Overall heat transfer to water-walls and wing

walls of commercial circulating fluidized bed

boilers

A DUTTA and P BASU

To find correlations for heat transfer to water walls and wing walls in commercial circulating fluidized bed
(CFB) boilers, data on overall heat absorption rates, bed temperature and suspension density from two
-1
commercially operating CFB boilers (100 and 700 t h capacities) were analyzed. A methodology was
developed to deduce heat transfer coefficients on wing walls from the measured overall heat duties of the
entire evaporator. Heat transfer coefficients on water walls are found to be higher than those on wing walls.
A correlation was developed relating the heat transfer coefficients, averaged over the entire height of the
heat absorbing wall, with operating parameters. This makes the correlation simple to use in boiler designs
and it reduces uncertainties in measured data. The correlation is tested against heat transfer coefficients
measured and reported by other researchers. This deviation is within 15% which is of the same order as the
uncertainty in measured data.
_________________________________________

1 Nomenclature

A projected area (m2)

h average heat transfer coefficient (W m-2K-1)
k proportionality constant
Q heat duty (W)
T average temperature (0C)
ρ average suspension density (kg m-3) Wing wall

Subscript

avg average
b bed
ewwall evaporator wing wall Water wall
g gas
max maximum operating
p particle convection
swwall superheater wing wall
wwall evaporator water wall

Superscript Fig: 1 Schematic Diagram of a 230 MWe CFB boiler unit (Nowak et. al 1999)
‘ superheater wing wall
“ evaporator wing wall

walls in the furnace. A wing wall can be either a part of the

2 Introduction evaporator or the superheater of the boiler. Typically wing
walls or platen surfaces hang from the roof and extend from
the front wall to some distance towards the opposite wall,
The furnace temperature of a circulating fluidized bed where the furnace exit is located (Fig. 1). To absorb the
(CFB) boiler is generally controlled by the heat transfer required amount of heat a CFB boiler often uses more than
to the enclosing water-walls and the suspended wing one wing wall. The amount of heat absorbed in the wing

