Date post: | 15-Sep-2016 |
Category: |
Documents |
Upload: | mohammad-asif |
View: | 213 times |
Download: | 0 times |
Minimum fluidization velocity and defluidization behavior of binary-solid
liquid-fluidized beds
Mohammad Asif *, Ahmed A. Ibrahim
Department of Chemical Engineering, King Saud University, P.O. Box 800, Riyadh-11421, Saudi Arabia
Received 18 March 2001; received in revised form 10 October 2001; accepted 28 February 2002
Abstract
The minimum fluidization behavior of five different binary-solid systems with a wide range of composition is experimentally investigated
by carrying out slow defluidization of an initially fluidized bed. The size ratios of these binaries vary from 4 to 10 while their buoyed-density
ratios vary from 0.22 to 0.52 such that their larger components are lighter and smaller ones are denser. The difference in the physical
properties of the two constituent solid phases of the fluidized bed is found to strongly influence the evolution of the packing structure, and
consequently the minimum fluidization velocity during the slow defluidization process. For binaries with the same size ratio, segregation
increases with the decrease in their buoyed-density ratios. On the other hand, even for binaries with large difference in their densities,
increasing the size ratio enhances the mixing. Depending upon the composition of the bed, a completely mixed defluidized structure
sometimes develops for high size-ratio binaries. Finally, a simple correlation is proposed that can better describe the present minimum
fluidization velocity data than other existing correlations.
D 2002 Elsevier Science B.V. All rights reserved.
Keywords: Binary-solid; Defluidization; Liquid fluidization; Minimum fluidization velocity
1. Introduction
The minimum fluidization velocity is an important
hydrodynamic feature of fluidized beds. It marks the tran-
sition at which the behavior of an initially packed bed of
solids changes into a fluidized bed, and is therefore a crucial
parameter in the design of reactors or other contacting
devices based on the fluidized bed technology.
Besides its potential application in the development of
multi-functional reactors involving simultaneous reaction
and adsorption [1–4], the binary-solid fluidization can help
to alter the basic hydrodynamic characteristics of a fluidized
bed by the addition of another solid phase, which has dif-
ferent physical properties than the resident solid phase of the
fluidized bed. Yang and Renken [5] have recently made one
such application in enhancing the mass transfer coefficients
by adding smaller but denser inert glass beads in a fluidized
bed containing active resin particles. In the gas–solid fluid-
ization, the use of binary-solid fluidized beds for thermo-
chemical processing of biomass is well established as can be
seen from the work of Narvaez et al. [6], Olivares et al. [7]
and Berruti et al. [8]. In this application, an inert solid
species, often sand, is used to achieve the fluidization of the
biomass and control its residence time besides improving
the heat transfer.
Our main interest here is to examine the behavior of
binary-solid liquid-fluidized beds close to the conditions of
the minimum fluidization by carrying out a slow defluidiza-
tion of an initially fluidized bed. Six different solid samples
involving five different binary-solid systems are considered
here. These binaries significantly differ in the size as well as
the density such that their size-ratios vary from 4 to 10 with
two of them possessing size ratios higher than 6.5. For the
size ratio of constituent solid phases over 6.5 is known to
show contraction of the specific volume in their packing
behavior [9,10]. To our knowledge there has not been any
specific study of the minimum fluidization behavior,
whether gas–solid or liquid–solid, in this range of size
ratios. On the other hand, two binaries with almost the same
size ratio but considerable difference in their buoyed-density
ratios are also considered in order to delineate the effect of
the density difference on their behavior. Awide range of the
bed composition is studied for each binary-solid system.
The experimental minimum fluidization velocity data, thus
obtained, are compared with existing correlations with the
0032-5910/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S0032 -5910 (02 )00061 -X
* Corresponding author. Fax: +966-1-467-8770.
E-mail address: [email protected] (M. Asif).
www.elsevier.com/locate/powtec
Powder Technology 126 (2002) 241–254
aim of improving their predictive capability particularly for
the binaries used in the present study.
2. Experimental
The test section of the fluidized bed consisted of a
transparent perspex column of 60-mm internal diameter
and 1.5-m length. A perforated plate with high density of
2-mm holes, located on a square pitch, and 4% fractional
open area was used as the distributor. Both its top and
bottom faces were covered with a fine mesh, and were
preceded by a 0.5-m-long calming-section packed with 3-
mm glass beads.
