Você está na página 1de 10

Modelling and simulation of an AFBC

steam heating plant using ASPEN/SP

P. L. Douglas and B. E. Young*


Department of Chemical Engineering, University of Waterloo, Waterloo, Ontario, Canada
fV2L 3GI
*Kurst & Steele Ltd., 2275 Speakman Drive, Mississauga, Ontario, Canada L5K lB1
(Received 25 May 1990; revised 74 September 7990)

A techno-economic study of the utilization of Eastern Canadian coal in steam production using atmospheric
fluidized bed combustion (AFBC) is presented. An industrial-scale AFBC boiler, located at Summerside
PEI, Canada, was modelled using the ASPEN/SP (advanced system for process engineering) process
simulator. ASPENjSP heat and material balance predictions were within 10% of plant measurements. A
sensitivity analysis showed that the combustion air flowrate has the greatest effect on the predicted carbon
burnout. An economic sensitivity analysis shows that the plant’s economic viability versus electric heating
is extremely sensitive to the price of the electricity. Rising electricity prices make this plant more attractive.

(Keywords: coal; fluidized beds; simulation)

In the 1970s the government of Canada recognized that required that the combustion efficiency, sulphur capture,
Eastern Canada’s consumption of expensive imported oil solid elutriation rate and the particle size distribution of
would be lessened if its large reserves of high sulphur the elutriated solids all be known u priori. In this work
coal were utilized in an environmentally safe manner. At these properties were calculated internally by predictive
that time a promising technology becoming commercially process models imbedded in the ASPEN/SP input file.
available outside Canada was atmospheric fluidized bed
combustion (AFBC). While the process of fluidization
THE PROCESS
was not new, the use of this process to burn coal for
energy production was. It was also found that by feeding There are two coal-fired AFBC boilers at CFB
limestone into the reactor with the coal and holding the Summerside which together produce 36 000 kg h- ’ of
reaction temperature around 87O”C, the emissions of 1070 kPa steam when operating at capacity. This
sulphur dioxide could be greatly reduced. As a operating rate consumes approximately 5000 kg of coal
demonstration of this technology, AFBC boilers were and 2050 kg of limestone per hour. The plant can most
erected at the Canadian Forces Base (CFB), Summerside, easily be regarded as three separate facilities working
Prince Edward Island. The AFBC units provided most as one. These are: the coal and limestone preparation
of the steam heating used at CFB Summerside during section; the gas/solids section and the steam/water
the 1985 heating season. section. Although the coal and limestone preparation
The long-term objectives of this research programme flowsheet, shown in Figure I, is essential to the operation
are to develop a common basis for the technical and of the plant, it is a minor portion of the overall plant
economic comparison of coal-based energy technologies and will not be discussed in detail here. For more
and to develop optimal strategies for using coal. The information the reader is urged to consult Young6.
short-term objective, and the one addressed here, is to The gas/solids flowsheet is shown in Figure 2. The coal
demonstrate the technical feasibility of using ASPEN/SP and limestone are transported from day-bins to the
to simulate AFBC-based power plants by simulating the combustor by remotely controlled weigh-feeders. Once
CFB Summerside steam heating plant. The key results inside the combustors coal and limestone particles are
include heat and material balances, equipment sizes, an rapidly mixed into a bed of ash and limestone particles
economic analysis and a sensitivity analysis of the that are suspended by an upward flow of combustion
process. air. The bed temperature is controlled at approximately
ASPEN/SP was chosen because it is ideally suited to 870°C by water-cooled walls and 18 cooling tubes in the
simulate coal processes and contains many features bed. As the coal burns, sulphur dioxide is produced and
essential to this type of workl. ASPEN/SP also contains captured as calcium sulphate by the calcined limestone,
economic analysis and optimization subroutines, making which makes up the bulk of the bed. Bed height is
it a valuable tool in this research. ASPEN/SP has been controlled by two drain pipes and water-cooled screw
used in coal conversion simulations in the pastze5. conveyors called bed-drain coolers, which are located
Methanol synthesis was studied by Schwint’ and below the combustors. Forced draft (FD) fans ensure
Knudsen3, Barker4 studied the tri-state indirect- adequate air velocity and pressure to elevate the bed
liquefaction process and Bajura5 simulated a pressurized particles. About 45 000 kg h- ’ of air at 3 m s- Is are
fluidized bed combustion process for electric power required by the boilers. The FD fans consume most of
generation. In BajuraV work, the ASPENjSP simulation the electric power required by this plant. Once through

