Professional Documents
Culture Documents
327-337, 1998
0 1998 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: SOO16-2361(97)00211-l 0016-2361/98 $19.00+0.00
ELSEVIER
A comprehensive model is developed for the combustion of coal in a circulating fluidized bed combustor (CFBC).
The proposed model integrates hydrodynamic parameters, reaction model and kinetic subroutines necessary to
simulate coal combustion in a CFBC. Kinetic expressions were developed for the char combustion rates and the
SO2 absorption in the bed using data from the literature. The reaction model, which considers only the important
steps of coal combustion, was simulated using four ASPEN PLUS reactor models and several subroutines. The
developed subroutines were then nested in the ASPEN PLUS input file, so that the CF’EK may be represented. The
validity of the model was demonstrated using 14 different sets of operating conditions for the CANMET
0.8 MWth CFBC pilot plant. 0 1998 Elsevier Science Ltd. All rights reserved.
(Keywords: ASPEN PLUS; CFBC; circulating fluidized bed reactors; fluidized bed combustion)
INTRODUCTION model which considers the riser as a plug flow reactor for
the gas phase and a CSTR reactor for the solid phase. The
Circulating fluidized bed combustors (CFBCs) are consid-
model also considers the feed particle size distribution and
ered as an improvement over the traditional methods
the attrition phenomena. Using a lumped-modeling
associated with coal combustion. The CFBC exhibits
approach, Arena et aL5 introduced the means for predictive
several advantages over conventional coal combustion
calculation by dividing the CFBC riser into four blocks,
methods, especially when high sulfur coal is used’.
each corresponding to a separate reactor. Three of these
Operation of CFBCs at industrial levels has confirmed
blocks related to the CFBC riser. The hydrodynamic
many advantages that include fuel flexibility, broad turn-
parameters were considered uniform within each section
down ratio, high combustion efficiency, low NO, emissions
and were used in various kinetic models to predict char
and high sulfur capture efficiency. These characteristics
conversion. Wong6 proposed a model for the hydrody-
assure an ever-increasing number of successful commer-
namics of CFBC risers to characterize the effect of the
cializations of CFBC in power generation applications.
internal flow structure within the riser, the particle size
Although CFBC technology is becoming more common
distribution and the operating conditions on CFBC behav-
from these commercial applications, there are some
ior. To estimate the axial voidage profile, a core-annulus
significant uncertainties in predicting their performance in
model was developed. The predictive hydrodynamic model
large-scale systems.
was then applied to a CFBC design. A comprehensive
Technical knowledge about design and operation of
review of relevant work on the hydrodynamics of circulat-
CFBC is widely available for pilot plant and large scale
ing fluidized bed risers is presented by Berruti et aL7
units. However, little has been done in the field of
Moreover, Senior’ conducted some theoretical and
mathematical modeling and simulation of combustion in
experimental investigations to improve the understanding
CFBCs. This might be attributed to the fact that the
of the fluid and particle mechanics in the CFBC riser and to
combustion process occurring in a CFBC involves complex
develop mathematical models to represent riser suspension
phenomena including chemical reactions, heat and mass
transfer, particle size reduction due to combustion, attrition, flows. On the other hand, a few CFBC modeling efforts have
fragmentation and other mechanisms, gas and solid flow been based on extension of bubbling AFBC hydrodynamic
structure, etc. Weiss et al.* introduced a CFBC model by concepts’-’ ‘.
Beyond those mentioned above, some modeling work
dividing it into 11 blocks, each corresponding to a CSTR
reactor for both gas and solid phase. Five of these blocks have been developed using ASPEN (advanced system for
related to the CFBC riser. Basu et aL3 developed a CFBC process engineering). ASPEN was developed at the
model in which a plug flow regime for both the gas and Massachusetts Institute of Technology (MIT) under a
solids is assumed. Lee and Hypanen4 presented a CFBC United States Department of Energy project to simulate
coal conversion processes. It has now become a powerful
tool for engineers to model chemical, power generation and
other processes. The work of Young’* entails the modeling
*Corresponding author and simulation of AFBC using ASPEN. Herein, the ‘black
box’ approach with one ASPEN PLUS stoichiometric general hypotheses of the hydrodynamic model along with
reactor was used to calculate the mass balances based on the modeling procedure are presented below.
