Sunteți pe pagina 1din 19

INTERNATIONALJOURNAL OF ENERGY RESEARCH, VOL.

19,535-553 (1995)

A DESIGN METHODOLOGY FOR SMALL-SCALE


BUBBLING FLUIDIZED-BED FURNACES
W. A. MACGREGOR, V.I. UGURSAL, AND F. HAMDULLAHPUR*

Depament of Mechanical Engineering, Technical University of Nova Scotia, Halifar, Nova Scotia, Canadn B3J u14

SUMMARY
A detailed methodology is presented for the design of small-scale fluidized-bed furnaces (SSFBF) in the 15-250 kW
capacity range for domestic space and hot water heating. These bubbling fluidized-bed furnaces burn a beneficiated
coal-water slurry mixture, and do not have heat exchange tubes in the bed or freeboard regions. Algorithms are
developed, based on this methodology, to calculate furnace design and performance parameters. The heat transfer
coefficients are calculated for the dense bed and freeboard of the SSFBF. These are then compared with theoretical
and experimental heat transfer coefficients in the published literature. Excellent agreement is found between those
in the SSFBF and those in the literature. The finding of this work is that the methodology and the heat transfer
coefficients obtained provide excellent design tools for the scale-up of bubbling fluidized-bed furnaces in the
15-250 kW range.
KEY WORDS fluidization; fluidized-bed-Combustion; domestic heating; heat transfer; furnace design

INTRODUCTION
Although the stability of price and security of the oil supply have decreased the demand for alternative
sources of energy, coal and coal-based fuels still maintain their importance for the generation of power
and heat. Coal-water slurry fuelled atmospheric pressure fluidized beds is one of the potential
alternatives to fuel oil for domestic applications. There is, however, very little published literature on the
design of small-scale bubbling fluidized beds for domestic space and hot water heating. In the not-too-dis-
tant past some work was done in Canada (Trivett et af.,1982; Hamdullahpur e? al., 1987) and in the
United States (Pagano, 19881, but neither heat transfer coefficient data nor theory were provided which
might lead to the thermal design of SSFBFs with no internal heat transfer tubes.
The problem of scale-up in a11 types of fluidized beds is well known and documented. Nevertheless,
industry needs reliable methodologies for the design and manufacture of fluidized-bed furnaces, includ-
ing reliable correlations for heat transfer. While most bubbling fluidized beds have heat transfer tubes
immersed in either the bubbling bed or in the freeboard, there is very little in the published literature on
the heat transfer in bubbling fluidized beds without immersed heat transfer tubes. The small scale
fluidized-bed furnace (SSFBF), developed at the Technical University of Nova Scotia (TUNS)in Halifax,
is just such a system (see Figure 1).The SSFBF provides an environmentally safe and efficient method for
burning indigenous coal for domestic space and hot water heating at two-thirds the cost of oil
(MacGregor et al., 1991, 1993):
In this paper a methodology for the design of SSFBF from 15 to 250 kW, including an algorithm
written (i) to calculate the heat transfer, and (ii) to design the fluidized-bed furnace are presented.
The heat transfer coefficients calculated in the model are compared with those in the published
literature, specifically with those compiled by Xavier and Davidson (19851, and with those resulting from

*Corresponding author.
'Figures based on the annual heating costs for Nova Scotia, Canada. Figures in $Can (1991).

CCC 0363-W7X/95/060535- 19 Received 9 February 1994


0 1995 by John Wiley & Sons, Ltd. Reuised 30 March 1994
536 W. A. MACGREGOR, V.I. UGURSAL AND F. HAMDULLAHPUR

Figure 1. Schematic of small-scale fluidized-bed furnace

Kunii and Levenspiel's (1991) general equation for the heat transfer coefficient in gas solid contactors.
Excellent agreement is found between the heat transfer coefficients in the literature and those calculated
for the SSFBF.

COMPONENTS OF THE SMALL-SCALE FLUIDIZED-BED FURNACE (SSFBF)


In an effort to provide a means for clean and efficient utilization of coal in small-capacity furnaces,
SSFBFs which burn coal in slurry form were developed at the Technical University of Nova Scotia
(TUNS) (Hamdullahpur and MacKay, 1986; Hamdullahpur and Trivett, 1985a, 1985b; Hamdullahpur et
al., 1989, 1986, 1987). The CWS-fired SSFBF comprises six main parts:

Combustor
Combustion of the CWS takes place in the dense bed and the heat of combustion is transferred to the
water jacket surrounding the bed and the freeboard. The fluidized-bed combustion zone includes an air
distributor plate, a sand bed (sand particles of 450 or 600 micrometres in diameter) and a freeboard
region in which second-stage combustion reactions and heat transfer take place. A uniform temperature
distribution is maintained within the bed and heat is transferred from the bed to the surrounding walls
and to the water jacket by the combined effects of radiation, convection, and particle conduction.

Air deliuery system


The air delivery system consists of a centrifugal blower, flow sensors connected to the automatic control
system, and associated duct work. Air is delivered into a plenum, and is forced upwards through the
distributor plate.