001~2361/91/020145-10
.f 1991 Butterworth-Heinemann Ltd. FUEL, 1991, Vol 70, February 145
Modeling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

Vibrating
rcrecn

Conveyor

LL
mcgnet
feeder
--I - --
&

(I
Conveyor

_ - -.-
Bucket elevator

Coal handling system

Limestone handling system

Figure 1 Coal and limestone preparation flowsheet

_Steom to base , p7~Boiler fee,d water

Limestone ---_E7

Cooling water

Figure 3 Steam/water process flowsheet


Stock
reinjection
system water in economizers. Large induced-draft fans pull the
gases through the economizers and the final particulate
Air
cleaning stage, the baghouses.
The steam/water flowsheet is shown in Figure 3. The
I.D. fan condensate returning from the base is supplemented with
make-up water and held in a large insulated holding tank
F.D. fan
called the hotwell. The hotwell is heated by water
Figure 2 Gas/solids process flowsheet returning from bed-drain coolers. Boiler feed water,
drawn from the hotwell, is first passed to a de-aerator,
the bed the air and any suspended ash, char and limestone where any air that leaked into the system is driven off
particles pass through an array of boiler tubes that cool by live steam supplied by the boilers. After de-aeration,
the combustion products to approximately 270°C. Most the boiler feed pump pressurizes the cooling water and
of these particles are returned to the bed by the in-bed then passes it on to the economizer. From there it travels
cyclones, through which the combustion gases must pass to the steam drum where it flows down into the
before leaving the combustors. Once through the water-walIs or the convection boiler bank and returns as
cyclones the gases exchange heat with the boiler feed steam. At this point about 250 kg h- ’ of water is purged

146 FUEL, 1991, Vol 70, February


Modelling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

flowsheet, the gas/solids flowsheet and the steam/water


Dumfeed
flowsheet, respectively. A detailed discussion of the
simulation flowsheets and their development is contained
in Young6. Statistics for the ASPEN/SP simulation are
presented in Table 1. Library models are available in
ASPEN/SP for all units except for an AFBC model and
some costing models.

Lime t t Cool AFBC model


ASPEN/SP does not have a library model to simulate
a fluidized bed. ASPEN/SP does, however, allow users
to input their own models. Fortran code, nested
Postscbl Postscbc within the ASPENjSP input file, controls the operation
of six ASPEN/SP blocks, which, taken together, simulate
the fluidized bed. Many fluidized bed models, of varying
complexity and accuracy, are available in the literature6.
These models range from simple charts that estimate
Feed1 design and performance parameters’ to detailed studies
that include gas concentration, reaction kinetics, carbon
Stream types concentration, heat transfer and temperature profiles’. It
- Material is necessary to define the quantities that must be predicted
---- Information by the model and to determine a prospective model’s
Figure 4 Coal and limestone preparation simulation flowsheet

Lpfeed

--(

, s
I !
Aerspply
Condi 1 @eup
Boilerht , t LpreturnT
-._ -._._
Hottank Dl Split1
(mixer) (splitter)
Creturn t Sbduty j Cool 1
Sbdcool outy1
(heater) (heater)
Pcoolout Scoolin Cool2
Pcoolin Fdrain
T;u”3p
.
Pbduty 7 Cwdrow i
II Fddraw Stream typL
_-___--___--_---__--_____^______
- Moteriol
---- Information
Stream types
----Energy
- Moterial
----Information Figure 6 Gas/solids simulation flowsheet
- -‘-‘- Energy

Figure 5 Steam/water simulation flowsheet


Table 1 ASPENjSP simulation statistics
into the blowdown tank, with 730 kg h-’ to supply the
43 Unit operation blocks
de-aerator and 17 100 kg h-’ to the base steam heating I Cost blocks
network. 4 Fortran blocks
9 Design specifications
51 Material streams
ASPEN/SP SIMULATION 19 Information streams (16 heat. 3 work)
885 Lines of ASPENiSP code
Simulation ,jlowsheets 365 Lines of Fortran code
Figures 4, 5 and 6 show the ASPEN/SP simulation 4 Megabytes of memory required
3 CPU seconds required for simulation (IBM 308l-G24 VM/SP)
flowsheets for the coal and limestone preparation