Water was used as the fluidizing medium with its temper-
ature carefully controlled at 20 jC. The water flow rate was
controlled using one of three calibrated flowmeters of
suitable range. An immersion cooler was used to remove
the heat generated by the water pump, and maintain the
water temperature constant in the water tank.
The bed heights were read visually with the help of a
ruler along the length of the column. The pressure drop
along the bed was measured using a manometer. The
observation included measuring the flow rate, the bed height
and the pressure drop across the bed.
2.1. Properties and fluidization behavior of solid particles
Six different types of solid samples were used in the
present study. Their physical properties are tabulated in
Table 1. The glass and sand samples consisted of the
fraction between the two adjacent sieves except SN257,
which had a relatively wider size distribution with 33% of
212–250-Am range and 67% of 250–300-Am range. The
mean particle diameter for the fraction between the two
adjacent sieves was taken to be the arithmetic mean of the
sieve openings while the volume-mean diameter of the
SN257 sample was computed to be 257 Am. Both the
volume-equivalent mean diameter and the shape factor of
the larger solid species are reported in Table 1. Although the
two larger solid species were almost of the same shape and
size, their densities were significantly different, thereby
leading to a 100% difference in their minimum fluidization
velocities and 50% difference in their terminal velocities.
The fluidization behavior of particle samples was indi-
vidually examined using water at 20 jC. The bounded
particle terminal velocities, Ut, reported in Table 1 were
obtained by fitting the mono-component expansion data
with the Richardson–Zaki [11] equation. On the other hand,
the minimum fluidization velocities were evaluated from the
pressure-drop versus liquid superficial velocity profiles.
It is of common knowledge that as the ratio of the particle
diameter to the tube diameter is lowered, wall-effects increas-
ingly influence the fluidization behavior. In order to quantify
wall-effects in the present case, the unbounded terminal
velocities of both the larger solid species were computed
using the correlation of Di Felice [12]. These were found to
be 97.5 mm/s for the PET and 154.7 mm/s for the PlG, and in
neither case the difference between the bounded and
unbounded terminal velocities exceeded 3.4%.
Binary systems considered here are shown in Table 2. It
is clear that there is a significant variation in the size of
binaries; their size ratios vary from 4 to 10 with those of
two binaries greater than 6.5. Densities of their constituent
solid phases are also substantially different. These differ-
ences in their physical properties are clearly reflected in the
ratios of their minimum fluidization velocities and terminal
velocities as seen in the table. Note that Umf ratios are much
bigger than their corresponding Ut ratios. While it is
desirable to have higher Umf ratios as seen later, it is not
so for Ut ratios in view of the fact that the upper range of
operation of a liquid-fluidized binary-solid-fluidized bed
will generally be limited by the lower terminal velocity of
the two components.
It is also worthwhile to point out here that all these
binaries show the phenomenon of the layer inversion. The
Table 1
Physical properties of individual particle species and their fluidization properties using water at 20 jC
Solids species Material Shape Size range
(Am)
Mean diameter
(Am)
Particle density
(kg/m3)
Umf
(mm/s)
Ut
(mm/s)
SN257 sand nearly spherical 212–300 257 2664 0.83 31.5
SN275 sand nearly spherical 250–300 275 2664 1.0 34.4
GB463 glass spherical 425–500 463 2465 2.5 58.0
GB655 glass spherical 600–710 655 2465 4.7 82.4
PET polyethylene
terephthalate resin
cylindrical
(w = 0.85)
2790F 60 1396 11.8 94.3
Plastic PlG cylindrical
(w = 0.87)
2880F 90 1760 26.5 149.5
Table 2
Binary systems studied
Binary System Ratio of
Component 1 Component 2 Size Buoyed
density
Umf Ut
1 PET SN257 10.8 0.22 14.2 3.1
2 PET SN275 10.1 0.22 11.8 2.8
3 PET GB463 6.0 0.27 4.7 1.7
4 PET GB655 4.3 0.27 2.5 1.2
5 PlG GB655 4.4 0.52 5.1 1.9
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254242
experimental overall bulk density profiles of individual
species, as obtained from overall expansion, are shown in
Fig. 1a and b, which can be used to gauge the segregation
tendency of a particular binary system. For example, it is
obvious from Fig. 1b that the segregation tendency of
Binary 4 will be much higher than that of Binary 5. In the
event of segregation, the denser layer, needless to say, will
tend to occupy the lower region of the binary-solid bed.