Fuel delivery system


The CWS fuel is stored in a fuel tank. A positive displacement pump delivers the fuel from the tank to
the fuel nozzles which are located at the top of the combustor. The fuel is forced through the nozzles in
the form of small droplets. These droplets fall through the freeboard region and splash onto the bed
where the turbulent action of particles suck the droplets into the bed.
DESIGN METHODOLOGY 537

Emission controks
The hot combustion gases laden with the particular matter (ash) leave the combustor through the
freeboard and enter the secondary heat exchanger where more heat is transferred to the surrounding
water jacket. Two water-jacketed cyclone separators are placed immediately downstream of the secondary
heat exchanger. These cyclones can capture all the ash and unburned fuel up to 10 micrometres in
diameter. Usually, particles smaller than 10 micrometres escape through the cyclone. A bag filter is used
to capture these particles. The ash and unburned carbon collected by the cyclones are dropped into a
small ash storage tank. The content of the small storage tank is pneumatically transported into a bigger
storage tank away from the combustor site. The large ash storage tank is replaced by an empty tank
during refuelling.
Control of SO, emissions is accomplished by adding limestone to the fuel before it is pumped into the
bed. Thus, the SO, components formed during the combustion react with limestone to form calcium
sulfate which is removed by the particulate collection equipment described above. The location and
mechanism for limestone addition, and the optimum Ca/S ratio were determined as a result of extensive
testing (Hamdullahpur et al., 1986).
The reduction of NO, and CO emmisions is accomplished by injecting secondary swirling air flow into
the freeboard. the use of swirling air results in improved combustion intensity and calcium utilization,
reduced freeboard height, and a reduction in both elutriation, and gaseous emissions, primarily of CO
and NO,.

Automatic dust /ash collection Jrstem


A positive pneumatic transport system is used in this design. The pressurized air for this system is
supplied by the combustion air centrifugal blower. The pressurized air for this system is supplied by the
combustion air centrifugal blower. The ash removal system is activated when the bed is slumped.

Controller
A microprocesser-based control unit is used to operate the system fully automatically. The overall
function of the controller is to initialize ignition, maintain the bed temperature in the operating range,
maintain a preset water jacket temperature, and quickly respond to demands for heat from a central
thermostat. In addition, dust/ash collection, and cleaning of the fuel nozzles during slump cycling are
performed by the controller. On critical conditions such as ignition failure or mechanical or electrical
malfunction, the controller stops the operation of the bed with appropriate indication of the alarm
condition.

SSFBF DESIGN METHODOLOGY


Once the design heat load of the building or space has been determined, the fuel flowrate and amounts
of combustion products are calculated. These products of combustion are then assumed to be the
fluidizing gas, and a perforated plate distributor is then designed (Basu, 1984). Next, the heat transfer
calculations are completed, and an energy balance is carried out. The heat transfer coefficients calculated
in the model are then compared with published data.

Thermal design of SSFBF


This section contains details of the thermal design of the SSFB. Three algorithms were written for the
thermal design:

(a) COMBUST (for combustion and fluidization calculations),


(b) DISTRIB (for distributor design calculations), and
(c) HEATRNSF (for heat transfer and energy balance calculations)
538 W. A. MACGREGOR. V. I. UGURSAL AND F. HAMDULLAHPUR

Combustion and fluidization calculations (COMBUST)


Once the design space heat load of any space has been determined, furnace design can be initiated. The
SSFBF design procedure makes the following preliminary assumptions:

(a) design efficiency is 85%,


(b) combustion efficiency is 98%
(c) sulfur capture efficiency is 90%, and any uncaptured sulfur is lost as SO,,
(d) negligible NOx are formed
(e) the products of combustion constitute the fluidizing gas,
(f) secondary air is not used, and
(g) excess air (EA) is 20% for small boilers ( I 55 kW) and 30% for large boilers (> 55 kW).

The mass flowrates of the combustion products are determined based upon the fuel's ultimate analysis
and the actual air-to-fuel ratio. These combustion products are assumed to constitute the fluidizing gas at
the bed design temperature. The minimum fluidizing velocity (Umf) is then calculated using the transport
and thermodynamic properties of the fluidizing gas at the bed design temperature:

where
Re,, -4
= - C,

C , = 27.1 and C , = 0.0408, (Grace, 1982) and


Pf( P, - Pf) gd;
Ar = (Archimedes number) (3)
4
A superficial velocity multiplication factor (SVM) is used to obtain a range of superficial velocities.
Later on, in concert with the heat transfer calculations, terminal velocity calculations, and the size
restrictions placed on the SSFBF, the most suitable U value is chosen:
u= Urn,sw (4)
The volume flow rate (QeJ of the exhaust gas, and the surface ( A b e d ) , and diameter @ b e d ) of the
fluidized bed are then calculated. Since the mass and volume flow rates of the SSFBF are fixed at any
steady-state condition, increasing the superficial gas velocity can be accomplished by decreasing the
cross-sectional area, and consequently the diameter, of the fluidized bed. If, however, the superficial gas
velocity exceeds the terminal velocity of the particle, the bed material may be entrained from the system
with subsequent emptying of the bed, a highly undesirable condition. The superficial velocity ( U ) must be
less than the terminal velocity (V,,,,) of the particles in the fluidized bed. Assuming a rigid spherical
particle, the Reynolds number, drag coefficient, and terminal velocity are calculated:
Pex d p u
Re = -
Pe X
24
c d = E (Re < 0.4) (6a)

(6b)

The terminal velocity is then compared with the superficial gas velocity to determine whether that
particular superficial gas velocity will result in substantial entrainment.
DESIGN METHODOLOGY 539

The mean bubble size formed at the distributor (d,,o) and the maximum bubble size attainable by
coalescence ( d , ), are estimated, using the correlation of Mori and Wen (Davidson et af.,1985). Both the
bed height ( H m f and
) the voidage at minimum fluidization ( E m f ) are design parameters.
d,, = 1-38 g-'" [ A,(U - Umf>] (8a)
0.4
d,, = 1-49 [Died (u- umf)] (8b)
The height of the expanded bed (Hex,) may then be determined from a particle balance:
( 1 - E m f ) H m f= ( 1 - &>He*, (9)

An iterative procedure is used to determine both the bubble size and the height of the expanded bed.

