Pergamon ( h,'m~, . I t i m/ m, ' . ' r m, / S: w m e. Vol 52. Nos. 21 22. pp. 3995 41~)3.

It~97
1'4q7 | qsc~.lcr , ~wn c e l . l d All r i ght s r escr ' . ed
Pr i nt ed in Gr e a t Br i t ai n
Pl h SO009-25A)9(97)OO242-X ~ ~ ~ 7 sl ~l ~ -~l~)
Mass transfer in packed beds at low Peclet
numbers wrong experiments or wrong
interpretations?
G. Rexwinkel,* A. B. M. Heesink and W. P. M. Van Swaaij
Depar t ment of Chemical Engineering, Twcnt e University of Technology, PO Box 217. 7500 AE
Enschcdc, The Netherlands
(Accepted 7 July 1997)
Abstract Much research has been focused on mass transfer phenomena in packed beds. i-'or
Peclet number s above 200, empirical relations have been derived that predict the valuc of the
mass transfer coefficient as a funct i on of the Reynolds number and the Schmidt number. These
rel at i ons are more or less similar to the well-known relation that Ranz and Marshall derived for
mass transfer ar ound a single sphere in an infinite medi um
Sh = ~ + f l Re' Sd' .
For packed beds of spherical particles an :~-value of 3. 89 can be calculated on basis of
fundament al consi derat i ons. However, Sherwood number s much lower t han this mi ni mum
value have been observed at Peclct number s below 100. Several expl anat i ons have been
proposed for this appar ent discrepancy, such as mi si nt erpret at i on of the experi ment al results
due to unjustified neglection of axial dispersion or wall channel i ng. In this work, a model that
predicts the combi ned effects of axial dispersion and wall channel i ng has been developed. With
this model, it is possible to explain the results obt ai ned with undi l ut ed beds in which all
particles are active in the process of mass transfer. However, such an expl anat i on is not possible
for the results obt ai ned with diluted beds in which not all particles arc active. Therefore, in the
case of di l ut ed beds ot her reasons for the appar ent drop in mass transfer rate must exist. In the
present investigation, it is demonst rat ed that the drop again originates from mi si nt erpret at i on
of the experi ment al results. It is shown, both experimentally and theoretically, that low
Sherwood number s can be obt ai ned when large differences cxist between the local concent ra-
tion, experienced by an active particle and the mixed cup concent r at i on of the whole bed
cross-section. (" 1997 Elsevier Science l,td
Keywor ds : Mass transfer; packed beds: low Peclet numbers; di l ut i on.
I. INTRODUCTION
The rate of mass transfer between particles and fluid
in packed-bed cont act ors in often predicted with the
help of a generalized, dimensionless correl at i on
Sh = funct i on (Re, Sc, Geometry!. (11
Several correl at i ons have been published, e. g. by
Wakao and Funazkr i (1978) and Gni el i nski (1978j.
Most of these were based on mass transfer measure-
ment s performed at high Peclet number s (Pe > 200),
and are modified versions of the well-known
Ranz-Mar s hal l equat i on, that was derived for a single
sphere in an infinite medium. Though these correla-
t i ons are accurate at high Peclet numbers, at low
*Corresponding aulhor. Tel . : 0031 534894338: filx:
0031534894774.
Peclet numbers (Pe < 100) Sherwood number s have
been measured that are less t han predicted. See Fig. 1.
in which the results of several studies are presented.
Sorcnscn and Stewart (1974) examined mass t rans-
fer in packed beds on a fundament al basis. They
demonst rat ed that, at low Reynolds number s
I Re < 10), the rate of mass transfer is onl y a funct i on
of the Peclet number. They also showed that the
mi ni mum value of the Sherwood number, which is
reached at Pe = O, amount s to 3.89 in the case of
ideally packed spherical particles. A similar result was
obt ai ned by Gun n (1978). Up to now, this result could
not bc confirmed experimentally, not even when ap-
plying s h a l l o t beds to minimize experi ment al inac-
curacies, that are inevitable due to two compl i cat i ons
that appear when operat i ng at low Peeler numbers:
(i) the number of mass transfer units becomes quite
large, maki ng it difficult to det ermi ne the dri vi ng force
for mass transfer in an accurate way, and (i i ) the
3995
3996 G. Rexwinkel e t a l .
1000 ~
100 ~ ~-
0 . . . . . . . "
0.1
0.01
0.1 1 10 100 1000 10000 100000
Williamson et al (1963)
Wilson and Geankoplis (1966)
Appel and Newman (1976)
Fedkiw and Newman (1982)
o Hsiung and Thodos (1977)
13 Hobson and Thodos (1951)
O Satterfield and Resnick (1954)
Hsiung and Thodos (1977)
Bar-Ilan and Resnick (1957)
Wilson and Geankop lis (1966)
- - Gn i e l i n s k i Sc= 1000 (liquid)
. . . . . . . Gnielinski Sc=2.5 (gas)
1000000
P e [ - ]
Fig. 1. Experimental Sherwood numbers from literature. Black symbols: undiluted bed--liquid; White
symbols; undiluted bed--gas; Gray symbols: diluted bed--gas (Bar-llan and Resnick, 1957; Hsiung and
Thodos, 1977) and diluted bed--liquid (Wilson and Geankoplis, 1966).
cont r i but i on of free convect i on to total mass transfer
may become significant. In order to overcome the first
compl i cat i on some researchers applied diluted beds
in which onl y a part of the particles is active in the
mass transfer process, e. g. Bar -l l an and Resnick
(1957), Wi l son and Geankopl i s (1966)and Hsi ung and
Thodos (1977). By doi ng so they could lower the
concent r at i on in the effluent and calculate the dri vi ng
force for mass transfer more precisely. Nevertheless,
their results are still not in accordance with expecta-
tions on basis of theory, as is illustrated by Fig. 1. The
correl at i on proposed by Gni el i nski is assumed to de-
scribe the act ual Sherwood number satisfactorily for
the whole Peclet range, since it predicts a limiting
Sherwood number of 3.8 at zero flow, which is in
agreement with the theoretical findings of Sorensen
and Stewart (1974), while at higher Peclet number s
this correl at i on has also proven to be accurate.
