simulation of NO Production in AirStaged Pulverized Coal
Fired Furnace
Robert Straka
1,
*
and Michal Beneˇs
2
1
Department of Heat Engineering and Environment Protection, Faculty of Metals Engineering and Industrial,
Computer Science, AGH University of Science and Technology, Aleja Mickiewicza 30, 30059, Cracow, Poland
2
Department of Mathematics, Faculty of Nuclear Sciences and Physical Engineering, Czech Technical University,
Trojanova 13, 12000, Prague, Czech Republic
Abstract: We describe behavior of the aircoal mixture using the NavierStokes equations for the mixture of air and coal
particles, accompanied by a turbulence model. The undergoing chemical reactions are described by the Arrhenian kinetics
(reaction rate proportional to
RT
E
e
). We also consider the heat transfer via conduction and radiation. The system of PDEs
is discretized using the finite volume method (FVM) and an advection upstream splitting method as the Riemann solver.
The resulting ODEs are solved using the 4thorder RungeKutta method. Simulation results for typical power production
level are presented together with the airstaging impact on NO production.
Keywords: Staged coal combustion, reactive flow modeling, NO production.
1. INTRODUCTION
The main motivation of the combustion model research is
its future inclusion in the current model of the steam
generator [1], to use this model for development of control
systems for the industrial installation. Another purpose for
the combustion model is the optimization of the production
of the nitrogen oxide, which strongly depends on the
temperature and unburned char distribution, and thus can be
controlled by intelligent distribution of fuel and oxygen into
the burners. Because the experiments on a real device are
prohibitively cumbersome and expensive, in extreme cases
even hazardous, the only way to test the behavior of the
furnace is mathematical modeling. The mathematical model
of combustion is based on the NavierStokes equations for a
mixture of multiple components where the coal particle are
treated as one of the phases. We chose to use this approach,
as it simplifies the model especially when dealing with
turbulence, and also removes several empirical relations and
constants, moreover it speedups the computational time of
the simulation.
2. MATHEMATICAL MODEL
Currently, the following components of the mixture are
considered:
• chemical compounds engaged in major thermal and fuel
NO reactions: nitrogen (N
2
), oxygen (O
2
), nitric oxide
(NO), hydrogen cyanide (HCN), ammonia (NH
3
), carbon
dioxide (CO
2
) and water (H
2
O);
*Address correspondence to this author at the Department of Heat Engineer
ing and Environment Protection, Faculty of Metals Engineering and Indus
trial, Computer Science, AGH University of Science and Technology, Aleja
Mickiewicza 30, 30059, Cracow, Poland; Tel: (+48)126172637;
Email: straka@metal.agh.edu.pl
• char and volatile part of the coal particles;
The gas phase is described by the following equations.
As stated above, the mass balance is described by equations
of mass balance of each subcomponent (the Einstein
summation is used)
t
(Y
i
) +
x
j
(Y
i
u
j
) =
r
J
i
+ R
i
, (1)
where t is time, x
j
is jth component of spatial variable,
is the flue gas mass density, Y
i
mass fraction of the ith
component, and u
j
are the gas velocity components. The
righthand side term
r
J
i
describe the laminar and turbulent
diffusion of the components and production or consumption
due to chemical reactions is described within the R
i
term.
The above equations of component mass balance are
accompanied by the equation of total mass balance
t
+
(u
j
)
x
j
= 0. (2)
Equations of momentum conservation are as follows
o
ot
(pu
i
) +
o
ox
j
(pu
i
u
j
) =
~
op
ox
i
+
o
ox
j
μ
eff
ou
i
ox
j
+
ou
j
ox
i
~
2
3
o
ij
ou
l
ox
l
[
\

¸
)
j








+ g
i
,
(3)
where p is pressure,
ij
is Kronecker's delta,
r
g = [g
1
, g
2
, g
3
]
is the external force acting on the fluid, in our case the
28 The Open Thermodynamics Journal, 2010, Volume 4 Straka and Beneˇs
gravity. The effective friction coefficient μ
eff
is calculated
from the turbulence model as
μ
eff
= μ + μ
t
= μ + C
μ
k
2
,
where μ is the laminar viscosity, k the turbulent kinetic
energy, and the turbulent energy dissipation rate. Constant
C
μ
, like additional constants mentioned later in the
description of the turbulence model, has to be chosen
empirically for the particular problem, in our case we use
C
μ
= 0.09 , which appears to give satisfactory results.
The last equation describes the conservation of energy
t
(h) +
x
j
(u
j
h) = n
coal
dm
p
dt
h
comb
+ q
r
+ q
c
+ q
s
, (4)
where h is specific enthalpy, n
coal
is the density of coal
particles (amount of particles in unit volume), m
p
is the
mass of the combustible matter within the one coal particle,
h
comb
is the heat of combustion, q
r
, q
c
and q
s
are
respectively heat transfer by radiation, heat transfer by
conduction, and heat source or sink. The heat transfer terms
are computed as follows
q
c
= T ( ),
for the transfer by conduction, which is described by the
Fourier law of heat conduction, where is the conductivity
and T is temperature,
q
r
= cT ( ),
for the transfer by radiation. The radiation heat transfer is
fully described by an integraldifferential equation of
radiation, which is very computationally expensive to solve.
However, as the flue gas can be considered as an optically
thick matter, the above approximation of the radiation flux
called the Rosseland radiation model can be applied. Here c
is socalled radiative conductivity and defined as
c =
16n
2
T
3
3
R
,
where n is the refractive index, is the StefanBoltzmann
constant and
R
is the Rosselandmean extinction
coefficient [2]. The heat sink term is nonzero only in the
edge computation cells and describes the energy exchange
with the walls of the furnace via conduction and radiation
q
s
= A(T
gas
T
wall
) B(T
gas
4
T
wall
4
),
where A and B are constants dependent on the properties of
the interface between the modeled region and its
surroundings. T
gas
and T
wall
are the temperatures of the gas
and walls respectively.
Our approximation of the coal particle combustion rate
has been based on the data in [11], where the dynamics of
combustion of a single particle of the coal used in the boiler
has been measured. This article also contained an empirically
postulated mathematical model of the particle combustion in
the form of an ordinary differential equation
dm
dt
=
DNt
(1+ Dt
N
)
2
,
where
N = a
1
d
i
a
2
Y
O
2
a
3
T
r
a
4
,
D = a
5
d
a
6
Y
O
2
a
7
T
0
a
8
T
r
a
9
,
T
r
=
T
0
T
z
T
0
,
where the symbols are as follows:
m burnt combustible matter [kg],
t elapsed time [s],
d particle diameter [mm],
Y
O
2
oxygen concentration [%],
T
r
relative temperature,
T
0
environment temperature [
o
C],
T
z
coal ignition temperature [
o
C].
The constants a
1
,..., a
9
are empirical and need to be
determined experimentally for the particular type of coal.
Although the above equations can be fitted to experimental
data by a careful choice of the a
1
,..., a
9
constants, the model
itself is questionable due to the time dependence of the right
hand side of the differential equation. Obviously, to be used
in any simulation, the combustion model needs to be time
invariant. To keep the computational complexity low, and to
avoid the time dependence on the right hand side, we opted
to use a simple approximative model consisting of two
Arrhenian exponentials. The rationale behind this choice is
that the coal combustion consists of the faster combustion of
the volatile matter, and slower combustion of the char.
dm
coal
dt
= ~A
vol
m
p
o
vol
Y
O
2
µ
vol
exp ~
E
vol
RT
p
[
\