Hi=Hmr 1 + ;a)
The iterative process for the bed height begins with an estimation for the height of the expanded bed, Hi,

(
and continues with an estimation of the bubble sue at the surface of the expanded bed,
(11)

The iterated values (Hi and d i ) are then compared with the previous values (Hmfand d,). If sufficient
convergences has not been achieved, the value of d , from equation (12) is substituted for d , in equation
(111, and both equations ( 1 1 ) and (12) are solved repeatedly until a sufficient level of convergence has
been reached. At that point the height of the expanded bed is set equal to the iterated value of the bed
height (H,,, =Hi) and the bubble diameter at the surface of the bed is set equal to the value of the
iterated bubble diameter:
d, =d, (13)
Hamdullahpur and MacKay (1986) give the transport disengagement height (TDH) as
TDH = 12db ( 14)
where d, (bubble diameter) is calculated at the bed surface. Using equations (8) through (14), the
expanded bed height, mean bubble size at the surface and TDH can be determined for any superficial gas
velocity.
To ensure the slugging velocity is not reached, the minimum slugging velocity (Urns) must be
calculated. Davidson et al. (1985) suggest
Urn,= Umf+ 0-07\/= if Hmf> 1.3D~;'~75 ( 15)
otherwise
Urn,= Umf+ 0 - 0 7 4 z +O-16(1.3D~;1d75
-Hmf)' (16)

Furnace construction and dimensions


Having determined the diameters of a range of furnaces, the remaining dimensions of the furnace can be
specified. For design conditions, the freeboard height is taken to be 20% greater than the TDH,
Hf,,, = 1.2TDH (17)

'The convergence level is an output to the computer model. In this work, the convergence level for both bubble size and bed height
was 1 mm.
540 W. A. MACGREGOR,V. I. UGURSALAND F. HAMDUUAHPUR

Now that a range of furnace sizes have been determined for the design conditions, the heat transfer
for each of those furnaces must be calculated.

Heat transfer calculations (HEATMSF)


Heat transfer in the SSFBF is modelled assuming that heat is transferred from the dense bed and
freeboard regions to the water jacket by convection and radiation. Heat transfer in the dense bed is
calculated first, followed by heat transfer in the freeboard region

Convective heat transfer in the dense bed: The convective heat transfer coefficient in the dense bed (h,)
can be approximated as the sum of the particle convective (hPJ and the gas convective ( h p , )components.
Xavier and Davidson (1985) suggest that the method of Carslaw and Jaegar (1959) will give the solution
within 5% accuracy by considering the contact or film resistance l / h f in series with the average packet
resistance 1/ h p . The particle convective heat transfer coefficient then becomes

where, from two-phase theory,

= 0'71 (20)
and d, is the bubble diameter. The bubble diameter at half the expanded bed height was calculated using
equation (12). This value was then used in equation (20) to provide the bubble rise velocity.
The thermal conductivity of the fluidizing gas and the particle diameter greatly influence the surface to
bed heat transfer. The film heat transfer can be expressed as

and while values of rn (a constant) range from 4 to 10 in the literature (Xavier and Davidson, 19851,
m = 6 is used for design purposes. The average heat transfer between the particulate phase and the
surface in the absence of a film resistance is given by

When calculating h, using this equation, the smaller of L and 8 is used. L is the characteristic vertical
dimension of the heater or surface, and 8 is the bubble sue. In the SSFBF there are no immersed
surfaces and all internal heat transfer occurs at the bed wall. Within the bed itself, the bubbles tend to
move to the centre as they rise with particle downflow at the wall, hence equation (22) reduces to

with
(24)
(25)
and
DESIGN METHODOLOGY 541

Xavier and Davidsion (1985) give the gas convective heat transfer coefficient, hgc,as
0-3Ar0'39k
( l o 3<Ar < 2 X lo6) (27)
hgc =
m
The minimal Archimedes number (lo3) is reached in the SSFBF with sand of approximately 600 p m in
diameter; however, the lack of any other correlation in the literature necessitates the use of this
correlation for diameters of sand as small as 200 pm. For 600 pm sand, h,, is but 10% of hcb,so the
error introduced is rather small. The overall convective heat transfer coefficient, hcb, can now be
expressed as the sum of the particle convective (hPJ and gas convective (hgc)components.
hcb = hpc + hgc (28)

Radiative heat transfer in the dense bed: Radiative heat transfer in the bed becomes significant above
600 "C.To determine the radiative heat transfer between the fluidized bed and the refractory wall, both
the bed and the wall are considered to be diffuse grey surfaces exchanging energy with each other. The
resultant heat transfer (q,,,,) becomes

Using the Stefan-Boltzmann law and assuming that

(a) the fluidized bed to refractory wall view factor (Fb,+ is equal to 1.0,
(b) the surface area of the refractory wall lining (A,) 1s equal to v D b e dHexp,and
(c) the surface area which the fluidized sand bed presents to the refractory wall ( A , ) is equal to the
surface area of the refractory wall presented to the sand bed (A,)

equation (29) simplifies to


5.67 X ( T t - T;)(E,E,)A,
4net = Es + &, - ESEW (30)

Assuming a refractoq wall temperature slightly lower than the bed temperature (approximately 150 K)
will result in a first approximation of qne,.At any instant, the radiative heat transfer coefficient between
the refractory wall and the fluidized bed ( h , ) can be calculated as

