close

Вход

Забыли?

вход по аккаунту

?

Numerical Simulation of NOx Emission from Char Combustion with Detailed Gas Phase Mechanisms.

код для вставкиСкачать
Dev. Chem. Eng. Mineral Process., 7(5/6), pp.513-524, 1999.
Numerical Simulation of NOx Emission
from Char Combustion with Detailed Gas
Phase Mechanisms
B. Feng, Y.B. Zhou, X.F. Shi, C.G. Zheng
National Laboratory of Coal Combustion, Huazhong University
of Science and Technology, Wuhan 430074, P.R. China
and K.Okazaki
Research Centerfor Carbon Recycling and Utilization, Tokyo Institute
of Technology, Ohokuyama, Meguro-ku 2-12-1, Tokyo 152, Japan
The present paper proposed a mathematical model for the simulation of NOx
emissionfiom char combustion. The one-dimensional two-phase model developed by
Kuo [9] was simplij?ed, and a detailed gas-phase reaction scheme was included,
based on the work of Miller and Bowman 1191 with updated reaction rates of some
important reactions. The scheme includes 289 reactions and 51 species. Two global
models for char oxidation were used and compared. The heterogeneousformation of
NOx fiom char oxidation, the reaction of NOx destruction on the char surface, and
the reaction of NO destruction by CO catalysed by char were taken into account
during char combustion. Our work showed that NOx destruction is sensitive to the
char oxidation rate. The heterogeneous reactions for both char oxidation and for
NOx formation and destruction requirefirrther investigation.
Introduction
Nitric oxide emission from coal or char combustion has been modeled by many
researchers [l-51. The models may be divided into two groups. First, models with
detailed fluid dynamics for the simulation of the practical combustion systems [ 1-31.
These two-phase models usually consist of some sub-models, such as a turbulence
model, heat transfer model, NOx chemistry model, etc. The models assume that the
NO concentration depends mainly upon the mixing effect, or diffusion, thus paying
special attention to the fluid dynamics rather than the NOx chemistry, and a relatively
simple NOx chemistry model is used. This group of models can provide some useful
513
B. Feng et al.
dormation on the effect of the design or operating factors on the NOx emission.
However they are not able to provide any information on the mechamsms of the NOx
formation and destruction, and the accuracy of the predictions is dependent strongly
on the accuracy of the reaction rates.
Another lund of model, unlike the above models, pay attention to the NO
chemistq, not the fluid dynamics. Peck et al. [4] developed a model including
detailed gas reactions and taking gas-solid reactions into account by incorporating
collision frequencies of some radicals with the char particle. Gustavsson et al. [6]
followed the method of Peck et al. [I] and simulated the
N20
reduction by
Afterburning. Kramlich et al. [ 5 ] included the contribution of char-N by assuming that
a part of char-N is converted to HCN, which has been suspected by Goel et al. [7]
recently. Through the models with very simple gas-solid reactions, the authors
concluded that the heterogeneous reactions were not important. However, Visna and
Stanmore [8] has found that the gas-solid reactions may be important.
The present paper proposes a model different from the above models. It is a
simphfied two phase model of Kuo [9]. The detailed lunetic gas phase reactions and
gas-solid reactions are included. Since as few as possible assumptions are adopted,
the accuracy of the prediction by this model depends mainly on the kinetic model.
Therefore hmodel can be served as a tool to evaluate the kinetic model, especially
the heterogeneous model, because the gas phase kinetic has been validated. The
model is solved by the public software DVODE, which is found robust for the M
problems. The present paper studied the sensitivity of the reaction rates of the gassolid reactions to the NOx emission.
MODEL DESCRIPTION
Mathematical model
The physical problem considered is an open vessel with coal fed from one end. The
coal particles are burned while moving in the vessel. This is a two-phase, multicomponent, one dimensional model, Kuo [9]. The model contains the conservation
equations of species, momentum, and energy for both gas and particle phase.
