You are on page 1of 7

BUBBLING BED MODEL

Model for the Flow of Gas through a Fluidized Bed


D A I Z O K U N I I

Chemical Engineering Department, University of Tokyo, Tokyo, Japan


OCTAVE LEVENSPIEL

Chemical Engineering Department, Illinois Institute of Technology, Chicago, Ill.

60616

This simple three-region model for the gas flow through fluidized beds views uniformly sized bubbles, surrounded by clouds and followed by wakes, rising through an emulsion of downward moving solids, Interchange of gas occurs continuously among bubble, cloud-wake, and emulsion regions. This model contains
one parameter, the effective bubble size, and all internal flows and interchanges in the bed are derived
from it. Qualitative and quantitative checks with reported bed behavior are presented.

incentive for much fluidization research at


present is the desire to develop procedures for design of
fluidized beds for chemical and physical operations. To date
the major obstacle thwarting these attempts is the fact that no
satisfactory flow model has been developed to represent the
actual flow and contacting pattern of gas with the solids in the
bed.
I n this paper we propose a simple model for the flow of gas
through fluidized beds. I t contains one parameter, the effective bubble size, and all internal flows and interchanges in the
bed are derived from it. In another paper (Kunii and Levenspiel, 1968b) we sholv that this model can fit the reported data
in all three areas-gas-solid
heat transfer. gas-solid mass
transfer, and conversion in catalytic reactions.
Now an actual operating fluidized bed consists of a lower
dense bubbling region and an upper region of decreasing density consisting of entrained, splattered, and settling solids.
Here we are concerned only with the dense bubbling region.
This consists of a continuous phase, called the emulsion,
through which rise bubbles which are usually surrounded and
accompanied by a cloud of rising gas.
N IMPORTANT

Bubble Phase

I n developing the model a number of simplifying assumptions are made.


Assumption 1. Behavior in Vicinity of a Single Bubble.
I n the vicinity of any one of the many rising bubbles in a
bubbling fluidized bed let us assume that the gas flow is given
by the Davidson model with its spherical bubble surrounded,
in the case of a large fast bubble, with a spherical cloud.
The velocity of rise of a single bubble of diameter d b , hence the
velocity relative to solids far from the bubble, is

A small, slow bubble rises more slo\vly than the gas percolating
through the emulsion? or

These bubbles are surrounded and accompanied by a cloud of


circulating bubble gas, distinct from the emulsion gas, and the
size of this cloud is given by

(4)

As the size and velocity of bubbles increase, the relative thick-

UAT

Uf

446

I&EC F U N D A M E N T A L S

(5)

> 5

the cloud thickness is less than 10% the bubble diameter.


Arbitrarily let us call these large fast bubbles with negligible
clouds.
Finally, gas flow upward within rising bubbles, and at the
maximum cross section of the bubble its velocity is ?urn,;
hence the upward volumetric flow rate in each bubble is

The derivation of these expressions. the justification of their


use, and considerable related information are given by Davidson and Harrison (1963).
Assumption 2. Bubble Size. Normally small bubbles
are observed at the bottom of a fluidized bed, larger ones near
the top. I n addition, a t any level in the bed the bubble size
is not uniform. For this treatment, however, we take the
bubble size to be uniform throughout the bed or section of bed
under consideration. \Ye call this the effective bubble size.
In a shallow freely bubbling bed this may be a rather poor
representation of the physical situation. I n large deep beds
where the maximum stable bubble size is reached, or in beds
containing internals or baffles to control the bubble size, this
should be a reasonable representation of the state of affairs.
Assumption 3. Bubble Velocity. Let us relate the velocity of rise of a crowd of bubbles to the velocity of rise of a
single bubble by
ub

and hence the faster rising emulsion gas u f shortcuts through


the rising bubble. A large fast bubble moves faster than the
emulsion gas, or

From Equation 4 we see that

ness of the cloud decreases.


when

= U,

U,f

f Ubr

= U,

- U,f

+ 0.711

(gdb)liz

(7)

This expression and arguments for its use were first presented
by Sicklin (1962) for gas-liquid systems, and later used by
Davidson and Harrison (1963) for fluidized beds.
Assumption 4. Voidage of Bubbles and Emulsion.
Although experimental evidence (Kunii and Levenspiel, 1968a)