The bed to refractory lining heat transfer coefficient then becomes the sum of the convective and
radiative coefficients:
hbw = hcb + hrb (32)
The bed to wall thermal resistance is then calculated by
R b w = hbw (33)
Assuming steady-state operation, the heat transferred from the bed to the refractory must then be
transferred to the water jacket. The thermal resistances of all other components are calculated and then
summed to provide overall thermal resistance (R,,,), and subsequently a new value for q;,,, the first
iteration in the heat transfer solution:
Rtot= Rsw + R f s o + R s + Rws + Rw + Rbw (34)
542 W. A. MACGREGOR, V. I. UGURSAL AND F. HAMDULLAHPUR

An iterative process was carried out using equations (30) through (35) until the qher did not change by
more than 5% through one iteration.
The convective heat transfer coefficient from the stack gases to the steel jacket is relatively simple to
calculate, while the radiative heat transfer coefficient is much more complex, owing to the interactions of
the sand particles, irradiation from H,O and C 0 2 , and radiation from the surface of the sand bed to the
freeboard region and the steel jacket. To estimate the radiative heat transfer in the freeboard region, the
Stefan-Boltzmann law is used, and an overall freeboard emmitance value is assumed for design purposes.

Heat transfer within the freeboard region: Both convective and radiative heat transfer occur within the
freeboard region. The average gas velocity in the freeboard region is quite low, resulting in laminar flow
within the freeboard, and consequently a low interior heat transfer coefficient. The radiative heat
transfer is somewhat more complicated. There are three regions of interest, each with their own average
temperature:

(a) the surface of the bubbling bed at the design temperature,


(b) the steel jacket in the freeboard region, whose temperature decreases with furnace height, and
(c) the stack gases, whose temperature decreases with increasing height, but act as interactive media, i.e.
they both absorb and radiate heat in the freeboard region.

Convective heat transfer in the freeboard region: The heat transfer across the steel lining in the freeboard is
determined by calculating the overall heat transfer coefficient from the stack gases to the water in the
water jacket. The mass flowrate of water in the water jacket (mw) is calculated using the design heat load:

The velocity and Reynolds number of the water in the jacket are calculated. The very low water velocity
results in laminar flow, and the Nusselt number in the water jacket is set to 4.0.
The steel jacket to water heat transfer coefficient ( h , ) is determined:

The same procedure is followed to calculate the stack gas to steel liner heat transfer coefficient ( h i ) ,
though the stack gas temperature used in the calculations is the average of the bed design temperature
and the assumed exit gas tempcrature. The velocity and transport properties of the stack gases are
calculated at this average temperature:
Nu,,= 4.0 (for laminar flow) (38)
Nu,,= 0-023 Re4/5Pr1/3 (for fully developed turbulent flow) (39)

The overall convective heat transfer coefficient can be calculated as


-1
Do +-R
Do DO
i+-ln (41)
Dihi Di f, 2k,,
The fouling resistances Rf, and Rf, are assumed to be negligible. Typically, the steel jacket is 4 mm in
thickness and greater than 300 mm in diameter, so Do/Di can be approximated as 1.0, reducing equation
(41) to
Uo= ( l / h i + l / h 0 ) - ' (42)
DESIGN METHODOLOGY 543

The log mean temperature difference is calculated as

where ATl is the Tbedminus the entering water temperature and AT2 is the assumed stack exit gas
temperature minus the exiting water temperature. The area of the freeboard is assumed to be the area of
an open-ended cylinder with diameter Dbed and height Hexp.The convective heat transfer from the
freeboard region to the water jacket is thus calculated as

Radiatiue heat transfer in the freeboard region: The method for calculating the radiative heat transfer in the
freeboard region is similar to that used in the fluidized-bed region. The steel jacket, and the CO, and
H,O are considered to be diffuse reflecting surfaces. In the literature, the emittance of the stack gases
with no solids is typically 0-14 to 0-18 (Lindsay et al., 1986); however, the freeboard emittance increases as
the amount of solids in the stack gas increases, with the freeboard emittance reaching 0.6 at a solids
loading of 1-0 kg/m3. For design purposes an overall freeboard emittance value of 0.25 is assumed. When
the SSFBF is operating at full load, the freeboard region will have a higher solids loading, and thus will
have a higher emittance. An assumed emittance of 0.25, through somewhat conservative, will ensure that
sufficient freeboard surface is available for heat transfer. The radiative heat transfer from the exhaust
gases to the water jacket can be calculated as
qf, = 0.25 X 5.67 X (45)
A, (T: - T,')
Then the total heat transfer in the freeboard (qf) can be expressed as
qf = qf, r + S f , conu (46)

Total heat transfer: The total heat transfer to tha water jacket (qro,) is the sum of the heat transfer in the
fluidized bed and in the freeboard region given by equations (36) and (47).

Energy balance analysis methodology


Verification of the thermal design can be achieved by conducting an energy balance on the SSFBF. If the
energy balance contradicts the thermal design, then the principle of conservation of energy has been
violated, and the design calculations must be repeated.
To calculate the heat transferred to the SSFBF, the higher heating value and mass flowrates of the
fuel, and the specific enthalpy and mass flowrates of combustion air, and combustion products need to be
specified. The power (kW) required to heat each of the exhaust gases from 25 "Cto the assumed exit gas
temperature (Texj,)is calculated using enthalpy data
(Stack loss) = X,[(h,- h&,, J (48)
The stack losses due to fly-ash and SO, were not included since they constitute less than 1% and 0.1%
respectively, of the entire mass of the system. Water constitutes 30% of the fuel mass, and is included in
the energy balance calculations.
q,,,, i s t = mf Hw, r),,, - (Stack 10SS) (49)
- (50)
q n e t , 1st - q u a i / - (Stack loss)
544 W. A. MACGREGOR,V.I. UGURSAL AND F. HAMDULLAHPUR

This value of qnel,Ist is compared with the calculated heat transfer (qto,),and if it is within 5% of qto,,
then it is assumed that the principle of conservation of energy has not been violated and the design is
valid. If qne,,Ist is not within 5% of qto, then the principle of conservation of energy has been violated
and the SSFBF should be redesigned with better assumptions.