¸
)
j
~A
char
m
p
o
char
Y
O
2
µ
char
exp ~
E
char
RT
p
[
\

¸
)
j
where
p
m is the particle's combustible mass, A
vol
, E
vol
,
vol
,
vol
, A
char
, E
char
,
char
,
char
are empirical constants,
Y
O
2
is the oxygen mass fraction, R is universal gas constant
and T
p
is the particle temperature. These equations are
accompanied by the equation of state, as usual
p = (k ~1)p e ~
1
2
u
2
[
\

¸
)
j
.
Here, is the Poisson constant and e is the gas specific
energy.
No Production in AIRStaged The Open Thermodynamics Journal, 2010, Volume 4 29
For the turbulence modeling, we use the standard k 
model, which describes the evolution of turbulence using
two equations — first one for the turbulent kinetic energy
o
ot
(pk) +
o
ox
j
(pku
j
) =
o
ox
j
μ +
μ
t
o
k
[
\

¸
)
j
ok
ox
j








+ G
k
~ pr,
(5)
and the second one for the turbulent kinetic energy
dissipation rate
o
ot
(pr) +
o
ox
j
(pru
j
) =
o
ox
j
μ +
μ
t
o
r
[
\

¸
)
j
or
ox
j








+ C
1r
r
k
G
k
~ C
2r
p
r
2
k
.
(6)
Constants in this model have to be determined empirically,
in our case we use the following values: C
1
=1.44 ,
C
2
=1.92 ,
k
=1.0 ,
=1.3.
Left hand sides of the equations describe passive
advection of the respective quantities by the advection
velocity
r
u . Right hand sides describe their spatial diffusion,
their production and dissipation.
The term G
k
, which describes the production of
turbulence, can be derived from the Reynolds averaging
process and written in the terms of the fluctuating part of the
velocity as
G
k
=¡
jl
ou
j
ox
l
= ~pu
j
'
u
l
'
ou
j
ox
l
= μ
t
S
2
, with S
ij
=
1
2
ou
i
ox
j
+
ou
j
ox
i
[
\

¸
)
j
,
where
jl
is the Reynolds stress tensor and S
ij
is mean strain
rate. However during practical computation, fluctuations u
j
and u
i
are unknown, we use the Boussinesq hypothesis that
the Reynolds stress is proportional to the mean strain rate.
Diffusion of the species consists of two processes —
laminar and turbulent, and the diffusion term in Eq. (1) can
be written in the form
r
J
i
= ~ pD
i,m
+
μ
t
Sc
t
[
\