3umf (Equation 6). Hence the relationship between the total


flow and that through the two phases is given by

Gas flow through bubble

ub + 3 &nf

Fraction
consisting

uo =

Fraction

/--consisting of

1
Average upward
velocity

- 6)urnf

+ 6 ( u b + 3urnf)

(9)

I n a bed of large fast bubbles with negligible clouds, or for


u b / u f > 5, each rising bubble carries its own gas up the bed
with it, so the net upward velocity of bubble gas is simply that
of the bubble itself. So, a material balance gives

emulsion
phase: I-.s

(1

wne;ion
Now at low gas velocity the bubble fraction is very small; a t
high u, this term dominates the term which follows. Hence
with little error we can write

Figure 1 . Typical cross section of a b e d containing


sma I I rising bubbles
All velocities measured an a superflcial (empty tube) basis

indicates that rising bubbles contain small amounts of solids,


we can with negligible error ignore this and take the bubble
void fraction cb = 1. I n addition, if 6 is the volume fraction of
bubbles in the bed, the average bed voidage el is related to the
voidage in bubbles and emulsion by
ef =

660

f (1

- 6)ee

= 6

+ (1 -

6)Ee

and measuring 6 and e \vi11 then give ce. Although ee may well
vary with position in the bed, let us assume the emulsion
voidage to be that of the bed a t minimum fluidizing conditions,
or
= e m f . Thus voidages and bed heights are related by

At this stage of our treatment these approximations for cb and


ce lead to little error. Later, in considering heat and mass
transfer and reaction we modify these assumptions.
Relationship among Bubble Phase Variables db, u b , a n d 6
Given u, a n d u m f . I n a bed of small slow bubbles, or ua/uf < 1,
the relationship between variables is found by considering a
typical cross section of a bed having a bubble fraction 6 and
cutting through the center of one bubble, as shown in Figure 1.
O n a superficial velocity basis the upward flow of gas in the
emulsion phase is u m f and through the rising bubble is u b

In the intermediate region where bubbles have appreciable


clouds, or 1 < u b / u f < 5, it suffices to interpolate the results for
the two bounding regions, as shown in Figure 2.
The above results (presented in Figure 2) show the relationship among the various quantities describing bubble conditions
in a bed. The solid lines represent Equations 9 and 10, and
the highest value of 6 shown is 0.5, corresponding to a bed completely full of bubbles.
Moving vertically upward on this figure, but keeping 6 <
0.5, indicates the possible conditions of the rising bubbles in the
bed. For example, with u, = I .3umf we may expect small slow
bubbles near the bottom of the bed for 6 = 0.1. By coalescence
these may grow to intermediate or large bubbles further up the
bed. O n the other hand at higher velocity, say u,, > burn/,
slow bubbles will not form, and practically cloudless bubbles
will form straightaway a little above the distributor by violent
coalescence.
I n summary, Equations 7, 9, and 10 relate the bubble phase
variables a t any point in the bed with the fluidizing conditions,
u, and u,
These equations show that measuring one quantity,
the bubble size, suffices to characterize the bubble phase of a
fluidized bed. Furthermore, for u, > 2umf, measured 6 values
are normally no greater than 0.1. Figure 2 then shows that
only large fast bubbles with negligible clouds may be expected
in beds when u, > 2urnf.

A
20
ub
Umf

10

uo / Urnt

Figure 2. Interrelationship of bubble velocity and bubble fraction of b e d with


superficial gas velocity and minimum fluidizing velocity
VOL. 7

NO. 3

AUGUST 1 9 6 8

447

dp t m m l
Figure 3. Relationship between
volume of wake and volume of
rising bubble
From Rowe and Partidge (1965)

