Professional Documents
Culture Documents
A new fluid-solid contacting technique known as Semifluidization has been studied. Mass
transfer data were obtained for a benzoic acid-water system, and correlations in terms of mass
transfer factors and over-all void fractions (for both packed and fluidized sections) are pre- depth and prevent the complete expansion
of the solid bed.
sented. Previous mass transfer correlations for fixed and fluidized beds are critically evaluated
with respect to the limitation and applicability and are discussed in relation to semifluidized-bedFor the second series of experiments the
mass transfer correlations. apparatus was modified.
In the first series only the inlet and out-
Semifluidization is a type of fluid- correlation of over-all mass transfer co- let concentrations were measured, whereas
solid contact operation in which the efficients and the comparison of the in the second series the axial concentration
expansion of a fluidized bed is partially correlations obtained with those of the gradient of the fluid was measured. Since
restricted. Under such conditions the fluidized and packed beds. the samples were obtained from inlet and
porosity of a semifluidized bed is inter- The relationships among the over-all outlet pipes in the first series of experiments,
mediate between the rigidly packed mass transfer coefficients, the mass they could be considered as the average
bed and the completely fluidized bed. transfer coefficients in each individual samples across the cross section of the
Some of the characteristics of a semi- bed, and the internal structure of the column. The sampling lines having l j 4 - h .
I.D. were employed in the second series of
fluidized bed, such as the simultaneous bed (that is, the relative size of the experiments. They were sufficiently large
formation of a packed section and a packed and fluidized sections in the to assume withdrawal of uniform average
fluidized section in the bed, were re- bed) were carefully considered. samples across the cross section of the
ported in a previous communication column, as evidenced by the fact that the
(9). The purpose of this paper is to EXPERIMENTAL concentration of outlet samples obtained
report the results of experimental work from the sampling line just above the top
The experimental apparatus consisted of of the screen was almost identical with
on the mass transfer aspect in a semi-
a main test column in which granular those obtained from the outlet of the
fluidized bed of a solid-liquid system. benzoic acid particles were dissolved by a column. The apparatus used in the latter
Aside from its practical significance measured stream of upward flowing water. series was the same as for the former
in mass transfer operations, the study A container was provided for retention of series, except the test section. This new
of rate of mass transfer is of general samples of each run for analysis. Figure 1 test section was made from a 2-in. I.D.
significance because it often leads to shows the diagram of the apparatus. standard iron pipe with cocks attached for
better understanding of other aspects Water was supplied from a 50-gal. con- sampling, at 2-in. spacings. A sheet of filter
of chemical engineering operations and stant-head tank ( A ) and was forced cloth was cemented to the inside wall over
heat transfer. through the system by a centrifugal pump the outlet of each cock to prevent the
Since the mechanism of interphase ( B ) . The water was initially directed plugging of the solid benzoic acid particles
through the globe valve and a flowmeter from the sampling lines. The sizes of
mass transfer in a fluidized bed is al- ( C ) as a means of controlling and measur- particles used (8 24 mesh) were much
N
ready complicated ( 4 , 2 4 ) , no particu- ing the rate of flow. The test column con- larger than the opening of the filter cloths,
lar attempts were made to analyze the sisted of two parts, a calming section ( D ) and the particles were freed from the dust
data in the light of the rigorous theo- and a test section ( F ) . The test section by a washing process suggested by Evans
ries of interphase mass transfer. Rather, was a vertical Pyrex tube 2 in. I.D. and and Gerald ( 8 ) . No escaping of the parti-
emphasis was placed on the empirical 24 in. high. Two brass plates clamping a cles from the sampling cocks was observed.
filter cloth ( E ) were placed between the Although the solubility of the benzoic
A WATLR TANK
calming and the test sections to establish acid in water is low, in order to minimize
(I?W? an even distribution of the upward flow. A the change of geometric shape of the par-
C ILOIYLTER
D C U Y l N D SCCTIOII
rigid close-fitting sieve plate ( G ) was at- ticles during the experimental runs the in-
E m . T U CLOTH tached to a long rod which could be moved dividual runs were not permitted to exceed
F TEST SECTION
B TO? SIEVE ?LATE
up or down to a fixed position. This was an 80-sec. duration. A decrease of less
M SAWLIMO OUTLET installed in the test column to adjust bed than 3% of the initial weight of the solid
I RCTLWTlON TbNK
particles for a single run was experienced.
Intermittent analysis of effluent liquid con-
centration during the run indicates that
steady state operation has been achieved.
....
...
-- ...
O
S I
I I
-.....
c
5
5 ...
:k ....
0..
...
I 11
e ..I 0.1
I
..I
11.
e.. ..I ...
Fig. 1. Diagram of experimental apparatus. Fig. 2. Axial concentration profile. Fig. 3. Axial concentration gradient.
.) 20
-x
z
\
;
;10
1 0
0
-
z
f a
4
NRe
point at 0.52 ft. of bed height corre- Since there exists a finite discontinu- Equation ( 7 ) may be transformed into
sponds to the boundary of fluidized ity in the axial concentration gradient
bed section and fixed bed section at
which a distinct discontinuity of con-
at the interphase of the fluidized and c, - c,
Gln -= k,, A
packed beds, (9)
centration gradient exists. c, - c,
dC '2 dC or
ANALYSIS AND CORRELATION
G f c c +Gl1= = G' (C, - C,)
OF RESULTS klm
(10)
To derive the rate equation for the A(AC),,
mass transfer between the solid parti- = (k,)Im a, r d z
In order to account for the effect of
cles and fluid in the semifluidized bed axial mixing, which was neglected, k,,
the following assumptions were made:
1. The concentration of the fluid in
+ (kps) In, up. JrdZ (5) must include a correction factor F sug-
gested by Epstein (7, 1 3 ) .
contact with the particle surface is In case the axial concentration data
or are available, as in the second series
equal to the saturated concentration.