In this paper, we will demonst r at e t hat the discrep-
ancy between measured and predicted Sherwood
numbers, which is observed at low Peclet numbers, is
caused by wrong cal cul at i on of the (local) dri vi ng
force for mass transfer. We will do so bot h for un-
diluted beds, in which all particles are active, and for
diluted beds, in which only part of the particles in
active. However, emphasis will be on diluted beds. It
will be shown that such beds need a completely differ-
ent approach t han undi l ut ed beds, and that the results
obt ai ned with diluted beds are of no value to those
interested in mass transfer phenomena in undi l ut ed
beds, at low Peclet numbers.
2. RE-EVALUATION OF EXPERIMENTAL DATA
Most Sherwood numbers, that were reported in
literature, where calculated while assumi ng plug flow
behavi or of the fluid flowing t hrough the bed, i.e.
while neglecting the effects of axial dispersion and
wall channel i ng. This may not always be justified,
especially not at small Peclet numbers. For example,
in the case of shallow beds, axial dispersion may have
a significant negative effect on the average dri vi ng
force for mass transfer, causi ng an appar ent drop in
Sherwood number as was illustrated by Wakao and
Funazkr i (1978). In the case of small di amet er beds,
wall channel i ng may result in overest i mat i on of the
average dri vi ng force in the bed, also resulting in an
appar ent drop in Sherwood number (Mart i n, 1978).
As most reported dat a were obt ai ned from
measurement s t hat were carried out with shallow beds
with a small diameter, it makes sense to re-evaluate
these dat a while correcting for the influences of axial
dispersion and wall channel i ng. We did so with the
help of the model developed by Mar t i n (1978), which
was expanded in the following ways:
Mass transfer is assumed also to occur in the wall
zone. The mass transfer coefficients in the wall
zone and in the center part of the bed are as-
sumed to be equal.
The fluid concent r at i ons in the effluents from
bot h zones are calculated on the basis of the
axially dispersed plug-flow model, in a similar
way as was done by Wakao and Funazkr i (1978).
The gas dispersion coefficient is calculated with
a correl at i on proposed by Gun n (1987), whereas
the liquid dispersion coefficient was calculated
using a correl at i on of Chung and Wen (1968).
These correl at i ons have been proven to be accu-
rate also at low Peclet numbers.
The results of the re-eval uat i on are shown in Fig. 2. It
appears that, in the case of undi l ut ed beds, the re-
eval uat ed Sherwood number s reasonabl y agree with
Mass transfer in packed beds 3997
1000
100
10
0.1
0.01
. . - ' ~ *" "
i, d J
[ ]
Ga"- .11'
. ' t
O
0 . I 1 10 100 I(XX) 1 0000 100( O) 1000000
P e [ - I
Williarnson et al (1%3)
Wilson and Geankoplis (1966)
Appel and Newman (1976)
Fedkiw and Newman (1982)
Hsiung and Thodos (1977)
Q Hobson and Thodos (1951)
o Satterfield and Resnick (1954)
# Hsiung and Thodos (1977)
o Bar-llan and Resnick (1957)
g Wilson and Geankoplis (1966)
Gnielin ski Sc=1000 (liquid)
. . . . . . . Gnielinski Sc=2.5 (gas)
Fig. 2. Re-evaluated Sherwood numbers. Black symbols: undiluted bed--l i qui d: White symbols: undiluted
bed--gas, Gray symbols: diluted bed gas iBar-Ilan and Resnick, 1957; Hsiung and Thodos, 1977) and
diluted bed liquid (Wilson and Geankoplis. 19661.
t heory. Appa r e nt l y, t he dr op i n Sher wood numbe r .
t hat is obs er ved i n un d i l ut e d beds at decr eas i ng Pecl et
n umbe r , is caused by wr o n g i nt e r pr e t a t i on of t he
exper i ment al resul t s, i.e. by unjus t i fi ed negl ect i on of
axi al di s per s i on a n d / o r wal l channel i ng.
Re -e va l ua t i on har dl y has an effect on t he da t a ob-
t ai ned wi t h di l ut ed beds. We t her efor e concl ude t hat ,
i n t he case of di l ut ed beds, ot he r p h e n o me n a t h a n
axi al di s per s i on or wal l c h a n n e l i n g cause t he dr op i n
Sher wood n umbe r whi ch is obs er ved at decr eas i ng
Pecl et numbe r . Below, we will t ry to i dent i fy t hese
p h e n o me n a . We st ar t wi t h a s t udy of a si ngl e act i ve
par t i cl e s ur r o un d e d by i nact i ve par t i cl es onl y. The n,
we will e xa mi ne beds c o n t a i n i n g mul t i pl e act i ve par -
ticles.