1
wall could be as high as 30% that absorbed in the furnace conditions. Wing walls, thus, remains the most reliable
walls. Much information (Basu and Nag, 1996, means for augmentation of furnace heat absorption. The
Glicksman 1997, Molerus and Writh 1997, Breitholtz present work, probably the first attempt in this area, might
2000) is available on heat transfer on membrane-tube help optimize the design of wing walls on commercial
type water-walls of CFB boilers. The density of gas- boilers.
solid suspension in the furnace exerts the strongest
influence on the heat transfer coefficient on the enclosing
walls. However, one can not neglect the importance of
the bed temperature on the heat transfer rate. The large 3 Data from commercial unit
contribution of radiation to the total heat transfer is a
good indicator of the role of bed or furnace temperature. The analyses of failures and performance predictions of
For example, in boilers operating with low suspension commercial boilers require meaningful scale up of data.
densities, (less than 20 kg m-3) (Divilio and Boyd 1994, Experimental data from bench-scale or laboratory-size units
Lackner and Andersson 1992, Couturier et al 1993, are often used to predict the performance of commercial
Andersson 1996), the radiation dominates the heat units. For a reliable prediction, one has to have the data in
transfer in a major part of the combustion chamber conditions physically compatible to those in large units.
(Divilio and Boyd 1994, Andersson and Lackner 1994 Thus, it is imperative that information on large commercial
and Writh 1994). Tang and Engstrom (1987) have units is used to validate correlations. Such a validation is
reported an increase in the contribution of radiation from very difficult and at times impossible. In scientific
60 to 90% of the total heat transfer when the load was experiments, the effect of one independent parameter on a
changed from a full to a minimum load. The suspension dependent variable is generally studied by measuring the
density reduces greatly when the fluidization velocity dependent variable for specific values of that influencing
was decreased from 5.2 at full load to 1.8 ms-1 in low parameter while keeping all other parameters constant. Very
load in a commercially operating CFB boiler where few commercial units operate in ideal conditions. In
associated things like circulation rate also changes with commercial CFB boilers, it is very difficult to achieve such
load. Basu and Konuche (1988), who measured the conditions. Most commercial boilers are operated by an
radiative and the total heat flux in a CFB pilot plant automated control system. Thus, when one parameter is
operating in temperatures in the range of 650 to 9000C, varied, several other parameters also change which makes it
found that when the bulk density decreased from 20 to 6 very difficult to assess the impact of the chosen parameter.
kgm-3, the radiative part of the heat transfer increased Secondly, very few commercial CFB boilers are
from 74 to 91%. A number of researchers (Basu & Nag; instrumented to measure all relevant parameters, like heat
1996, and Breitholtz et. Al., 2000) proposed correlations transfer coefficients or suspension densities etc.
for estimating average heat transfer coefficients on water
wall using the cross-section average suspension density These difficulties do not, however, diminish the importance
as a variable. All of these data and studies pertain to the of generation of data in commercial units, because that is the
enclosing walls. Currently there is no correlation ultimate test of engineering models which are to be used for
available in the open literature, let alone any studies on industrial designs. With this in mind, efforts were made to
average heat transfer on the wing walls, which operates collect data from industrial units. Commercial secrecy and
in different hydrodynamic condition than that of water lack of proper instrumentation made it difficult to get access
walls. to a decent set of data. However, some data on heat transfer
coefficients were deduced from a set of measured operating
Furnaces as large as 250 MWe capacity routinely use parameters in two large commercial circulating fluidized bed
wing walls. Inadequate knowledge about heat transfer boilers. One unit is in the range of 20 MWe while the other
rates on bed to wing walls forced manufacturers to revise one produces more than 170 MWe power. Like any
wing wall surfaces in some instances even after commercial unit, these boilers are operated under different
manufacture. With time, the manufacturers have largely loads at different times. Data logged at different loads were
overcome design uncertainties they had in the initial analyzed to deduce heat transfer coefficients at different
stage. Still, no analysis of heat transfer to wing walls, conditions.
theoretical or empirical is available in published
literature. Only recently, some research (Dutta and Basu
2001) has started in this area. 4 Methodology for estimating heat
Some manufacturers use division walls in the furnace as transfer coefficient
an alternative to wing walls. It is a vertical wall that
effectively separates the furnace into two beds. Unlike laboratory test rigs, commercial boilers are not
Theoretically, this appears to be a sound idea, but its designed to gather scientific data. As such they do not use
practical operation could be difficult: For instance, the pressure transducers, thermocouples, heat flux meters, or
air-solid flow may flip-flop between two sections flow meters at every location. Thus, it is difficult to get a
creating non uniform hydrodynamic and heat transfer direct measure of the heat transfer coefficient on a

2
commercial boiler. It has to be estimated indirectly. Heat
balance over different sections of the boiler allowed us to h′p +
( k Tg4 − Tswwall
4
)
measure heat absorption in individual group of heating hswwall (T g − Tswwall )
=
k (T )
surfaces (panels). The larger boiler, in question, used (1)
wing walls as both evaporative surface as well as hewwall 4
− Tewwall
4