The fractional porosity, 0,of the material is defined as
514
Numerical simulation of NOx emissionfrom char combustion
+l---
n p m p - volumeof void
total volume
f P
where np is the number density of the particles, mp is the mass of each particle, and
pp is the mass density of particles. In addition, we define A, as:
related to Sb,the burning surface of the spherical particle whose radms is rp. Assume
that the particles have constant density during combustion, AEcan be expressed as
3
A, = - ( 1 - 4 3 ( 1 I;,
2
-
Using these definitions, the one-dimensional fluid equations for the density of the
gas and particulate, the equations for momentum conservation for the gas and the
particles, and the equations for energy conservation for the gas and the particles, are
written as in Kuo [91.
Some simplifications are made as follows:
1. The temperatures of gas and char particles are known
01:set
as parameters.
Although the gas and the particle energy equations can be solved for the gas and the
particle temperatures, it is suggested that the measured temperatures be used because
the heat transfer and heat losses are not well known and the predicted temperatures
are usually not accurate. This eliminates the necessity of the solution of the gas and
the particle energy equations.
2. The char particles have the same velocity as the gas. It is usually reasonable to
assume the velocity slip is negligible in the char burnout process. Thls eliminates the
solution of the particle momentum equation. Also the drag force between the gas and
the particle due to the velocity slip is zero.
3. The stresses are ignored, and the char particles have the &om
dmneter.
The equations are reduced to:
515
B. Feng et al.
Add the conservation equation of species:
Here, wk is the production rate of species k
Chemical kinetic model
To complete the problem definition, wk and rb have to be supplied. The former
depends on both the homogeneous lunetics and the heterogeneous reactions while the
latter depends on the heterogeneous reactions.
Gas phase mechanism
The gas phase kinetic model is mamly that of Miller and Bowman[l9] for the
carbon hydrogen oxidation with NOx formation and destruction submodel. However,
the submodel CMIOIN is replaced by that of Lindstedt et al. [101. They compiled the
latest reaction rates of some important reactions for NOx destruction in the CI& flame
and got good results using the mechanism.
As discussed in Feng et al. [ll], the mechanism applied the newly determined
reaction rates of the reaction of HCCO and NO which were found to be the most
important reaction for NO destruction. The reaction rates of some other important
reactions were also measured by some researchers and the rate constants were used.
The mechanism consists of 289 reactions and 5 1 species.
Char oxidation model
There are two semi-global char oxidation model available, and was used and
compared by H.K.Chelliah [12]. One is employed by Bradley et al. [13] for modeling
oxidation of nonporous graphte particles. Another one is developed by Makino et al.
[ 141 for modeling oxidation of porous graphite particles. The reaction rates of Makino
et al. are much faster that those of Bradley, 100 times faster for the reaction carbon
gasification by C 0 2 and 550 times faster for the reaction of carbon oxidation by 0..
516
Numerical simulation of NOx emissionfrom char combustion
The present study first modeled the oxidation of carbon using the two models
respectively. then different rates of NO destruction by char was included and
compared.
NO creation and destructionbv char
The char particle produces and destructs NO during char burnout by the following
reactions:
1
N + - 0 , +NO
2 -
1
(?+NO+-N,+CO
(9)
1
2
(10)
2
-
C+NO+CO+-N2+2C0
It is assumed that NO is produced directly from char oxidation with a rate in
proportion to the char burning rate.
RN
= qrb
(11)
where q is the molar fraction of N to C in char.
The burning rate of the char particles is the total amount of the reaction rates of six
reactions, i.e. C+02, C+C02, C+H20, C+OH, C+H and C+O:
rb=(s1+s2+sJ+s4+ss+s6)/Mc
(12)
where Mc is the atom weight of carbon.
NO is destructed by the reaction with char particles and the reaction of NO with
CO catalyzed by char surface. The reaction rate of NO with char has been measured
by some researchers. Visna and Stanniore [ S ] compared the reaction rates by the
different authors and found that Chan’s rate was much faster than those of the others.