¸
)
j
VY
i
.
First term corresponds to linear laminar diffusion, the second
one to turbulent diffusion. Given the fact that the turbulent
diffusion generally predominates the laminar one, and the
term D
i,m
is difficult to determine, the laminar diffusion can
be usually ignored. The coefficient Sc
t
is the turbulent
Schmidt number and we put Sc
t
= 0.7 .
3. SIMPLIFIED MODEL OF NO CHEMISTRY
This model has been developed to approximately
describe the amounts of NO emissions leaving a coal
combustion furnace. The real mechanism of coal flue gas
production seems to be very complicated, so that just the
most important phenomena and reaction paths were
considered to provide maximum possibility of using this
model in realtime control and operation systems.
There are two major processes attributing to the total NO.
The former is known as Thermal NO or Zeldovich and
simply consists of oxidation of atmospheric nitrogen at high
temperature conditions. The latter is called Fuel NO and
describes NO creation from nitrogen, which is chemically
bounded in coal fuel. Fuel NO is usually the major source of
NO emissions. These are the only mechanisms involved,
although a few more could be considered (such as Prompt
NO (Fenimore) or Nitrous oxide (N
2
O) intermediate
mechanisms).
3.1. Thermal NO
Thermal NO generation mechanism attributes only at
high temperature conditions and is represented by a set of
three equations, introduced by Zeldovich [3] and extended
by Bowman [4]
O+ N
2
k
1
N+ NO
N+ O
2
k
2
O+ NO
N+ OH
k
3
H+ NO
All these reactions are considered to be reversible. Rate
constants were taken from [5] and are as follows
k
1
+
= 1.8 10
8
exp
38370
T
k
1
= 3.8 10
7
exp
425
T
k
2
+
= 1.8 10
4
T exp
4680
T
k
2
= 3.8 10
3
T exp
20820
T
k
3
+
= 7.110
7
exp
450
T
k
3
= 1.7 10
8
exp
24560
T
where “+” and “–” in superscript denotes forward and
reverse rates, respectively. T means temperature in Kelvin.
In order to compute the NO concentration, concentrations
of nitrogen radical [N], oxygen radical [O] and hydroxyl
radical [OH] must be known. It is useful to assume [N] to be
in a quasisteady state according to its nearly immediate
conservation after creation. In fact, this Nradical formation
is the rate limiting factor for thermal NO production, due to
an extremely high activation energy of nitrogen molecule,
which is caused by a triple bond between two nitrogen
atoms. Hence, the NO formation rate can be stated as
d[NO]
dt
= 2k
1
+
[O] [N
2
]
1
k
1
k
2
[NO]
2
k
1
+
[N
2
]k
2
+
[O
2
]
1+
k
1
[NO]
k
2
+
[O
2
] + k
3
+
[OH]
.
30 The Open Thermodynamics Journal, 2010, Volume 4 Straka and Beneˇs
Concentrations of O and OH can be profitably described by
following partial equilibrium approach [6, 7].
[O] = K
1
[O
2
]
1/2
T
1/2
, [OH] = K
2
[O]
1/2
[H
2
O]
1/2
T
0.57
.
Equilibrium constants K
1
and K
2
are as follows
K
1
= 36.64 exp
27123
T
, K
2
= 2.129 10
2
exp
4595
T
3.2. Fuel NO
Composition analysis show, that nitrogenbased species
are more or less present in coal, usually in amounts of tenths
to units of percent by weight. When the coal is heated, these
species are transformed into certain intermediates and then
into NO. Fuel itself is therefore a significant source of NO
pollutants. When a coal particle is heated, it is presumed,
that nitrogen compounds are distributed into volatiles
and char. A parameter is introduced to describe the
distribution of the coalbounded nitrogen between the
volatiles and char part of the coal particle.
m
vol
N
= m
tot
N
,
m
char
N
= (1) m
tot
N
,
where < 0,1 > , m
tot
N
is the total mass of nitrogen, m
vol
N
is
the mass of bounded nitrogen in volatiles and m
char
N
is the
mass of bounded nitrogen in char.
As already mentioned, nitrogen transforms to pollutants
via intermediates, which usually are ammonia NH
3
and
hydrocyanide HCN. To proceed further, we must define
four parameters to describe complex partitioning of the fuel
bound nitrogen.
• is amount of volatile bounded nitrogen which converts
to HCN .
•
1
is distribution of char bounded nitrogen which
converts to HCN .
•
2
is distribution of char bounded nitrogen which
converts to NH
3
.
•
3
is distribution of char bounded nitrogen which
converts to NO.
• < 0,1 >,
1
+
2
+
3
=1.
Different parametric studies should be carried out to find
the best values of , ,
1
,
2
and
3
suitable for specific
type of coal. Five overall reactions of either NO formation or
depletion were incorporated in the combustion part of the
numerical code.
3.2.1. NO, HCN, NH
3
Reactions
According to [8], we have following reactions and its
formation rates
HCN+ O
2

R
1
NO+ Z R
1
=1.0 ·10
10
· X
HCN
· X
O
2
a
· exp
~33732.5
T
[
\

¸
)
j
NH
3
+ O
2

R
2
NO+ Z R
2
= 4.0 ·10
6
· X
NH
3
· X
O
2
a
· exp
~16111.0
T
[
\

¸
)
j
HCN+ NO
R
3
N
2
+ Z R
3
= ~3.0 ·10
12
· X
HCN
· X
NO
· exp
~30208.2
T
[
\

¸
)
j
NH
3
+ NO
R
4
N
2
+ Z R
4
= ~1.8 ·10
8
· X
NH
3
· X
NO
· exp
~13593.7
T
[
\