given combustion and sulfur capture efficiencies. CFBC
simulation work was also initiated at CERCHAR13 to General hypotheses of the hydrodynamic model. For
provide the technical information required for the evalua- steady state conditions, the assumptions regarding the
tion and optimization of CFBCs under steady-state con- hydrodynamic model are the following:
dition in power generation applications. The study of
(1) The CFBC is naturally divided into two hydrody-
pollutant emissions such as SO*, NO, and N20, as well as
namic regions:
the ash composition leaving the CFBC was not included in
this work; instead several ASPEN PLUS user subroutines (i) a lower region-turbulent fluidized bed (dense bed);
were used in order to study the hydrodynamic, combustion (ii) an upper region (dilute bed).
and heat transfer phenomena in the bed. The approach used
at CERCHAR is similar to that of Young’s, but was The boundary between the two regions is defined by the
extended to cover CFBCs. Combustion Engineering Inc.14 height of the secondary air injection point.
also used ASPEN in modeling a Lurgi circulating fluid bed. (2) There is perfect mixing of solids (individual ash, char
The approach, similar to that of Young’s’* used here, has a particles and sorbents) in the lower region and in each
low level of complexity since the goal was the calculation of zone of the upper region17. This assumption is justified by
the mass and energy balances for the CFBC. Up to now, the high internal and external recirculation of solids in the
modeling of CFBCs using a process simulator (such as bed.
ASPEN PLUS) has been limited to simple mass and energy (3) Plug flow regime for gas is assumed in the bed. This is
balances, without predictive capabilities. consistent with the results of gas backmixing ex eriments
ASPEN PLUS is widely accepted in the chemical in the CFBC risers as reported in the literature p7.
industry as a design tool because of its ability to simulate (4) The gas velocity throughout the bed is uniform and
a variety of steady-state processes ranging from single unit constant for each region of the bed.
operation to complex processes involving many units. (5) For a given superficial gas velocity, the mean voidage
Consequently, ASPEN PLUS was chosen as a framework in the lower region of the CFBC is constant. This assump-
for the development of a CFBC process simulation. Since tion is justified by the results of experiments of
there is no CFBC model provided by ASPEN PLUS, we Chehbouni et al. l8 for group B particles considering the
must develop our own using the tools offered by ASPEN lower region to be operated under the turbulent fluidiza-
PLus15,16. In addition to its conventional reactor models, tion condition.
ASPEN PLUS has the flexibility to allow the insertion of (6) In the upper region of the CFBC, the voidage
Fortran blocks and user kinetic subroutines into the decreases with the vertical position along the riser.
simulation.
In this work, several ASPEN PLUS reactor models Modeling procedure. The model considers that the
interact with their corresponding user-written kinetic CFBC is divided into two regions: a dense lower region
subroutines to perform calculation during the simulation. with a constant suspension density (turbulent fluidized
This flexible structure of ASPEN PLUS permits handling of bed) and a more dilute upper region with a decaying suspen-
complex processes, such as those occurring in a CFBC. sion density with height. Detail related to the gas-solid
Hence, an attempt is made to develop a model which structures chosen to represent two regions of the riser are
includes several features that were neglected or simplified in given below:
the previous studies as outlined above, in order to produce a
predictive tool. This paper presents the details of the
modeling approaches taken to obtain a process simulation Lower region of the CFBC. The lower region is fluidized
programme for coal combustion in a CFBC. by the primary air supply. Kunii and Levenspiel’, Saraiva et
al. ‘O,and Kwaulk et al. ’’treated the lower region of CFBC
using the models developed originally for bubbling flui-
MODELING APPROACHES dized beds. This is inconsistent with the fact that the gas
In a typical CFBC used for coal combustion, crushed coal superficial velocity in this region is usually higher than a
together with limestone or dolomite and ash particles are certain critical value, U,, where the region becomes turbu-
fluidized by the combustion air entering at the bottom of the lent18. At this condition, solid velocity, bubble diameters
bed and at one or several secondary air injection points. A and velocities are quite different from the bubbling
large portion of the bed particles exits the riser of the CFBC regime7,18. However, for simulation purposes, perfect
with the flue gas due to the high superficial gas velocities mixing between the solids and the gas phases is assumed
utilized. The particles are then separated from the exhaust in this region. Under these conditions, the mean voidage of
gas in a gas/solid separator (often a cyclone) and recycled the dense region is considered constant and may be obtained
into the riser to promote complete combustion of the coal. using the correlation proposed by Kunii and Levenspiel’.