They recommended the rate by Chan et al. [ 151 because the prediction using this data
was more reasonable. Visna and Stanmore found that the rates by Levy et al. [ 161, De
Soete [ 171 and Song et al. [ lS] were close to each other, but the rates by Levy et al.
[ 161 was a little slower. Ths study used the reaction rate of Song et al. and Levy et al.
as the reaction rate of (9), DN1, while using that of Chan et al. as that of (9) and (10).
DN2.
Chan et al. have also derived a hgmuir-Hinshelwood model for the reaction of
517
B. Feng ef al.
NO with carbon in the presence of CO. The reaction rate was used to study the effect
of CO or COz on the destruction of NO. Another method was used by IHI to give the
reaction rate of (10). The rate was estimated to be 12Ox(CO) times of the rate of (lj),
where x(C0) is the mole fraction of CO. The present study also used this rate and
compared with Chan’s rate.
;
I
;
rLI
DN = A, exp(--I-)mchPNO
RTP
32700
D N ~= 7.5x lo6 exp(--)m,P,,
RTP
DN2 = k,PNo
k A 0 +k3
klPNO
+ k2PC0
+ k3
where,
kmol
)
m2satm
95600
hol
exp(--),(
k2 = 7.5 x
1
Tp
m2satm
20100
kmoz
k3 = 1.5 xlO-’ exp(--),(
1
Tp
m’satm
13100
k, = 2.1 x lo-’ exp(--),(
Tp
Production rates of some suecies
The production rate of a species is the total amount of the production rate of the
species from each reaction involving the species. The production rates of the species
from the gas phase reactions are calculated using the CHEMKM software. For some
species, the production rates have to be revised because they also involve in the
heterogeneous reactions according to:
518
Numerical simulation of NOx emissionfrom char combustion
pHlo
PH,O,g
(19)
-'5
mo, = Po2,&?
- s , - RN
+ s2 + s, + sj +s,+D,, + DN2
mco = mcoXg
+ S,
mCH,
mN2
mNO
= mCH,
-
- mNz,g
.g
+ '3
+D
~ -k
l DN2
= mNO,g + Rhr- D , , - DAr2
(20)
(2 1)
(23)
(24)
(25)
where wLs is the production rate of species i by the gas phase reactions whch is
calculated by CHEMKIN.
Numerical method
The equations (4)-(7) were solved simultaneously and the variations of the volume
fraction, gas velocity, gas density and species mole fractions with the distance of the
reactor were obtained. A robust software, DVODE, was used to solve the typical shf€
o r d m q differential equations. As stated above, CHEMKIN was used to handle the
gas phase scheme and to calculate the production rates of the species. The
temperatures of gas and particle were set to be 1473 K and 1573 K in most of the
cases, but they also were changed to study the temperature effect. The code also
allows the temperature to be a function of the distance or time.
RESULTS AND DISCUSSION
Figure 1 shows a comparison of the experimental CO molar fractions with the
calculated ones using Bradley's model and Maluno's model respectively. The
experimental results were obtained by burning char particles in a drop tube reactor.
The char particles were derived from several Chmese coals holding at the temperature
of 1173K in a N2 stream for half an hour. The CO, 02,C 0 2 and NOx concentrations
were measured for comparison with the predicted ones.
The results for char A show that Bradley's model underestimates the CO
519
B. Feng etal.
productions whereas Maluno’s model over-predicts. The results for the the other chars
are similar. It indicates that the reaction rates of char oxidation in Bradley’s model are
slow while those in Makino’s model are too fast. What is interested is that whether
the NOx emission is mfluenced by the reaction rates of char oxidation or not. This is
studied by calculating the NOx concentration using the two models.
12
h
5
8
2
4
10 -
A
8
.Bradley’s model
A Makino‘s model
6 - A
.-0
- 0 4
2
&
2
0
0
1
2
3
4
Experimental CO (%)
Figure 1. A comparison of the experimental CO molarjPactions for char A with the
predicted ones by BradIey et al. ’s model and by Makino ‘s model.