¸
)
j
where X
*
is the mole fraction, Z are other products we
simply neglect and a is the oxygen reaction order
X
O
2
4.110
3
Ka = 1
4.110
3
X
O
2
1.1110
2
Ka = 3.95 0.9 ln X
O
2
1.1110
2
X
O
2
0.03Ka = 0.35 0.1 ln X
O
2
X
O
2
0.03Ka = 0
3.2.2. Heterogeneous NO Reduction On Char
Present char allows following adsorption process to occur
Char + NON
2
+ Z
Levy [9] uses pore surface area to define NO source
term
S
ads
NO
= k
5
c
s
A
BET
M
NO
p
NO
,
where k
5
= 2.27 10
3
exp
17168.33
T
is the rate constant,
S
ads
NO
is the NO source term, c
s
is the concentration of
particles, A
BET
is the pore surface area and p
NO
is the partial
pressure of NO.
In order to evaluate overall NO source term, single
source terms have to be summarized. This overall source
term can be further used in transport equations. As for HCN
and NH
3
source terms, it is possible to determine them from
coal burnout rate. It is assumed, that nitrogen from both char
and volatiles transforms to intermediate species quickly and
totally.
4. NUMERICAL ALGORITHM
For numerical solution of the equations, finite volume
method is used. For left and right hand sides in Eqs. (1), (2),
(3), (4), (5), (6), advection upstream splitting method (see
[10]) is used to approximate fluxes in the FVM formulation,
and edge dualvolume approximation is used to approximate
the second order derivatives respectively. For detailed
description of the solution procedure see [1].
5. RESULTS
Computational results obtained for different excess air
values for the burners are summarized in Figs. (14). In
Table 1 are values of various parameters used in simulation
and summarized numerical values of air distribution among
the burners. In Figs. (1, 2) one can see temperature and NO
mass fractions profiles for cases indicated in Table 1. In Fig.
No Production in AIRStaged The Open Thermodynamics Journal, 2010, Volume 4 31
(3) is comparison of the NO mass fractions for four cases of
air staging. In Fig. (4) are temporal profiles of concentrations
of NO in flue gas. The studies cases are named Normal
operation – this means that the whole combustion air is
distributed equally among burners, Airstaging 1 – the
combustion air is staged as indicated in Table 1 ; 50% to the
first(bottom) row of burners, 20% to the second and third
row of burners and 10% to the fourth(top) row of burners,
Airstaging 2 – in this case the combustion air is staged
again, but with reverse ordering i.e. 10% to the first row of
burners and 50% to the fourth row of burners, Over Fire Air
(OFA) – the combustion air is distributed equally among
four burner rows and one row of OFA slots i.e. 20% to each
row. In each case the fuel distribution is 25% to every row of
burners. In Fig. (5) is validation of the computed temperature
field with boiler K5TOT, which is installed in heating
facility in Otrokovice (Czech Republic). Mean values
together with standard deviations of the temperature are
depicted in this figure. Those temperatures were measured at
three different locations using optical pyrometers and
thermocouples [11].
6. DISCUSSION
One can see that the airstaging has essential influence on
the NO production, one reason is creation of oxygenrich and
oxygenlean zones and second reason is the different
location of the maximal temperature region. In the case of
Normal operation (see Fig. 1), with air distributed equally,
great amount of NO is produced by both the Thermal and the
Fuel mechanism, with maximum concentrations in the lower
region of the boiler. Further decrease in NO concentration is
mainly due to the diffusion and mixing with the air from
burners. Airstaging 1 and Airstaging 2 cases (see Figs. 1,
2) show the opposite ordering of the combustion air and the
different NO production pattern. Here the oxygenrich zone
is located in the bottom section of the boiler, but due to
higher velocities, the maximum temperature region is shifted
upwards, where the oxygenlean zone is so less ThermalNO
is produced for the case of Airstaging 1. The Airstaging 2
case, with lower velocities in the bottom region of the boiler,
shows the location of the maximum temperatures in the
lower region and due to oxygenlean zone in that region, we
Fig. (1). Profiles of temperature and NO mass fractions for (from top to bottom) Normal operation, Airstaging 1.
32 The Open Thermodynamics Journal, 2010, Volume 4 Straka and Beneˇs
Fig. (2). Profiles of temperature and NO mass fractions for (from top to bottom) Airstaging 2, Over fire air.
Fig. (3). Profiles of NO mass fractions for (from left to right) Normal operation, Airstaging 1, Airstaging 2, Over fire air.
No Production in AIRStaged The Open Thermodynamics Journal, 2010, Volume 4 33
Fig. (4). Flue gas NO concentrations for Normal operation, Airstaging 1, Airstaging 2, Over fire air.
Table 1. Simulation Parameters and Fuel and Air Setup for Studied Cases. Total Air is Distributed among Rows of Burners,
Percentage Distribution is Indicated in the Table
Parameter Symbol Value Unit
StefanBoltzmann constant
5.678 · 10
8
W m
2
K
4
refractive index n 1.7 
gas heat capacity
V
c
1037.0 J kg
1
K
1
molar mass of gas M 0.02896 kg mol
1
raw coal density
1345.0 kg m
3
coal particle radius
r
coal
0.2 · 10
3
m
coal heat capacity
c
V
2048 J kg
1
K
1
boiler crosssection S 49.0 m
2
coal feeding rate
& m
coal
16.4 kg s
1
initial pressure
p
ini
1.0 · 10
5
Pa
initial temperature T
ini
800
o
C
mixture temperature at the inlet T
mix
400
o
C
pressure at the inlet
p
in
1.0 · 10
5
Pa
pressure at the outlet
p
out
pin  100 Pa
excess air coefficient
1.3 
temperature of the wall T
wall
300
o
C
width of the side wall a, b 7.0 m
first burners row position B
1
2.74 m
second burners row position B
2
3.84 m
0
100
200
300
400
500
600
700
0 5 10 15 20 25 30 35
N
O
c
o
n
c
e
n
t
r
a
t
i
o
n
[
m
g
/
m
3
]
Time [s]
Normal operation
Air staging 1
Air staging 2
OFA
34 The Open Thermodynamics Journal, 2010, Volume 4 Straka and Beneˇs
Table 1. Contd….
Parameter Symbol Value Unit
third burners row position
B
3
4.93 m
fourth burners row position
B
4
6.03 m
number of burners in a row N
B
4 
burner diameter
D
B
0.8 m
OFA row position
B
OFA
8.78 m
OFA slot diameter
D
OFA
0.4 m
pore surface area A
BET
25000
m
2
coal nitrogen distribution
0.5 
0.5 
1
0.5 
2
0.25 
3
0.25 
gas  wall heat transfer coef. (conduction) 30.0 W m
2
K
1
gas  wall heat transfer coef. (radiation) 0.6 · 10
8
W m
2
K
4
coal analysis:
ash A 9.5 %
water W 30 %
carbon C 45.72 %
hydrogen H 3.69 %
sulfur S 1.16 %
oxygen O 9.42 %
nitrogen N 0.5 %
Air percentage for burners Case
B1 B2 B3 B4 OFA
Normal operation 25 25 25 25 0
Airstaging 1 50 20 20 10 0
Airstaging 2 10 20 20 50 0
Over fire air 20 20 20 20 20
can see less NO production there. The last case of Over Fire