FUEL, 1991, Vol 70, February 147


Modelling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

ELUTRIATED CHAR
Equation (16)
CALCULATE SEGREGATION FACTOR

DRAIN SOLIDS

CALCULATE PSD OF
ELUTRIATED CHAR

Equation (9)

CALCULATE MASS OF CHAR


AND BED SOLIDS
Equations (14) and (15)

I
1
( STOP )

Figure 7 Algorithm of the simulation program

applicability to the system under study. The important described above, information from two sources was used.
quantities for this study are: calculation of combustion Checks on the overall heat and material balances of the
heat release; modelling of solids size distributions and plant. as well as the particle size distributions of the
elutriation rates; prediction of sulphur capture efficiency. product solids streams, the production of pollutants and
These requirements eliminate those models that predict the electricity consumption of the plant are all made
gas phase behaviour or temperature profiles of the bed against actual plant operating data. Equipment sizing
and freeboard, since these models are both time- data were checked against the information supplied in the
consuming and provide more information than required design manuals”. Six 24 h combustion trials were
for our purposes. Heat transfer models were not performed at CFB Summerside and portions of these trial
considered because the bed had enough heat transfer results are used for comparison with the ASPEN/SP
surface to collect all the recoverable heat generated by simulation model. Table 2 presents a comparison
combustion of the coal. The AFBC Fortran model was between the plant data, running at 100% load
developed from a model presented by Kunii and (approximately 18 000 kg h-i) using Devco coal and
Levenspiel’. Peripheral programming steps have been Havelock limestone, and the model at the same
added around this model to speed up the calculation conditions. This agreement suggests that the model
sequence and allow the computation of other required predicts the plant operation closely. The stack gas
values. An algorithm of the program is shown in Figure 7. emissions are significantly less than the government limits
The main elements of the Fortran code are: superficial of 2.952 and 0.36 kgGcal_’ for sulphur dioxide and
velocity and excess air; sulphur capture calculation; an particulates, respectively. All of the significant errors seen
underbed withdrawal correction; a cyclone collection in Table 2 can be attributed to the fact that the model
efficiency for each size fraction; limestone abrasion predicts a higher carbon burnout rate.
model; a solids population balance. Rather than present Figure 8 is a plot of the plant steam flow versus the
a detailed discussion of the model equations a reference predicted steam flow with the model running at the same
in the algorithm is given. conditions. The solid reference line on the graph depicts
perfect agreement and the broken lines represent f 10%
deviation from the reference line. Going from left to right
RESULTS AND DISCUSSION
on the graph, the points represent the following plant
Heat and material balance operating loads and coal types: Evans coal, 65% load
To test the capabilities and limitations of the model (approximately 13 200 kg h- ‘); Devco coal, 65% load;

148 FUEL, 1991, Vol 70, February


Modelling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

Table 2 Calculated values versus plant data model assumes that abrasion, which produces only
elutriable particles, is the only size-reducing mechanism
Difference
Parameter Plant Model
that acts on the limestone. This mechanism does not
(%)
produce particles in the size range lo&700 pm.
Sulphur dioxide capture (%) 84.56 84.44 -0.14 Particles in this range must be produced by fracture of
Baghouse: large limestone pieces, a mechanism not considered here.
flow (kgh-‘) 451.2 458.1 +1.53 If a simple fracture model becomes available its inclusion
carbon (%) 1.24 6.55 -9.53
would increase the accuracy of the model in the
Bed drain: 100-700 pm particle size range.
flow (kgh-‘) 303.3 301 .o -0.76
carbon (%) 0.18 0.122 - 32.22
Economic analysis
Steam flow (kg h ‘) 17053 18 605 +9.10
Elec. consumption fans (%) 88.7
The economic analysis of a project includes capital and
85-90
operating cost estimates. Many methods exist for the
Dry gas analysis (wt%)
15.25 18.30 + 20.00
escalation of the delivered costs of the ‘main plant items’
CO,
02 - 8.53 1.35 - 13.83 to the total capital required by a project”.r4. The costing
N2 76.08 74.24 -2.42 equations in ASPEN/SP would be used wherever
so2 0.0966 0.1062 +9.94 possible” ’ 3. To reduce the error associated with
Stack emissions (kgGcal_‘): literature cost correlations a composite correlation was
particulates 0.136 0.138 1.4-l
developed by linear regression of all the relevant literature
SO, 1.689 1.692 0.18
cost data. The economic results were compared with the
contract quote awarded for the construction of the plant
addition’ ‘. The simulation predicted a construction cost
of $17.8 million (1982 Canadian dollars) and the quote
20000
was for $13 million (1982 Canadian dollars). The
estimated total fixed capital produced by the two schemes
came to within 4% of one another. A summary of the
y=x operating data used in the economic analysis is contained
/
F 18000 in Table 3. Table 4 contains the cost estimates for a plant
\ built today at CFB Summerside, assuming 90%
y” availability and 100% load operation. These results
V suggest that fluidized bed combustion is an attractive
3 16000 alternative to electric heating for the base.
0
Z
Sensitiviqj analysis
A sensitivity analysis using the ASPEN/SP simulation
model was performed. The main reason for performing
a sensitivity analysis is to identify the key variables