Because coal combustion in a CFBC is directly affected by
its hydrodynamic parameters, both hydrodynamic and Upper region of the CFBC. The upper region is suspended
combustion models must be treated simultaneously to both by the combustion gases from the lower region and the
yield a predictive model for the CFBC. The description of secondary air supply which determines the boundary
the method followed in developing the hydrodynamic and between the two regions. Hydrodynamic models, as pro-
reaction models is given below. posed in most CFBC literature regarding the upper region,
are classified into three broad groups’: (1) those predicting
Hydrodynamic model the axial profile of the solids suspension density but failing
The hydrodynamic model enables the variation of the to predict the radial variation; (2) those assuming two or
void fraction with height in the riser to be determined. The more regions considering either the core annulus or the
cluster models to predict the radial variation; (3) those (4) Since char combustion is slower, it is assumed to
applying the fundamental equations of fluid mechanics to occur after all the volatile products have been burned2’.
model gas-solid flow structure. Type 1 and 2 models, which This is an acceptable hypothesis considering the very
are lumped models, can be easily coupled with reaction short time required for volatile combustion,
models to simulate a CFBC reactors. On the other hand, (5) Burning coal particle and gas temperatures are con-
type 3 models, which are differential models, become rapidly sidered constant and equal to the bed temperature. This is
tedious when coupled to reaction models because of the a simpliying hypothesis considering the fact that the coal
numerical complexity. particle temperature is higher than the temperature in gas
For simulation purposes, we chose to apply the type 1 media.
model to predict the mean axial voidage profile in the upper (6) The contribution of the cyclone and the circulation
region of the CFBC, assuming that this region consists of loop on the overall combustion process is neglected.
two zones: an acceleration zone and a fully developed zone. Arena et al5 have considered the cyclone as a reaction
In the acceleration zone, the axial voidage decreases with block in their simulation, but due to the small particle
the vertical position along the riser’: residence time in the cyclone and the lack of excess
oxygen in the recirculation loop, the hypothesis appears
e * - E(Z) -az
=e (1) reasonable.
E* -El (7) Char particles are assumed to bum with a constant
In the lumped modeling approach used in this work, the riser diameter. This diameter is the mean char particle diam-
will be divided into a discrete number of intervals. Based on eter based on the experimental particle size distribution.
the void fraction variation in the acceleration zone given by (8) The attrition rate constant for char particles in the
eqn (I), the mean value of voidage in a certain riser interval CFBC is smal15. Therefore, the attrition-assisted combus-
between height Zi_l and Zi can be calculated using the tion rate is deemed negligible.
expression proposed by Kunii and Levenspiel’: (9) The effects of the primary fragmentation of coal and
the secondary fragmentation of char in the overall coal
1 combustion process are neglected5.
Ei=E* - -&El -e*)(exp-“z -exp-“&-I) i=2,3 (10) Any char particle size reductions caused by ash par-
ticles or the walls of the CFBC are neglected.