-Bubble
phase

Wake

and Grummer, 1952; Talmor and Bennenati,-1963), and in


terms of a diffusion-type process (Bart, 1950; Brotz, 1965;
Lewis et al., 1962; May, 1959; Mori and Nakamura, 1965).
Further experiments by Sutherland (1961) and Rowe and
Sutherland (1964) questioning the influence of bubbles helps
to explain this phenomenon. Thus, each of the crowd of
rising bubbles carries behind it a wake of solid particles, the
size of which is reported by Rowe and Partridge (1965) in
Figure 3. To balance this upward sweep of solids the rest of
the particles, the emulsion solids, must move downward in the
bed.
These findings suggest that the rate of circulation of solids
can be estimated from the number and size of bubbles passing
through the bed, and that from this we can find the flow rate of
gas through the emulsion. This is the key idea used here.
Assumption 5. Solid Movement in Bed. Every rising
bubble has an associated wake of material rising behind it,
The ratio of wake to bubble volume, V,,/Vb, is found by experiment-Le., Figure 3-and
we take the void fraction of the
wake to be that of the emulsion phase.
Just above the distributor solid is entrained by rising bubbles
to form the bubble wake. This solid is carried up the bed at
velocity ub and is continually exchanged with fresh emulsion
solid. At the top of the bed the wake solids rejoin the emulsion to move down the bed at velocity us.
Assumption 6. Gas Velocity in Emulsion. The relative
velocity between upward percolating emulsion gas, ue, and
downward flowing solid, us, is given by the minimum fluidizing
conditions, or

Emulsion
phase

t t t
"0

Figure 4. Main features of solid movement and


gas flow as visualized b y bubbling b e d model

Emulsion Phase

The treatment of bubbles took the upward velocity of


emulsion gas to be the minimum fluidizing velocity, or

This is equivalent to assuming that solids do not circulate in


the bed. Further on we show that this is a reasonable simplification, because the contribution of the emulsion gas is small
compared to the total flow through a bubbling bed.
For an adequate treatment of emulsion gas flow, however,
any gross movements of solids must be considered. If there
is such a movement, even with an unchanged relative velocity
between gas and solid u, = umf/Emf? we find that the upward
velocity in the emulsion is
UB

# u/ = em f

Experiments by Rowe and Partridge (1962) and Toei et al.


(1967) sho\v a distinct pattern of solids movement in a bubbling
bed, involving a slow downward drift countered by an occasional rapid upward sweep of groups of particles in the bed.
Alternative descriptions for the movement of solids have been
given in terms of a turnover rate (Katz and Zenz, 1954; Leva
448

l&EC FUNDAMENTALS

This expression shows that if the downward velocity of solids


is sufficiently high, as may be the case in vigorously bubbling
beds, the emulsion gas will reverse its direction of flow. This
result may seem surprising to some; nevertheless tracer studies
by May (1959) and Kunii et al. (1967) in vigorously bubbling
beds support this finding.
Relationship among Variables do, ub, 6, u,, and us,
Given uo and u,,.
Figure 4 illustrates the essentials of this
model. Let us develop some of its consequences. \Yith 6
as the volume fraction of the bed consisting of bubbles, a material balance for the solids crossing any horizontal plane gives
fraction of cross
section where
"(downward,
solids are moving
hence)"

At

fraction of cross
section where
~
~
~
~

everywhere but
bubble and wake
In symbols this becomes
(1 - 6

- aS)u,

= a8uo

where a is defined as
volume of emulsion transported
upward behind a bubble
E
a =
volume of a bubble

V,

Hence the down\vard velocity of solids in the emulsion is

(13)

For large fast bubbles Lvith negligible clouds (ub > 5 u f , or


u, > 2umf) the relationship between the total flow of gas and
the flow in emulsion phase and bubble phase with its wake is

uo =

void volume
of emulsion ue
(bed volume)

void volume of
bubbles and wakes ub
bed volume

or
u, = (1 -

- ff8)emfu,

+ (6 +

66mf)Ub

(15)

Combining Equations 12, 14, and 15 to eliminate u, and


u, gives an expression relating bubble velocity with uo and
Umf;

thus

axial and radial dispersion coefficients D, and D, to the movement of gas, and by fitting a variety of two-region models with
and without gas interchange between bubble and emulsion
regions.
The bubbling bed model indicates that the gas in a vigorbubble gas
ously bubbling bed is in three regions-namely,
surrounded by cloud gas, both rising through emulsion gas,
with the characteristics of these three regions given in terms of
u,, unLf,and one bed parameter, the bubble size.
T o be able to use this model for heat transfer, mass transfer,
and reaction operations one more characteristic of bubbling
beds is needed-namely, the interchange rate from region to
region. This is here developed. For beds of small slow bubbles (ub < u,, or u,
2urn,) these bubbles simply supply a
short cut for the faster moving percolating gas. This situation
is easy to treat, but is not of much practical importance. Hence
we develop only the expressions for vigorously bubbling beds
having bubbles with negligible clouds (ub > Emf, or uo > 2urnf).
This is the troublesome situation where gas bypassing is serious,
and it also is the situation of practical importance.
Since there are as many definitions of gas interchange as
there are interpreters of two-region models, let us first clearly
and unambigously define a set of gas interchange measures for
the three-region model. So, recognizing bubble, cloud, and
emulsion regions, consider a single bubble of volume Vb.
Based on unit volume of bubble the interchange coefficient
between bubble and cloud, &c, cloud and emulsion, K,,, and
the over-all coefficient between bubble and emulsion, Kb,, can
be defined as

