Mass Transfer Between Flowing Fluid and Sphere

Buried in Packed Bed of Inerts

J. R. F. Guedes de Carvalho, J. M. P. Q. Delgado, and M. A. Alves
Dept. de Engenharia Quı́mica, Faculdade de Engenharia da Universidade do Porto, Rua Dr. Roberto Frias,
4200-465 Porto, Portugal

The equations describing fluid flow and mass transfer around a sphere buried in a
packed bed are presented, with due consideration given to the processes of transverse and
longitudinal dispersion. Numerical solution of the equations was undertaken to obtain
point values of the Sherwood number as a function of the Peclet and Schmidt numbers
over a wide range of values of the relevant parameters. A correlation is proposed that
describes accurately the dependence found numerically between the values of the Sher-
wood number and the values of Peclet and Schmidt numbers. Experiments on the
dissolution of solid spheres (benzoic acid and 2-naphthol) buried in packed beds of sand,
through which water was forced, at temperatures in the range 293 K to 373 K, gave values
of the Sherwood number that were used to test the theoretical results obtained. Excellent
agreement was found between theory and experiment, including the large number of
available data for naphthalene–air, and this helps establish the proposed correlation as
“general” for mass transfer between a buried sphere and the fluid (liquid or gas) flowing
around it. © 2004 American Institute of Chemical Engineers AIChE J, 50: 65–74, 2004
Keywords: mass transfer, dispersion, packed bed, active sphere, liquid flow

Introduction In a recent study on transverse dispersion in liquids, Delgado

In several situations of practical interest a large solid mass and Guedes de Carvalho (2001) showed that there is a signif-
interacts with the liquid flowing around it through the interstices of icant dependence between D T and Sc, in the range Sc ⬍ 550.
a packed bed of inerts. Examples are the leaching of buried rocks Since the rate of mass transfer around a buried sphere exposed
and the contamination of underground waters by compacted bur- to a flowing fluid is strongly determined by D T (see Guedes de
ied waste. The dissolution of a slightly soluble sphere buried in a Carvalho and Alves, 1999), it may be expected that this mass
packed bed of sand, through which water flows, is a useful model transfer will show a significant dependence on Sc.
for such processes, and it is considered in the present work. The only data available on mass transfer between a sphere
This study may be seen as a natural extension of work made immersed in a packed bed of inerts and the liquid flowing
by our team (Coelho and Guedes de Carvalho, 1988b; Guedes around it seem to be those of Guedes de Carvalho and Delgado
de Carvalho and Alves, 1999) on mass transfer around a sphere (1999), who worked with water near ambient temperature.
buried in a packed bed through which gas flows, which is Their results showed that the approach of Coelho and Guedes
important in the analysis of char combustion in fluidized beds. de Carvalho (1988b) could be extended to predict the rates of
From a fundamental point of view, working with water has mass transfer in liquids at high values of the Peclet number,
great interest, since it allows a considerable variation in the Pe⬘ (based on the diameter of the sphere).
value of the Schmidt number, Sc. Indeed, whereas Sc is of The present work was undertaken in an effort to both widen
order 1 for most gas mixtures, it is possible to cover the the range of application of the theoretical analysis and provide
approximate range 50 ⬍ Sc ⬍ 1000 by working with com- experimental data over a substantial range of values of Sc.
mon solutes in water, at temperatures between 290 and 373 K.