Comparison with published heat transfer coeficients


The heat transfer coefficients obtained from the model were compared with those of Xavier and
Davidson (19851, who estimate the heat transfer for an immersed cylinder and the maximum convective
heat transfer. The overall heat transfer coefficient in the dense bed was also compared with results
obtained using a correlation proposed by Kunii and Levenspiel (1991). Good agreement was obtained in
all cases.

Gas conuective heat transfer: Xavier and Davidson (1985) demonstrate a relationship between

and Rep&, where h,, is the gas convective heat transfer coefficient. They suggest the use of

For a flat plate, the second term ( k a / 2 r s )would equal zero. In the SSFBF, 2rs is twice the radius of the
heat transfer surface, which is the diameter of the bed (Dbed).However, the bed radius is of the order of
0.5 to 1.5 metres, and this second term represents less than 2% of the h,, as calculated. Thus, in the
SSFBF h,, could be approximated by

However, to allow for the effect of heater length in the SSFBF, the correlation given by Xavier and
Davidson (1985) was employed (equation (27)).In the SSFBF the Archimedes number ranges from 650 to
950, which is less than the lower limit on the correlation. However, nothing else could be found in the
literature which suggested how to account for the length of the heat transfer surface. Given that h,, is
but 5% of the heat transfer coefficient in the dense bed, and that the Archimedes number in the SSFBF
is close to the lower limit of equation (28), the error introduced was assumed to be quite small. The
agreement between the calculated SSFBF heat transfer coefficients and those in the literature later
validated this assumptiong

Maximum convective heat transfer


For the case of very small particles, Xavier and Davidson (1985) calculate the maximum convective heat
transfer coefficient as

where L in the equation is the smaller of the column diameter or vertical heater dimension. In the
SSFBF, L would always be the height of the expanded bed, Hexp.
However, as the particle size increases, a film resistance term must be included. Assuming that
U - Um = ub for maximum heat transfer, a unique relationship appears between
(54)

'Note that above h,, a 1/ 6and as bed depth increases, h,, as calculated does in fact decrease.
DESIGN METHODOLOGY 545

suggesting the use of

The results of Xavier and Davidson and of our model using Equation (54) are presented in Figures 2(a)
(450-pm particles) and 2(b) (600-pm particles)." The broken lines in these figures mark f 40% from their
theoretical results.ll More than 90% of the experimental values shown in their figure lie within the dashed
lines. As can be seen in Figures 2(a) and 2(b), the values calculated for equation (54) using the present
model for the SSFFJF are in close agreement with both the theoretical results and the findings of many
experimenters. Using either equation (27)or equation (52) to calculate the film heat transfer coefficient,
h f , provides no significant difference in the results. It is interesting to note that while the values
computed for equation (54) are a function of SSFBF capacity (and thus, superficial velocity, bed voidage,
etc., etc.) they do not appear to be influenced by bed design temperature, remaining essentially
unaffected by design temperature, changing by less than 2% from 800 to 950 "C.
The film heat transfer coefficient ( h f )for the SSFBF is calculated using

with values of m ranging from 4 to 10 in the literature. Xavier and Davidson recommend that rn = 6 be
used for design purposes, noting that m = 6 for horizontal cylinders and spheres, while m = 10 for
vertical surfaces. A sensitivity analysis was carried out to determine the effect of the value of m.Setting
m = 6 gave more conservative results which agreed more closely with those of Kunii and Levenspiel.
Therefore m = 6 was used in this model.

Ovemll heat transfer coefficient: Once the thermal design was completed, and the first law analysis was
satisfactory, the overall heat transfer coefficient in the dense bed was calculated using
q'net
- T,)
hbw = A, (Tb
(57)

where A, is the surface area of a right circular cylinder of diameter D b e d and length ITexp.
This value is compared with that calculated using Kunii and Levenspiel's (1991)
h'bw = [ +hg)] bubble at surface

1 ( h p a c k e t ) ( hr 2k,9, + awepg
+ dp pguo
9 1.0 \ I
I emulsion a t surface
Aemulsion
where

The bubble frequency term n, is given by Fung and Hamdullahpur (1993):

'The data points shown in these figures are for selected SSFBS, and represent single design points, specifically, that bed diameter
which results in a superficial gas velocity approximately 20% below the slugging velocity. The rationale for this design decision is
explained in the section titled 'Consideration and results of SSFBF design'.
"Thesedashed lines are included in Xavier and Davidson and Davidson (1985); see their Figure 13.6, page 454.
546 W. A. MACGREGOR. V. I. UGURSALAND F. HAMDULLAHPUR

1 I 1
I . 1 1 1 I I I 1
J '/
-
hcmax
hP=

0.5

hf + hscc
hf + 2hpmax hpmax

0.1
0.0s 0.1 0.5 1
(4
1 T I I I I I I

-
hcmax
hpmax

0.5

0.1 I I I , * . I
0.05 0.1 0.5 1
(b)
Figure 2. Comparison of maximum convective heat transfer coefficients in SSFBF with published values. (a) 450-pm particles; (b)
600-pmparticles. (c) 0,15 kW (equation (53)); 0,15 kW (equation (28)); A , 250 kW (equation (53)); A , 250 kW (Equation (28)).
DESIGN METHODOLOGY 547

600

m
401)

300

zm
im
" I ' I 0
0.4 05 0.6 0.7 0.8 0.9 1.0 03 0.4 03 0.8
(P) k(P) 0.7
Figurc 3n (@JOT) Figure 3c (W0"C)

0.4 03 0.6 0.7 08 0.9 03 0.4 0.6 0.1


(P) Re0&

.,
Figure 3b (850°C) Figure 3d (USO'C)
Figure 3. Heat transfer coefficient versus Re (450-pm particles). (a) 800 "C; (b) 850 "C;(c) 900 'C; (d) 950 "C A , 15 kW (MacG);
A , 15 k d (K&L); 0 , 2 5 0 kW (MacG); 250 kW (K&L)

The bubble diameter ( d , ) used in equation (60) is calculated at half the expanded bed height from
equation (12). The heat transfer coefficients from eauation (57) are then compared with those from
equation (58).