3. DI Ll ? TED BEDS: A S I N G LE ACTI VE P ARTI CI , E
3.1. Them'etical
Ra nz a nd Mar s hal l (1952) e xa mi ne d mass t r ansfer
a r o un d a si ngl e spher e l ocat ed in a n i nfi ni t e me di um,
in whi ch no ot her par t i cl es ar e pr esent . The y showed
i n a si mpl e way t hat t he mi n i mum val ue of t he Sher-
wood n umbe r (at Pe = 0) a mo un t s to 2. I n t he case
t hat t he spher e is l ocat ed i n a n i nfi ni t e bed of i nact i ve
par t i cl es t he val ue of t he mi n i mum Sher wood n umbe r
is by defi ni t i on o bt a i n e d by mul t i pl yi ng t hi s val ue of
2 wi t h t he por os i t y ~: of t he bed a nd by di vi di ng it by
t he t or t uos i t y r of t he bed:
S h mi n = - - 2 . ( 2 )
T
The t or t uos i t y of a r a n d o ml y packed bed c a n be
cal cul at ed accor di ng to ( Pun c o c h a r a nd Dr ahos ,
1 9 9 3 ) :
1
r - (3)
yi el di ng, Sh,,~, = 2t: 1 s. Thi s is i n agr eement wi t h t he
resul t s obt a i ne d by Mi yauchi (1971). The mi ni ma l
val ue t hus a mo un t s to a bo ut 0.5 whi ch is l ower t ha n
t he mi ni ma l val ue pr edi ct ed for undi l ut e d beds, i n
whi ch t he c o n c e n t r a t i o n profi l e a r o un d a par t i cl e does
not rcach be yond t he ne i ghbor i ng part i cl es.
3.2. Experimental
Sol i d-gas mass t r ansfer coeffi ci ent s of a si ngl e ac-
tive sphere i n a n i nact i ve bed have been de t e r mi ne d
by me a s ur i ng t he e va p o r a t i o n rat e of a c a mp h o r
spher e wi t h a pur i t y 96% a n d a di a me t e r of 1 cm in
a ni t r oge n gas st r eam. A c a mp h o r spher e was wei ghed
before t he exper i ment a nd s ubs e que nt l y pl aced i nsi de
t he cent er of a r egul ar l y packed bed of I cm di a me t e r
glass spheres, havi ng a di a me t e r of 10. 5cm a nd
a hei ght of 12c m [ Fi g. 3(a)] . Before e nt e r i ng t he
packed bed, t he ni t r oge n gas was led t hr ough a di s-
t r i but o r wi t h a hei ght of 6 cm, cons i s t i ng of glass
beads of 1 mm, in or der to ens ur e a n evenl y di st r i b-
ut ed gas flow. After a cer t ai n t i me of oper at i on, d ur i n g
whi ch t he gas flow was di r ect ed ei t her upfl ow or
downfl ow t hr ough t he bed, t he c a mp h o r spher e was
wei ghed agai n. Fr o m t he change i n mass t he mass
t r ansfer coefficient was cal cul at ed:
k = m o - m . . . . ( 4 )
t ~ . ~ x d 2 ( C , - C b ) M "
I n a first exper i ment a c a mp h o r spher e was wei ghed,
pl aced i nsi de t he packed bed a nd aft er a while, not
a ppl yi ng a ni t r oge n gas st r eam, t a ke n out of t he bed
3998 G. Rexwinkel e t al.
Sin~l-~--
active
particle
Vent ]
--L
12 cm
Distributor
10.5 cm
~ UV-meter J
Inert layer of
glass beads
Inert layer of
glass beads
5 c m
i 1
i 125 cmCm
).
10.5 cm
a) b)
Fig. 3. Schematic of the experimental setup for two different types of experimens: (a) Single active sphere in
an inactive packed bed of glass spheres; (b) Multiple active spheres in a bed of inactive glass spheres which
is located between two layers of inactive glass spheres.
Table 1. Physical properties (at 25~C) together with design data and operating conditions
applied during the experiments
Single a c t i ve Multiple active
particle particles
Transferred species Camphor Methylbenzoate
Fluid Nitrogen gas Water
Temperature 20-25 'C 20- 25:C
Molecular mass of transferred species 0.15224 kg/mol 0. 13615 kg/mol
Diffusion coefficient of transferred species 6.2 10 -6 m/ s * 9.3 10- lo m2/s ,
Solubility of transferred species -- 2.15 g/l:
Vapor pressure of transferred species 30.2 Pa ~ ---
Particle diameter 10 mm 0.63-0.71 mm
Bed porosity 0.40 + 0.01 0.40 + 0.01
Percentage active particles -- 0.35%
Sc 2.54 1099
Re 0.11-141 0.001-- I
*Yaws (1995); *Wilke and Chang (1955); tGetzen e t al . (1992); Presser (1972)
and weighed again. It appeared t hat the reduct i on of
mass dur i ng this experiment could be neglected com-
pared to the reduct i on observed after a nor mal experi-
ment of several hours. Dur i ng all experiments thc
decrease in particle di amet er was negligible whereas
the increase of the camphor concent r at i on in the ni-
trogen stream was small. Therefore, the bul k concen-
t rat i on, Cb, could be considered equal to zero. The
sat ur at i on concent r at i on, Cs, was det ermi ned from
the ideal gas law and the camphor vapor pressure,
Pr e p , t hat was calculated for the applied t emperat ure,
which varied between 20 and 25 C but remained
const ant dur i ng a single experiment. Wi t h these dat a
and the dat a from Tabl e 1 available, the Sherwood
number was calculated accordi ng to
S h - R T ( m o - mr , , , ) (5)
t c x p n d p P v a p M D "
In Fig. 4, the experimentally obt ai ned Sherwood
number s are plotted versus the Peclet number. The
Mass transl~cr
solid line in this figure represents the correl at i on pro-
posed by Ranz and Marshall (1952). The upflow and
downflow experiments gave equal results i ndi cat i ng
that the influence of free convect i on on the mass
transfer rate can be neglected. Below a Peeler number
of 50 the presence of inert particles becomes notice-
able causi ng the measured Sherwood number of fall
below the value predicted by the Ra nz Marshall
equat i on. At very low Peclet number s the Sherwood
number indeed approaches the theoretical mi ni mum
of approxi mat el y 0.5, which is lower t han the mi ni mal
value of 3.89 in undi l ut ed beds. We may thus conclude
that diluted beds cannot be used to exami ne mass
transfer phenomena in undi l ut ed beds at low Peeler
numbers. We also concl ude that the drop in Sher-
wood number to wdues below the limit of 0.5, as
observed with diluted beds. e.g. Bar-i l an and Rcsnick
(1953), must bc caused by i nt eract i on of the active
particles present in the diluted bed. Wc will examine
this possible i nt eract i on in the next section.