h′′ +
g
superheater surface. Thus, the total evaporative load of
the boiler does not give or measure the heat absorbed in
p
(T g − Tewwall )
the water walls. To compare the heat transfer to the wing After simplification equation (1) can be written as
walls with that to the water walls, one needs to calculate
=
2
(
hswwall h ′p + k Tg + Tswwall (Tg + Tswwall )
2
)
the heat transfer coefficient separately for each case at a
specific operating condition. The heat transfer coefficient (
hewwall h ′p′ + k Tg2 + Tewwall
2
)
(Tg + Tewwall ) (2)
on the superheater wing wall could be easily deduced
from its surface area as the temperatures of the steam The overall heat transfer coefficient for the superheater wing
entering and leaving the superheater wing wall were wall was calculated by using measured value of
recorded in the control room. Although both types of Qswwall , Tg , Tswwall
wing walls are exposed to the same hydrodynamic
conditions, the heat transfer coefficient for superheater Qswwall
hswwall =
Aswwall (Pr ojected ) (Tg − Tswwall )
wing wall may not be applicable directly to the (3)
evaporative wing wall because the surface temperature of
the evaporative wing walls could be much lower than Using assumption (2) the particle convection for the
that of the superheater wing walls. So, it needs to superheater wing wall was estimated as
determine the heat transfer coefficient for the evaporative
wing walls from that on the superheater wing walls. h′p = 0.2 × hswwall (4)
Following assumptions were made to estimate the
evaporative wing wall heat transfer coefficient.
Considering the total heat transfer coefficient is the addition
of particle convection and radiation (Basu and Nag 1987,
1. The superheater and evaporator wing walls are
Wu et. al 1987), the constant K of the radiative term into
arranged side by side in the upper parts of the
wing walls (Stephen Boltzman constant, emmisivity, etc)
furnace. Thus both are exposed to the same
was estimated from the equation (5).
hydrodynamic condition. As the particle convection
component of the heat transfer does not depend
significantly on the surface temperature, it is hswwall − h ′p
k=
assumed to be same on both types of wing wall. (T g
2
+ Tswwall
2
)
(Tg + Tswwall )
(5)

2. The particle convection is constant and 20% of the

total heat transfer in the upper zone of the CFB riser Using this value of k and the assumption (1) i.e., h ′p = h ′p′ ,
(Basu and Konuche 1988).
the evaporative wing wall heat transfer coefficient was then
deduced from the equation (2).
3. As per boiler design norm (Basu et al 1999). the
The amount of heat absorbed by the evaporator
average surface temperature of the superheater is
water wall was estimated by subtracting estimated
500C above the average steam temperature.
evaporator wing wall duty from the total calculated
evaporator duty.
4. The average surface temperature of the evaporator is
250C above the saturation temperature as per boiler
design norm. (Basu et al 1999) Qwwall = Qevaporator − Qewwall (6)

The total heat transfer coefficient in a CFB is the sum of The average water wall heat transfer coefficient was then
convective and radiative component heat transfer (Basu estimated by using the same equation as that of step 1.
and Nag 1987, Wu et. al 1987, Basu & Fraser, 1991).
The ratio of the total heat transfer coefficient for the Qwwall
hwwall =
Awwall (Pr ojected ) (Tg − Twwall )
superheater wing wall, hswwall , and evaporator wing (7)
wall, hewwall , is, therefore, written as
This gives the average heat transfer coefficient over the
entire height of the water wall.

3
 1.1 1.2

1.0
1.0

0.9
1.5046
h  T 
0.8 = 0.8944 b 
0.8 hmax  Tb max 

h/hmax
h/hmax

0.7
0.485 0.6
h  ρ 
= 1.048 
0.6
hmax  ρ max 

0.4
0.5

0.4 0.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.5 0.6 0.7 0.8 0.9 1.0 1.1
ρ Tb
ρ max Tb max
(a) (a)
1.1 1.2

1.0
1.0

0.9
1.4531
h  T 
0.8
= 0.8984 b 
0.8
hmax  Tb max 
h/hmax
h/hmax

0.7 0.4756
h  ρ  0.6
= 1.0538 
hmax  ρ max 
0.6

0.4
0.5

0.4 0.2
0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.5 0.6 0.7 0.8 0.9 1.0 1.1
ρ Tb
ρ max Tb max
(b) (b)

Fig. 2 Heat transfer coefficient with non-dimensional suspension Fig. 3 Heat transfer coefficient for (a) water walls and (b) wing walls
density: (170-MWe boiler): (a) water walls and (b) wing walls. with non-dimensional bed temperature (170-MWe boiler)