(2)
In the fully developed zone, the mean axial voidage is esti- Modeling procedure. For simulation purposes, the com-
mated by the following equation: bustion of coal particles can be modeled using the following
1 reactions:
E4 = (3)
(PG, (1) devolatilization and volatile combustion;
1+
U2Ps (2) char combustion;
(3) NO, formation;
where the slip factor Q is19:
(4) SO2 absorption.These reaction steps occur in the differ-
ent regions of the riser which can be divided into a
+ = 1+ $ + o.47fi4’ (4) number of individual reactors. To carry out the required
r calculations for each of those reactors, ASPEN PLUS
reactor blocks were selected and combined in a program
The variation of void fraction with length in the riser is flowsheet representing the CFBC. The CFBC riser is
illustrated in Figure 1. divided into two regions: a lower region and an upper
region. The lower region is represented by a single
Reaction model CSTR (continuous stirred tank reactor), while in the
The reaction model allows for the determination of the upper region a plug flow regime for both gas and
chemical changes and the heat released during combustion. solid phases is assumed. A series of CSTR reactors
Since coal combustion in the CFBC is quite complex, only are used to simulate the corresponding plug flow
the major steps of coal combustion are considered in the regime in the upper region2’. The number of CSTR in
model with some simplifying hypotheses. The general series is determined based on the hydrodynamic
hypotheses of the reaction model along with the modeling description of the upper region. As mentioned pre-
procedure is presented below. viously, this region is divided into two zones, a fully
developed zone and an acceleration zone. One CSTR
General hypotheses of the reaction model. For reactor simulates the fully developed zone. Since the
steady-state conditions, the assumptions regarding the height of the acceleration zone predicted by model is
reaction model are the following: relatively high with a considerable solid fraction varia-
tion, this zone is then modeled using two CSTR reactors
(I) The coal and limestone are fed into the bottom of the with a different mean solid fraction for each reactor.
bed at a uniform temperature”. This is largely encoun- With these arguments, the use of four CSTR reactors
tered in industrial units operating at high feed rate, in ASPEN PLUS is justified. Figure I shows the mean
because in these conditions the temperature gradient solids fraction (1 - Ei) corresponding to the four reac-
within the feed is negligible. tors (lower region and three sections of the upper
(2) Since the time required for volatile combustion is very region). The reactors are numbered as 1 for the lower
short, the devolatilization process is considered instanta- region reactor and 2, 3 and 4 for the three upper region
neous and to take place at the bottom of the bed’. reactors. Description of reaction steps involved in each
(3) Char is uniformly distributed throughout the circulat- reactor, with the corresponding ASPEN PLUS unit
ing bed. operation blocks, are presented below.
_-- c+co*=+2co
These reactions occur in the entire riser, hence in the four
1(l-e,) ;
CSTR reactors, and RCSTR (ASPEN PLUS CSTR reactor)
I Lowetregion 0
I is used to model this process. This block requires the knowl-
I
I edge of the reaction kinetic model which is presented below.
I + b The first and third reactions are heterogeneous and
Solid fraction,(1e)
the second is homogeneous. Since the temperature of the
burning particles in the CFBC is not sufficiently high, the
effect of the third reaction on the combustion rate is 10w*~,
Figure 1 Variation of void fraction with height in the riser
and this reaction is neglected in the model. For the
remaining two reactions, the reaction rate expressions
Devolatilization and volatile combustion. When coal is must be developed. The first reaction is a gas-solid reaction
introduced into a CFBC, it decomposes into two parts: and the chemical changes take place on both the external
hydrogen-rich volatile and char. The char remains in the and the internal surface of the char particles’. The following
bed and burns slowly. Based on the plume model, coal expression for the char combustion rate, to form CO, per
devolatilization and complete combustion of the volatile unit volume of the ith interval is obtained25.
occur at the feed entry point23. Two steps will then be con-
sidered in the simulation: decomposition and volatile 3v02kcrFchar, i( 1 - fi)
rl,i = CO, (5)
combustion. PcharrC( 1 - EC)Fsolid, i
Decomposition. In this step, coal is converted into its k,, can be expressed by an Arrhenius form as follows:
constituting components such as carbon, hydrogen, sulfur,
nitrogen and ash. This step occurs in the lower region reac- k,, = kolexp
tor only and RYTELD (ASPEN PLUS yield reactor) is used
to model this process by speciying the yield distribution
vector according to the coal ultimate analysis. The following equation is used to calculate the mean par-
ticle radius based on the experimental particle size
Volatile combustion. To simulate the volatile combus- distribution:
tion step, three reactions are considered in the model: 1
rc= m (7)
c+$**co
s+o**so* t r,(k)
Carbon monoxide produced during the heterogeneous com-
H2 + $02 * H20 bustion of char reacts with 02 in the homogeneous gas phase
reaction to form CO*. Factors contributing to CO emission
These reactions occur in the lower region reactor only where levels are the bed temperature, 02, CO and Hz0 concentra-
the coal is introduced and RSTOIC (ASPEN PLUS Stoi- tion. The following ex ression is used for the CO combus-
chiometric reactor) is used to model the volatile combustion tion rate in the model 2g :
process. The combustion of the volatile matter is based on
the following hypotheses:
(i) Considering that the volatile matter (VM) in the coal,
rco,i= 1.18* 10’3f&6~~o(
&)exp
(obtained from a proximate analysis) consists exclusively
of carbon, hydrogen and sulfur, the fraction of total coal x (-F)CEi (8)
carbon associated to volatile combustion is given by X, =
VM - H - S, where H and S are the fraction of hydrogen
and sulfur in the coal. This supposes that the entire NO, formation. Staged combustion remains an attractive
method for reducing NO, emissions in various combustion V205, is often present in fly ash from heavy oil combustion,
systems. For staged combustion in a CFBC, the air used for for example. However, in our case such an element was not
the combustion is divided into two or more streams: the first present and it is therefore correct to assume negligible SO2
one is supplied through the bottom air distributor and sec- to SO3 conversion.