2. The velocity distribution and
Gln - Gln-
c, - c, + c, - c, of experiments, the value of k in each
concentration of fluid are uniform at bed can be directly calculated by use
any cross section of the test section.
c,- c, c, - c, of the integrated mean driving force
as
3. The surface area of the particles
per unit volume in each bed is uni-
= (k,)Im ath, + ( h e ) l m a p a h p a
form. G' (C1- CO)
(6)
4. Axial turbulent diffusion coeffi-
cient, E,, in each bed remains constant But Equation can be further
under definite flow conditions. simplified: and
5. The process is isothermal.
The following material balance could Gln-
c.. - c, = (k,),,,,A, +
be obtained for any bed height: c. - cz
(kpa)imApe (7) The over-all mass transfer coefficient
based on integrated mean driving force
If the weighted mean value of over- may be evaluated as
all mass transfer coefficient is defined
as
When one assumes that longitudinal or
axial mixing in the bed can be negligi-
ble
other investi ators ( 3 , 8) for the estimated by an equation including the J d = 1.340 1- (7)
(21)
%
packed- and uidized-bed mass trans- independent variables of the experi-
fer. The effect of natural convection on ment and the characteristics of ordi- The over-all correlation of Evans and
the rate of mass transfer which would nary fluidization under similar flow Gerald's (8) data may be expressed as
become appreciable for the large parti- conditions. Such an equation was al-
cles ( D , > '/4 in.) at the very low
(7)
-8.480
ready reported by the authors in the N'Re
Jd = 1.680 1 - (22)
Reynolds numbers as indicated by previous communication (9).
Gaffney ( 1 2 ) was neglected, and no It was also reported in the commu-
attempt was made to introduce Grashof nication that the over-all mass transfer Comparison of Equations (22) with
number in the correlation of data. coefficients based on the over-all loga- (15) reveals that within the range of
The broken line in Figure 4 represents rithmic drivin forces can be correlated the variables of the present investiga-
the generalized correlation used by P
by use of J d actor suggested by Chu tion and the errors of experimentation
Chu et al. ( 4 ) to correlate the mass and his co-workers ( 3 ) within the the semifluidization data may be cor-
transfer data for both packed and range of maximum probable deviation. related as well as the data for ordinary
fluidized beds. The solid line correlated Referring to Figure 4, one can see that fluidized and packed beds by use of
all the semifluidization data with the Chu's original Jd factor correlation the Id factor VS. W R e / ( l- E ) plot
magnitude of the deviations compara- yields J d values consistently higher based on the over-all logarithmic mean
ble to the correlation used by Chu than the present experimental data. mass transfer coefficient, k,,.
et al. ( 4 ) . The equation for the solid Chu's correlation pertains only to the The data for the benzoic acid-water
line was found to be ordinary fluidized and packed beds. system were not included in Shirai's
The maximum deviation of the present ( 2 3 ) generalized correlation covering
d] z= 1.865
N)&#
l--E
(-) (15) data based on his correlation appears the data of the fluidized bed, fixed
to be 22% in the range of Reynolds bed, and a sin le particle. Evans and
numbers used. Evans and Gerald's (8) Gerald's (8) ata were again recor-
Shirai ( 2 3 ) presented the following data for the benzoic acid-water system related by the method suggested by
f
correlation of the mass transfer data in were not included in the original cor- Shirai ( 2 3 ) as illustrated in Figure 5 ,
the particle beds as suggested by the relation by Chu et al. ( 3 ) . They were which indicates that the experimental
boundary-layer theory (10, 16): recalculated in terms of J d VS. W R e / data were consistently lower than the
N ~ , -= +
E 2.0 0.75 Nge1" Nsdl'
(16)
Figure 5 shows the similar correlation
for the present data and a comparison
Correlation linr of aulhorb data
with Shirai's equation ( 2 3 ) . The re-
gression line through the data is ex-
__-__ Wnlation line of Evom8 CornIda data
pressed as
Nshc = 2.0 + 0.46 N',,O.MNs,"."
(17) mesh, second srrir
To take the effect of molecular dif-
fusion and natural convection into ac-
count the Id factor can be modified as
(23)
or
Nke, (I - 8 ,
Nsh = 2.0 + l.51[ ( 1 - -E)N'R~]''*
Fig. 6. I'd factor vs. Nm/(l-e) correlation of the experimental
( N s c ) 'I8 (19) data.
(23) 0
For fluidized bed:
Nshe = 2.0 + 0.230 NR80.868 NS:ls
(Nsc)’18 (26)
[
kl, = 1-
p,S(l--EPY)(h,.-h)
W
accuracy for mass transfer calculations.
But there is good indication that axial
mixing in the liquid-solid systems is
and the fluidized-bed data are corre- much less extensive than in the gas-
lated as solid systems (21, 2 6 ) . In Figures 4, 5,
and 6 the mass transfer coefficients cal-
Nsa = 2.0 + 1.031[ ( 1 - e)N’Re]’’’
p a S,(1- ep.) (h,,? -h) culated by use of the integrated mean
( N s c )‘Is(27)
+ W
driving force are compared with the
correlation obtained in the first series
An. over-all correlation equation may (1 - .,) of experiments. They indicate that the
be expressed as (kpa) Im (30) deviation of the second series experi-
(6, - CPB)
mental data from the correlation are
Ns, = 2.0 + 1.202[ ( 1 - e)wRe]112
An equation similar to Equation (30) within the range of the first series of
( Nsc)Iia (28) to estimate k,, can be derived by a experiments. This implies that the
Vol. 6. No.
~-3 A.1.Ch.E. Journal Page 487