5 Results and discussion Non-dimensional heat transfer coefficients to water walls

and wing walls are also plotted against the non-dimensional
Applying the above methodology, the average heat bed temperatures in Fig: 3a and 3b. If radiation is the
transfer coefficients to water walls and wing walls are dominant factor, one would expect the heat transfer
4
estimated at different loads of the 170 MWe boiler. The coefficient to be proportional to Tg . But the exponent is
average heat transfer coefficients, suspension densities found to be 1.5 instead of 4 as expected for pure radiation.
and bed temperatures are presented in non-dimensional Yet, this suggests that furnace temperature has a greater
form by dividing its maximum operating values. Fig. 2a influence on the heat transfer coefficient than the suspension
and 2b show that non-dimensional heat transfer density has on it. Instead of choosing data for any
coefficients to the water wall and wing wall increases temperature, if one uses value for a narrow range of bed
with non-dimensional suspension densities, measured temperature (930-940 0C), the exponent for the suspension
across the entire upper section of the furnace at different density is found to be 0.21 for the water wall (Fig. 4a). It is
loads. The effect of the bed temperature is not considered interesting that this value is similar to that reported by
while fitting the curve as h ~ ρsus 0.48. The exponent for Breitholtz et. al (1999), who proposed a correlation for
the fitted curve is found to be 0.48. This exponent is commercial boilers by minimisation of the square of the
close to that observed by Basu and Nag (1996), deviations between the measured data and the correlations
Glicksman 1988, Ebert et al., 1993, who found it to be without considering bed temperature. The exponent for the
0.5. wing wall in this case is found to be 0.11 (Fig. 4b), which

4
suggests a less dependence on the average suspension
1.02
density.
1.00

1.02
0.98

1.00
0.96

h/hmax
0.98 0.94
0.2131
h  ρ 
= 0.9971 
0.96
hmax  ρ max  0.92 0.4728
h  ρ 
h/hmax

= 1.0005 
hmax  ρ max 
0.94 0.90

0.88
0.92

0.86
0.90
0.75 0.80 0.85 0.90 0.95 1.00 1.05
ρ
0.88 ρ max
0.6 0.7 0.8 0.9 1.0 1.1
ρ (a)
ρ max
1.02
(a)

1.005 1.00

0.98
1.5247
1.000 h  T 
= 1.0005 b 
0.96 hmax  Tb max 
0.1137
h  ρ 
h/hmax

0.995 = 0.9976  0.94

hmax  ρ max 
h/hmax

0.990 0.92

0.90
0.985

0.88

0.980
0.86
0.90 0.92 0.94 0.96 0.98 1.00 1.02
0.975 Tb
0.84 0.86 0.88 0.90 0.92 0.94 0.96 0.98 1.00 1.02 Tb max
ρ
ρ max (b)
(b)
Fig. 6 Water walls heat transfer coefficient (a) with non-dimensional
Fig. 4 Heat transfer coefficient for (a) water walls and (b) wing suspension density and (b) with non-dimensional bed temperature
0
walls of 170 MWe boiler (at930-940 C) (20-MWe boiler)

0.4
The heat transfer curves for both evaporative wing wall and
water wall are nearly parallel to each other (Fig: 2 and 3).
The absolute value of heat transfer coefficient on the wing
0.3
walls is found to be consistently lower than that on the water
(hwwall-hewwall)/hwwall

walls. In Fig: 5, the ratio of this difference and heat transfer

0.2 to the water walls is plotted against non-dimensional
suspension density and bed temperature. The ratio is found
to be about 0.22 or 22% at different loads of the boiler which
0.1 is more than the particle convective component of a CFB
furnace (20%) operating at 800-9000C (Basu and Knouche
1988). Even an assumption that particle convective heat
0 transfer on the wing wall is negligible cannot account for
0 0.2 0.4 0.6 0.8 1
this larger difference. So, one could speculate that both
( ρ/ρ max )*(T b /T bmax )
particle convective and radiative heat transfer are lower on
Fig. 5 The percentage reduction of heat transfer coefficient the wing walls than what they would be on water walls. The
between water walls and wing walls at different operating hydrodynamics on wall tubes are different from that on wing
conditions walls, which are exposed to relatively dilute up-flowing

5
solids. The heat transfer on wing walls is found to be 1
(Dutta and Basu, 2001) dominated by forced convection
from dilute suspension, while that on the water walls is 20 Mw e Boiler 170 Mw e boiler
dominated by down-flowing particle convection. As a 0.8
result, there is higher particle convection on the water

