Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
To cite this article: H. Groenewold & E. Tsotsas (1999) PREDICTING APPARENT SHERWOOD NUMBERS FOR FLUIDIZED BEDS,
Drying Technology: An International Journal, 17:7-8, 1557-1570, DOI: 10.1080/07373939908917635
Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
DRYING TECHNOLOGY, 17(7&8), 1557-1570 (1999)
ABSTRACT
Mass transfer data of bubbling fluidized beds have been reevaluated with a
new model which is completely predictive. The model bases on a two-phase ap-
proach with active bypass, formally plug flow for the suspension gas and a consid-
eration of backmixing in the main kinetic coefficient, i.e. in the apparent panicle-
to-fluid Shenvood number. A good agreement with experimental results of various
authors with a broad range of Reynolds numbers and particle diameters is demon-
strated.
INTRODUCTION
It is a well known phenomenon - see Kunii el al., 1992 -that Sherwood num-
bers for particle-to-gas mass transfer in fluidized beds do not approach a tinit value
but fail to zero for small Reynolds numbers, when a simple plug flow model is used
to evaluate the experimental results. Many attempts have been made to improve
models for the mass transfer in fluidized beds in this respect. The most important
approaches assume
I. bypassing effects (resulting from bubbles) or
2. backmixing (resulting from turbulence)
or combinations o f these. However. i t is characteristic for the main body o f previous
literature, that one or the other crucial model parameter has been used to fit mea-
sured data, without real consideration of its physical meaning. A promising way out
of this situation is the complete calculation of fluidized beds by means o f CFD. As
Downloaded by [McMaster University] at 09:13 20 November 2014
long as this method is still developing, attention should be paid to the difference
between apparent Sherwood numbers, which can fit anything, and real Sherwood
numbers: only the latter result from real fluid dynamics and real driving forces.
It is. therefore, reasonable to s t m a new model with a real Sherwood number
following well accepted physical rules:
I.The real Sherwood number tends to a finit value at low Reynolds numbers;
2. Thc real Sherwood number is independent o f bed height (hold-up);
3. The real Sherwood number is independent o f the superficial Reynolds number.
The third rule results from the similarity o f heat or mass transfer to momentum
transfer. Since pressure drop i s approximately independent o f the fluidization ve-
locity. the same should hold for heat and mass transfer coefficients.
THE N E W MODEL
What is here denoted ;ls the new model has been introduced by Groenewold et
al.. 1997, so that only few aspects have to be briefly repeated. Drying experiments
with porous ceramic particles belonging to Geldan group D provided the basis of
this model. The new model is a relatively simple two-phase approach with an active
bypas. The necessary parameters are:
I.An apparent Sherwood number Sh,,,, for the mass transfer from the particles to
the suspension gas;
2. The fraction o f the bypass flow rate v;
3. A number o f transfer units NTUi to describe mass transfer from the suspension
gas to the bypass.
FLUIDIZED BEDS 1559
Any appropriate method may be used to take the intraparticle resistances during the
second drying period into account. The three parameters are determined as follows:
The new model starts with the Sherwood number as calculated from the equa-
tions of Gnielinski. 1980, (see appendix). These equations are o f proven validity
for fluidized beds at high Reynolds numbers and fulfill all the demands mentioned
Downloaded by [McMaster University] at 09:13 20 November 2014
above. Calculations at low Reynolds numbers have shown (refer again to Groe-
newold et a]., 1997) that, in a model with perfect backmixing of the suspension
gas, the Sherwood number according to Gnielinski can be used to predict suitably
the drying rate o f the first drying period. However, the backmixing model failed
i n describing the second drying period correctly. This was only possible by means
o f a model with ideal plug flow o f the suspension gas and. as i n literature usual,
low values of the Shenvood number. The novelty o f the model is that this apparent
Sherwood number was not estimated or adapted but calculated from
This equation results from an equivalence between perfect backmixing and ideal
plug flow for a boundary condition o f the first kind, with the invariance o f effi-
ciency (of outlet concentration of the transferred species) as the equivalence crite-
rion. I t transforms Gnielinski's Sherwood number Shc,,i,, which is regarded as a
real Shenvood number following the requirements mentioned above, to the appar-
ent Shenvood number Shop,,, which is suitable for the new model. The consequence
o f this transformation is that a backmixing effect is introduced into the kinetics, not
into the driving force.
Bypass Rario v:
Hilligardt et al.. 1986. (see also Kunii et al.. 1992) have measured the local
bubble fraction i n fluidized beds. Their data for the bypass ratio is correlated to the
1560 GROENEWOLD AND TSOTSAS
reduced excess gas velocity with a group-specific relative bypass ratio v, according
to 110 - llmf
v = v r . (2)
I(,"/
The respective values of v, have been used without any change i n the present work.