Air technique shows that the maximum NO level is located
in the near burners region, but less combustion air creates the
oxygenlean (and the nitrogenlean zone for the Thermal
mechanism) zone and an unburned char is able to remove
NO by the adsorption process. The overall comparison of
NO concentrations for all cases are depicted in Fig. (3).
Other important information is the concentrations of NO in a
flue gas leaving the boiler, comparison of the temporal
change in NO concentrations in the flue gas for the cases
considered is in Fig. (4). Here one can see that the OFA
technique gives the best result considering the NO
concentrations. The last part is the validation of the
temperature profile (as the temperature has the greatest
influence on the NO production) with the real values
measured in the heating facility in Otrokovice, results are
depicted in Fig. (5). Inaccuracy for the second spot is
probably caused by the simplified radiation model, thus
further refinement of the radiation heat transfer is needed,
but very good accuracy is achieved for the burners and outlet
regions of the boiler.
No Production in AIRStaged The Open Thermodynamics Journal, 2010, Volume 4 35
7. CONCLUSION
A numerical model of pulverized coal fired boiler with
staged air design has been presented. We developed this
model keeping in mind the balance between the computation
complexity and the precision of the simulation. This was also
the reason why we had used simplified reaction kinetics
instead of full reaction set for the production of NO and
simple kinetic approach of the coal particle combustion. The
model can be used as airstage designing tool for coal fired
boilers with wall burners. Presented results with emphasis on
the NO production are adequate when compared with real
life devices, as well as model behavior for different cases.
Comparison of the measured and computed temperature
profiles showed sufficient accuracy for the burner region as
well as for the flue gas outlet area. Still there are places
where the enhance the model is needed, mainly the radiation
heat transfer modeling, which can improve the accuracy of
the temperature profile above burners, and more accurate
coal particle combustion model.
ACKNOWLEDGMENTS
This work has been supported by the project "Applied
Mathematics in Physical and Technical Sciences" MSM
6840770010 and "Jindich Neas Center for Mathematical
Modelling" of the Czech Ministry of Education, Youth and
Sport.
REFERENCES
[1] J. Makovika, V. Havlena, and M. Bene, “Mathematical
modelling of steam and flue gas flow in a heat exchanger of a
steam boiler”, in ALGORITMY 2002. Proceedings of Contributed
Papers, Slovak Technical University: Bratislava, 2002, pp. 171
178.
[2] M. F. Modest, Radiative Heat Transfer, McGrawHill, Inc., 1993.
[3] J. B. Zeldovich, “The oxidation of nitrogen in combustion and
explosion”, Acta Physicochimica, vol. 21, pp. 577628, 1946.
[4] C. T. Bowman, D. J. Seery, Emissions from Continuous
Combustion Systems, Plenum Press: New York, 1972.
[5] NIST, “Chemical Kinetics Database on the Web”, National
Institute of Standards and Technology, 2000. Available from:
http://www.kinetics.nist.gov
[6] J. Warnatz, NOx Formation in High Temperature Processes,
University of Stuttgart: Germany, 2001.
[7] D. L. Baulch, C. J. Cobos, R. A. Cox, C. Esser, P. Frank, Th. Just,
J. A. Kerr, M. J. Pilling, J. Troe, R. W. Walker, J. Warnatz,
“Evaluated Kinetic Data for Combustion Modelling”, Journal
of Physical and Chemical Reference Data, vol. 21, pp. 411734,
1992.
[8] G. G. De Soete, “Overall reaction rates of NO and N formation
from fuel nitrogen”, In Proceedings of the 15th Symposium
(International) on Combustion, Tokyo, Japan, 1975, pp. 10931102.
[9] J. M. Levy, L. K. Chen, A. F. Sarofim, J. M. Beer, “NO/Char
reactions at pulverized coal flame conditions”, In Proceedings of
the 18th Symposium (International) on Combustion, University of
Waterloo: Canada, 1981, pp. 111120.
[10] M. S. Liou, C. Steffen Jr., “A New Flux Splitting Scheme”, Journal
of Computational Physics, vol. 107, pp. 2329, 1993.
[11] Z. Maek , I. Bezinová, “Description of a model of a combustion
chamber and a design of a model of heat exchanging surfaces” (in
Czech), Technical Report 96107/3, MORE s.r.o., 1997.
Received: October 07, 2009 Revised: October 26, 2009 Accepted: November 11, 2009
© Straka and Beneˇs; Licensee Bentham Open.
This is an open access article licensed under the terms of the Creative Commons Attribution NonCommercial License (http://creativecommons.org/licenses/by
nc/3.0/) which permits unrestricted, noncommercial use, distribution and reproduction in any medium, provided the work is properly cited.
Fig. (5). Comparison of computed and measured temperatures (from [11]) for the K5TOT boiler installed in Otrokovice  Czech Republic.
400
500
600
700
800
900
1000
1100
1200
1300
1400
0 5 10 15 20 25 30
T
e
m
p
e
r
a
t
u
r
e
[
C
]
Height [m]
computed
measured
oxygen concentration [%]. in our case we use Cμ = 0. However. These equations are accompanied by the equation of state. Constant Cμ . The heat sink term is nonzero only in the edge computation cells and describes the energy exchange with the walls of the furnace via conduction and radiation Echar RT p where vol 2 m p is the particle's combustible mass. 2 Here.. qr = (c T ). particle diameter [mm].28 The Open Thermodynamics Journal. Tgas and Twall are the temperatures of the gas and walls respectively. where the dynamics of combustion of a single particle of the coal used in the boiler p=( 1) e 1 2 u . has to be chosen empirically for the particular problem. dt (1 + Dt N )2 where a a a where μ is the laminar viscosity. char . The heat transfer terms are computed as follows N = a1 di 2 YO 3 Tr 4 . R is universal gas constant and T p is the particle temperature. a9 are empirical and need to be determined experimentally for the particular type of coal.. as usual where A and B are constants dependent on the properties of the interface between the modeled region and its surroundings. the combustion model needs to be timeinvariant. Here c is socalled radiative conductivity and defined as The constants a1 . Volume 4 Straka and Beneˇ s gravity. as the flue gas can be considered as an optically thick matter.. Achar . DNt dm = . and slower combustion of the char. qc and qs are respectively heat transfer by radiation. Although the above equations can be fitted to experimental data by a careful choice of the a1 . Avol . to be used in any simulation. Echar . 