4. DI LUTED BEt ) S: M I . I . TI PI , E ACTI VE P ARTI CI . ES
4.1. Theoretical
As has been stated earlier, it is i mpor t ant to know
the local dri vi ng force for mass transfer in order to be
able to calculate the local mass transfer coefficient. In
1(~)
I 0
-r , ~ f . -
D
0.1
1
Single particle u p f l o v .
O Smgte particle downtlow
I~ Bar-Ilan and R~ni ck (19571
Ranz-Mar~hall e q u a t k m .~x=-2.5 (~L~,)
0 . 0 1 L , , ,
O. I I I 0 I ~ 10 0 0 1(X](30
Pe [ - I
Fig. 4. Experimental Sherwood numbers for a single active
particle together with the results of multiple acti~,e particles
of Bar-ilan and Resnick 11957).
in packed beds 3999
the case of a single active sphere in an infinite bed of
inert particles, the dri vi ng force for mass transfer
experienced by the active particle in exactly known
and trivial. Det er mi nat i on of the dri vi ng force experi-
enced by an i ndi vi dual active particle in a diluted
packed bed. in which more active particles are pres-
ent. is less trivial.
In Fig. 5, several hypothetical di st r i but i ons of ac-
tive particles in a t wo-di mensi onal packed bed are
shown. The dri vi ng force experienced by the active
particles will be much different in each si t uat i on. In
si t uat i on 1 all active particles experience the same
dri vi ng force which has not been influenced by the
presence of any active particles upstream. In si t uat i on
It all active particles have been placed directly on top
of each other. The influence of particles upstream on
the driving force experienced by the particles dov<n-
stream will bc significant. The average mass transfer
rate in si t uat i on II will therefore be lower t han the
average mass transfer rate in si t uat i on 1. In si t uat i on
I11 the active particles have also been placed on top of
each other, but now inactive particles arc placed in
between. In this si t uat i on the average mass transfer
r a t e will also bc lower t han in si t uat i on !, but higher
t han in si t uat i on il due to the extra radial dispersion
of the transferred species. Although the particle distri-
but i ons shown in l-ig. 5 are rather exceptional, thev
do indicate that the average rate of mass transfer will
bc overestimated if it is assumed that no radial con-
ccnt rat i on gradi ent s occur.
4.2. Experimental
l ' he spread in experimental results obt ai ned with
diluted beds at low Peclet number s is considerable.
Hsiung and Thodos (1977) in their diluted packed bed
observed only slightly lower Shewood number s t han
the values found for a single sphere, whereas Bar-l l an
and Resnick (1957) have measured much lower Sher-
wood numbers. Since the cause for this spread is
unknown, wc also carried out some experiments in
diluted packed beds. Mass transfer experiments wcrc
conduct ed by measuri ng the di ssol ut i on rate of
mct hyl bcnzoat c in water. Met hyl benzoat e was im-
pregnated in spherical porous Amberlite XAD-2
particles with an average di amet er 0.66 mm. In all
experiments tl.Sg of impregnated particles was
I II I I I
C) I n a c t i ve par t i cl es 0 Ac t i ve par t i cl es
D i r e c t i o n
o f f l ow
I"ig. 5. Several hypothetical distributions of active particles in a t',vo-dimensional packed bed.
4000
t horoughl y mixed with spherical glass particles of the
same size. The mixture, of which 0.35% of all particles
was active, was brought i nt o the bed resulting in an
active layer of 2 cm high. This layer was enclosed by
two 5 cm thick layers of inert glass beads. See Fig.
3(b). The met hyl benzoat e concent r at i on at the exit of
the packed bed was cont i nuousl y measured using an
UV-spect rophot omet er. The met hyl benzoat e concen-
t rat i on in the effluent remai ned const ant dur i ng the
first hour of operat i on, i ndi cat i ng that pore l i mi t at i on
in the porous Amberlite XAD-2 particles could be
neglected. The Sherwood number s were calculated by
assumi ng plug-flow behavi or
S h - d 2 ~ v p P In ( C s - C o u t )
6 D i n , \ C s - ~ " (6)
The results are shown in Fig. 6 and are similar to
those of Bar-Il an and Resnick (1957): Much smaller
values are obt ai ned t han with a single active particle.