Design conditions
The following design conditions were modelled:

(a) particle size: 450 p m 600 p m


(b) design heat load: 15 kW 55 kW 140 kW 250 kW
(c> bed design temperature: 800 "C 850 "C 900 "C 950 "C
(d) m (factor in equation (21)): 6 10

The results obtained from those simulations are compar d to those from Kunii and Levensuiel(199 ) and
are presented in Figures 3 and 4. Figures 3(a) to 3(d) are the results for 450-pm particles: while Figures
4(a) to 4(d) are for 600-pm particles. There is excellent agreement between this work and the results are
predicted by equation (58). Equation (58) exhibits the well-documented dependence of heat transfer
coefficient on Reynolds number. However, this correlation does not account for the size of the heat
transfer surface. The decrease in h as heat transfer surface size increases is also well documented, as is
the fact that h approaches a constant vaIue as the heat transfer surface length increases beyond about 15
548 W. A. MACGRELGOR, V. I. UGURSALAND F. HAMDULLAHPUR

1.0 12 1.4 1.6 18 20 22 24 08 1.0 If 1.4 1.6 IB 20


Re 0 Re (P)
Figure 4a (800°C) Figurc 4c ( W C )

18 12 1.4 1.6 I8 20 22 08 1.0 1.2 %(,,) 1.4 1.6 18


(P)
Figure 4b (850°C) Figure 4d (950°C)
Figure 4. Heat transfer coefficient versus Re, (600-pmparticles). (a) 800 "C; (b) 850 "C;(c) 900 "C;(d) 950 "C.

em. Our work shows an essentially constant value for h. This is to be expected, given that the heat
transfer surface is the refractory lined wall of the combustor, which for the 250 kW unit is 60 to 80 cm
high, and more than 1 m in diameter.

Considerations and results of SSFBF design


Clearly, the selected SSFBFs must fit into the building for which they are purchased. Since large
apartment buildings generally have mechanical rooms with larger entrances than do houses, the SSFBF
in an apartment building can have greater physical dimensions than can the SSFBF in a house or
four-unit apartment building. Accordingly SSFBS for house and four-unit apartment buildings were
limited to an assembled width of 71 ern (in order to fit through a standard 76 cm wide door) and a total
height of less than 122 cm (to fit into basements). In the small domestic units (I 55 kW) it is assumed
that the refractory lining, the water jacket and the exterior casing with insulation are all 2-5 cm thick.
When constructed using these dimensions, the diameter of these domestic units was limited to 64 cm.
Given the large mechanical rooms in larger appartment buildings, sue restrictions on the SSFBF were
not as critical, and bed diameters ranged from 70 ern to 151 ern for these structures. The dimensions of
the SSFBFs for all sizes of buildings are given in MacGregor et al. (1991).
Given that the mass (and hence volume) flow rate of air into the SSFBF is fixed at any design
temperature by the design heat load, clearly, a smaller bed diameter is highly desirable from structural
design, manufacturing and installation perspectives.
The limit on reducing the bed diameter is reached when (i) the bed begins to slug, (ii) significant
DESIGN METHODOLOGY 549

Table 1. Selected SSFBF dimensions for 450-pm particles

Bed design Heating Diameter Total Heat transfer coefficients Difference


temperature capacity (kW) bed (m) height (m) eq. (79) eq. (80)

850 "C 15 033 090 433 478 -5%


250 1*40 2.59 315 360 - 8%

900 "C 15 034 080 478 503 - 5%


250 158 2.04 394 405 - 3%

Table 2. Selected SSFBF dimensions for 600-pm particles

Bed design Heating Diameter Total Heat transfer coefficients Difference


temperature capacity (kW) bed (cm) height (m) eq. (79) eq. (80)

850 "C 15 029 0.93 437 436 0%


250 1.20 321 321 345 - 5%

900 "C 15 030 091 479 464 +3%


250 1.21 3.05 362 381 -5%

entrainment occurs, or (iii) the SSFBF becomes too large (diameter or height) and will not fit into the
space for which it is purchased. As far as practical bed operation is concerned, the optimum design point
occurs when the fluidization velocity is about 20% below the slugging velocity. The design factor ensures
the bed will not slug, while still achieving a small bed diameter. The dimensions of selected SSFBFs
designed using our methodology are shown in Table 1 (450-pmparticles). These data correspond to
Figures 3(b) and 3(c) (for 450-pm particles) and Figures 4(b) and 4(c) (for 600 p m particles). The heat
transfer coefficients calculated using equations (58) and (59) are also listed for each design.** The final
column in Tables 1 and 2 headed ('Difference') is calculated using equation (62) and represents our heat
transfer coefficient compared to that of Kunii and Levenspiel. As can be seen from Tables 1 and 2, our
results are very close to those predicted by Kunii and Levenspiel's General Equation.