0.7

4.0
L Evans 65% load
10000
g 14.0
12000 14000 16000 18000 P)
Predicted steam flow (kg/h) *$ 34.0
&
Figure 8 Plant steam llowrate versus model predictions
g 61.0

.-F
Brogan coal, 65% load; Devco coal, 100% load. The G 82.0
measured values of the plant steam flow are always lower J
than those predicted because the model assumes no heat E
losses from the boiler to the surroundings. =J 95.0
0
Figure 9 is a plot of the predicted and actual particle
size distributions (PSDs) of the baghouse ash and the
bed drain streams. The predicted and measured PSDs of
the baghouse ash agree very well, suggesting that the
cyclone and baghouse models are representing the actual
collection efficiency quite well. On the other hand, there 10 100 1000 10000
is a large gap between the predicted and actual PSD of
the bed drain stream at small particle diameters. This is Particle diameter (pm)
caused by two factors. First, the logarithmic vertical scale Figure 9 Predicted and actual particle size distributions for bed drain
magnifies errors at small diameters and, second, the and baghouse solids

FUEL, 1991, Vol 70, February 149


Modelling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

Table 3 Economic data in combustion efficiency. This is due to the higher carbon
Coal cost: $25/tori
loading in the bed, since the larger particles are not
Limestone cost: $36/tori elutriable. Of course this trend cannot be expected to
Electricity cost: $O.O85/kWh continue indefinitely because very large particles would
Number of operators: 25 not burn efficiently. What cannot be shown on a graph
Average operator salary $23 700/year is the increased efficiency brought about by altering coal’s
Timmerhaus“’ provided the following data: PSD by pre-processing. To simulate pre-processing we
Operating hours: 7884 h/year
used the limestone’s double screened distribution
Plant life: 25 years
Construction time: 4 years function and the coals larger mean particle diameter. This
Make-up water cost: $0.0072/1000 Ibs provided a coal feed stream with the same mean particle
The following assumptions were made based on the current economic diameter, but with a much narrower range than the
conditions: original feed. Using this coal the overhead carbon losses
Interest rate 12% were reduced by 99% of the nominal value. Since a price
1.35 Canadian dollars per American dollar
for double screened coal was not available, an economic
study of the viability of using pre-processed coal was not
possible.
Table 4 Economic predictions
-___
Effect of limestone flowrate and PSD on carbon losses
Total capital required: $27 million
Operating expenses: $11 million/year Figure 11 is a plot of changes in the overhead carbon
Savings versus electric heating over plant lifetime: $65 million loss versus limestone properties. Carbon losses are not
sensitive to the flowrate of limestone. This demonstrates
that sulphur capture by limestone and combustion
efficiency are uncoupled, making optimization easier
affecting the performance and economics of the plant.
because each of the considerations may be optimized
While a sensitivity analysis can be performed using an
separately. The overhead carbon loss is, however,
operating plant, it is preferable to use a computer
increased significantly by a reduction in the mean particle
simulation for a number of reasons: faster results from
diameter of the limestone. The sensitivity of carbon loss
the simulation; no off-specification product produced
to changes in the limestone PSD is unexpected and
during analysis; less expensive; more combinations of
suggests that the increased sulphur capture efficiency of
plant conditions can be studied; isolation of manipulated
smaller limestone particles would have to be weighed
variable.
We chose the overhead carbon loss as the dependent
variable because it is a major contributor to plant
inefficiency and is sensitive to the way in which the plant 100,
is operated. Independent variables studied were: coal
flowrate, PSD and chemical analysis; limestone flowrate
and PSD; combustion air flowrate and bed temperature. 80
An economic sensitivity study was performed by L
comparing the savings made in the lifetime of the project
against the cost of heating the base electrically for the
same period. Independent variables studied were the unit
costs of coal, limestone and electricity.