3 R dmcoal = Avol m p vol YO vol exp 2 dt Achar m p char YO char exp 2 Evol RT p is the StefanBoltzmann where n is the refractive index. and to avoid the time dependence on the right hand side. elapsed time [s].. coal ignition temperature [o C]. energy. which is described by the Fourier law of heat conduction. m p is the mass of the combustible matter within the one coal particle. 2010. To keep the computational complexity low. a a T0 8 2 Tr 9 . Our approximation of the coal particle combustion rate has been based on the data in [11].. 2 D = a5 d Tr = T0 T0 a6 YO 7 . This article also contained an empirically postulated mathematical model of the particle combustion in the form of an ordinary differential equation . The rationale behind this choice is that the coal combustion consists of the faster combustion of the volatile matter. char are empirical constants. k the turbulent kinetic energy. a Tz where the symbols are as follows: m t d YO 2 burnt combustible matter [kg]. is the Rosselandmean extinction constant and R coefficient [2]. . where is the conductivity and T is temperature. environment temperature [o C]. Evol . The effective friction coefficient μeff is calculated from the turbulence model as μeff = μ + μt = μ + Cμ k2 has been measured. relative temperature. heat transfer by conduction. (4) xj t dt where h is specific enthalpy. we opted to use a simple approximative model consisting of two Arrhenian exponentials. The radiation heat transfer is fully described by an integraldifferential equation of radiation. and the turbulent energy dissipation rate. the above approximation of the radiation flux called the Rosseland radiation model can be applied. which is very computationally expensive to solve. vol . qs = A(Tgas 4 Twall ) B(Tgas 4 Twall ). like additional constants mentioned later in the description of the turbulence model. and heat source or sink.. ncoal is the density of coal particles (amount of particles in unit volume). Tr T0 Tz qc = ( T ). the model itself is questionable due to the time dependence of the right hand side of the differential equation. YO is the oxygen mass fraction... a9 constants. which appears to give satisfactory results. qr . for the transfer by radiation. Obviously.09 . is the Poisson constant and e is the gas specific . for the transfer by conduction. hcomb is the heat of combustion. c= 16n 2 T 3 . The last equation describes the conservation of energy dmp ( h) + ( u j h) = ncoal hcomb + qr + qc + qs .
and the term Di. (5) and the second one for the turbulent kinetic energy dissipation rate t ( )+ xj ( uj ) = xj simply consists of oxidation of atmospheric nitrogen at high temperature conditions. Right hand sides describe their spatial diffusion. Hence.1. due to an extremely high activation energy of nitrogen molecule. + xj xi k1+ k1 + k2 = 1. These are the only mechanisms involved. can be derived from the Reynolds averaging process and written in the terms of the fluctuating part of the velocity as Gk = uj jl O + N2 k1 N + NO N + O2 N + OH k2 O + NO H + NO k3 All these reactions are considered to be reversible. T means temperature in Kelvin. Given the fact that the turbulent diffusion generally predominates the laminar one. (6) Constants in this model have to be determined empirically. which describes the production of turbulence. this Nradical formation is the rate limiting factor for thermal NO production.m is difficult to determine. fluctuations u j and ui are unknown.3 . The real mechanism of coal flue gas production seems to be very complicated. First term corresponds to linear laminar diffusion. Volume 4 29 For the turbulence modeling. and the diffusion term in Eq.44 . SIMPLIFIED MODEL OF NO CHEMISTRY This model has been developed to approximately describe the amounts of NO emissions leaving a coal combustion furnace. oxygen radical [O] and hydroxyl radical [OH] must be known.7 .8 10 3 T exp 7. respectively. 2010. introduced by Zeldovich [3] and extended by Bowman [4] μ+ μt xj 2 + C1 k Gk C2 k .92 . 3. The latter is called Fuel NO and describes NO creation from nitrogen. which is chemically bounded in coal fuel. the second one to turbulent diffusion. we use the Boussinesq hypothesis that the Reynolds stress is proportional to the mean strain rate.1 10 7 exp 1. we use the standard k model.8 10 7 exp 1. Diffusion of the species consists of two processes — laminar and turbulent. = 1.7 10 8 exp 38370 T 425 T 4680 T 20820 T 450 T 24560 T where is the Reynolds stress tensor and Sij is mean strain = rate. the laminar diffusion can be usually ignored. concentrations of nitrogen radical [N]. In fact. C2 = 1. with Sij = 1 2 uj ui . The term Gk . It is useful to assume [N] to be in a quasisteady state according to its nearly immediate conservation after creation. The coefficient Sct is the turbulent Schmidt number and we put Sct = 0. In order to compute the NO concentration. Thermal NO Thermal NO generation mechanism attributes only at high temperature conditions and is represented by a set of three equations. so that just the most important phenomena and reaction paths were considered to provide maximum possibility of using this model in realtime control and operation systems.No Production in AIRStaged The Open Thermodynamics Journal. k = 1.m + μt Sct Yi . Fuel NO is usually the major source of NO emissions.8 10 8 exp 3. Rate constants were taken from [5] and are as follows xl jl = u j ul uj xl = μt S 2 . (1) can be written in the form = k2 + k3 = r Ji = = Di. which describes the evolution of turbulence using two equations — first one for the turbulent kinetic energy t ( k) + xj ( ku j ) = xj μ+ μt k k + Gk xj . However during practical computation. which is caused by a triple bond between two nitrogen atoms. Left hand sides of the equations describe passive advection of the respective quantities by the advection r velocity u . their production and dissipation.8 10 4 T exp 3. There are two major processes attributing to the total NO. The former is known as Thermal NO or Zeldovich and k3 = where “+” and “–” in superscript denotes forward and reverse rates. the NO formation rate can be stated as d[NO] = 2k1+ dt k1 k2 [NO]2 + k1+ [N 2 ]k2 [O 2 ] [O] [N 2 ] . 3. in our case we use the following values: C1 = 1. k [NO] 1+ + 1 + k2 [O 2 ] + k3 [OH] 1 .0 . although a few more could be considered (such as Prompt NO (Fenimore) or Nitrous oxide (N2O) intermediate mechanisms).
