You are on page 1of 8

Chemical Engineering Science 55 (2000) 4563}4570

The K-L reactor model for circulating #uidized beds


Daizo Kunii , Octave Levenspiel *
1-25-16 Nakamachi, Meguro-ku, Tokyo 153-0065, Japan
Chemical Engineering Department Oregon State University, 103 Gleeson Hall, Corvallis, OR 97331, USA
Received 18 November 1999; received in revised form 15 February 2000; accepted 28 February 2000

Abstract

In this paper we present a model for determining the performance behavior of a CFB reactor. This model is not sophisticated and
does not require computer calculations, but is realistic and convenient for engineers engaged in the development of new processes.
A number of parameters must be evaluated to use this model. This paper shows what they are, and thus suggests where future research
of CFB should be. Four examples in the di!erent #ow regimes illustrate the use of this model.  2000 Elsevier Science Ltd. All rights
reserved.

Keywords: CFB reactors; Conversion; Two-region two-zone model

1. Introduction 2. Vertical distribution of solids

Today the CFB plays the dominant role in FCC and in For a super"cial gas velocity u , and solid mass velo-

a few other large-scale industrial catalytic processes. It city G , a reasonable representation of observation shows
Q
most likely will play an important role in future processes that the solids are distributed in two regions in the vessel,
for converting natural gas produced in remote sites to a constant solid fraction f in the lower dense region of
B
transportable liquids. Much e!ort has gone into studying height H , and an upper lean region of height Hl in
B
its behavior, and many models have been proposed to which the solid fraction fl falls exponentially with height
explain how gas and solid contact and react with each from f towards the saturation carrying capacity of the
B
other. The "nal goal of these studies is to predict reactor gas f H. Fig. 2 shows the symbols used to describe the bed.
behavior, as shown in Fig. 1. For a "xed gas #ow rate Fig. 3 shows how the height of
The models that have been proposed to date are inva- the two regions depend on the solid #ow rate.
riably computer-based and require the user to accept the As mentioned, the fraction of solids in the lean region
developed programs to make behavior predictions. decreases exponentially to the saturation carrying capa-
This paper develops a model which is transparent in city of the gas f H, thus at any point zl in the lean region
that it is based on assumptions which the user can check we have
and modify at will. The reactor behavior can be deter-
fl "f H#( f !f H)exp[!az ], (1)
mined directly, and then its predictions are compared B D
with the ideal of straight plug #ow of gas through the where a is called the decay constant for the solids.
CFB. At the top of the vessel Hl the fraction of solids
We "rst deal with the vertical distribution of solids in becomes f , and zl becomes Hl , in which case Eq. (1)
the reactor, then the radial distribution. We follow this 
becomes, when rearranged,
by developing the performance equations for gas reac-
tions on catalytic solids, and we end up with some illus- 1 f !f H
trative examples to show how the predictions of this Hl " ln B . (2)
a f !f H
model match the experiment. 
Also the mean fraction of solids in the lean region is


1 &l
* Corresponding author. Tel.: 1-541-753-9248; fax: 1-541-737-4600. fM l " fl dz. (3)
E-mail address: levenspo@peak.org (O. Levenspiel). Hl

0009-2509/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 9 - 2 5 0 9 ( 0 0 ) 0 0 0 7 3 - 7
4564 D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570

Fig. 1. The CFB reactor problem.


Fig. 4. The shape of the vessel exit in#uences the solid fraction at the
top of the vessel.

reported correlations of f or u with the material


 Q
balance expression
G "o (u !u ) f , (5)
Q Q  Q 
where u is the slip velocity of the solids. Unfortunately,
Q
the reported values di!er widely, so which should we
trust?
Finally, if the vessel has a smooth exit we may be able
to approximate the slip velocity by the terminal velocity
of the particles u , in which case
Fig. 2. Symbols used to describe the geometry of CFBs. R
G "o (u !u ) f . (6)
Q Q  R 
Fig. 4a shows a sharp ine$cient exit which causes a sig-
Combining with Eq. (1) gives on integration
ni"cant percolation of solids down the vessel. Fig. 4c
f !f shows a smooth exit.
fM l "f H# B  (4) With a large through#ow rate of solids
aHl
Hl (calculated)(H , (7)
To predict the reactor behavior of the CFB we need R
to know the fraction of solids at the top of the vessel f , In this situation the vessel will have a lower dense

see Fig. 4. region and an upper lean region as shown by the curves
It is best to have a direct measurement of f . If in Figs. 3a}c. These curves represent the fast yuidization

this information is not available one may have to use regime.