(Equation (57)) - (Equation (58))


Difference = x 100% (61)
(Equation (58))

CONCLUSIONS
The findings of this work can be summarized as follows:

(a) A new procedure has been developed to calculate the heat transfer coefficients in bubbling
fluidized-bed furnaces with no internal heat transfer surfaces. This procedure includes both thermal
and mechanical design procedures for the scale-up of SSFBFs from 15 to 250 kW capacity.
(b) This procedure is validated by the close agreement found between the heat transfer coefficients
calculated in the design and the theoretical and experimental values in the published literature. This

** The 55-kW unit did not meet these size restrictions for either particle diameter at 800 "C. We feel that, given the small
proportion of 55-kW units ( > l%),this is not a significant hindrance to commercialization. Further details can be found in
MacGregor et a[., 1993.
550 W. A. MACGREGOR, V. I. UGURSAL AND F. W D U L L A H P U R

is significant since the heat transfer surface in the SSFBF is 50 to 100 times the diameter and as
much as 10 times the height of the probes generally used to measure heat transfer in experimental
units.
(c) The SSFBFs are of a practical size for the residential and commercial housing markets.

It can be concluded from these results that the design methodology is valid for the scale-up of
bubbling fluidized-bed furnaces in the 15-250-kW range. Previous research has shown that CWS-fired
SSFBFs are economically superior to oil-, wood- and coal-fired domestic space and hot water heating
systems (MacGregor et al., 1991, 1993). Their low emission levels, practicality and favourable economics
should encourage countries with coal resources to focus on this environmentally friendly technology.

ACKNOWLEDGEMENTS
This work was funded in part by a grant from Canada Mortgage and Housing Corporation, under the
terms of the External Research Program. W. A. MacGregor also received funding from Canada
Mortgage and Housing Corporation in the form of a University Scholarship for Graduate Studies.

NOMENCLATURE

= coal-water slurry (fuel for the SSFBF)


= small-scale fluidized-bed furnace
= transport disengagement height, m
= cross-sectional area of a fluidized bed of diameter Dbed,m
= freeboard area, m2
= Archimedes number
= surface area of the expanded bed in contact with the refractory wall, m2
= surface water of the refractory wall in contact with the expanded bed, m2
= cross-sectional area of the water jacket, m2
= area of distributor associated with each orifice (reciprocal of No,), m2
= fluidizing gas heat capacity, kJ/kg K
= particulate phase heat capacity, kJ/kg K
= specific heat capacity of water, kJ/kg K
= specific heat capacity of bed material, kJ/kg K
= drag coefficient
= bubble diameter, m
= mean bubble size at the bed surface, m
= diameter of a bubble formed at the distributor, m
= maximum bubble size attainable by coalescence, m
= iterated mean bubble size at the bed surface, m
= orifice diameter, rn
= particle diameter, m
= diameter of the bed, m
= freeboard diameter, m
= inside diameter of the steel jacket, m
= outside diameter of the steel jacket, m
= black-body emissive power of the fluidized bed, W
= black-body emissive power of the refractory surface, W
= bed to wall view factor
= fluidized bed to refractory wall overall heat transfer coefficient, W/m2 K
DESIGN METHODOLOGY 55 1

= overall heat transfer coefficient (Equation (80), W/m2 K


= fluidized bed to refractory wall convective heat transfer coefficient, W/m2 K
= gas film heat transfer coefficient, W/m2 K
= gas convective heat transfer coefficient within the fluidized bed, W/m2 K
= stack gas to steel jacket heat transfer coefficient, W/m2 K
= steel to water heat transfer coefficient in the freeboard region, W/m2 K
= average heat transfer coefficient between particulate phase and surface in the absence of
film resistance, W/m2 K
= particle convective component of h c B ,W/m2 K
= fluidized bed to refractory wall radiative heat transfer coefficient, W/m2 K
= maximum value of h,, w/m2 K
= maximum value of hpc in the absence of a film resistance, W/m2 K
= specific enthalpy of component of exhaust gas (CO,, H,O, etc.), W/kg K
= specific enthalpy of component of exhaust gas (CO,, H,O, etc.) at 25 "C, kJ/kg K
= height of the expanded bed, m
= iterated height of the expanded bed, m
= height of the bed at minimum fluidization, m
= fuel higher heating value, kJ/kg
= apparent thermal conductivity of packed beds, W/m K
= effective thermal conductivity of packed beds with motionless fluid, W/m K
= thermal conductivity of exhaust gases, W/m K
= thermal conductivity of gas, W/m K
= thermal conductivity of particulate phase, W/m K
= thermal conductivity of sand, W/m K
= thermal conductivity of the steel jacket, W/m K
= thermal conductivity of the water in the water jacket, W/m K
= characteristic vertical dimension of heating surface, m
= air mass flowrate
= mass flowrate of the exhaust gases, kg/s
= fuel mass flowrate, kg/s
= mass flowrate of water in the water jacket, kg/s
= bubble frequency, s-'
= number of orifices
= Nusselt number of the stack gases
= Nusselt number of the water in the water jacket
= heat transferred to furnace assuming reactants and products at 25 'C, W
= design space heat load, W
= total heat transfer from the exhaust gases to the steel jacket, W
= convective heat transfer from the exhaust gases to the steel jacket, W
= radiative heat transfer from the exhaust gases to the steel jacket, W
= radiative heat transfer from fluidized bed to refractory wall, W
= iterated total heat transfer from fluidized bed to water jacket, W
= heat transfer from fluidized bed to water jacket from the approximate first law analysis, W
= total heat transfer from the furnace to the water jacket, W
= volume flowrate of exhaust gases at the bed design temperature, m3/s
= fluidized bed to refractory wall thermal resistance, m2 K/W
= fouling resistance on the steel from the stack gas, m2 K/W
= fouling resistance on the steel from the water, m2 k/W
= thermal resistance of the steel jacket, m2 K/W
= steel jacket to water thermal resistance, m2 K/W
= total thermal resistance (sum of all thermal resistances), m2 K/W
552 W. A. MACGREGOR, V.I. UGURSAL AND F. HAMDULLAHPUR