Effect of coalflowrate, PSD andanalysis on carbon losses


Figure 10 is a plot of overhead carbon losses versus
coal flowrate, ash and water content in the coal, coal PSD
and carbon content. The carbon losses are not sensitive
to either the water or ash content in the coal, indicating
that the use of poorer quality fuels in a fluidized bed is
possible. The fluidized bed’s ability to burn fuel of various
qualities is the technology’s greatest advantage over its
competitors.
Carbon losses are, however, sensitive to coal flowrate,
average particle size and carbon content. The rapid rise
in carbon loss with an increase in the flowrate of coal is
due to an increase in superficial gas velocity required, diameter \
since the percentage excess air was kept constant. The -80
effect of increasing superficial gas velocity on carbon
losses is discussed later. Increasing the carbon content
decreases the overhead carbon loss because this promotes
lower superficial velocities in the freeboard and higher -50 -30 -10 10 30 50
carbon loading in the bed. The mean diameter of the feed
coal particles and the shape of the distribution function Change in coal property (%)
play a large role in the efficiency of the combustor. Figure 10 Change in overhead carbon losses versus changes in coal
Increasing the average coal particle diameter, without flowrate, carbon, ash and water content, and average coal particle
changing the distribution function, gives a large increase

150 FUEL, 1991, Vol 70, February


Modelling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

rate due to temperature changes is unknown, the reaction


rate at the surface of the char particle increases four fold
over the bed temperature range studied in the sensitivity
analysis”. The change in reaction rate due to increased
oxygen concentration in the fluidized bed is believed to
be smaller than that for temperature changes. The model
assumes a constant reaction rate, thus the steepness of
these curves is exaggerated. Even so, we believe that the
carbon loss is very sensitive to the air feed rate. These
results suggests that the plant should be run at the lowest
air feed rate possible to promote high combustion
efficiencies. However, operation at these conditions
would lead to poor mixing in the bed, which would lead
to lower heat transfer and possibly the formation 6f
clinker. A trade-off between economics and operational
stability has to be made.

Effect of unit costs on process economics


Figure 13 is a plot of change in discounted savings
versus a change in the price of coal, limestone, and
utilities. The economics are not very sensitive to coal and
limestone unit costs. This is largely due to the large loan
repayment, which amounts to more than 61% of the
total yearly expenses. If interest charges are neglected,
the coal and limestone unit price has a significant inverse
effect on the economic viability of this process, i.e. the
lower the coal prices the higher the savings.
Since the viability of the combustor is being compared
-50 -30 -10 10 30 50 with the cost of electric heating, it is expected that a small
increase in the price of electricity would have a very large
Change in limestone property (%) positive effect on the relative economics of the plant; this
Figure 11 Change in overhead carbon losses versus changes in
limestone flowrate and average limestone particle diameter
100

against higher carbon losses. The amount of coal leaving


IAir
flow
rote

!
the combustor is indirectly related to the carbon loading
in the bed. In the model, the carbon loading in the bed
is assumed to be the same as that in the bed drains, which
is calculated by the equation from Anthony”. In this
equation, a decrease in average bed particle size, which
is related to the average limestone particle size, will
produce a decrease in the carbon losses to the bed drain
and hence a lower carbon loading in this stream.
Therefore a decrease in the mean diameter of the
limestone feed will produce lower carbon loading in the
bed, which will decrease the overall efficiency of the plant.
The model always translates a decrease in plant efficiency
as an increase in the overhead carbon loss.