Fig. 3. Behavior of a CFB for various solid #ow rates.


D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570 4565

3. Ideal plug 6ow of gas

Consider a "rst-order catalytic reaction


1 dN
APR, !r"!  "kC
 < dt 
Q
1 dN
or !r "!  "kC . (9)
 = dt 
In general, for plug #ow of gas through a catalyst bed of
height H having a constant solid fraction f, the perfor-
mance equation is
C kf H
ln  " . (10)
C u
CV 
But in a bed where the solid fraction changes with height
z, but is uniform across the cross section, the performance
equation is


Fig. 5. Measured solid distributions of a CFB, for runs A and G of C k &
Schoenfelder et al. (1996). ln  " f (z) dz . (11)
C u
CV  
We use these equations, obtained from Levenspiel (1999),
in deriving our equations.
When the through#ow rate of solids is very small The fast #uidized reactor has a lower dense region and
then the vessel will not have a lower dense region. an upper lean region, as shown in the curves of Figs.
The whole vessel becomes lean, with maybe a 3a}c. For the dense region
shallow dense solid acceleration zone by the solid
C kf H
inlet. Fig. 3d and e illustrate this behavior, which occurs ln  " B B (12)
C u
when B 
and for the lean region
Hl (calculated)'H . (8)
R
 
C kfM H k f !f H
The dense solid acceleration zone is di$cult to represent. ln B " l l " f H# B Hl (13)
C u u aHl
In any case, we have here what is called the pneumatic CV  
transport regime. All nine runs reported by Schoenfelder, So for the whole vessel either add Eqs. (12) and (13), or
Kruse and Werther (1996) are in this low solid #ow else "nd fM overall experimentally and use the following
regime, and Fig. 5 shows two of their results. equation:
Now that we have developed the equations for the C kfM H
vertical distribution of solids in a vessel, we need numer- ln  " R (14)
C u
ical values for the various #ow constants: CV 

u the slip velocity of particles


Q 4. The CFB reactor model - FF regime
u the terminal velocity of a solid particle in the gas
R
stream
f the mean fraction of solids in the dense region of Observations and measurements show that the lower
B region has a lean core which becomes progressively
a CFB
f the fraction of solids in the wall zone of a CFB richer in solids as one approaches the wall. Let us ap-
U proximate this by two distinct zones, a central core zone
f H the saturation carrying capacity of solids in the gas
stream with solid fraction f , and a wall zone rich in solids f .
BA BU
a the solid decay constant in the lean region of a CFB Also let the volume fraction of core be d (m core/m
B
reactor).
Values for some of these constants were collected and The upper lean region exhibits clumps of rising and
summarized from the literature by Kunii and Levenspiel falling solids plus solids at the wall. Let us again approx-
(1991, 1997). This will not be repeated here. imate this by two distinct zones: a core zone f , and a wall
JA
We now develop the equations in turn for plug #ow of zone f which includes all rising and falling clumps of
JU
gas in the reactor, for the FF regime, then for the PT solids. Again let d be the volume fraction of this core. Of
A
regime. course, the wall zone gets thinner on rising up the vessel.
4566 D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570

Fig. 7. Mechanism of reaction in the dense region of a CFB.