Formation, rise. and growth of the bubbles have been i n the interest o f many
researchers (e.g. Davidson el el.. 1985). so that i t is known that bubbles behave i n
a different way for different particle diameters, bed heights. and bed diameters. O f
course, mass transfer from the suspension to the bubbles is influenced by lhese pa-
rameters as well. Many uncertainties arising from this behavior make the definition
of a lumped parameter NTU; reasonable, which includes both a mass transfer co-
cfticicnt and the total surface area of the bubbles. The number of transfer units has
been estimated to NTU; = 1.0 for a bed height o f 50 mm. A linear adaptation to
the actual bed height is included i n the new model. Some additional discussion on
NTU; will be given at the end o f the present paper.
In Groenewold el al.. 1997, the new model has hecn developed on the basis
of own drying experiments with a significant second period. Here, it w i l l be com-
pared with data from literature, gained either from drying experiments in the first
period, or from sublimation o f naphthalene, i.e. without intraparticle resistances.
From the vast amount o f published work many have been rejected for reasons o f
unsufficient documentation, so that, finally, six sources have been selected and are
used, namely: ( I ) Zabeschek, 1977, (2) Schwarzbach, 1989. (3) Subramanian et al.,
1977, (4) Hsiung et al., 1977. (5) Ramirez el al., 1981. and (6) Resnick et al.. 1949.
The basic variable of comparison is the efficiency q which is equivalent to the
concentration of the transferred component in the outlet gas stream. Both, exper-
imental results and model predictions were convened to q i n a first step. In many
cases, the experimentally determined efficiency was given so that no conversion was
necessary. I n the other cases, the model o f the author was used together with the pa-
FLUIDIZED BEDS 1561
rameters lilted to each single experiment to get q. The model predictions- applying
the new model with NTUi, v, and Sh,,,,, as stated before - were also expressed in
terms o f q.
Thesecond stepof comparison is theconversion o f q to the Shenvood number
o f the simple (homogeneous) plug flow model ShplUg,which is not equal lo either
She.;? or Sh,,,,. This Sh,l,, has the advantage o f higher sensitivity (the difference
between q = 0.90 and q = 0.95 is small (5.5%) but gives Shp~,, = 2.3 and Shplug=
3.0 (30%). rcspectivcly (with the fictive value o f & = I,compare to Equation
(3)).
Downloaded by [McMaster University] at 09:13 20 November 2014
This method surely produces some errors, but its results were approximately within
the range stated by Zabeschek for the bed height. Figure I compares experimental
and calculated Sherwood numbers. The agreement is very good with an average
relative error o f only 22.8 %. Using the original model of Zabeschek an error o f
' i s obtained.
28.3 %
Schwarzbach. 1989, has conducted drying experiments with porous particles
wetted with water or mixtures of water and isopropanol. and reported plug flow
Shewood numbers. After having been treated in the described way, his results com-
pare with the predictions o f the new model as shown in Figure 2.
GROENEWOLD AND TSOTSAS
Downloaded by [McMaster University] at 09:13 20 November 2014
.experimental results
o calculated with the new model
FIGURE I Experimental and calculated plug flow Sherwood numbers for dry-
ing experiments of Zabeschek, 1977. Average relative errors: 22.8 % (new model).
28.3 % (model of the author).
.
o
experimental results
calculated with the new model
FIGURE 2 Experimental and calculated plug Row Sherwood numbers for drying
experiments o f Schwarzbach, 1989. Average relative errors: 49.2 90(new model).
25.9 % (model o f the author).
FLUIDIZED BEDS 1563
Subramanian et a]., 1977. Figure 3, have used the same apparalus as Zabeschek
(see above) with slightly larger panicles. Hsiung el al.. 1977. have measured the
sublimation o f naphthalene spheres in a fluidized bed. which was diluted with inert
spheres of the same size and density. Figure 4 shows the results o f the reevalua-
tion. Ramirez et al.. 1981, have dried silica gel and derived heat and mass trans-
fer coefficients from over-all balances. I n the present reevaluation. only the mass
transfer results are used. I n Figure 5. the results of the calculations are plotted. Fi-
nally, Resnick et al.. 1949, Figure 6. have measured the sublimation of naphthalene
spheres in a fluidized bed.
Downloaded by [McMaster University] at 09:13 20 November 2014
Table 1 summarizes the ranges for which the reevaluations have been con-
ducted, while Figure 7 shows all results in one diagram. The plot is complered by
curves calculated from the equations of Gnielinski (see appendix) for a bed a1 min-
imum fluidization (E = 0.4) and for a single panicle (E = I).
1w
.experimental results
o calculated with the new model
--
..
1 10 r
z
5
Downloaded by [McMaster University] at 09:13 20 November 2014
FIGURE 3 Experimental and calculated plug flow Shewood numbers for dry-
ing experiments of Subrnmanian et al., 1977. Average relative errors: 33.8 % (new
model). 13.5 90(model oflhe author)
*.