Z are other products we simply neglect and a is the oxygen reaction order Composition analysis show. and edge dualvolume approximation is used to approximate the second order derivatives respectively. Fuel itself is therefore a significant source of NO pollutants. To proceed further. it is presumed.2.1 10 3 Ka = 1 1.1 10 3 2 4. Different parametric studies should be carried out to find the best values of . [OH] = K 2 [O]1/2 [H 2 O]1/2 T Equilibrium constants K1 and K 2 are as follows 0. which usually are ammonia NH3 and hydrocyanide HCN. usually in amounts of tenths to units of percent by weight. Five overall reactions of either NO formation or depletion were incorporated in the combustion part of the numerical code. NO. In Fig.1 >. 3.9 ln XO XO 2 2 XO 2 1. When a coal particle is heated.129 10 2 exp T 4595 T NH 3 + NO R4 N2 + Z R4 = 1. This overall source term can be further used in transport equations. For left and right hand sides in Eqs.1 > . 2 and 3 suitable for specific type of coal.0 1012 XHCN XNO exp K1 = 36.11 10 2 0. HCN. it is possible to determine them from coal burnout rate. 7].11 10 2 Ka = 3. that nitrogenbased species are more or less present in coal. • 2 where k5 = 2.64 exp 3. HCN + NO R3 R3 = 3. XO 4.1 ln XO 2 XO 2 0.2 T [O] = K1 [O 2 ]1/2 T 1/2 . (5). K 2 = 2.35 0.0 T 30208.7 T where X* is the mole fraction. (3).2. • is distribution of char bounded nitrogen which 3 converts to NO . . • is distribution of char bounded nitrogen which 1 converts to HCN . . cs is the concentration of particles. single source terms have to be summarized.27 10 exp 17168.33 T is the rate constant. 2010.8 10 8 XNH 3 XNO exp 13593. that nitrogen compounds are distributed into volatiles is introduced to describe the and char. In order to evaluate overall NO source term. 1 + 2 + 3 = 1. (6). nitrogen transforms to pollutants via intermediates. • < 0. 2) one can see temperature and NO mass fractions profiles for cases indicated in Table 1. m is the total mass of nitrogen. Heterogeneous NO Reduction On Char Present char allows following adsorption process to occur Char + NO N tot N tot N2 + Z m N char = (1 ) m . < 0. RESULTS is distribution of char bounded nitrogen which converts to NH 3 . For detailed description of the solution procedure see [1].95 0.2.1. (2). ABET is the pore surface area and pNO is the partial pressure of NO. NO Sads is the NO source term.03Ka = 0.0 1010 XHCN XO 2 exp 33732. 1 . advection upstream splitting method (see [10]) is used to approximate fluxes in the FVM formulation.0 10 6 X NH 3 a XO 2 exp 16111. m N char N vol where is Levy [9] uses pore surface area to define NO source term NO Sads = k5 cs ABET M NO pNO . (4). It is assumed. 5. Volume 4 Straka and Beneˇ s Concentrations of O and OH can be profitably described by following partial equilibrium approach [6. NH 3 Reactions According to [8]. A parameter distribution of the coalbounded nitrogen between the volatiles and char part of the coal particle. When the coal is heated.03Ka = 0 3. In Table 1 are values of various parameters used in simulation and summarized numerical values of air distribution among the burners. • is amount of volatile bounded nitrogen which converts to HCN . (1).5 T Computational results obtained for different excess air values for the burners are summarized in Figs. In Figs. Fuel NO 27123 . finite volume method is used.2. we have following reactions and its formation rates HCN + O 2 R1 NO + Z a R1 = 1. N mvol = N mtot . (1. we must define four parameters to describe complex partitioning of the fuel bound nitrogen.57 . NH 3 + O 2 R2 NO + Z N2 + Z R2 = 4. that nitrogen from both char and volatiles transforms to intermediate species quickly and totally. 4. As for HCN and NH 3 source terms. these species are transformed into certain intermediates and then into NO. is the As already mentioned.30 The Open Thermodynamics Journal. (14). 3 the mass of bounded nitrogen in volatiles and m mass of bounded nitrogen in char. NUMERICAL ALGORITHM For numerical solution of the equations.
with maximum concentrations in the lower region of the boiler. (1). 2) show the opposite ordering of the combustion air and the different NO production pattern.e. Airstaging 1 – the combustion air is staged as indicated in Table 1 . 50% to the first(bottom) row of burners. 20% to the second and third row of burners and 10% to the fourth(top) row of burners.e. with lower velocities in the bottom region of the boiler. Airstaging 2 – in this case the combustion air is staged again. Mean values together with standard deviations of the temperature are depicted in this figure. The Airstaging 2 case. 1). (3) is comparison of the NO mass fractions for four cases of air staging. but due to higher velocities. but with reverse ordering i. which is installed in heating facility in Otrokovice (Czech Republic). 20% to each row.No Production in AIRStaged The Open Thermodynamics Journal. the maximum temperature region is shifted upwards. Over Fire Air (OFA) – the combustion air is distributed equally among four burner rows and one row of OFA slots i. (4) are temporal profiles of concentrations of NO in flue gas. one reason is creation of oxygenrich and oxygenlean zones and second reason is the different location of the maximal temperature region. 1. 2010. where the oxygenlean zone is so less ThermalNO is produced for the case of Airstaging 1. Profiles of temperature and NO mass fractions for (from top to bottom) Normal operation. In Fig. Volume 4 31 Fig. shows the location of the maximum temperatures in the lower region and due to oxygenlean zone in that region. Here the oxygenrich zone is located in the bottom section of the boiler. In the case of Normal operation (see Fig. (5) is validation of the computed temperature field with boiler K5TOT. we . 10% to the first row of burners and 50% to the fourth row of burners. Airstaging 1. with air distributed equally. great amount of NO is produced by both the Thermal and the Fuel mechanism. Airstaging 1 and Airstaging 2 cases (see Figs. Further decrease in NO concentration is mainly due to the diffusion and mixing with the air from burners. Those temperatures were measured at three different locations using optical pyrometers and thermocouples [11]. In Fig. 6. DISCUSSION One can see that the airstaging has essential influence on the NO production. The studies cases are named Normal operation – this means that the whole combustion air is distributed equally among burners. In each case the fuel distribution is 25% to every row of burners.
(2). Fig. . Over fire air. Profiles of NO mass fractions for (from left to right) Normal operation. Airstaging 2. Profiles of temperature and NO mass fractions for (from top to bottom) Airstaging 2. Volume 4 Straka and Beneˇ s Fig. 2010. (3). Airstaging 1.32 The Open Thermodynamics Journal. Over fire air.