This represents a series/parallel process. So the perfor-


mance equation for the dense region is

 
Fig. 6. The distribution of solids in the CFB model. C 1 H
ln  " f Hd k# B .
C B 1/d K #1/f (1!d )k u
B B AU U B 
(21)

In addition, let us make the following two simplifying For the lean region we have a similar expression except
assumptions: that d is replaced by dM l , which is given by Eq. (19),
B

 
f "f "f (15) C 1 H
BU JU U ln B " f HdM l k# J
C 1/dM l K #1/f (1!dM l )k u
and CV AU U 
(22)
f "f "f H. (16)
BA JA
For the reactor as a whole
If actual data or if better assumptions are available,
certainly use them in the place of Eqs. (15) and/or (16). C C C
ln  "ln  #ln B , (23)
This represents our model, and Fig. 6 shows the situation C C C
visualized. CV B CV
A material balance at any level of the lower region where the two terms on the RHS are given above.
gives
f !f 5. For lean solids reactors + pneumatic transport
f "f Hd #f (1!d ) or d " U B (17)
B B U B B f !f H
U
For a low solid circulation rate, Eq. (8) applies and the
and at any level of the upper region gives
whole reactor is in the pneumatic transport regime, as
fl (z)"f Hdl (z)#f (1!dl (z)) (18) shown in Figs. 3d}e, or Fig. 8.
U
If no solid fraction data is available estimate f from
For the upper region as a whole 
Eqs. (5) or (6), then fM l from (4), and dM l from Eq. (19).
f !fM l Finally, apply Eq. (22) to "nd the conversion of gas.
fM l "f HdM l #f (1!dM l )or dM l " U . (19)
U f !f H If some solid fraction data is available, such as f or fM l ,

U certainly use it in preference to the unreliable u or f .
In addition, assume that gas only rises in the core Q 
This analysis assumes that there is no solid accumu-
zones, but is stagnant in the wall zone. Finally, let K be lation near the feed injection point where solids are
AU
the gas interchange coe$cient between core and wall accelerating up the bed. The experimental results of
zones, de"ned as Schoenfelder et al. (1996) of Fig. 5 show that this is not
m gas going from c to w, or w to c/s a good assumption for their vessel. Probably this is due
K " . (20) to the design of their solid distributor.
AU m of reactor

4.1. Conversion equations 6. Illustrative examples

In the lower dense region the rising gas reacts with The following examples, based on reactor experiments
solids in the core. It also transfers to the wall zone where reported by Schoenfelder et al. (1996), show how to use
it also reacts. This is shown in Fig. 7. this model for the various #ow regimes (see Fig. 9). But
D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570 4567

Table 1
Outline of the conditions used in the four examples

Example Solid #ow rate, Representative Solid distribution data


number G curves of Fig. 3
Q
1 17 kg/m s e Represents run A of
Schoenfelder et al.
(1996); the data is shown
in Fig. 5
2 24 between c Represents run G of
and d Schoenfelder et al.
(1996), the data is shown
in Fig. 5
3 100 b No data taken by
Fig. 8. Solid distribution in a lean solids CFB. Schoenfelder et al.
(1996) at these G
Q
4 300 a values

Example 1. G "17 kg/m s - run A of Schoenfelder et


Q
al. (1996)

For G "17 kg/m s and u "3.1 m/s, Schoenfelder


Q 
et al. (1996) measured the vertical distribution of solids
shown in Fig. 5a, and a conversion of gas of X "16%.

They had to choose k"2.72 m/m cat s to get these
results. Let us see what our model gives.
Since G is very small the bed is lean. So use the solid
Q
distribution data directly. The additional data needed to
use our model is

f "0.023 (see Fig. 5a)



fM "0.027 (see Fig. 5a)

f "0.4 (*).
U
Eq. (19) then gives
Fig. 9. Experimental reactor setup.
0.4!0.027
dM l " "0.9564,
0.4!0.01
"rst, Table 1 outlines the regimes studied. From Schoen-
felder et al. (1996) we have d "50;10\ m, o " Eq. (22) gives
N Q
1420 kg/m, 203C, 100 kPa, u "1.9;10\ m gas/m
KD
 