Note that in the present study the denser layer at lower
liquid velocities is the one of the smaller components, which
has a lower minimum fluidization velocity than its larger
counterpart. This configuration is clearly different from the
equal-density coarse-fine combination where the coarse
solid species with higher Umf will constitute the lower layer.
It was found during experiments that the initial packing
of the bed of particles substantially affected its pressure-
drop profile. This is shown in Fig. 2 that different pressure-
drop profiles were obtained when the flow rate of the liquid
was progressively increased in an initially packed bed
depending upon the porosity of the packing structure, being
higher for the denser packing. On the other hand, hardly any
difference was seen in the pressure-drop profiles when the
bed was slowly defluidized by decreasing the flow rate in an
initially fluidized bed. Moreover, such an equilibrium
defluidized bed structure yielded reproducible pressure-drop
profiles when the bed was fluidized again. Therefore, most
experimental runs involved recording the pressure-drop
profile by slowly defluidizing an initially fluidized bed.
Fig. 1. (a) Experimental bulk density profiles of mono-component bed of solid particles of SN275, GB463 and PET. (b) Experimental bulk density profiles of
mono-component bed of solid particles of GB655, PET and PlG.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254 243
Once the defluidized structure started to develop in the bed,
the procedure of the slow defluidization involved decreasing
the liquid velocity in small increments and allowing enough
time for the pressure drop to stabilize between successive
readings. This was important in view of long transients of
particle movement observed during this phase of partial
defluidization. Needless to say, as commonly suggested in
the literature the minimum fluidization velocity was
obtained where pressure-drop profiles of the packed and
the fluidized bed intersect each other. For example, it was
taken to be about 5.6 mm/s in Fig. 2.
3. Existing correlations for the prediction of the
minimum fluidization velocity
There are several correlations proposed in the literature
for the prediction of the minimum fluidization velocity of
fluidized beds containing two or more different solid spe-
cies. Recently, Wu and Baeyens [13] have listed some of
these. However, only three main approaches, which in a
broader sense underlie most existing correlations, are dis-
cussed in this section.
The most common approach to predict the hydrodynamic
features of fluidized beds containing two or more solid
species is to modify the correlation used for the mono-
component beds to account for the presence of two compo-
nents in the fluidized bed. For the prediction of the mini-
mum fluidization velocity, mean values of particle
properties, i.e. diameter and density, can be used directly
in the Ergun equation [14]. Owing to the difficulty involved
in characterizing the particle shape factor and the porosity of
the bed, it is, however, quite common to introduce some
modifications in the Ergun equation, whereby terms involv-
ing these quantities are replaced by constants [15]. The
generality of such modifications, however, remains doubtful
as pointed out by Lippens and Mulder [16]. Even greater
uncertainty is likely in the event of their extension to binary
systems. Another important question in this context is the
appropriate averaging procedure for the particle properties.
Different types of property averaging are found in the
literature. For example, Thonglimp et al. [17] and Noda et
al. [18] used the following,
1
qs
¼ X1
qs1
þ X2
qs2
ð1Þ
1
d̄ qs
¼ X1
d1qs1
þ X2
d2qs2
ð2Þ
where quantities with an over-bar indicate mean values, and
X1 and X2 are fractions of components 1 and 2, respectively.
As far as the question of the averaging of particle
densities is concerned, the volume-averaged mean density
appears more appropriate if deduced from the pressure-drop
consideration in a binary-solid-fluidized bed. On the other
hand, averaging of particle diameter is perhaps more flex-
ible. But, it is the surface-to-volume averaging which has
been commonly used [19–21].
In view of the above the discussion, the following
general form of the Ergun equation with mean particle
properties is used for the minimum fluidization velocity,
1:751
e3mf
� �qfUmf d̄
l
� �2
þ1501� emf
e3mf
� �qfUmf d̄
l
� �
¼ d̄3qf ðqs � qf Þgl2
� �ð3Þ
where the term on the right-hand side is the dimensionless
Galileo number (Ga). The following definitions of the mean
Fig. 2. Effect of the initial packing structure on the pressure-drop profiles in a binary-solid bed containing SN257 and PET for X1 = 0.75.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254244
Fig. 3. (a) Effect of the liquid velocity and the bed composition on the height of the lower pure component layer of sand for Binary 2 (Fixed SN275weight = 528 g).