(h/hmax )Predicted
walls. Also, the water wall is exposed to the entire cavity
of the furnace whereas the wing wall is exposed to the
0.6
relatively narrow cavity between wings. For this reason,
the mean beam length of the gas radiation in water walls
is higher resulting in higher radiative heat transfer
coefficient. 0.4

The above discussions relate to data from the 170 MWe

unit. The data from the 20 MWe boiler were analyzed in 0.2
the same fashion. The 20 MWe boiler does not have the 0.2 0.4 0.6 0.8 1
wing wall. The results on water wall support the findings (h/h max )Estimated
for the water wall of the 170 MWe boiler. Results of heat
transfer on the water walls of the 20 MWe boiler are (a)
shown in Fig. 6a and 6b. The exponent for the fitted 1
curve for non-dimensional heat transfer coefficients with
non-dimensional suspension densities is 0.47 (Fig. 6a) 170 Mw e boiler

whereas it is 1.52 for non-dimensional heat transfer 0.8

coefficients with non-dimensional bed temperatures (Fig.

(h/hmax )Predicted
6b). These results are in the same order that is found for
170 MWe (Fig. 2a and 3a). 0.6

6 Empirical correlation of heat transfer

coefficient 0.4

Presently no empirical correlation on heat transfer to

wing walls is available in published literature. So, an 0.2
effort is made here to develop a correlation which can be 0.2 0.4 0.6 0.8 1
used to predict average gross value of heat transfer to (h/h max )Estimated
wing walls and water walls. Fig. 1 and 2, show that total
(b)
heat transfer coefficient depends on both suspension
density and bed temperature. The average heat transfer Fig. 7 Heat transfer coefficient to (a) the water walls and (b) the wing
coefficients, to the water wall and wing wall is correlated walls; estimated vs. predicted.
to the average suspension density ρ sus and average bed
250
temperature Tg , by a potential function,
h = αρ avg
a
Tgb (8) 200
HTC Predicted (W/m 2.K)

Using Statistical Analysis Software (SAS), and by

150
converting the non-linear equation to multiple linear +15%

equations, the values of the parameter in the correlation,

α, a and b were found by regression analysis. 100
-15%

For the water wall the correlation is 50

watt
hwwall = 5 × ρ avg
0.391
× Tg0.408 (9) 0

m 2 .K 0 50 100 150 200 250

where ρ avg is in kgm-3 and Tg in 0C. HTC Measured (W/m 2.K)

whereas for wing wall, the correlation is Basu & Fraser (1991)
Anderson (1996)
Couturier (1993)
Werdermann (1994)
Blumel (1992)
Jestin (1992)

watt
hewwall = 3.6 × ρ avg
0.37
× Tg0.425 (10) Fig. 8 Experimental vs. predicted overall heat transfer coefficient for
m 2 .K water walls.