Correspondence concerning this article should be addressed to J. R. F. Guedes de

Theory
Carvalho at jrguedes@fe.up.pt.
In terms of analysis, we consider the situation of a slightly
© 2004 American Institute of Chemical Engineers soluble sphere of diameter d 1 buried in a bed of inert particles

AIChE Journal January 2004 Vol. 50, No. 1 65

of diameter d (with d Ⰶ d 1 ), packed uniformly (void fraction This integral may be evaluated numerically for each set of
⑀) around the sphere. The packed bed is assumed to be “infi- conditions from the discretized concentration field that is ob-
nite” in extent, and a uniform interstitial velocity of liquid, u0, tained through the numerical solution of Eq. 5 (see Appendix
is imposed at a great distance from the sphere. A). It is convenient to express the rate of dissolution in terms
Darcy’s law, u ⫽ ⫺Kgradp, is assumed to hold, and if it is of the Sherwood number, Sh⬘ ⫽ kd 1 /D⬘m , where k ⫽
coupled with the continuity relation for an incompressible n/[ ␲ d 21 (c* ⫺ c 0 )] is the mass-transfer coefficient for the
fluid, div u ⫽ 0, Laplace’s equation ⵜ2␾ ⫽ 0 is obtained for the sphere. (The effective coefficient of molecular diffusion for the
flow potential, ␾ ⫽ Kp, around the sphere. solute is D⬘m ⫽ D m / ␶ , where ␶ is the tortuosity; in all primed
In terms of spherical coordinates (r, ␪ ), the potential and dimensionless groups, D⬘m is used instead of D m ).
stream functions are, respectively (see Currie, 1993) In the analysis of the results of the numerical computations,
it is convenient to consider two separate ranges of Pe⬘p (⫽

冋 冉 冊册
3 u 0 d/D⬘m ).
1 R
␾ ⫽ ⫺u0 1 ⫹ r cos ␪ (1) 1. Low Pe⬘p (say, Pe⬘p ⬍ 0.1). For very low Pe⬘p , dispersion
2 r is the direct result of molecular diffusion, with D T ⫽ D L ⫽
D⬘m , and the numerical solution obtained by Guedes de Car-

冋 冉 冊册 valho and Alves (1999) applies. Those authors suggest that

3
u0 R
␺⫽ 1⫺ r 2sin2 ␪ (2) their results are well approximated (with an error of less than
2 r
1% in Sh⬘) by

and the velocity components are

冋 册
1/ 2
Sh⬘ 4 4
⫽ 4 ⫹ 共Pe⬘兲 2/3 ⫹ Pe⬘ (7)
⑀ 5 ␲
⭸␾
冋 冉 冊册
3
R
ur ⫽ ⫽ ⫺u0 cos ␪ 1 ⫺ (3)
⭸r r where Pe⬘ ⫽ u 0 d 1 /D⬘m is the Peclet number for the sphere.
2. Intermediate to high Pe⬘p (say, Pe⬘p ⬎ 0.1). As Pe⬘p is

冋 冉 冊册
increased above 0.1, values of D T and D L start to deviate from
1 ⭸␾ 1 R 3
u␪ ⫽ ⫽ u 0sin␪ 1 ⫹ . (4) D⬘m , and this has to be taken into account in the integration of
r ⭸␪ 2 r Eq. 5. For gas flow, it is generally accepted that good approx-
imate values are given by
The analysis of mass transfer is based on a steady-state material
balance on the solute crossing the borders of an elementary ud DL Pe⬘p u
volume, limited by the potential surfaces ␾ and ␾ ⫹ ␦␾, and D L ⫽ D⬘m ⫹ or ⫽1⫹ (8)
Pe L共⬁兲 D⬘m PeL 共⬁兲 u0
the stream surfaces ␺ and ␺ ⫹ ␦␺. The resulting equation is
(see Guedes de Carvalho and Alves, 1999)
ud DT Pe⬘p u
D T ⫽ D⬘m ⫹ or ⫽1⫹ (9)
Pe T共⬁兲 PeT 共⬁兲 u0
冉 冊 冉 冊
D⬘m
⭸c ⭸ ⭸c ⭸ ⭸c
⫽ DL ⫹ D T␻ 2 (5)
⭸␾ ⭸␾ ⭸␾ ⭸␺ ⭸␺ for the entire range of values of Pe⬘p , with Pe L (⬁) ⫽ 2 and
Pe T (⬁) ⫽ 12, for flow through beds of approximately isomet-
where ␻ is the distance to the flow axis, and D L and D T are the ric particles (see Wilhelm, 1962; Coelho and Guedes de Car-
longitudinal and transverse dispersion coefficients, respec- valho, 1988a).
tively. The numerical solution of Eq. 5, with D L and D T given by
The boundary conditions to be observed in the integration of Eqs. 8 and 9, was worked out by Guedes de Carvalho and
Eq. 5 are: (1) the solute concentration is equal to the back- Alves (1999), who observed that the values of Sh⬘ obtained
ground concentration, c 0 , far away from the sphere; (2) the from the numerical solution are well represented (within 2% in
solute concentration is equal to the equilibrium concentration, Sh⬘) by
c ⫽ c*, on the surface of the sphere; and (3) the concentration