(b) Effect of the liquid velocity and the bed composition on the height of the lower pure component layer of glass for Binary 3 (Fixed GB463 weight = 500 g). (c)
Effect of the liquid velocity and the bed composition on the height of the lower pure component layer of glass for Binary 4 (Fixed GB655 weight = 500 g).
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254 245
diameter and the mean density are used in the above
equation,
d̄ ¼ 1
X1
w1d1þ ð1� X1Þ
w2d2
ð4Þ
qs ¼ X1qs1 þ ð1� X1Þqs2 ð5Þ
Besides the property-averaging approach mentioned
above, approaches based on the averaging of the minimum
fluidization velocities of the two components are also
proposed in the literature. Among early researchers, Otero
and Corella [22] proposed the arithmetic averaging of
minimum fluidization velocities of the constituent solid
phases. On the other hand, assuming the binary-solid-fluid-
ized bed as consisting of two completely segregated mono-
component layers, some others recommended using the
harmonic averaging of minimum fluidization velocities
[23]. The model, thus obtained, is similar to the serial model
used for the prediction of the overall bed void fraction.
Though inherently perceived to be applicable for segregated
beds, this approach has been shown to hold good even if the
components are substantially mixed [24]. Introducing a
more general expression for such averaging approaches,
we can write
Umfp ¼ X1U
pmf1 þ ð1� X1ÞUp
mf2
� �ð6Þ
where p =� 1 yields the harmonic averaging, whereas p = 1
leads to the arithmetic averaging. It will be interesting to see
whether other values of p can improve predictions and will
be considered here later.
Although based on minimum fluidization velocities of
constituent solid phases, Cheung et al. [25] proposed a
slightly different approach. Their empirical correlation is
given by,
Umf
Umf2
¼ Umf1
Umf2
� �X 21
ð7Þ
where Umf1 and Umf2 are the minimum fluidization veloc-
ities of components 1 and 2, respectively. And, X1 is the
fluid-free volumetric fraction of the larger component. The
experimental data of Formisani [20] were found to show
good agreement with the above equation.
It is worthwhile to point out here that approaches based
on Eqs. (6) and (7) benefit from the incorporation of the
mono-component Umf data as against the property-averag-
ing approach based on the Ergun equation. Moreover, the
latter needs information about the particle shape factors and
the porosity at the minimum fluidization conditions. While
the issue of the difference of the shape factors can be
addressed by using the equivalent particle diameter for
individual components in Eq. (4), the porosity of the
packing of two particle mixtures of different sizes is known
to show substantial contraction, and therefore need to be
properly accounted for in the prediction of Umf of binary
particle mixtures.
4. Results and discussion
It was observed that the difference in particle properties
significantly affected the defluidization mechanics and,
consequently, the minimum fluidization behavior of bi-
nary-solid fluidized beds. This issue is therefore first dis-
cussed in the following. The predictive capability of existing
Fig. 4. Effect of the bed composition on the minimum fluidization velocity of the binary-solid fluidized bed for Binary 1 and Binary 2.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254246
correlations is examined next in the light of the present data
and an attempt is made to improve their predictive capability.
4.1. Defluidization mechanics of binary-solid fluidized beds
During the process of slow defluidization, the evolution
of the defluidized packing structure of each binary system
differed significantly from the other due to the difference in
their physical properties. Close to the minimum fluidization
conditions, unlike the commonly studied gas fluidization of
the coarse–fine combination of equal-density solids where
fines naturally tend to migrate towards the upper part of the
fluidized bed, the fines or smaller component, being denser
in the present case, showed a tendency of migration towards
the lower region of the bed. As a result, the lower layer
sometimes consisted only of the denser component, and
sometimes it was composed of both components. The
mixing and segregation behavior of the two components
were found to mainly depend upon their size and density
differences besides being affected by the overall composi-
tion of the bed.