. experimental results
o calculatedwith the new model
FIGURE 4 Experimental and calculaled plug flow Sherwood numbers for the ex-
periments of Hsiung el al., 1977. Average relative errors: 27.6 % (new model).
77.1 % (model of the author).
FLUIDIZED BEDS
0.1
0
. experimental results
calculated with new model
001
.c
"l
OW1
0
Downloaded by [McMaster University] at 09:13 20 November 2014
0
I
FIGURE 5 Experimental and calculated plug flow Sherwood numbers for the ex-
periments of Ramirez el al.. 1981. Average relative errors: 81.6 % (new model).
33.2 % (model of the author).
.experimental results
o calculated with the new model .-
.r:
FIGURE 6 Experimental and calculated plug flow Sherwood numbers for the ex-
periments of Resnick et al., 1949. Average relative errors: 59.8 % (new model).
2 1.5 % (model o f the author).
GROENEWOLD AND TSOTSAS
1 (4) .?.
, , Hsiune 11.32-2071
I I
0.25-2.0 1
I
B.D 1
I
0.15-6.5 1
I
(5) Ramirez 0.3-3.23 0.125-0.35 1 A.B 0.0009-0.073 23
(6) Resnick 0 . 6 2 - 121 1 0.27- 1.16 B 1 0.027 - 3.9 102
experimental results
0 calculated with new model
-i
Downloaded by [McMaster University] at 09:13 20 November 2014
0.m: 1
0.0 0.2 01 0.6
bed height [rn]
FIGURE 8 Fitted parameter NTU; for all experiments compared to NTU; calculated
as within the new model.
Author
(I ) Zabeschek
new model
original version
(NTU; = I)
22.8
new model 1 new model
without bypass inactive bypass
(NTU; 4 -)
23.6
(NTU, = 0)
36.3
1 author's
model
28.3
~
(2) Schwarzhach 29.2 69.7 53.7 25.9
(3) Subramanian 1 33.8 21.4 54.7 13.5
(4) Hsiung 27.6 79.4 30.5 77.1
(5) Ramirez 1 81.6 23.1 95.3 33.2
(6) Resnick 59.8 25.2 78.1 21.5
all 33.6 38.5 45.2 35.4
1568 GROENEWOLD AND TSOTSAS
are given in Table. Comparison of these values indicates a preference for the model
in its original version (with NTUi = I).
CONCLUSIONS
Thc reevaluation shows that the new model can be used for bubbling fluidized
beds over the whole range of Reynolds numbers and panicle groups. The parameter
Downloaded by [McMaster University] at 09:13 20 November 2014
NTU; could not be correlated to any propeny of the panicles or the bed. Its variation
from zero to infinity showed that the original assumption of NTUj = I at L= 50mm.
with lincar adaption to other bed heights. i s quite good. The main conclusion to be
drawn from the presented reevaluation is that the most imponant parameter of the
model. thc apparent Sherwood number Shop,, can be reliably predicted by Equation
(1).
NOTATION
m1
panicle surface area per volume fraction 7
panicle diameter mm
bed height m
number of transfer units describing mass transfer between -
bubbles and suspension
Reynolds number
Schmidt number
apparent Sherwood number, parameter of the new model -
Sherwood number after Gnielinski. 1980
plug flow Sherwood number, valid for the homogeneous -
plug flow model
gas velocity E
Greek Lzrrers
e bed porosity
Y, - Y .
11 efficiency. rl = $$f
Subscriprs
0 superficial
lam laminar
SP single particle
adiabatic saturation
Downloaded by [McMaster University] at 09:13 20 November 2014
US
mf at minimum fluidization
rurb turbulent
in at inlet
0111 at outlet
In the frame of the new model all calculations are conducted at minimum fluidiza-
tion conditions.
LITERATURE CITED
Davidson, J.F., Clift. R., and Harrison. D., 1985. Fluidization, 2nd ed., Academic
Press. London. 733 pp.
1570 GROENEWOLD AND TSOTSAS
Groenewold, H. and Tsotsas, E.. 1997, A new model for fluid bed drying, Drying
Technology, 15 pp. 1687-1698
Hillignrdt, K. and Wenher, I., 1986, Local bubble gas hold-up and expansion of
gadsolid fluidized beds, German Chemical Engineering, 9 pp. 215-221
Hsiung, T.H. and Thodos, G., 1977. Mass transfer in gas-fluidized beds: Measure-
ment of actual driving forces. Chem. Eng. Sci.. 32 pp. 581-592
Downloaded by [McMaster University] at 09:13 20 November 2014
Ramirez, I.. Ayore, M. and Vizcarra, M.. 1981, Study of the behavior of heat and
mass transfer coefficients in gas-solid fluidized bed systems at low reynolds num-
bers, ACS Symposium Series 168, pp. 185-200
Resnick, W. and White, R.R.. 1949, Mass transfer in systems of gas and fluidized
solids, Chem. Eng. Prog.. 45 pp. 377-390