6
Fig. 7a and 7b compares heat transfer coefficients 5. BASU P. and KONUCHE F. in P. Basu and J.F. Large
measured with those calculated by the above correlations. (eds.), Circulating Fluidized Bed Technology II,
The agreement between measured and correlated values Pergamon, Oxford, 1988, pp 245
is within ± 5% limit which is a good agreement for most 6. BASU P., KEFA C. and JESTIN L. Boilers and burners
empirical correlation. design and theory, Springer, New York, 1999.
The correlation of average heat transfer coefficient to 7. BASU. P and NAG. P.K. An investigation into heat
water walls is also validated with average water wall heat transfer in circulating fluidized beds, International
transfer coefficients for a number of commercial boilers Journal of Heat and mass Transfer 30 (11) 2399-2409.
reported by a number of researchers (Anderson 1996, 8. BREITHOLTZ C. Heat transfer in circulating fluidized
Werdermann 1994, Couturier 1993, Jestin 1992, and bed boilers, PhD thesis, Chalmers University of
Blumel 1992, and Basu and Fraser 1991). Fig. 8 shows Technology, Goteborg, Sweden, 2000.
that the correlation is in good agreement with the 9. BREITHOLTZ C, LECKNER B. and BASKAKOV A.
reported values with in 15%. As no measured data are Wall average heat transfer in CFB-boilers, Proc. III
available for wing walls in commercial boilers in the European Conf. on Fluidization, Toulouse-France, 2000
open literature, the correlation is could not validated 10. COUTURIER M.F, STEWARD F.R. and POOLPOL S.
against data from other independent source. However, Experimental determination of heat transfer coefficients
one can expect the similar agreement, like water walls. in a 72 MWt circulating fluidized bed boiler, in L.
Rubow and G. Commonwealth (eds.), Proc. 12th Int.
Conf. Fluidized Bed Combustion, ASME, 1993, pp
7 Conclusions 1215-1222.
11. DIVILIO R.J. and BOYD T.J. Practical implications of
the effect of solids suspension density on heat transfer in
1. The influence of suspension density and furnace
large-scale CFB boilers, Proc. 4th Int. Conf. Circulating
temperature on the heat transfer on wing wall is
Fluidized Beds, Hidden Valley, PA, 1994, pp 334-339.
similar to that on heat transfer on furnace water
12. DUTTA. A, and BASU. P. An Experimental
walls.
Investigation into the Heat Transfer on Wing Walls in a
Circulating Fluidized Bed Boiler (Communicated)
2. Heat transfer coefficients on wing walls are 20%
13. EBERT. E.R., GLICKSMAN. L.R., and LINTS. M.
lower than those on water walls.
Determination of particle and gas convective heat
transfer component in circulating fluidized bed.
3. Two empirical correlations are developed for
Chemical Engineering science 48, 1993, pp. 2179-2188.
estimation of the average heat transfer coefficients to
14. GLICKSMAN L.R. Heat transfer in circulating
water walls and that wing walls. These relate heat
fluidized beds, in: J.R. Grace, A.A. Avidan, TM
transfer coefficients to average suspension density
Knowlton (Eds.), Circulating Fluidized Beds, Chapman
and average bed temperature in corresponding parts
& Hall, London, 1997, pp. 261-310.
of the furnace
15. GLICKSMAN, L.R. Circulating Fluidized Bed Heat
Transfer, in Ciculating Fluidized bed Technology II
4. Average values of heat transfer coefficients for water
(Edited by P. Basu and J.F. Large) Pergamon Press,
wall, predicted by the above correlation, agree with
Oxford 1988, pp 13-29
15% of those reported data in the literature.
16. JESTIN L, MEYER P., SCHMITT C. and MORIN J.X.
Heat transfer in a 125 MWe CFB boiler, Presented at the
Engineering Foundation conf. Australia, April 1992.
8 References 17. MOLERUS O. and WRITH K.-E. Heat transfer in
fluidized beds, 1st ed., Chapman & Hall, London, 1997,
1. ANDERSSON B.-A. and LECKNER B. Local pp. 111-151.
lateral distribution of heat transfer on tube surface of 18. NOWAK W., Z. BIS., LASKAWIEC J.,
membrane walls in CFB boilers, in A.A. Avidan KRZYWOSZYNSKI W., WALKOWIAK R., Proc. 15th
(ed.), Proc. 4th Int. Conf. Circulating Fluidized Beds, Int. Conf. Fluidized Bed Combustion, ASME, Georgia,
Hidden Valley, PA, 1994, pp 311-318. 1999.
2. ANDERSSON B.-A. Effect of bed particle size on 19. TANG J.T. and ENGSTROM F. Proc. 9th Int. Conf.
heat transfer in circulating fluidized bed boilers, Fluidized Bed Combustion, 1, ASME, New York, 1987,
Powder Technology, 87 (1996) pp 233-238 pp 38
3. BASU P and NAG P.K. Heat transfer to walls of a 20. WU. R. L., LIM. C. J., CHAOUKI. J., GRACE. J. R.
circulating fluidized-bed furnace, Chemical Heat transfer from a circulating fluidized bed to
Engineering science, 1996 51 (1), pp 1-26. membrane waterwall surfaces, AIChE Journal 33 (1987)
4. BASU P. and FRASER S. Circulating fluidized bed 1888-1893.
boiler-design and operation, Butterworths-
Heinemann, Stoneham, 1991.