= thermal resistance of the refractory wall, m2 K/W


= refractory wall to steel thermal resistance, m2 K/W
= minimum fluidization Reynolds number
= thickness of the water jacket, m
= bulk absolute temperature of the sand bed, K
= average absolute stack gas temperature, K
= average absolute steel temperature in the freeboard, K
= absolute temperature of the refractory wall surface, K
= log mean temperature difference, K
= difference between high and low water temperature set points, K
= design temperature minus low water temperature set point, K
= exhaust gas temperature minus high water temperature set point, K
= rise velocity of bubble or slug in isolation, m/s
= superficial gas velocity, m/s
= superficial gas velocity for minimum fluidization, m/s
= minimum slugging velocity, m/s
= superficial velocity of air, m/s
= overall convective heat transfer coefficient in the freeboard region, W/m2 K
= terminal velocity of particle, m/s
= velocity of water in the water jacket, m/s
= bed voidage
= fraction of the bed volume occupied by bubbles or slugs
= voidage at minimum fluidization
= emissivity of the bed material
= emissivity of the refractory wall
= actual combustion efficiency
= bubble or slug length, m
= viscosity of exhaust gases, Pa s
= density of the exhaust gases at the bed design temperature, kg/m3
= density of the fluidizing gas, kg/m3
= particulate phase density, kg/m3
= particle density, kg/m3
= viscosity of fluid, Pa s

REFERENCES
Basu, P. (Editor) (1994). Fluidized bed boilers: design and application, Pergamon Press, Toronto.
Davidson, Clift, and Hamson (Eds) (1985). Fluidization, 2nd edn., Harcourt, Brase, Jovanovich, New York.
Fung, A. and Hamdullahpur, F. (1993). A gas and particle flow model in the freeboard of a fluidized bed based on bubble
coalescence, Powder Technology, 74 121-133.
Hamdullahpur, F. and MacKay, G. D. M. (1986). Two-phase flow in the freeboard of a fluidized bed, MChE J., 32,2407-2055.
Hamdullahpur, F. and Trivett, G. S. (1985). Extended performance tests of small scale AFBC using coal/water mixtures,
Proceedings of the Fifth International Workshop on Coal-Liquid Fuels Technology, Halifax, NS, 1985.
Harndullahpur, F. and Trivett, G. S. (1985). A multi-functional control system for CWM slurry combustion in a small scale fluidized
bed, Proceedings of the Second Annual Coal Conference, Pittsburgh, PA, 1985.
Hamdullahpur, F., Zayed, R.S., and El Khawaja, E. (1989). NO, reduction in fluidized bed combustors, Proceedings of the
American Flame Research Committee’s 1989 International Symposium on Combustion in Industrial Furnaces and Boilers, New
Jersey, September 1989.
Hamdullahpur, F., Trivett, G. S. and Zhou, Z. Q. (1987). Design and operation of a CWS fueled small scale atmospheric fluidized
bed combustor, Proceedings of the 9th International Conference on Fluidized Bed Combustion (Poster session), ASME, Boston,
MA. May 1987.
Hamdullahpur, F., Trivett, G. S., Boak, R. B. and Campbell J. C. (1986). Performance tests of a light industrial (150 kW) AFBC
fueled with coal-water slurry, Proceedings of the Sixth International Workshop on Coal-Liquid and Alternate Fuels Technology,
Halifax, NS, 1986.
DESIGN METHODOLOGY 553
Lindsay J. J., Morton, W. and Newey, D. C. (1986). Radiative heat transfer in the freeboard region of a fluidized bed, Proceedings of
the Fifth Engineering Conference on Fluidization, Elsinore, Denmark, 18-23, May 1986, inFfuidiurron V, Engineering
Foundation, New York, pp. 385-392.
MacGregor, W. A., Hamdullahpur, F. and Ugursal, V. I. (1991). A technoeconomic assessment of the feasibility of using coal slurry
fired small scale fluidized bed furnaces for space and domestic hot water heating in single- and multi- family dwellings in Nova
Scotia, Final report prepared by Dept. of Mechanical Engineering, Technical University of Nova Scotia, for Canada Mortgage
and Housing Corporation.
MacGregor, W. A., Hamdullahpur, F. and Ugursal, V. I. (1993). ‘Space heating using small fluidized beds: a technoeconomic
evaluation’, Znt. J. Energy Rex, 17, 443-466.
Pagano, M. A. (1988). ‘Developmentof a residential scale fluidized bed combustor for space heating, ASME Papers, 88-WA/MET-I,
Chicago, IL.
Trivett, G. S., Mackay, G. D. and Pegg, M. J. (1982). Operating experience with a fluidized bed combustor in a single family
residence, Proceedings 7th International Conference on Fluidized Bed Combustion, Vol. 1. pp. 429-438.
Van Heerden, C., Novel, A. P. and Van Krevelen, D. W. (19.. ‘Mechanism of heat transfer in fluidized beds’, Znd. Eng. Chem., 45,
1237.
Xavier, A. M. and Davidsion J. F. (1985). In Fluidisation, 2nd edn, Davidson, Clift and Harrison (Eds), Harcourt, Brace, Jovanovich,
New York, Chapter 13A.

S-ar putea să vă placă și