ondary air streams are injected in the upper region of the Since CaCOs is unstable under CFBC conditions, the
riser. NO, formations in combustion processes result from a calcination process is assumed to occur instantaneously and
combination of a thermal generation process and fuel nitro- completely in the lower region of the bed. The second
gen oxidation. At very high temperatures, thermal genera- reaction, representing the SOi capture in the riser, is
tion of NO, from the air nitrogen becomes very important, considered to occur in both the lower and the upper regions.
while at low temperatures found in a CFBC, the dominant The corresponding fractional conversion for SO2 (Xso,, i) is
source of NO, is fuel nitrogen oxidation. It is also important calculated using a Fortran program according to the CaO
to emphasize the complexity involved in NO, chemistry conversion model presented below. RSTOIC is then used to
within the CFBC because of many catalytic reactions invol- model the capture of sulfur in the riser from the calculated
ving, for example, char, ash and sorbent particles. value Of Xso*,i. The fractional conversion of CaO to CaS40
Therefore, the total NO, formation in the CFBC can be is strongly affected by the physical and chemical properties
calculated by the following formula which considers the of limestone, hydrodynamic parameters, mass transfer
thermal generation and fuel nitrogen oxidation as detailed resistance, temperature, reactive concentration, particle
below: size distribution (PSD) and operatin conditions, and can
%
be calculated according to Couturier from the following
]N0,1,,,,1 = ]NOx]thermar+ ~1WQlfuel expression:
where
3acyS02,i
1+ (e”” - 1)
cy, = Overall fuel nitrogen to NO, conversion factor Vcao R,Kv
x CaO,i’ -I __,
X(O<cr, < 1) RS -- 1
‘I 3aCYso,,i KV
(9)
Thermal generation ([NO,] thermal). Three main reactions Considering the ideal gas law for the combustion products
are used to represent this process in the model: in the riser, the total gas concentration is expressed by:
;N~+&*NO P
C=---.- (10)
RlTb
The values of parameters a I and a! may be written as*‘:
N2 + &02 * N20 a, = 3.33 * 10e4eyR\ (11)
REQUIL (ASPEN PLUS equilibrium reactor) is used to
predict the amount of thermal NO, formed during coal com-
bustion based on equilibrium conditions considering the
CY= 35D”.3
P (12)
nitrogen present in the riser. Using eqn (9), the moles of SO2 removed per unit v,Ollume
become
Fuel nitrogen oxidation ([NO,] fuel). The NO, formation
via fuel nitrogen oxidation is modeled using the following VCaOFl
overall reaction: rso~*i
= 1 _ E,Au * 100
3oCYsoz,,
1+ (e”” - 1)
R&v
RSTOIC block is used to calculate the fuel nitrogen oxida-
X (13)
tion with a given value of o, which is taken from the litera-
ture. The NO, formation calculations are then applied in 3aCYS02,i - j&
each of the four reactors using the combined REQUIL and
I
RSTOIC blocks.