In spite of the difference in their size distribution, the
behavior of Binaries 1 and 2, both with size-ratios greater
than 6.5, was similar. Depending upon the overall compo-
Fig. 5. (a) Effect of the bed composition on the minimum fluidization velocity of the binary-solid fluidized bed for binaries in the present study. (b) Effect of the
bed composition on K for binaries in the present study.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254 247
sition of the bed, three different types of defluidization
dynamics were observed for these binaries and are discussed
in the following.
For beds with composition, X1, as high as 0.6 and so long
as the liquid velocity was close to the Umf of the PET, a pure
component layer of the sand was seen developing in the
lower zone of the bed with occasional straying of the PET
from the upper mixed layer consisting of PET and sand into
the lower layer of pure sand. As the liquid velocity was
gradually decreased, the lower interface of the upper mixed
layer started to stabilize and became more distinct. At this
stage, the sand from the lower mono-component layer
appeared moving into the upper mixed layer where both
components apparently developed uniform concentration.
The motion of the sand through the interstices of the PET
imparted movement to the PET as well. This situation
prevailed till the liquid velocity was decreased to a value
slightly higher than the Umf of the sand when the much
slower motion of the sand could still impart vibratory motion
to the PET and the whole bed appeared to be fluidized. The
notable difference in beds with different compositions is seen
in the height of the lower pure component layer, which for
the same liquid velocity decreases as the fraction X1 is
increased. This can be seen in Fig. 3a. The reason is obvious.
The bed with higher PET fraction can accommodate greater
amount of sand, and since the amount of the sand is fixed in
all these runs (weight of sand = 528 g), this leads to smaller
height of the lower layer of the sand.
As the fraction X1 was increased to 0.75, its initial
defluidization behavior was similar to what has been dis-
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254248
cussed above. This, however, changed as the liquid velocity
was decreased. As seen in Fig. 3a, no lower layer of pure sand
was observed with the progress of the defluidization process.
Marked with long transients, a defluidized zone containing
mainly the PET with the entrapped sand slowly started to
develop at the bottom of the bed. The sand appeared moving
up through the interstices imparting its momentum to the
PET. Thus, two visually distinct zones existed in the binary-
solid-fluidized bed: the lower defluidized zone marked with a
complete absence of any particle motion, and the upper
fluidized zone in which the motion of the sand caused overall
particle motion however small. With the further decrease in
the liquid velocity, the upper fluidized zone gradually
decreased in size while the lower defluidized zone increased
and ultimately engulfed the whole bed. At this stage, the sand
was seen dispersed throughout the bed with apparently higher
concentration at the top of the bed.
When the fraction X1 was increased to 0.86 for either
Binary 1 or 2, no lower layer of the pure sand is found as
depicted by Fig. 3a. Rather, a complete mixing of the two
components was observed with the sand dispersed through-
out the bed. In this case, no PET motion was visible once
Fig. 6. (a) Comparison of predictions of Eq. (6) for different values of parameter p with the experimental data for Binary 2. (b) Comparison of predictions of
Eq. (6) for different values of parameter p with the experimental data for Binary 3. (c) Comparison of predictions of Eq. (6) for different values of parameter p
with the experimental data for Binary 4. (d) Comparison of predictions of Eq. (6) for different values of parameter p with the experimental data for Binary 5.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254 249
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254250
the liquid superficial velocity was decreased below the Umf
of the PET. The sand entrapped in the interstices of the PET
nonetheless showed localized, albeit little, movement.
The defluidization behavior of Binary 3 with the size
ratio of 6.0 is shown in Fig. 3b. The layer of the glass
owing to its higher bulk density constituted the lower layer.
Unlike the earlier case, however, no substantial reduction in
the size of the lower glass layer is noticed for small
fractions of the PET in the bed. Especially, up to X1 = 0.4,
the size reduction can be considered not too significant.
Comparing this with the earlier case of PET-sand binary
systems, it is obvious here that as the size ratio is decreased,
the capacity of the bed of the larger component to absorb its
smaller counterpart is substantially reduced. The intermix-
ing of the two components was observed mainly at the
interface with a gradually decreasing concentration of the
glass away from the interface.
Fig. 3c shows the height of the lower layer of the pure
glass in the case of Binary 4, which has the smallest of all
size-ratios considered here. Hardly any reduction in the
size of the lower layer of the glass is observed in this case.