In order to confirm that these smaller Sherwood
number s are found because of i nt eract i on between the
different active particles in the bed, extra experiments
were performed. These were done at a fixed Peclet
number of 2.89 using 0.5 g of i mpregnat ed Amberlite
XAD-2 particles. The number of mass transfer units
was thus kept const ant . The height of the bed, includ-
ing the inert layers at the bot t om and the top, always
was 12 cm. In order to simulate si t uat i on I of Fig. 5 all
active particles were posi t i oned in a hori zont al layer
which was enlosed by two layers of inert glass beds
havi ng the same di amet er of 0.66 mm as the active
particles. In this way about 20% of the bed cross-
section was covered with active particles. Si t uat i on III
of Fig. 5 was also simulated. First all active particles
were mixed with equally sized inert glass beads. The
obt ai ned mixture was then dumped i nt o a t ube with
a di amet er of 1.35 cm and a height of 8 cm, which was
placed in the center of the bed, which was already
filled with a 2 cm thick layer ofi ner t glass beads. Next,
the a nnul a r space in between the t ube and the bed
wall was filled with inert glass beads and the t ube was
slowly removed. Finally, a 2 cm thick layer of inert
G. Rexwinkel e t al.
glass beads was added to complete the bed. Addi-
t i onal to these experiments, tests were performed us-
ing diluted layers of varyi ng thickness [ accordi ng to
the set-up shown in Fig. 3(b)]. By appl yi ng a const ant
amount of i mpregnat ed Amberlite XAD-2 particles of
0.5 g, the degree of di l ut i on was varied. The results of
the i nt eract i on experiments were eval uat ed on basis of
eq. (61. The calculated Sherwood number s are shown
in Fig. 7.
If the assumpt i on of a radially well-mixed fluid
would hold in all cases, all experiments should have
given the same result, since equal amount s of active
particles have been applied. However, the results
show a significant influence of the particle di st ri bu-
tion, which is in conflict with this assumpt i on. The
calculated Sherwood number s appear to increase
when the fraction of active particles is decreased,
especially at high degrees of di l ut i on. This is in ac-
cordance with expectations, since the average axial
distance between the active particles strongly depends
on the degree of di l ut i on at high di l ut i on degrees,
whereas this dependence is relatively small at lower
degrees of di l ut i on. So, the extent of radial mi xi ng will
increase with the degree of di l ut i on, in part i cul ar at
high degrees of di l ut i on. The results obt ai ned with the
particle di st r i but i ons resembling si t uat i ons I and III
of Fig. 5 also confirm that i nt eract i on between active
particles is an i mpor t ant item in diluted beds. This
i nt eract i on, which results from radial concent r at i on
profiles, is not account ed for when assumi ng plug-
flow behavior. We will illustrate this with the model
discussed below.
4.3. T h e m o d e l
The i mpor t ance of radial concent r at i on profiles
and the resul t i ng i nt eract i on between different active
particles in the bed can also be demonst rat ed in a the-
oretical way. To do so, the diluted bed is modelled by
a t wo-di mensi onal network of i nt erconnect ed con-
t i nuousl y stirred t ank reactors (CSTR); see Fig. 8.
Each CSTR cont ai ns exactly one particle which is
100
10
0.1
0. 01
0.1
, Y
Smgte part i cl e. Thi s wor k
I I I 0 Mul t i pl e part i cl es. Thi s wor k
Bat -Il an and Resni ck ( 1957)
13 t i s i ung and Thodos ( 1977)
Ranz-Mar~,hall equat i on Sc.=2.5 (gas)
I I I
I 10 100 1000 10000
Pe [-]
0. 6'
0.5
0.4
i
, ~ 0. 3
0. 2
0.I
Si ngl e mono l a ye r r e s e mbl i ng
s i t ua t i on I o f Fi g. 5.
o
Tube n ' ~ ur e m e n t r e s e mbl i ng
s i t ua t i on I n o f Fi g. 5.
. . . . . . . . . . . . . : : ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
o d~luted bed meas ur ement s
0. 01 0. 02 0. 03
fraction of partciles active [-]
0. 04
Fig. 6. Experimental Sherwood numbers for single particle Fig. 7. Experimental Sherwood numbers for several distri-
and multiple particles in a diluted bed. butions of active particle over the bed. Pe = 2.89; Sc = 1099.
Mass transfer
Fig. 8. Schematic representation of the interconnected
CSTRs model where the grey dots represent active particles;
Flowdirection is downward, n x = number of particles in
horizontal direction and ny = number of particles in vertical
direction.
ei t her act i ve or inactive. Whet her a par t i cl e is act i ve
or not is r a ndoml y det er mi ned. A s omewhat si mi l ar
a ppr oa c h was appl i ed by van den Bleek (1969), who
i nvest i gat ed t he effect of di l ut i on on t he degree of
conver si on in fi xed-bed cat al yt i c react ors.