effect of AFBC operating conditions on carbon losses


Figure 12 is a plot of carbon loss versus air flowrate
and bed temperature. It is obvious that both the air flow
and the bed temperature have a large direct effect on the
carbon loss. Both of these phenomena are a result of an
increase in the superficial velocity in the freeboard. In
the case of increased air flow, it is simply the increased
mass of air that causes the increase in superficial velocity.
The higher bed temperature decreases the density of the
fluid and therefore for the same mass flowrate an -8O- ’ ’ ’ ’ ’ ’ ’ ’ ’
increased volumetric flow is observed, causing an equal -50 -30 -10 10 30 50
increase in superficial velocity. It should be noted that
in both of these cases some increase in the reaction rate, Change in operating parameter (%)
hence a decrease in carbon loss, would occur at the plant. Figure 12 Change in overhead carbon losses versus changes in air
While the absolute magnitude of any changes in reaction flowrate and bed temperature

FUEL, 1991, Vol 70, February 151


Modelling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

6. Given the promising results shown in this work,


Utility ASPEN/SP should be used as the simulation system
.. ,
of choice in the remaining studies of coal-based energy
technologies for the development of optimal strategies
30 for using coal.

20 ACKNOWLEDGEMENTS
The authors acknowledge the financial assistance of
10 CANMET (Canadian Centre for Materials & Energy
: Technology) and Energy Mines and Resources Canada.
0 The technical input from E. J. Anthony and H. A. Becker
is gratefully appreciated.
-10
REFERENCES
-20 1 ‘Aspen/SP User’s Manual’, JSD Inc., Denver, USA, 1986
2 Schwint, K. T. ‘Final Topical Report-Great Plains ASPEN/SP
Model Development: Methanol Synthesis Section’, Scientific
-30 Design Co., Inc., New York, USA, 1985
3 Knudsen, R. A. AIChE Symp. Ser. 1982, 78, 214
4 Barker, R. E. October Report, Oak Ridge National Laboratory,
-40
Oak Ridge, USA, 1983
5 Bajura, R. A. AIChE Meeting, Orlando, Florida, USA,
February 1982
-50
Young, B. MSc Thesis Queen’s University, Kingston, Canada,
-50 -30 -10 10 30 50 1986
Ganapathy, V. Hydrocarbon Processing 1981, 60(1 l), 269
Change in unit costs (%) Gordon, A. L. and Amundson, N. R. Chem. Eng. Sri. 1976,31,
1163
Figure 13 Change in savings versus changes in coal, limestone and Kunii, D. and Levenspiel, 0.. Robert E. Krieger Publishing
utility unit costs Company, 1977
Yagi, S. and Kunii, D. 5th Int. Symp. Combustion 1954, p. 231
Friedrich, F. Proc. potential applications of fluidized bed
is clearly shown in Figure 13. This result may lead one combustion in Canadian industry workshop, Calgary, Canada,
to have reservations about the viability of the plant when 1982
subject to a reduction in power costs. However, it is more 12 Humphreys, K. K. and Katell, S. ‘Basic Cost Engineering’,
Marcel Dekker, Inc., 1981
probable that the price of electricity will increase, relative 13 Ham, R. W. MSc Thesis Queen’s University, Kingston, Canada,
to that of coal, especially in an area that has little or no 1986
hydro-electric power, thus improving the overall 14 Timmerhaus, K. and Peters, M. ‘Plant Design and Economics
economics of the process. for Chemical Engineers’ (3rd Edn.), McGraw-Hill, 1980
15 Anthony, E. J. 8th Int. Conf. Fluidization, Houston, USA, 1985
16 Congalidis, J. P. and Georgakis, C. Chem. Eng. Sci. 1981,36,6
CONCLUSIONS 17 Ray, A., Berkowitz, D. A. and Sumaria, V. H. Energy 1978,
2(5), 269
1. By using ASPEN/SP, the CFB Summerside AFBC 18 Ray, A.. Berkowitz, D. A. and Sumaria, V. H. Trans. ASME
steam heating plant has been simulated at the heat 1980, 102
19 Charagundla, S. and Rao, R. Cornbust. Sci. Technol. 1977, 16
and material balance level and an economic analysis 20 Donsi, G., Massimilla, L. and Miccio, M. Combusr. Flame 1981,
has been performed. 41, 57
2. A Fortran AFBC model was developed and embedded 21 Chen, T. P. and Saxena, S. C. Fuel 1977, 56, 401
in the ASPEN/SP simulation to simulate an AFBC. 22 Razbin, V. Energy Mines and Resources Canada, 555 Booth
Street, Ottawa, Canada, personal communication, 1986
3. The accuracy of the simulation is presented in the 23 Yagi, S. and Aochi,T. Sot. Chem. Eng. (Japan), Spring Meeting,
following comparisons between the plant’s operating 1985
data and the simulation results: less than 10% error
in total steam production; the predicted particle size
distributions and carbon contents of the overhead and
NOMENCLATURE
underbed solids streams closely approximated the
plant data; the total capital cost predicted by the cc1 Carbon concentration (wt%, dry basis)
simulation came to within 37% of the contract. CGI Fixed carbon concentration (wt%, dry basis)
4. Results show that the ASPEN/SP system can analyse DP Diameter of particle (cm)
AFBC boilers on a heat and material balance Ievel. DPC Cut diameter of particle (cm)
5. A sensitivity analysis has shown that: carbon loss is D Maximum diameter of particle (cm)
not a strong function of coal ash or water content and F:: Feed flowrate of char (g s- ‘)
limestone flowrate; carbon loss is a strong positive F coal Feed flowrate of coal (g s-‘)
function of coal flowrate, air flowrate and bed Fc, Flowrate of bed drain char (g s-‘)
temperature and a strong negative function of carbon Fc, Flowrate of elutriated char (g s- ‘)
content, coal PSD and limestone PSD; the economic F,(R) Feed flowrate of size fraction R (g s-‘)
attractiveness is moderately sensitive to the unit cost F, Feed flowrate of solids (g s- ‘)
of coal and limestone and strongly affected by the cost FI Flowrate of bed drain solids (g s-l)
of electricity. F, Flowrate of elutriated solids (g s- ‘)