冋 册冉 冊
field is symmetric about the flow axis. 1/ 2 1/ 2
Sh⬘ 4 4 Pe⬘p
The procedure followed in the numerical solution of Eq. 5, ⫽ 4 ⫹ 共Pe⬘兲 2/3 ⫹ Pe⬘ 1⫹ (10)
⑀ 5 ␲ 9
with the appropriate boundary conditions, is an improved ver-
sion of the method described by Guedes de Carvalho and Alves
(1999), and it is detailed in Appendix A. Now, for liquids, since Eqs. 8 and 9 are still good approxima-
The numerical solution of Eq. 5 gives the concentration field, tions at low to intermediate values of Pe⬘p (say, up to Pe⬘p ⬇ 1,
and from it the instant rate of dissolution of the sphere, n, is for D L , and up to Pe⬘p ⬇ 10, for D T ), Eq. 10 will be adequate,
obtained by integrating the diffusion/dispersion flux over the but only for a narrow range of Pe⬘p . Beyond that range, more
surface r ⫽ R accurate values of D T and D L are required, in the numerical
solution of Eq. 5.
Values of D T for liquid flow have been reported recently by

n⫽⫺ 冕 冉冊␲

0
⑀DT
⭸c
⭸r r⫽R
2␲␻Rd␪ (6)
Delgado and Guedes de Carvalho (2001) and Guedes de Car-
valho and Delgado (2000) in what seems to be the only
available study on the influence of Sc on D T . Their data for the

66 January 2004 Vol. 50, No. 1 AIChE Journal

 For Sc ⬎ 550 the transverse dispersion coefficient is found to
be independent of the Schmidt number, and the correlation
reduces to

DT Pe⬘p
⫽1⫹ ⫺ 8.1 ⫻ 10 ⫺3Pe⬘p1.268 (12)
D⬘m 12

As for D L , it is fortunate that its value is not needed with

accuracy, since for Pe⬘p ⬎ 1, the boundary layer for mass
transfer around the sphere is thin, provided that the approxi-
mate condition d 1 /d ⬎ 10 is observed. Indeed, for Pe⬘(⫽
Pe⬘p d 1 /d) ⬎ 10, the boundary layer is thin and the term with
D L in Eq. 5, can be neglected (Coelho and Guedes de Carvalho,
Figure 1. Experimental values of DT/Dⴕm for Peⴕp < 2000. 1988b); numerical computations were undertaken in the
present work that confirm the insensitivity of Sh⬘ to D L , for
Pe⬘ ⬎ 10.
For each value of Sc indicated in Figure 1, Eq. 5 was solved
range of values of Pe⬘p of interest in the present study are
numerically, with the point values of D T given by the corre-
shown in Figure 1, where it is possible to observe the depen-
sponding fitted curve (Eqs. 11 and 12). From the numerical
dence of D T on the Schmidt number for the range Sc ⱕ 550.
simulations, plots of Sh⬘/ ⑀ vs. Pe⬘ were prepared for given
An empirical correlation was found to describe the measured
values of d/d 1 in a similar fashion to what was done by Guedes
data of D T for Sc ⱕ 550 (shown in Figure 1 as dashed lines)
de Carvalho and Alves (1999); in the present case, a set of plots
had to be made for each value of Sc. The results of the
DT
D⬘m
⫽1⫹
Pe⬘p
12
⫺ 冉 冊
Sc
1500
4.8
Pe⬘p4.83⫺1.3log10共Sc兲 (11)
numerical computations are shown as points in Figure 2a–2d,
and an expression was sought to describe the functional de-