<<

ub

1
= - [u,
6

(1

- 6 - a6)Umfl

At high gas velocities u, is the dominant term in the large


brackets, lvhile a t low uo the bubble fraction 6 becomes very
small; hence as a good approximation we have in the whole
range of flows

This expression is identical to Equation 10, derived assuming


stationary soiids. Hence the relation of U b , 6, and db with u,
and urn is approximately the same lvith both treatments, and
is given by Equations 7 and 10. This finding justifies the statement just after Equation 11 that the description of the bubble
phase is essentially the same whether the solids are taken as
circulating or stationary.
The upward velocity of gas in the emulsion is found directly
by combining Equations 12, 14, and 16 to give

cAc,

Now, when the emulsion solids descend faster than the gas can
percolate through the solids, the direction of gas flow reverses
itself and becomes downward in the emulsion. From Equation 18 this occurs when ue < 0 or when

where cab,
and CAeare the mean concentrations of gaseous
component A in the bubble, in the gas cloud and wake, and in
the emulsion phase, respectively. The relationship between
interchange coefficients is then
1

- N -

Kbe-

Approximating a: by v w v b from Figure 4, or cr = 0.2


0.4
for d , < 0.2 mm, taking e m f = 0.5 and 6 to be small, we find
that this flow reversal occurs when

From a physical standpoint the interchange coefficient can be


looked upon as a flow of gas from bubble to emulsion with an
equal flow in the opposite direction; thus

Kbe

T o summarize: At low gas velocities solid circulation is


negligible, so Equations 7 to 10 describe the bubble behavior.
,4t high gas velocities the circulation of solids becomes appreciable; however, an accounting of this circulation gives substantially the same expressions as the simpler treatment without
solid circulation. Hence Equations 7 to 10 describe the bubble
phase for the whole range of gas flow rates. At low gas velocities the emulsion condition is given by Equation 11 and by
us = 0 ; a t high gas velocities it is given by Equalions 18 and
14.
Gas Interchange between Phases

The over-all pattern of gas flow through bubbling beds has


been interpreted in various ways: by dealing with stimulusresponse curves alone, by fitting a diffusion-type model with

Kbc

+-Kce1

(volume of gas going from bubble )


\to emulsion or emulsion to bubble)
=
(volume of bubbles in the bed) (time)

Alternatively we may express the gas interchange in terms of


a dimensionless crossflow ratio, defined with respect to either
bubble volume or bed volume. Thus

number of times the bubble gas


is replaced on passage of bubble
through the bed

ub/L f

For uniform conditions throughout a bed I; remains constant


but X varies linearly with bed height. Various other bases
may be used to define the interchange coefficient, and in comparing the results of different studies one must be careful to
convert all these coefficients to the same base.
From the results of the analyses of the bubble and emulsion
phases let uq present expressions and values for these interchange coefficients. First consider the interchange between
VOL. 7

NO. 3

AUGUST 1 9 6 8

449

and in terms of bubble diameter, and with Equation 4, we have

where r refers to velocities with respect to the solid and De


is the effective diffusion coefficient of gas in the emulsion phase.
Replacing Equation 27 in Equation 25 and matching with
Equation 21 gives
Kce

kceSbc ( d c / d b )

Vb

or

Figure 5. Individual contributions to flow of gas


between bubble and emulsion

bubble and cloud. This will involve both bulk flow across the
boundary and mass transfer between gas in the bubble and in
the cloud. So, referring to Figure 5 , we have, for a single
bubble,

- - dNAb
_

(q f

kbcsbc)(CAb

dt

- CAC)