Mass t r ansfer bet ween CSTRs t akes pl ace in axi al
di r ect i on t hr ough convect i on and axi al di sper si on
and in l at er al di r ect i on by di spersi on. The a mount of
mat er i al t ransferred by di sper si on bet ween t wo neigh-
bor i ng CSTRs is cal cul at ed accor di ng t o
N~.j.~.i~ t = - D~ Ci.i+ t - C,.j (7a)
d ,
C i - l , j - C i . j
N c i ~ i - 1.~ = - Dl~, (7b/
t i p
A s t at i onar y di mensi onl ess mass bal ance for CSTR i.
j yi el ds
2 2 C + 1
0= ( - I B o , ~ , B - ~ , , x ) ~ ' J + ( 1 B-~a,)C~.j
1 1
+ zo~- c ' ' i ~ ~ + __Bol~-- C~_ ~.j
1 . . r t S h
+ __Bo,~--Ci + t , ~ + n,~ ' j ~: Pe (ACi4)' (8)
Wi t h C~.j defi ned as C ~ . j / C . , , C~ bei ng t he concent r a-
t i on at t he surface of an act i ve par t i cl e which is ei t her
di ssol vi ng or evapor at i ng. In this equat i on n~, 4 can
a da pt t he val ues 0 or 1 and i ndi cat es whet her or not
an act i ve par t i cl e is present in CSTR i , j . The CSTRs
in packed beds 4001
t hat are l ocat ed agai nst t he packed-bed wall are onl y
connect ed t o CSTRs in t he i nt er i or of t he bed.
An i mpor t a nt t er m is AC~,~. which represent s t he
dr i vi ng force experi enced by an act i ve par t i cl e in
CSTR i, j. In t he case of undi l ut ed beds t hi s dr i vi ng
force shoul d by defi ni t i on be based on t he mi xed cup
concent r at i on of t he whol e bed cross-sect i on. How-
ever, it is not to be expect ed t hat t he same defi ni t i on
hol ds for di l ut ed beds because of r adi al concent r at i on
gradi ent s. As we are i nt erest ed in demons t r at i ng t he
effects of negl ect i ng these r adi al concent r at i on gr adi -
ents, in the present model t wo defi ni t i ons of dr i vi ng
force are appl i ed: (ij mi xed cup dr i vi ng force, which is
cal cul at ed from t he mi xed cup concent r at i on in
a hor i zont al row, and liil CSTR dr i vi ng force, which is
cal cul at ed on basis of t he concent r at i on wi t hi n t he
CSTR in which t he act i ve par t i cl e is l ocat ed. In mat h-
emat i cal form
AC, j - - I - ~ "'" (9a)
t I t 2. ' (
ACcj --- 1 - C~.j. (9b)
Equat i on (8) has been sol ved using bot h expr essi ons
for t he dr i vi ng force by the successive over r el axat i on
met hod. As it makes no sense t o appl y expressi ons
t hat were der i ved for t hr ee-di mensi onal beds and t hat
predi ct a final val ue of Sher wood at P e = 0. t he Sher-
wood number was cal cul at ed somewhat ar bi t r ar i l y
accor di ng to
S h = 0.3 P e 1 3 . (10)
The val ues of Bot,, and Bo~x were cal cul at ed from
cor r el at i ons r epor t ed by Gunn (1987). Fi gur e 9 shows
t he cal cul at ed l at er al concent r at i on profi l es at t he exit
of a packed bed. cont ai ni ng 200 x 40 par t i cl es with
a di amet er of 0.5 mm of which 300 ar e active, using
t he t wo different defi ni t i ons of dr i vi ng force.
Fi gur e 9 cl earl y shows t hat l arge l at er al concent r a-
t i on gr adi ent s may exist in di l ut ed packed beds. Fur -
t hermore, it shows t hat negl ect i ng these l at er al
concent r at i on gradi ent s, which is done when cal cul at -
i ng t he local dr i vi ng force for mass t ransfer accor di ng
to eq. (9al, can result in l ocal di mensi onl ess concent r a-
t i ons which are l ar ger t han unity, which by defi ni t i on
is i mpossi bl e. The l ocal dr i vi ng force for mass t ransfer
in a di l ut ed bed is t hus over est i mat ed when t he r adi al
concent r at i on pr ot i l e in t he bed is neglected, as is
done when t he pl ug-fl ow model is appl i ed to eval uat e
t he exper i ment al results. Thi s is al so i l l ust r at ed by
Fig. 10 in which t hc pr cdi ct ed Sher wood number is
shown as a funct i on of t he Peclet number for bot h
defi ni t i ons of t he local dr i vi ng force ACi4. The pre-
sented Sher wood number s were cal cul at ed while
assumi ng plug-flow behavi or of t he fluid, which is
common pr act i ce when eval uat i ng mass t ransfer
exper i ment s
I 1 X c; ~' ~i 1 i.n~,
S h - P e l n I . I11)
t / a t : i t , c 7C ~ . ~ /
o r
4O02
1.6
1.4
!.2
1
0 . 6
0 . 4
0.2
0
G. Rexwinkel et al.
I Ma Ji mu m corv~mlrlUon I~SStbl
& ~ r u ~ o ~ l l l l ~ o n BI [ he (
~it o f t l ~ I ~ d
2 0 40 60 80 10 0
: : t , _ . . . . . . .o._ ,.,e
0.2
A l t l ' l g e Ol l t ' l l l l l l i oil l [
I ~ t ot a,. I ,~
0
2 0 40 60 80 10 0
lateral position [ram] lateral position [ram]
Fig. 9. Cal cul at ed radial concent r at i on profiles at the exit of t he packed bed: (a) Mixed up dri vi ng force:
(b) CSTR dri vi ng force. Pe = 10: Sh = 1.5; Boil, = 10; Bo~, = 4: d r = 0.5 mm; nx = 200: n y= 40:
,z~,,,, = 300: ~: = 0.48 = 1 - n:6.
10
s
,ll
0.1
O M ixed cup driving force
CSTR driving force
10 100 1000
Pe 1-1
Fig. 10. Sherwood numbers calculated using eq. (11) from
the mixed cup concent r at i on at the exit of the bed as cal-
culated by the i nt erconnect ed CSTR model using t he two
different definitions of the dri vi ng force. S o = 1000;
dp = 0.5 mm; nx = 2(X): ny = 40; n~,,,, = 300: ~: = 0.48.