152 FUEL, 1991, Vol 70, February


Modeling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

F,(R) Flowrate of bed drain solids of size fraction R elutriation of fines. Each of these terms must be calculated
(g s-‘) before Equation (2) can be solved. The particle size
F,(R) Flowrate of elutriated solids of size fraction R distribution of the inlet stream was obtained from plant
(gs-‘) data22.
9 Gravitational constant (9.80 m se2)
Mb Mass of bed material (g) Segregation correction factor
MC Mass of bed char (g)
Mass of bed solids (g) Because a wide size distribution exists within the bed
MS the upper portion of the bed will have more fines, and
P,(R) Size distribution of feed solids (cm - ‘)
Size distribution function of bed drain solids the lower portion will have more coarse particles. To
P, (RI
(cm-‘) reflect this segregation in the composition of the overflow
Size distribution of bed solids (cm ‘) streams, a segregation correction factor, $(R), is
P,(R)
R Particle radius (cm) calculated’ :
Rer’ Particle Reynolds Number
R, Radius of largest particle in mixture of solids (3)
(cm)
R, Radius of smallest particle in mixture of solids $(R) was calculated from plant data as follows2’:
(cm) $(R)=O.l for R 60.0075 cm
UO Superficial gas velocity (cm s-l)
4 Terminal particle velocity (cm s- ‘) t,+(R)= 50.0 for R>0.3 cm
W Weight of solids in bed (g)
IC/(R)=464.294 R’.9047 0.0075 < Rd0.3 cm
w, Weight of bed solids (g)
2 Average reaction rate of burning char
(mol cme2 s-l) Cyclone collection ef$ciency
CI Volume of particle corresponding to 1 mol of
The cyclone collection efficiency for each size fraction
carbon (cm3)
is required in Equation (2). The correlation used here is
Abrasion constant
from ASPEN/SP’ :
Elutriation constant (s-l)
Segregation constant r(R) = v,(R)= 1 for D,/D,,>8 (4)
Collection efficiency (bed solids) q(R)=0.988-0.4863D~‘D,c for D,/D,,68 (5)
Collection efficiency (char)
Sulphur collection efficiency rlc(R) = 0.988 - 0.4863°.839609 Dp/D,,
Viscosity of gas (g cm- ’ s- ‘)
for D,lD,,<B (6)
Density (g cms3)