Although the mono-component overall bulk density pro-
files of the two binaries, i.e. 3 and 4, are not much
different as seen in Fig. 1, yet Binary 4 showed much
stronger segregation behavior at all bed compositions. The
upper layer of the larger component fails to absorb its
smaller counterpart apparently due to the fact that their
size difference is not large enough to allow the passage of
the smaller component through the interstices of the larger
ones.
On the other hand, Binary 5 showed a different type of
behavior than all others discussed before. Note that Binary 5
has almost the same size ratio as Binary 4 but the buoyed-
density ratio of the former is almost twice that of the latter
Fig. 7. (a) Comparison of correlation predictions with the experimental data for Binary 1. (b) Comparison of correlation predictions with the experimental data
for Binary 2. (c) Comparison of correlation predictions with the experimental data for Binary 3. (d) Comparison of correlation predictions with the experimental
data for Binary 4. (e) Comparison of correlation predictions with the experimental data for Binary 5.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254 251
and the others. No lower layer of a distinct solid phase is
observed in this case. As the process of the slow defluidiza-
tion was carried out from high liquid velocities, the upper
layer of GB655 slowly started receding into the lower layer
consisting mainly of the PlG. This led to an increase in the
glass concentration in the lower mixed layer. With a further
decrease in the liquid velocity, a defluidized layer was
observed to develop from the bottom of the fluidized bed
containing mostly the larger component trapping the smaller
component alongside. Above the defluidized layer, the
smaller component imparted its motion to the larger par-
ticles and kept them fluidized. As the liquid velocity was
decreased further, the lower defluidized zone slowly
increased in the size and finally covered the whole fluidized
bed. From the visual observation, it appeared that the
concentration of the smaller component was lower in the
bottom and progressively increased along the bed height.
While for smaller X1 the presence of the glass beads was
clearly visible close to the distributor, no glass was, how-
ever, seen in the distributor region and even close to it when
X1 was as high as 0.86.
4.2. Minimum fluidization velocity of binary mixtures
Fig. 4 presents the experimental data for Binaries 1 and
2, which differ only in their size distribution. Apparently, no
effect of the size distribution is seen on the minimum
fluidization velocity of the binary mixtures. It is noteworthy
here that as long as the fraction of the larger component in
the bed is less than 0.6, the minimum fluidization velocity of
the binary mixture remains unchanged. It is only when
X1 = 0.75, is there any clear difference observed. Moreover,
when the fraction of the larger component is further
increased to 0.86, the minimum fluidization velocity of
the mixture is almost the same as that of the mono-
component bed of the PET.
This dependence of the minimum fluidization velocity on
the bed composition is in fact closely related to the defluid-
ization behavior as discussed before. So long as enough
sand is available to fill all the interstices of the upper PET
layer and still available to constitute a lower pure compo-
nent layer, the behavior at the minimum fluidization is
virtually controlled by the smaller component. This situation
changes when X1 is increased. At X1 = 0.86 the amount of
the smaller component is not enough to fully occupy all the
available interstitial space of the PET layer, and are there-
fore not able to exert their influence on the bed behavior.
From the standpoint of modifying the behavior of a bed
containing the PET by the addition of another solid, it is
quite obvious from this figure that a sixfold decrease in the
minimum fluidization velocity of the larger component is
achieved by the addition of the smaller but denser compo-
nent in the present case.
Fig. 5a presents the experimental data for all the five
binaries. The overall trend is seen to be similar to what has
been seen before in Fig. 4. The Umf of the binary mixture is
significantly lower than that of the larger component. It is
nonetheless noteworthy that as the size ratio is decreased,
the steepness of the Umf profile also decreases. Now, the
minimum fluidization velocities of binary mixtures are
lower than those of the larger component even when the
fraction of the larger component in the bed is as high as 0.86
for low size-ratio binaries. At the same time even for beds
with low fractions of the larger components, the minimum
fluidization velocity is affected due to their presence. Since
using absolute values of Umf for the sake of comparison
appears little unwieldy, the following dimensionless param-
eter is defined
K ¼ Umf � Umf1
Umf2 � Umf1
� �: ð8Þ
The comparison is shown in Fig. 5b. Salient features of Fig.