Figure 2 shows the effect of the particle density on the NOx emission by using
Chan’s reaction rate for NOx destruction and Bradley or Makino’s char oxidation
model. The NO behaviors are quite different. According to Bradley’s model, nitric
oxide is destroyed slowly in the particle densities of llmgA and 33mgA. NO is
reduced relatively rapidly in the particle density of SrngA, which can be explained by
the faster homogeneous destruction rate in the case of fuel rich. About half of the NO
is destructed. However, NO is shown to be reduced much faster in Makino’s model.
More than half of the NO is reduced in 0.03 sec.. The NO chemistry is the same in
both of the models. This means that the NO behavior is dependent on the char
oxidation reaction rates, or the NO conversion from char-N depends on the coal type.
520
Numerical simulation of NOx emissionfrom char combustion
It may be due to the production of large amount of active radicals from char
combustion.
The present study also found that the reaction rate of C+H and C+OH is very fast,
and usually faster than the reaction rate of C+02.C&, CO and H2 are assumed to be
produced from the two reactions, which are all important for the destruction of NO.
More work on the reaction rates of C+H, C+O, C+OH etc. is needed.
I
n
E
1100
2 1000
v
=0
'=
c
3
L.
0
2
s
0
900
800
700
600
500
400
0
0.1
0.2
0.3
0.4
0.5
Time (sec.)
Figure 2 Efect of the particle density on the NOx emission from char combustion
(calculated). Conditions: Tg 1473K Tp l573K,
02,
21%,
N2
78%, CH4 I%, NO
IOOOppm, Residence time OSsec.. a, b, c - Bradley's model, A40 I h g A , 33mgA and
SSmg/l respectively. d, e, f - Makino's model, MO IlmgA, 33mg/l and SSmg/l
respectivelv
The rapid destruction of NO may be due to: a, the rapid consumption of oxygen by
char oxidation reactions. The O2 concentration profiles calculated using the two
models are presented in Figure 3. At the particle density of llmg/l, The O2
concentration decreases to 3% in 0.03 seconds by Maluno's model, and 16% in 0.5
seconds by Bradley's model. The low oxygen concentration favorites the destruction
of NO. b, the lugh concentration of CO, thus the fast reaction rate of NO destruction
by the reaction (10). The CO profiles are shown in Figure 4. The lughest
concentration by Bradley's model is 0.6%, which is much lower than the prediction
521
B. Feng et al.
by Makino’s model, 10%. Experiments showed that 3 - 5% of CO could be produced
at the particle density of 55mg/l. The difference between the NO profiles at 33mg/l
y come from the reaction of NO with char in the presence of
and 55mg/l should d
CO because the oxygen concentration in the two cases are close to each other.
a
o
0.005 0.01 0.015 0.02 0.025 0.03
Time (sec.)
Figure 3 Efect of the particle density on the O2 concentration (calculated).
Conditions: Tg 1473K, Tp 1573K,
02,
21%, Nz 78%, CH, 1%, NO lOOOppm,
Residence time 0.03sec.. a, b, c, d, e,f- the same as in Figure 2.
Conclusions
A mathematical model is proposed in the present paper for simulation of the N o s
emission from char combustion. The char oxidation reactions have apparent effect on
the NOx chemistry during combustion. It suggests that errors may occur when using
inaccurate char oxidation rates, thus the measured rates are preferred.
522
Numerical simulation of NOx emission from char combustion
0.008
a
0.004
=
__---
0.002
I
0.1
0
H--
0.2
0.3
0.4
0.5
0.1
$
0.08
5
0.06
5
0.04
E
8 0.020
0
0.005
0.01
0.015
0.02
0.025
0.03
Time (sec.)
Figure I Effect of the particle density on the CO concentration (calculated).
Conditions: Tg 1473K, Tp 1573K, 02,21%, NZ 78%, CH4 I%, NO IOOOppm,
Residence time 0.5sec.. a, b, c, d, e,f - the same as in Figure 2.