Limestone abrasion

APPENDIX The limestone particles abrade while in the bed. This


phenomenon has been modelled by assuming that the
Mass balance and size distribution of solids in thejluidized abrasion of non-elutriable size fractions will produce a
bed decrease in this fraction and a corresponding increase in
The following outline of the fluidized bed model, like elutriable size fractions. This process keeps the total mass
those of others16-21 is based on a formulation by Kunii supplied to the bed constant. The abrasion constant, p,
and Levenspiel’. For more detail including the Fortran was determined to be 0.16 by fitting the model to plant
code the reader is urged to consult Young6. data22.
The overall model is based on separate calculations of
the mass flowrate and size distributions of the bed solids F,(F) = F,(F) + BFo(C) (7)
and char. First, we will consider the bed solids. This F,(C) = F,(C)- BFo(C) (8)
involves the calculation of four key quantities: the
F represents the fines and C represents the coarse size
flowrate of bed drain solids; the flowrate of elutriated fractions.
bed solids; the PSD of bed drain solids; the PSD of the
elutriated bed solids. F = R < 0.0075 cm and C = R 2 0.0075 cm.

Flowrate of bed drain solids


Elutriation rate
The overall material balance around the fluidized bed
The elutriation rate of each size fraction was calculated
is given by
from Kunii’. The following equations were developed
F,=F, +F, (1) from the data of Yagi and Aochi23 presented in Kunii’:
Kunii’ developed a relationship between the flowrate of K(R)= (uo -u:W&)y (9)
bed drain solids, F,, and the feed flowrate F,:
where log,, y=[-2.0556+ 1.18642Re,(R)+0.124757
po(R)dR Re,(R)2 -0.05743 Re,(R)3].
Fl=FoRr (2) Once these terms are calculated Equation (2) can be
Rm 1+ W)U -vIW~CWIF,W)I solved. Because F, appears on both sides of Equation (2)
Other terms in the equation represent the particle size an iterative solution is required. Here we used the simple
distribution of the inlet stream; segregation correlation half-interval method in combination with an Euler
factor; cyclone collection efficiency; limestone abrasion; integration routine.

FUEL, 1991, Vol 70, February 153


Modefling of an AFBC steam heating plant: P. L. Douglas and B. E. Young

Flowrate of elutriated solids been calculated the flowrate of elutriated char is


The flowrate of elutriated solids is then calculated by calculated. The equation must consider the distribution
difference from: of sizes as well as the fact that the char particles are
shrinking”.
F,=F,-F, (10)

PSD of bed drain solids


The PSD of bed drain solids was calculated from’:
(16)

1 (11) I(D,, D,,J is calculated from”:

I@,, D~~)=expi-a;~~[~+K(D,)]d(D~)} (17)


PSD of eiutriated solids
The PSD of the elutriated solids is calculated by
difference : Elutriation rate constant
Fo(R)Fo-F,UW, The elutriation rate constant for char is calculated in
FAN= _ (14 a similar manner as that for bed solids. The only
difference is that the sphericity of the char was assumed
Rate of char withdrawal from the bed to be 0.6. This affects the calculation of the drag
coefftcient C,,, which in turn affects the calculation of the
The rate of char withdrawal from the bed is calculated
particle Reynolds number Re,.
by”:

F J.5 x 10-6[~.~F~~+F.,,,(~c’~~cf’)~ (13) PSD of bed drain char


The size distribution function for the withdrawn char
CL is calculated by lo :
(u,/100)0.93
3

Update estimates of mass of solids and char


Assuming the bed has the same composition as the
withdrawal stream and that the total bed weight is ~(DpDpJ4Dp) (18)
constant, new masses of bed solids and char are
calculated: PSD of elutriated char
%F, The size distribution function for the elutriated char
MC=----- (14) is calculated by”:
F,+F,
M,=M,-M, (15) (19)
A convergence test is performed. If M, and M, are
reasonably close to the last iteration, then proceed to the
next set of calculations, otherwise repeat the calculations Sulphur capture calculation
for the mass and size distribution using the newly The sulphur capture is based on plant data as
calculated values of M, and M,. f0110ws2*:
ylsoz= 0.95 for Ca/S > 3.5 (20)
Flowrate of elutriated char
Once the mass of char and limestone in the bed have rSoz = 0.74 + 0.0551 (Ca/S) for CaSd3.5 (21)

154 FUEL, 1991, Vol 70, February

Você também pode gostar