5a are more pronounced here. While the effect of the size
ratio on the steepness of the K-profile is clearly evident,
increasing the buoyed-density ratio, on the other hand, tends
to mitigate the effect of the size ratio for lower X1, but
enhances the same for higher X1. Recall that the size ratios
of Binaries 4 and 5 are almost the same with the buoyed-
density ratio of the latter being almost twice that of the
former. Now, the presence of even a small amount of the
larger component can be felt on the minimum fluidization
velocity of the smaller component for Binary 5. Its behavior
at higher X1 on the other hand is quite close to Binary 3,
which has a higher size ratio.
4.3. Prediction of the minimum fluidization velocity of bi-
nary solids
As pointed out before that though there are several
correlations available in the literature to predict the mini-
mum fluidization velocity of a binary-solid mixture, three
approaches can nonetheless represent a majority of them.
The simplest of these appears to be the direct averaging of
the minimum fluidization velocities represented by Eq. (6)
with its generality enhanced by the introduction of the
parameter p. For the arithmetic averaging p is 1, whereas
p =� 1 is the harmonic averaging. Predictions of these
existing averaging rules are compared with p =� 0.5 as
shown in Fig. 6 for different binaries. It can be seen here
that the prediction of the latter is clearly superior in most
cases except for Binary 4 where its predictions are slightly
poor when compared with the harmonic averaging. A more
judicious choice of parameter p will perhaps incorporate
the size ratio and buoyed-density ratio in assigning its
value.
The predictive capability of the property averaging in
conjunction with the Ergun equation and that of Eq. (7) [25]
is also examined here in the light of the present data. Also,
presented along with in Fig. 7 are predictions of Eq. (6) with
p =� 0.5. It is obvious here that the predictions of the latter
are, in general, superior to others.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254252
5. Conclusions
Five different binaries were considered in the present
study. While Binaries 1 and 2 differed in their size distri-
bution, no apparent difference is observed in their defluid-
ization behavior. As the size ratio is decreased, the
segregation tendencies increase for binaries with a substan-
tial difference in their densities. This is seen from the
behavior of Binaries 1 to 4. While complete mixing devel-
ops for Binaries 1 and 2, Binary 3 shows partial mixing. The
lowest size ratio binary, i.e. Binary 4, on the other hand,
exhibits complete segregation of the two components in its
equilibrium defluidized structure. This picture, however,
completely changes when the buoyed-density ratio is
increased. For almost the same size ratio binaries, i.e. 4
and 5, much enhanced mixing is observed for Binary 5,
which has twice the buoyed-density ratio of Binary 4.
A key feature of the defluidization dynamics described
above is the complete absorption of the smaller component
by the matrix of the larger component for the largest size
ratio binaries considered here. This phenomenon, though
known to exist above the size ratio of 6.5 in the packing
structures of binary particle mixtures, is clearly seen for the
size ratios of 10.1 and 10.8 in the present case. The same
effect is seen to be much weaker for the size ratio of 6.0 and
almost non-existent for the size ratio as high as 4.
It is abundantly clear that the addition of the smaller
component of higher density can considerably lower the Umf
of a bed of larger particles. The bed composition, however,
needs to be carefully controlled, if the size ratio of the two
components exceeds 6.5. For such binaries, so long as the
fraction of the larger component is kept below 0.6, the
minimum fluidization velocity of the mixed bed is virtually
controlled by the Umf of the smaller component. On the
other hand, the Umf of the mixed bed is unaffected by the
presence of smaller component if its fraction is low enough
as to be completely absorbed by the matrix of the larger
component. As the size ratio of binaries of the same buoyed-
density is decreased, even a small fraction of the smaller
component is seen affecting the Umf of the mixed bed, but
the magnitude of the reduction may not be sizeable.
From the comparison of the minimum fluidization data, it
is evident that the proposed generalized averaging procedure
can better describe the present experimental data by using
the value of parameter p =� 0.5. A more judicious choice of
the value of this parameter is expected to incorporate the
property difference of the binaries.