If the sorbent particles in the bed are well mixed, their
SO2 absorption. The SO2 capture by limestone can be residence time is independent of particle size*‘. Therefore,
represented by the following ureactions: the mean residence time is expressed by
CaCOs =+ CaO + CO2 ALI
?I =W%p-
6
and
CaO + SO2 + & 3 CaS04
Table 1 The reactor models description utilized in the OnceGo,, i is calculated for the ith reactor, it is used as the
simulation’5”6 fractional conversion of the key component necessary to run
Reactor block Description
the RSTOIC block corresponding to this reactor. Calcula-
tions start at the first reactor (lower region) and continue
RYIELD To simulate a reactor by speciying
upward until the top of the riser.
yield distribution data or correlation
when reaction stochiometry and
kinetics are unknown
RSTOIC To simulate a reactor with the SIMULATION RESULTS AND MODEL VALIDATION
unknown or unimportant reaction
kinetic and Imown stoichiometry by The process simulation program for a CFBC was developed
speciying the extent of reaction or the using four ASPEN PLUS reactor blocks-KYIELD,
fractional component of the key RSTOIC, RCSTR and REQUIL-to represent the phenom-
component ena identified in the coal combustion process. A detailed
To handle any number of simulta-
description of the ASPEN PLUS reactor blocks along with
neous or series reactions
RCSTR To model CSTR reactors with known their re uirement is given in the ASPEN PLUS user
reaction kinetic manualsq5,t6. A brief description of the reactor blocks is
To require user supplied kinetics given in Table 1. Table 2 presents the reactor model
subroutine when solids, such as char,
parameters and input variables required for the simulation.
are participating in the reactions
REQUIL To calculate simultaneous phase
Considering the reaction and hydrodynamic models, the
chemical equilibrium by solving entire CFBC system is divided into three sub-flowsheets:
stoichiometric chemical and phase lower, upper and sep-flowsheets. The first sub-flowsheet
equilibrium equations represents the dense region, the first reactor of the riser,
To be convenient to the known reac-
where the phenomena associated with coal devolatilization
tion stoichiometry when only a few
reactions approach equilibrium and volatile combustion, char combustion, NO, formation,
limestone calcination and SO2 capture take place. The
Since SO2 is well mixed in each interval of the bed, an second sub-flowsheet represents the dilute region of the riser
overall SO* balance gives which is divided into three intervals; each one represented
by an individual reactor. Since the devolatilization and
-5 @so,,
I - rso,,I)
volatile combustion process is assumed to take place
Ysoz,I = (17)
GUI exclusively in the lower region, only char combustion,
and NO, formation and SO2 capture are considered to occur in
each reactor of the upper region. The last sub-flowsheet
YSO*,i
WRso,,i - rso2,J i contains two unit operation blocks, CYCLONE and
(18)
z 1
R so*,i =
(1 - Xso,, i - I Yc wsi#l (20)
to satisy the material balance.
32AAL Along with the unit operation blocks provided by ASPEN
PLUS, several complete Fortran programs and an external
The value of Yso2,i is calculated by simultaneous solving of
subroutine for the kinetic models were used in the
eqns (13) and (17) for i = 1, or eqns (13) and (18) for i f 1.
simulation. The Fortran codes contain the following four
The fractional sulfur capture for each reactor (Xso,J can
blocks which are required to complete the model. The first
then be calculated from
block, ‘KINETIC’, is the external kinetic subroutine
developed for both heterogeneous gas/solid and homoge-
(21) neous gas phase reactions. The second Fortran block, called
‘KESTIME’, when integrated with the ASPEN PLUS input
file, calculates the residence time of char particles in the
CFBC. This calculation is required in order to execute the
‘KINETIC’ subroutine. The third block, ‘HYDRO’, is
inserted into the ASPEN PLUS input file to calculate the
&02.i=1- [ Fc,j$fzi_,J i# 1 (22) mean void fraction in each section of the upper region and in
the dense bed of the riser. The fourth block of Fortran codes,
s4
B2 ~2, B3 &B4 - BS
Rl
1 :
Bl7 B16 $1 ‘I B9 z BS
a s2n #s&y - S1Q
FSPLIT CYCLOM R4 RSm1c - REQUIL
Table 3 Value of the fixed parameters used in the simulation Table 5 The CFBC operating conditions”
Q8
1000
!z
B
2 800
F
.fi 800
400
0 I I 1 I I I I
These parameters were obtained by fitting the CANMET observed between predicted and experimental SO* concen-
experimental data. However, more data from various tration in the flue gas.
sources are required in order to confirm this correlation.