Nomenclature
External area of particles
m2/m3
Drag due to the relative velocity
N
Molecular weight of carbon
kg/mol
Mass of each particle
Number density of particles
kg
1/m3
Total pressure of gas
atm
Molar fraction of N to C in c h q
Burning rate of char particles
kg/(m2-s)
Release rate of N in char
kg/(m2-s)
Radius of particles
m
Burning surface of particle
m2
Reaction rate of C+OH+CO+H
kg/(m2s>
Reaction rate of C+O+CO
kg/(m2
523
B. Feng et al.
s3
Reaction rate of C+4H-CH4
kg/(m2s)
s4
Reaction rate of C+CO,+2CO
kg/(m2 s)
s5
Reaction rate of C+H,O+CO+H,
kg/(m2s)
S6
Reaction rate of C+ %O,+CQ
kg/(m2s)
k
3
Velocity of gas
d S
Wk
Production rate of species k
molls
YL
Molar fraction of species k
Q
Fractional porosity
PP
Mass density of particles
kg/m3
Subscripts
0
initial
P
particle
g
gas
References
I . Fiveland, A and Latham, C.E. 1993. Combust. Sci. Tech., 93,53-72.
2. Coimbra, C.F.M., Azevedo, J.L.T. and Carvalho, M.G. 1994. Fuel, 73(7), 1128-1134.
Int. Cod. Combustion Technologies for a Clean
3. Antifora, A, Sala, M. and Xgnenano, L. 1993.
Environment, paper 19.2, Lisboa.
4. Peck, R.E., Glarborg, P., and Johnsson, J.E. 1991. Combust. Sci. Tech., 76,81-109.
5. Kramlich, J.C., Cole, J.A, McCarthy, J.M., Lanier, S.W. and McSorley, J.A 1989. Combust. Flame,
77,375-384.
6. Gustavsson, L., Glarborg, P., and Lecher, B. 1996. Combust Flame, 106,345-358.
7. Goel, G., Zhang, B., and Sarofim, AF. 1996. Combust. Flame, 104,213-217.
8. Visona, S.P., and Stanmore, B.R. 1996. Combust. Flame, 105,92-103.
9. Kuo, K. 1986. Principles of combustion, Wiley, New Yo&.
10. Lmdstedt, R.P.,Bowman, C.T. andSelim, M.A 1995. Combust. Sci.Tech, 108,231-254.
11. Feng, B., Ando, T. and Okazaki, K. 1996. Proceedings of the Thirty-Fourth Japanese Symp. On
Combustion, Japanese Society of Combustion, Hiroshima, Nov. 27-29,468470.
12. Chelliah, H.K. 1996. Combust. Flame, 104,81-94.
13. Bradley, D., Dixon-Lewis, G., El-Din Habik, S., and Mushi, E.M.J. 1984. Twentieth Symp. (Int) on
Combustion, The Combustion Institute,Pittsburgh, 93 1-940.
14. Makino, A., Araki, N., and Mihara, Y. 1994. Combust. Flame, 96,261-274.
15. Chan, L.K., Sarofim, AF., and Beer, J.M. 1983. Combust Flame, 52,3745.
16. Levy, J.M., Chan, L.K., Sarofim, AF., and Beer, J.M. 1981. Eighteenth Symp. ( I d ) on Combustion,
The Combustion Institute, Pittsburg, 111-120.
17. De Soete, G.E. 1990 . Twenty-third Symp. (Int.) on Combustion, The Combustion Institute, 12571261.
Pohl, J.H., Beer, J.M., and Sarofim, A.F. 1982. Combust. Sci. Tech., 28,31-39.
18. Song, Y.H.,
19. Miller, J.A and Bowman, C.T. 1989. Prog. Energy Combust. Sci., 22,275-332.
524
Документ
Категория
Без категории
Просмотров
1
Размер файла
456 Кб
Теги
nox, simulation, mechanism, numerical, emissions, detailed, gas, combustion, chars, phase
1/--страниц
Пожаловаться на содержимое документа