Symbols used
di Diameter of ith particle species [mm]
GB463 Glass beads with properties described in Table 1
GB655 Glass beads with properties described in Table 1
g Gravitational acceleration [m s� 2]
p Exponent p in Eq. (6)
PET Polyethylene terephthalate resin with properties
described in Table 1
PlG Plastic with properties described in Table 1
SN257 Sand sample with properties described in Table 1
SN275 Sand sample with properties described in Table 1
Umf Minimum fluidization velocity [mm s� 1]
Uo Liquid (superficial) velocity [mm s� 1]
Ut Particle terminal velocity [mm s� 1]
X1 Fluid-free volume fraction of particle species 1 [–]
Greek symbols
e Overall bed void fraction [–]
c Buoyed density ratio=[(qs1� qf)/(qs2� qf)] [–]
K Parameter defined in Eq. (8) [–]
l Fluid viscosity [kg m � 1 s� 1]
qf Fluid density [kg m� 3]
qs Solid density [kg m� 3]
w Shape factor or sphericity [–]
Subscript
1 Larger but lighter component
2 Smaller but denser component
mf Minimum fluidization condition
Acknowledgements
Authors gratefully acknowledge the Research Center at
the College of Engineering, King Saud University, Riyadh,
Saudi Arabia for its support of the project.
References
[1] B.H. Davison, C.D. Scott, Biotechnol. Bioeng. 39 (1992) 365–368.
[2] A. Srivastava, P.K. Roychoudhary, V. Sahai, Biotechnol. Bioeng. 39
(1992) 607–613.
[3] L.A.M. van der Wielen, P.J. Diepen, A.J.J. Straathof, K.Ch.A.M.
Luyben, Ann. N. Y. Acad. Sci. 750 (1995) 482–490.
[4] L.A.M. Wielen, M.H.H. van Dam, K.Ch.A.M. Luyben, Chem. Eng.
Sci. 52 (1997) 553–565.
[5] J. Yang, A. Renken, Chem. Eng. Process. 37 (1998) 537–544.
[6] I. Narvaez, A. Orio, M.P. Aznar, J. Corella, Ind. Eng. Chem. Res. 35
(1996) 2110–2120.
[7] A. Olivares, M.P. Aznar, M.A. Caballero, J. Gil, E. Frances, J. Corella,
Ind. Eng. Chem. Res 36 (1997) 5220–5226.
[8] F. Berruti, A.G. Liden, D.S. Scott, Chem. Eng. Sci. 43 (1988) 739–
748.
[9] D.J. Cumberland, R.J. Crawford, The Packing of Particles, Elsevier,
Amsterdam, The Netherlands, 1987.
[10] A.B. Yu, N. Standish, Powder Technol. 55 (1988) 171–186.
[11] J.F. Richardson, W.N. Zaki, Trans. Inst. Chem. Eng. 32 (1954) 35–52.
[12] R. Di Felice, Int. J. Multiphase Flow 22 (1996) 527–533.
[13] S.Y. Wu, J. Baeyens, Powder Technol. 98 (1998) 139–150.
[14] S. Ergun, Chem. Eng. Prog. 48 (1952) 88–94.
[15] J.R. Grace, in: G. Hetsroni (Ed.), Fluidization in Handbook of Multi-
phase Systems, Hemisphere Publishing, NY, 1982, pp. 8.1–8.228.
[16] B.C. Lippens, J. Mulder, Powder Technol. 75 (1993) 67–78.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254 253
[17] V. Thonglimp, N. Hiquily, C. Laguerie, Powder Technol. 39 (1981)
223–239.
[18] K. Noda, S. Uchida, T. Makino, H. Kamo, Powder Technol. 46 (1986)
149–154.
[19] A. Kumar, P. Sengupta, Indian J. Technol. 12 (1974) 225–228.
[20] B. Formisani, Powder Technol. 66 (1991) 259–264.
[21] B. Formisani, G. De Cristofaro, R. Girimonte, Chem. Eng. Sci. 56
(2001) 109–119.
[22] A.R. Otero, J. Corella, An. R. Soc. Esp. Fis. Quim. 67 (1971) 1207–
1219.
[23] E.H. Obata, E.H. Watanbe, N. Endo, J. Chem. Eng. Jpn. 15 (1982)
23–28.
[24] J. Rincon, J. Guardiola, A. Romerero, G. Ramos, J. Chem. Eng. Jpn.
27 (1994) 177–181.
[25] L. Cheung, A.W. Nienow, P.N. Rowe, Chem. Eng. Sci. 29 (1973)
1301–1303.
M. Asif, A.A. Ibrahim / Powder Technology 126 (2002) 241–254254