(23)

where q is the volumetric gas flow into and out of a single bubble
(cubic centimeters per second), given by Equation 6, and k b c
is the mass transfer coefficient between bubble and cloud
(centimeters per second).
Assuming a spherical cap bubble with nose angle e = 100'
and the Higbie penetration model with diffusion limited to a
thin layer at the interface, Davidson and Harrison (1963) derived the following expression for the mass transfer coefficient
between bubble and cloud,

Substituting the above expression and Equation 6 into Equation 23 and matching with Equation 21 gives the interchange
coefficient between bubble and cloud as

For a quiescent emulsion in which gas and solid do not move


relative to each other, the effective diffusivity of gas can be
taken as E ,
However, a bubbling bed has an active moving
emulsion in which the relative velocity between gas and solid is
u f and this should result in an increase in effective diffusivity
above the quiescent value. I n the absence of experimental
data and until such data becomes available, let us assume the
numerical value of De to be comparable to as follows:
De =:

Discussion

Bubble Size from Frequency Measurements. T h e size


of bubbles is a key quantity in this treatment, and there are
many ways to measure it. Here we suggest a simple method
using a frequency probe. Now the frequency of bubbles
passing the tip of a probe is given by

-I

450

l&EC FUNDAMENTALS

ub

(24)

where kc, is the mass transfer coefficient between cloud and


emulsion. Since the exposure time of all elements of interface
moving from the top to the bottom of the bubble is the same,
this process is best represented by the Higbie penetration model.
Analogous to the contact of a bubble with a liquid, the characteristics of this system are equivalent to the contact of a
vertical cylinder with the same diameter and height as the
spherical cloud (Higbie, 1935). Thus

a,

Equations 24 and 28 are the expressions which when combined using Equation 22 give the bubble model predictions of
Kbeand X. As with the previously obtained expressions for the
bubble and emulsion phase properties, we see that these interchange coefficients are completely determined by one bed parameter, the effective bubble diameter.

n=-

Next, estimate a coefficient between cloud and emulsion.


Because there is no flow of gas between these regions, diffusion
will be the only mechanism of transfer operating; thus

,/a)

where h is the height between two successive bubbles to register


on the probe. If the probe registers all bubbles whose centers
pass within one bubble radius of the probe, then on the average
h is related to 6 by

Combining the above two expressions and eliminating 6 with


Equation 10 gives, in a bed with large bubbles,
1.5ub

1.5
.--uo - u m f - (uo - urn/)
ub

umf

(29)

Thus in a fluidized bed where bubbles are rising uniformly


across a section under consideration, the bubble size is related
in a simple way to the bubble frequency as measured by the
probe.

bed, including the flow in each phase and the extent of gas
interchange between phases.
With the basic machinery as developed here we are now in a
position to account for the behavior of the bed when used for gassolid heat transfer and mass transfer or for solid-catalyzed gasphase reactions. This is reported in a follow-up article (Kunii
and Levenspiel, 1968b).
Recent experiments by Kunii et al. (1967) on solids circulation, gas flow, flow reversal in the emulsion, and other characteristics of freely bubbling beds indicate that the type of flow
visualized and assumed for this model is approximated in beds
where the height is no greater than the diameter. In taller
beds more bubbles tend to rise near the bed axis, inducing a
channeling effect. This is not accounted for in the present
model. O n the other hand, this model should be a reasonable
representation of a bed containing adequate internals.

h
0

Nomenclature

At
Cab,

db

[cml

= cross-sectional area of fluidized bed, sq. cm.

CAo CAc= concentrations of gaseous component A in

D,,D,

Figure 6. Over-all gas interchange for different


bubble sizes

53, De

Comporison of bubbling bed model predictions of Equations


22, 2 4 , and 2 8 with experimental measurements of Koboyarhi
et a/. (1 967)

db

dbad

= 0.4

dP

.i?