5. CONCLUSIONS
Th e d e vi a t i o n be t we e n p r e d i c t e d a n d r e p o r t e d
S h e r wo o d n umbe r s i n p a c ke d be d s a t l ow Pe c l e t n um-
be r s is c a us e d by t h e wr o n g i n t e r p r e t a t i o n of t he
e xp e r i me n t a l da t a . I n t he c a s e of un d i l ut e d be ds c or -
r e c t i ons s h o ul d be ma d e for a xi a l d i s p e r s i o n a n d wal l
c h a n n e l i n g . Th e S h e r wo o d n umbe r s wh e n o bt a i n e d
r e a s o n a bl y a gr e e wi t h t he or y. Suc h c o r r e c t i o n s d o n o t
suffi ce i n t h e c a s e of d i l ut e d beds. Th e mi n i ma l She r -
wo o d n umbe r i n d i l ut e d be d s is l owe r t h a n i n un-
di l ut e d beds. Fur t h e r mo r e . i n d i l ut e d be ds t he
e xi s t e nc e of r a di a l c o n c e n t r a t i o n pr ofi l es s h o ul d be
c o n s i d e r e d , wh i c h is n o t pos s i bl e wh e n t h e d i s t r i bu-
t i o n of t he a c t i ve pa r t i c l e s is r a n d o m a n d t h e r e fo r e n o t
e xa c t l y kn o wn . Ne g l e c t i o n of t he s e r a d i a l c o n c e n t r a -
t i o n pr ofi l es by a p p l yi n g of t he pl ug-fl ow mo d e l re-
s ui t s i n un d e r e s t i ma t e d S h e r wo o d n umbe r s .
Th e r e fo r e , d i l ut e d be ds s h o ul d n o t be us e d t o e xa mi n e
ma s s t r a n s fe r p h e n o me n a i n un d i l ut e d be ds at l ow
Pecl et n umbe r s .
Th e va l ue s o bt a i n e d wh e n c a l c ul a t i n g t h e d r i vi n g
for ce o n ba s i s of t he mi xe d c up c o n c e n t r a t i o n a r e
a l mo s t e q ua l t o t h o s e p r e d i c t e d by eq. (10). ( On l y t h e
va l ue at P e = 1 is s o me wh a t less d ue t o axi al di s pe r -
s i on whi c h is i n c l ud e d i n t h e model . ) Th e va l ue s o b-
t a i n e d wh e n c a l c ul a t i n g t he d r i vi n g for ce o n bas i s of C
t h e CS TR- c o n c e n t r a t i o n a r e muc h l ower . We, t he r e -
fore, c o n c l ud e t h a t t he d e vi a t i o n be t we e n me a s ur e d C
a n d p r e d i c t e d S h e r wo o d n umbe r , wh i c h is o bs e r ve d i n
d i l ut e d be ds a t l ow Pecl et n umbe r s , is, at l eas t pa r t l y, AC~.j
c a us e d by t h e ne gl e c t i on of r a di a l c o n c e n t r a t i o n pr o-
files i ns i de t he bed. d r
Due t o i t s s i mpl i ci t y, t he p r e s e n t mo d e l is n o t s ui t e d D
t o q ua n t i fy t he effect s of r a di a l c o n c e n t r a t i o n pr ofi l es
o n t he a ve r a g e r a t e of ma s s t r a n s fe r i ns i de a di l ut e d D~x
bed. Fo r t hi s. CFD s i mul a t i o n s ar e neces s ar y. Dlat
Ac k n o wl e d g e me n t s
This investigation was support ed by the Dut ch Ministry of
Economi c Affairs. The aut hor s acknowledge J. G6r t en and
J. M. Meerdink for t hei r assistance in the experi ment al
wo r k
NOTATION
c o n c e n t r a t i o n of t r a n s fe r r e d speci es i n t h e
fl ui d, mo l / m 3
d i me n s i o n l e s s c o n c e n t r a t i o n C/ C~, d i me n -
s i onl e s s
d r i vi n g for ce e xp e r i e n c e d by a n a c t i ve p a r -
t i cl e i n CS TR i , . j , d i me n s i o n l e s s
pa r t i c l e d i a me t e r , m
di ffus i on coeffi ci ent o f t r a n s fe r r e d speci es,
m 2 / s
axi al d i s p e r s i o n coeffi ci ent , mZ/ s
l a t e r a l d i s p e r s i o n coeffi ci ent , mZ/ s
k
m
M
n~ .J
Ilactivc
n x
ny
N
Pvap
R
tcxo
T
r
mass t ransfer coefficient, m/ s
mass, kg
mol ecul ar mass, kg/ mol
r a ndom fact or (0 or 1) ofeq. 8, di mensi onl ess
number of act i ve part i cl es, di mensi onl ess
number of par t i cl es in hor i zont al di rect i on,
di mensi onl ess
number of part i cl es in vert i cal di rect i on, di -
mensi onl ess
flux, mol / m2s
vapor pr essur e of camphor , Pa
gas cons t ant , J/ mol K
dur a t i on of exper i ment , s
t emper at ur e, K
i nt er st i t i al fluid velocity, m/ s
Greek letter.s
~: bed por osi t y, di mensi onl ess
r/ viscosity, Pa s
v fluid flOW, m3."s
p densi t y, kg/ m 3
r t or t uosi t y, di mensi onl ess
Subscripts
min mi ni mal
0 i ni t i al l y
h in t he bul k
f with respect t o fluid
i r adi al pos i t i on of CSTR
in at t he inlet
j axi al pos i t i on of CSTR
out at t he out l et
p with respect to part i cl e(s)
s s at ur at i on
Dimensionless numbers
Boax
BOla!