0 Unit W m2 K4 J kg1 K1 kg mol1 kg m3 m J kg1 K1 m2 kg s1 Pa o Parameter StefanBoltzmann constant refractive index gas heat capacity molar mass of gas raw coal density coal particle radius coal heat capacity boiler crosssection coal feeding rate initial pressure initial temperature mixture temperature at the inlet pressure at the inlet pressure at the outlet excess air coefficient temperature of the wall width of the side wall first burners row position second burners row position cV M rcoal 0. 2010.0 2.0 · 105 800 400 1. Airstaging 1.0 16.2 · 103 2048 49.No Production in AIRStaged 700 The Open Thermodynamics Journal.3 cV S & mcoal pini Tini Tmix C C o pin pout Pa Pa o Twall a. Volume 4 33 600 Normal operation Air staging 1 Air staging 2 OFA 500 NO concentration [mg/m3] 400 300 200 100 0 0 5 10 15 Time [s] 20 25 30 35 Fig. Total Air is Distributed among Rows of Burners.84 C m m m B2 .4 1.7 1037. Simulation Parameters and Fuel and Air Setup for Studied Cases. Table 1. b B1 300 7.74 3. Flue gas NO concentrations for Normal operation. Over fire air. Airstaging 2.678 · 108 n 1.0 · 105 pin . Percentage Distribution is Indicated in the Table Symbol Value 5.100 1.0 0.02896 1345. (4).
5 30 45. Other important information is the concentrations of NO in a flue gas leaving the boiler.25 30.16 9. (5).69 1.25 0.93 6. but very good accuracy is achieved for the burners and outlet regions of the boiler. 2010. Contd….42 0. Inaccuracy for the second spot is probably caused by the simplified radiation model.0 0. Straka and Beneˇ s Parameter third burners row position fourth burners row position number of burners in a row burner diameter OFA row position OFA slot diameter pore surface area coal nitrogen distribution Symbol Value 4. The last part is the validation of the temperature profile (as the temperature has the greatest influence on the NO production) with the real values measured in the heating facility in Otrokovice.5 Air percentage for burners B3 25 20 20 20 % % % % % % % B4 25 10 50 20 OFA 0 0 0 20 can see less NO production there. Volume 4 Table 1.5 0. . thus further refinement of the radiation heat transfer is needed.wall heat transfer coef.5 0.34 The Open Thermodynamics Journal. comparison of the temporal change in NO concentrations in the flue gas for the cases considered is in Fig. The last case of Over Fire Air technique shows that the maximum NO level is located in the near burners region.8 8. results are depicted in Fig.72 3. (conduction) gas . (3).4 25000 0. Here one can see that the OFA technique gives the best result considering the NO concentrations.6 · 108 2 3 gas . (4). (radiation) coal analysis: ash water carbon hydrogen sulfur oxygen nitrogen Case B1 Normal operation Airstaging 1 Airstaging 2 Over fire air 25 50 10 20 B2 25 20 20 20 A W C H S O N 9.wall heat transfer coef.78 0.03 4 0.5 Unit m m m m m m W m2 K1 W m2 K4 2 B3 B4 NB DB BOFA DOFA ABET 1 0. but less combustion air creates the oxygenlean (and the nitrogenlean zone for the Thermal mechanism) zone and an unburned char is able to remove NO by the adsorption process. The overall comparison of NO concentrations for all cases are depicted in Fig.
Emissions from Continuous Combustion Systems. Radiative Heat Transfer. A. 2000. D. Journal of Computational Physics. Just. 577628. Esser. Cox. Cobos.. National Institute of Standards and Technology. s This is an open access article licensed under the terms of the Creative Commons Attribution NonCommercial License (http://creativecommons. M. “NO/Char reactions at pulverized coal flame conditions”. V. Kerr. 107. 2009 Accepted: November 11. “Description of a model of a combustion chamber and a design of a model of heat exchanging surfaces” (in Czech). Technical Report 96107/3. G. M.nist. De Soete. 1981. 7. 21. NIST. Seery. Volume 4 35 computed measured 1300 1200 1100 Temperature [C] 1000 900 800 700 600 500 400 0 5 10 15 Height [m] 20 25 30 Fig. 1993. Makovi ka. F. mainly the radiation heat transfer modeling. Presented results with emphasis on the NO production are adequate when compared with real life devices. Still there are places where the enhance the model is needed. Sarofim. “Chemical Kinetics Database on the Web”.0/) which permits unrestricted. Bowman. C.org/licenses/bync/3. 2002. M. and more accurate coal particle combustion model. Plenum Press: New York. Comparison of computed and measured temperatures (from [11]) for the K5TOT boiler installed in Otrokovice . pp. Levy. J. B ezinová. Frank. 1993. C. 2001. Modest. 1992. J. “A New Flux Splitting Scheme”. J.o. I. [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] Received: October 07. Pilling. “Overall reaction rates of NO and N formation from fuel nitrogen”. vol. Proceedings of Contributed Papers. Liou. “Mathematical modelling of steam and flue gas flow in a heat exchanger of a steam boiler”. Havlena. G. Inc. Th. W. vol. In Proceedings of the 15th Symposium (International) on Combustion. McGrawHill. 111120. which can improve the accuracy of the temperature profile above burners. Available from: http://www. 411734. 171178. K. Tokyo. pp. and M. REFERENCES [1] J. T.Czech Republic. C. L. in ALGORITMY 2002. Walker. Troe. J. R. 1972. distribution and reproduction in any medium. C. P. Acta Physicochimica. J. Baulch. L. (5). provided the work is properly cited. Chen. 2010. Warnatz. M. “The oxidation of nitrogen in combustion and explosion”.r.No Production in AIRStaged 1400 The Open Thermodynamics Journal. 2009 © Straka and Beneˇ . Journal of Physical and Chemical Reference Data. S. CONCLUSION A numerical model of pulverized coal fired boiler with staged air design has been presented. Beer. Licensee Bentham Open.. NOx Formation in High Temperature Processes. vol. R. Z. 1946. J.kinetics. 2009 Revised: October 26. “Evaluated Kinetic Data for Combustion Modelling”. Bene . The model can be used as airstage designing tool for coal fired boilers with wall burners. J. Youth and Sport. We developed this model keeping in mind the balance between the computation complexity and the precision of the simulation. University of Stuttgart: Germany. Warnatz. B. 21. University of Waterloo: Canada. pp. In Proceedings of the 18th Symposium (International) on Combustion. pp. Japan. pp. D. ACKNOWLEDGMENTS This work has been supported by the project "Applied Mathematics in Physical and Technical Sciences" MSM 6840770010 and "Jind ich Ne as Center for Mathematical Modelling" of the Czech Ministry of Education. A. MORE s. J. M. Ma ek . Steffen Jr. A. J. 1997. . 10931102. 2329. Comparison of the measured and computed temperature profiles showed sufficient accuracy for the burner region as well as for the flue gas outlet area.gov J. as well as model behavior for different cases. Zeldovich.. pp. This was also the reason why we had used simplified reaction kinetics instead of full reaction set for the production of NO and simple kinetic approach of the coal particle combustion. noncommercial use. Slovak Technical University: Bratislava. 1975. F.
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.