The CO emission levels predicted by the model ranged NO, emission levels. As mentioned earlier, two forma-
from 119 to 271 ppm, while the experimental data varied tion mechanisms were considered in the modeling of NO,
between 112 and 3 16 ppm. Therefore, the model based on formation in the CFBC: thermal generation and fuel nitro-
the new correlation estimates CO emission levels relatively gen oxidations. Thermal generation was calculated con-
well. The comparison between predicted and experimental sidering equilibrium conditions, while NO, formation
CO levels is presented in Figure 4. from fuel nitrogen oxidation was calculated using an overall
conversion factor (cY~)from the literature. Since the aim of
SO* emission levels. The Ca/S molar ratio ranged this work was not to study the NO, formation and reduction
between 1.6 and 2.3 for the various runs considered. Such processes in detail, this overall approach was taken to simu-
values of Ca/S are usually considered sufficient to achieve late the fuel nitrogen oxidation. Fuel nitrogen conversion
reasonably high sulfur capture efficiencies. Without using entails relatively complex reactions schemes involving sev-
any fitting parameters, Figure 5 shows the close agreement eral heterogeneous reaction steps and therefore attains a
lower overall conversion26. Typical values of fuel N2 to
NO, conversion factors, as reported by Legros et CZZ.*~ and
Becker et d3’, vary approximately between 0.05 to 0.25,
depending on coal properties, feed particle size distribution,
excess air level and operating conditions. In our simulation,
a value from that interval, which gave the best agreement
between predicted and experimental NO, was chosen as the
overall conversion factor. As reported in most CFBC
literature, the results also confirmed that thermal NO, for-
mation, which leads to between 18 and 65 ppm of NO,, is
ii
‘: 200 unimportant compared to that of fuel nitrogen oxidation
8 which approximately lies between 84 and 104 ppm of
NO,. The predicted emission levels ranged from 130 to
150
% 267 ppm, while NO, emissions for the experimental
CFBC ranged from 107 to 309 ppm.27. Figure 6 appears
%
100 to indicate a reasonable agreement between predicted and
experimental NO,. The difference between experimental
data and those of the simulation prediction is attributable
to the fact that a constant value of CY
I = 0.05 is used through-
out the entire simulation.
0 In recent years, several comprehensive studies have been
0 50 100 150 200 250 300 350 reported6.3 ’-34 regarding NO, formation and reduction
processes. These were conducted to develop an improved
Ew=tm~tNO,(ppm)
understanding of the fundamental nature of NO, chemistry
Figure 6 Comparison between the predicted and experimental and underlying physical processes in CEBCs, and to support
NO, the needs for experimental work in this field. Emphasis is
8 6- 0 _ 2500
Redi*edCOConcartration(ppm) p
.g 5- - 8
E 2000 i
E
$!' 3- - 1500 E
0"
0 - 1000 "
2-
..
*.'..Q - 500
l-
_.
....
0 .-...._._..
I I I I I I 00
0 1 2 3 4 5 6
BedHeight
Figure 7 O2 and CO concentration profiles within the CFBC predicted by the model
also given to develop reliable techniques to control NO, providing a scholarship to Mr R. Sotudeh-Gharebaagh.
emissions from fluidized bed combustion. For example, Helpful discussions from F. Preto and E. J. Anthony are also
many published studies suggest that NO, emissions could be appreciated. We greatly acknowledge Aspen Technology
controlled by adding chemical components such as carbon for having granted special permission for the use of the
monoxide, char hydrogen, ammonia, unburnt hydrocarbons ASPEN PLUS under the condition of the academic
and limestone due to the catalytic reactions found in licensing agreement.
CFBC32V33.
21 Rajan. R., Krishnan, R. and Wen, C. Y., Simulation of Fat,I primary air mass flowrate (kg h-‘)
fiuidized bed combustors-part II. In Fluidizatioti Fwr-2 secondary air mass flowrate (kg h-‘)
application to coal conversion process. AIChE Sym., Fc mass flowrate of coal (kg s-‘)
flux of the char particles entering the ith interval (kg s-‘)
1978, 74, 112. FdlK,
mass flowrate of coal (kg h-‘)
22 Levenspiel, O., Chemical Reaction Engineering. 2nd ed. FL-,,,