C 1 14
bed s and K "0.3 s\. Reaction is APR, "rst order.  " 0.01(0.9564)(1.5)#
AU ln
C 1 1 3.1
To "t their measured conversion data, Schoenfelder et al. CV #
(1996) had to pick di!erent k values for each of their nine 0.9564(0.3) 0.4(1!0.9564)(1.5)
runs. We picked one value for all four of our examples
k"ko "1.5 m/m cat s. "0.1645,
Q
The calculated values and their sources:
Therefore
u "0.0990 m/s (from Kunii & Levenspiel, 1997, p. 2478) C
R CV "0.8483 or X "15.13%.
C 
f H"0.01 (from Kunii & Levenspiel, 1997, p. 2475) 
For plug yow, Eq. (14) gives
f "0.2 (from Kunii & Levenspiel, 1997, p. 2474)
B C fM kH 0.027(1.5)14
ln CV " R" "0.1829.
a"0.4 m\(*). C u 3.1
 
4568 D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570

Therefore Next, the height of the lean region is given by Eq. (2)
C 1 0.2!0.01
 "0.8328, X "16.72%. Hl " ln "6.33 m,
C   0.4 0.0251!0.01
CV
Therefore
Example 2. G "24 kg/m s - run G of Schoenfelder et
Q
al. (1996) H "14!6.33"7.67 m.
B
This result shows that curve b of Fig. 3 represents the
The data of Fig. 6b show that the bed is barely com-
reactor. Then Eq. (4) gives
pletely in the lean regime. Also
0.2!0.0251
u "3.0 m/m bed s(given by Schoenfelder et al., 1996), fM l "0.01#
 0.4(6.33)
"0.0791.
f "0.035 (from Fig. 5b)
 Eq. (19) gives
fM "0.07 (from Fig. 5b),
0.45!0.0791
f "0.40 (*) dM l " "0.8430
U 0.45!0.01
Then Eq. (19) gives and Eq. (17) gives
0.4!0.07 0.45!0.2
dM " "0.8462. d " "0.5682.
0.4!0.01 B 0.45!0.01
Eq. (22) gives the reactor behavior as We are now ready to use the reactor performance equa-
tions. For the dense region, Eq. (21) gives
 
C 1 14
ln  " 0.01(0.8462)(1.5)#
C 1 1 3.0
 
CV C 1 7.67
# ln " 0.01(0.5682)1.5#
0.8462(0.3) 0.4(1!0.8462)(1.5) C 1 1 3.0
B #
"0.3751. (0.5682)(0.3) 0.45(1!0.5682)1.5

Therefore "0.2966. (i)

C Therefore
CV "0.6872 X "31%.
C  C
 B "0.7433.
For plug yow, Eq. (14) gives C

C 0.07(1.5)14 For the lean region, Eq. (22) gives
ln " "0.4900.
C 3.0
 
CV C 1 6.33
ln B " 0.01(0.8430)1.5#
Therefore C 1 1 3.0
CV #
0.8430(0.3) 0.45(1!0.8430)1.5
C
CV "0.6126
C "0.1843. (ii)

or X "39%. Therefore
 
C
Example 3. G "100 kg/m s. CV "0.8317.
Q C
B
This solid #ow rate is higher than any used by Schoen- Finally Eq. (23) gives
felder et al. (1996). Without solid distribution data means
C C C
that we must estimate more constants so the data used is CV " CV ) B "0.8317(0.7433)"0.6182.
C C C
as follows:  B 
u "3.0 m/m bed s (selected condition), f "0.2(*), Therefore
 B
u X "38%.
f "0.45 (*), Q "2(*). 
U u
R For plug yow we need fM . Since this is not available put
Then Eq. (5) gives K "R. Then Eq. (i) becomes
AU

 
G 100 C 1 7.67
f " Q " "0.0251. ln  " 0.0085# "0.767.
 o (u !u ) 1420(3.0!0.099;2) C 3.4309 3.0
Q  Q B
D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570 4569