L, = 10
100 cm.
e m f E 0.5
umf = 2.1 cm./sec.
uo = 9 N 18 cm./sec.
d b measured at mid height

d,
h

Kb,,

Kce

of bed

Relationship between Effective Bubble Size and Measured Size. Because of the many simplifying assumptions
made throughout this analysis, from the assumption of the
Davidson bubble onward, it would be mere coincidence if the
effective bubble size (calculated from the model) was found
to match the actual measured size. I n any case it is expected
that a simple relationship holds.
A first comparison can be made from the reported data of
Kobayashi et al. (1967), who found both Kbe and db in the same
experiment. Their reported data are shown as the points in
Figure 6, while the predictions of the model, obtained from
Equations 22, 24, and 28, are shown as lines. They also calculated the approximate values of both KO, and db based on the
experimental results reported by Gilliland and Mason (1949,
1952), Overcashier et al. (1959), and Ioffe and Pismen (1960),
which are included in Figure 6. The seemingly perfect agreement is better than can be expected.
Additional checks with other types of measurements such as
axial and radial dispersion of gas, the movement of solids, etc.,
are needed. Such a program is under way and will be reported
on soon.

1
N A

t
Ub, Ubr

Summary

Quantities descriptive of the bubble phase, emulsion phase,


and the gas interchange rate between these phases have been
developed for a bubbling fluidized bed. These expressions
indicate that knowing u,,, uo, and the properties of the materials, only one parameter, the effective bubble size, is needed to
supply an adequate description of the flow of gas through the

vb

Vu
X

bubble, in gas cloud-wake region, and in


emulsion phase, respectively, g.-moles/cc.
= axial and radial dispersion coefficients of gas,
sq. cm./sec.
= diffusion coefficient of gas and effective diffusion coefficient of gas in emulsion phase, respectively, sq. cm./sec.
= effective diameter of bubble, cm.
= effective diameter of cloud, cm.
= particle diameter, cm.
= acceleration of gravity, cm./sec.2
= vertical distance between two successive bubbles to register on a frequency probe, cm.
= interchange coefficient between bubble and
cloud and between cloud and emulsion,
respectively, based on volume of bubble,
sec. -l
= over-all
interchange coefficient between
bubble and emulsion based on volume of
bubble, sec.-l
= mass transfer coefficient between bubble and
cloud and between cloud and emulsion,
respectively, cm./sec.
= mean height of fluidized bed, cm.
= height of static bed, cm.
= height of bed at minimum fluidizing conditions, cm.
= vertical distance from distributor, cm.
= g.-moles A
= frequency of bubbles passing a probe located
a t a given point in bed, set.?
= volumetric gas flow rate into and out of a
single bubble, cc./sec.
= effective radius of bubble and of cloud, respectively, cm.
= boundary area of a single bubble, and outer
boundary area of its cloud, respectively,
sq. cm.
= time, sec.
= rise velocity of a bubble in a bubbling bed
and velocity of bubble with respect to
emulsion ahead of it, cm./sec.
= upward gas velocity within emulsion phase,
cm./sec.
= u m f / e m f ,cm./sec.
= superficial gas velocity a t minimum fluidization conditions, cm./sec.
= superficial gas velocity (based on an empty
tube), cm./sec.
= downward velocity of solids in emulsion phase,
cm./sec.
= volume of single bubble, cc.
= volume of wake of single bubble, cc.
= number of times bubble gas is replaced on
passage of bubble through bed
VOL. 7

NO. 3

AUGUST 1968

451

ff

= volume

=
=
=
=
=

Ee
Ef
Cnl

Emf

of emulsion transported upwards


behind a bubble per volume of a bubble
volume fraction of bubbles in fluidized bed
void fraction of emulsion phase
average void fraction in fluidized bed
void fraction of static bed
void fraction of bed at minimum fluidization
conditions

literature Cifed

Bart, R., Ph.D. thesis, Massachusetts Institute of Technology,


Cambridge, Mass., 1950.
Brotz, W.,
Chem. Zngr.-Tech. 28, 165 (1965).
Davidson, J. F., Harrison, D., Fluidized Particles, Cambridge
University Press, New York, 1963.
Gilliland, E. R., Mason, E. A., Znd. Eng. Chem. 41, 1191 (1949);
44, 218 (1952).
Higbie, R., Trans. A.Z.Ch.E. 31, 365 (1935).
Ioffe, I. I., Pismen, L. M., Prom. Khim. 4,287 (1960).
Katz, S., Zenz, F. A,, Petrol. Rejner 33, 203 (1954).
Kobayashi, H., Arai, F., Sunakawa, T., Kagaku-Kogaku (Chem.
Eng., Japan) 31, 239 (1967).
Kunii, D., Levenspiel, O., Fluidization Engineering, Wiley,
New York, 1968a.