Sh
Pe
Re
Sc
axi al Bodenst ei n number ( = vdp/D~d
l at er al Bodenst ei n number ( = vdp/D~,,)
Sher wood number ( = kdp/D)
i nt er st i t i al Peclet number ( = vdp/D)
Reynol ds number ( = plvdp/qy)
Schmi dt number ( = q//pyD)
REFERENC ES
Appel , P. W. and Newman, J. (1976) Appl i cat i on of
t he l i mi t i ng cur r ent met hod t o mass t ransfer in
packed beds at very low Reynol ds numbers.
A.I.Ch.E.J. 22, 979-984.
Bar-Il an, M. and Resnick, W. (1957) Ga s phase mass
t ransfer in fixed beds at l ow Reynol ds numbers. Ind.
Engng Chem. 49, 313-320.
Bleek van den, C . M., Wi el e van der, K. and Berg van
den, P. J. (1969) The effect of di l ut i on on t he degree
of conver si on in fixed bed cat al yt i c react ors. Chem.
Engnq Sci. 24, 681-694.
Mass transfer in packed beds 4003
Chung, S. F. and Wen, C. Y. (1968) Longi t udi nal
di sper si on of l i qui d flowing t hr ough fixed and
fluidized beds. A.I.Ch.E.J. 14, 857-866.
Fedki w, P. S. and Newman, J. (1982) Mass-t r ansfer
coefficients in packed beds at very l ow Reynol ds
number. Int. d. Heat Mass Transfer 25, 935-943.
Get zen, F., Hefter, G. and Maczynski , A. (1992) Solu-
bility Data Series, Vol. 48. Per gamon Press, Oxford.
Gni el i nski , V. (1978) Gl ei chungen zur Berechnung des
W~irme- und St offaust auches in dur chs t r 6mt en
r uhenden Kugelschi. ittungen bei mi t t l er en und
grossen Pecl et zahl en. VT-verfahrenstechnik 12,
363--366.
Gunn, D. J. (1978) Tr ansfer of heat or mass t o par -
ticles in fixed and fluidised beds. Int..I. Heat Mass
TransJer 21, 467 476.
Gunn, D. J. (1987) Axi al and r adi al di sper si on in fixed
beds. Chem. Engn,q Sci. 42, 363-373.
Hobson, M. and Yhodos, G. (1951) Lami nar flow of
gases t hr ough gr anul ar beds. Chem. Enqng Pro.q. 47,
370 375.
Hsi ung, T. H. and Thodos, G. (1977) Mass-t r ansfer
fact ors from act ual dr i vi ng forces for t he flow of
gases t hr ough packed beds (0.1 < Re < 100). Int. J.
Heat Mass TransJer 20, 331-340.
Mar t i n, H. (1978) Low Peeler number par t i cl e-t o-fl ui d
heat and mass t ransfer in packed beds. Chem.
Enyn 9 Sci. 33, 913-919.
Mi yauchi , T. (1971} Fi l m coefficients of mass t ransfer
of di l ut e spher e-packed beds in l ow flow rat e re-
gime. J. Chem. Engng Japan 4, 238 245.
Presser, K. H. (1972) Di e St ofi i ber t r agung fest-gasf6r-
mi g bei l ami nar er freier Knovect i on. Wiirnw- und
Sto~iibertragunq 5, 220--228.
Puncochar , M. and Dr ahos, J. (1993) The t or t uos i t y
concept in fixed and fluidized bed. Chem. Engng Sci.
48, 2173 2175.
Ranz, W. E. and Mar shal l , W. R. Jr (1952) Evapor -
at i on from dr ops. Par t I. Chem. Engng Prog. 48.
141-146.
Sat t erfi el d, C. N. and Resnick, H. (1954) Si mul t aneous
heat and mass t ransfer in a di ffusi on-cont r ol l ed
chemi cal react i on. Chem. Engng Prog. 5 0 , 504 510.
Sorensen, J. P. and St ewart , M. E. (1974) Comput a -
tion of forced convect i on in slow flow t hr ough
duct s and packed be d s - - I l l . Heat and mass t r ans-
fer in a si mpl e cubi c ar r ay of spheres. Chem. En#ng
Sci. 29, 827 832.
Wakao, N. and Funazkr i , T. 11978) Effect of fluid
di sper si on coefficients on par t i cl e-t o-fl ui d mass
t ransfer coefficients in packed beds. Chem. Engn.q
Sci. 33, 1375-1384.
Wilke. C. R. and Chang, P. (1995) Cor r el at i on of
diffusion coefficicnts in di l ut e sol ut i ons. A.I.Ch.E.J.
1, 264-270.
Wi l l i amson, J. E., Bazai re, K. E. and Geankopl i s , C. J.
(19631 Li qui d-phas e mass t r ansfer at low Reynol ds
numbers. Ind. Engnq Chem. Fund. 2, 126--129.
Wi l son, E. J. and Geankopl i s , C. J. (1966) Li qui d mass
t ransfer at very low Reynol ds number s in packed
beds. Ind. En,qn.q Chem. Fundam. 5, 9 14.
Yaws, C. L. (1995) ltandook ~" Transport Property
Data. Gul f Publ i shi ng company, Houst on.