Therefore Therefore
C C
B "0.4645. CV "0.5664 or X "43%.
C C 
 
Also, Eq. (ii) becomes For plug yow, fM "f "0.2, so Eq. (14) gives
B

 
C 0.2(1.5)14
ln
C
B " 0.0126#
1 6.33
"0.251. ln CV " "1.4,
C 3.0
C
CV
9.4362 3.0 
Therefore Therefore
C
C CV "0.2466 or X
CV "0.7786. C  
"75%.
C 
B
Combining gives, for plug #ow
6.1. Final comments
C C C
CV " CV ) B "0.7786(0.4645)"0.3618. 1. To use this model we need information or reason-
C C C
 B  able assumptions about
Therefore
a, f H, f , f , K and k
B U AU
X "64%.
  2. These four examples illustrate reactors operating in
various #ow regimes
Example 4. G"300 kg/m s Example 1: lean pneumatic #ow.
Example 2: lean FF regime alone.
This G value represents a very high solid through#ow Example 3: both lean and dense FF regime.
Q
rate. Additional data needed are Example 4: dense FF regime alone.
u "3.0 (selected condition), 3. Table 2 and Fig. 10 compare the performance of
 these reactors with that of plug #ow and of mixed
f "0.2 (*), #ow. These clearly show that the CFB reactor be-
B
haves worse than both the plug- and mixed-#ow
f "0.5 (*),
U reactors.
u
Q "2 (*).
u
R Notation
First of all Eq. (5) gives a decay constants for solids, m\
C or C concentration of reactant A, mol/m
300 
f " "0.0754. CFB circulating #uidized bed
 1420(3.0!0.099;2) f volume fraction of solids, m solids/m vessel
fH saturation carrying capacity of gas,
Eq. (2) gives
m solids/m vessel
1 0.2!0.01 FF fast #uidization regime
Hl " ln "3.03 m. G mass velocity, kg/m s
0.2 0.1888!0.01
H height, m
This is practically a dense bed throughout, as sketched in k "rst-order reaction rate constant, m/kg cat s
Fig. 3a. So assume that this is a total dense bed. Then Eq. k "rst-order reaction rate constant, m/m cat s
(17) gives
0.5!0.2
d " "0.6122. Table 2
B 0.5!0.01 Performance from the model, from plug #ow and from mixed #ow

Then the reactor performance expression of Eq. (21) gives kfM H X (%) X (%) X (%)
R "y       
u "y/(1#y)

 
C 1 14
ln CV" 0.01(0.6122)1.5# 0.1829 15.17 16.72 15.46
C 1 1 3
 # 0.4823 31.0 38.8 32.5
0.6122(0.3) 0.5(1!0.6122)1.5 1.0174 38.0 64.0 50.4
1.4 43.3 75.3 58.3
"0.5684.
4570 D. Kunii, O. Levenspiel / Chemical Engineering Science 55 (2000) 4563}4570

u slip velocity of a particle with respect to the


Q
rising gas, m/s
u terminal velocity of a falling particle, m/s
R
d volume fraction of core, m core/m bed
o density, kg/m
* indicates that the value is arbitrarily chosen

Subscripts
c core zone
d lower dense region
ex leaving the vessel
l upper lean region
s solid
t total height
top at the top of the vessel
w wall zone
Fig. 10. Comparison of the conversion from a CFB reactor with that z height
from plug #ow and from mixed #ow.

References

Schoenfelder, H., Kruse, M., & Werther, J. (1996). Two-dimensional


model for circulating #uidized bed reactors. A.I.Ch.E. Journal, 42,
K interchange coe$cient between c and w, 1875}1888.
AU Kunii, D., & Levenspiel, O. (1991). Fluidization engineering, (2nd edn).
m transferred from c to w/m vessel s, or vice
versa Boston, MA, USA: Butterworth-Heinemann.
Kunii, D., & Levenspiel, O. (1997). Circulating #uidized beds. Chemical
PT pneumatic transport regime Engineering Science, 52, 2471}2488.
r reaction rate, m converted/m cat s
 Levenspiel, O. (1999). Chemical reaction engineering (3rd Ed.). (p. 396)
u super"cial gas velocity, m gas/m bed s New York, NY, USA: Wiley.


You might also like