Kunii, D., Levenspiel, O., IND. ENG. CHEM.PROCESS


DESIGN
DEVELOP.
7, in press (1968b).
Kunii, D., Yoshida, K., Hiraki, I., Proceedings of International
Symposium on Fluidization, Netherlands University Press, Amsterdam, 1967.
Leva, M., Grummer, M., Chem. Eng. Progr. 48, 307 (1952).
Lewis, i V . K., Gilliland, E. R., Girouard, H., Chem. Eng. Progr.
Symp. Ser. 58 (38), 87 (1962).
May, \V. G., Chem. Eng. Progr. 55, 49 (December 1959).
Mori, Y . , Nakamura, K., Kagaku-Kogaku (Chem. Eng., Japan)
29, 868 (1965).
Nicklin, D. J., Chem. Eng. Sci.17, 693 (1962).
Overcashier, R. H., Todd, 0. B., Olney, R. B., A.Z.Ch.E. J . 5 ,
54 (1959).
Rowe, P. N., Partridge, B. A., Proceedings of Symposium on
Interaction between Fluids and Particles, Institution of Chemical Engineers, p. 135, June 1962.
Rowe, P. K.,Partridge, B. A., Trans. Znst. Chem. Engrs. 43, T157
(1965).
Rowe, P. N., Sutherland, K. S., Trans. Znst. Chem. Engrs. 42, T55
119641.
Sutherlakd, K. S., Trans. Znst. Chem. Engrs. 39,188 (1961).
Talmor, E., Bennenati, R. F., A.Z.Ch.E. J.9, 536 (1963).
Toei, R., Matsuno, R., Nagai, Y . , Ka,caku-KoRaku (Chem. Eng.,
J a j a n ) 31, 457 (1967).
RECEIVED
for review Novcmber 13, 1967
ACCEPTED
April 17, 1968

ELUTRIATION FROM A MULTISIZE


PARTICLE FLUIDIZED BED
D E R A N H A N E S I A N A N D ALBERT

R A N K E L L

hbwark College of Engineering, Aewark, .V. J. 07102

Elutriation in a multisize particle fluidized bed was studied by fluidizing spherical glass beads of size range
less than 595 and greater than 88 microns, in a 3-inch diameter column using air as the fluidizing medium.
The variables studied were particle size distribution and superficial gas velocity. Particle distributions were
made to follow a straight line on log-probability coordinates. The rate of elutriation could be described
( 1 - b)e-k28in which X and X, are the conmathematically by an equation of the form: X / X , =
centrations by weight of particles below a given size in the bed at times 0 and 0, respectively; b is a constant
kz, both constants having dimensions of reciprocal time units. k l increased with
with value 0 5 b 5 1 ; k l
increasing superficial gas velocity and decreased with increasing particle size. The size distribution of the
elutriated particles formed a log-normal distribution in which the geometric mean diameter increases with
elutriation time, while the standard deviation decreases with time.

<

LUTRIATION

of particles from a fluidized bed is usually en-

E countered whenever the bed is composed of particles which

do not have the same diameter. The elutriation phenomenon


results from the fact that the gas velocity required to fluidize
the largest particle usually exceeds the terminal velocities of
the finer particles in the bed. Thus, the finer particles become entrained in the fluidizing gas and are removed from
the system.
Previous studies of this phenomenon have investigated artificial systems consisting of one coarse and one or more fine
components (Leva, 1951; Osberg and Charlesworth, 1951 ;
Wen and Hashinger, 1960). This work, however, is not directly applicable to operating conditions in full-scale fluidized
bed processes. This observation has been confirmed by
Thomas, et al. (1961a).
1

452

Present address, University of IVisconsin, Madison, IVis.


l&EC FUNDAMENTALS

Since there are few published data on the elutriation of


particles from fluidized beds with wide ranges of particle size,
it is hoped that this study (Rankell, 1965) will add to the available information.
Review of Literature

One of the first attempts to correlate experimental elutriation


rates was the work of Leva (1951). The systems he investigated consisted of binary mixtures composed of 80% of coarse
component and 20y0 of fine component. H e expressed the
rate of elutriation by an equation of the form:

= Xoe-ka

(1)

Equation 1 was compared to the rate constant for a firstorder chemical reaction. The elutriation rate constant was
found to be almost independent of the composition and composite size of the coarse component, but did increase rapidly

You might also like