Figure 2. Dependence of Shⴕ/å on Peⴕ for different values of d/d1 at (a) Sc ⴝ 52; (b) Sc ⴝ 142; (c) Sc ⴝ 288; (d) Sc ⴝ
550: the points were obtained from the numerical solution of Eq. 5 and the solid lines represent Eq. 13.

AIChE Journal January 2004 Vol. 50, No. 1 67

pendence observed, with good accuracy. The following equa-
tion is proposed for Sc ⱕ 550

冋 册
1/ 2
Sh⬘ 4 4
⫽ 4 ⫹ 共Pe⬘兲 2/3 ⫹ Pe⬘
⑀ 5 ␲

⫻ 1⫹ 冋 Pe⬘p
9
⫺ 冉 冊冉 冊
Sc
1500
4.8
4
3
Pe⬘p 册
4.83⫺1.3log10 共Sc兲 1/ 2

(13)

For Sc ⬎ 550, the value of Sh⬘ is independent of Sc, since

D T /D⬘m is independent of Sc. Substituting Sc ⫽ 550 in Eq. 13
leads to

冋 册
1/ 2
Sh⬘ 4 4
⫽ 4 ⫹ 共Pe⬘兲 2/3 ⫹ Pe⬘
⑀ 5 ␲ Figure 3. Dependence of ␩ on Peⴕp for different values of

冋 册 Sc: the points were obtained from the numer-

1/ 2
Pe⬘p
⫻ 1⫹ ⫺ 1.16 ⫻ 10 ⫺2Pe⬘p1.268 (14) ical solution of Eq. 5.
9

and this can be expected to predict mass-transfer coefficients and

for Sc ⱖ 550. In the plots in Figure 2, the solid lines represent

冋 册
1/ 2
either Eq. 13 or Eq. 14, and it may be seen that they describe Pe⬘p
the results of the numerical computations with very good ␩⫽ 1⫹ ⫺ 1.16 ⫻ 10 ⫺2Pe⬘p1.268 for Sc ⬎ 550.
9
accuracy.
It is worth emphasizing some important features of Eqs. 13 (17b)
and 14. First of all, the fact that the first term on the righthand
side of both equations gives the dimensionless mass-transfer It can be seen that ␩ approaches unity as Pe⬘p is reduced, and
coefficient for low Pe⬘p when both longitudinal and transverse for gas flow, with Sc ⬵ 1, Eq. 17b is virtually coincident with
dispersion are due to molecular diffusion alone (D L ⫽ D T ⫽ ␩ ⫽ (1 ⫹ Pe⬘p /9) 1/ 2 , which was the expression suggested by
D⬘m ). This result was obtained by Guedes de Carvalho and Guedes de Carvalho and Alves (1999). The values of ␩ ob-
Alves (1999), who emphasized the fact by writing (for Pe⬘p ⬍ tained from the numerical solution of Eq. 5 are represented as
0.1, say) points in Figure 3, alongside the lines corresponding to Eqs.
17a,b. The scale is linear and it can be seen that the agreement

冋 册
1/ 2
is excellent.
Sh⬘ Sh⬘md 4 4 Equations 16 and 17 therefore can be seen as a general result
⫽ ⫽ 4 ⫹ 共Pe⬘兲 2/3 ⫹ Pe⬘ (15)
⑀ ⑀ 5 ␲ derived from first principles for mass transfer between a sphere
of diameter d 1 buried in a packed bed of inerts of diameter d,
The second term (with square brackets) on the righthand side and the fluid flowing through the interstices of the bed with
of Eqs. 13 and 14 is therefore the “enhancement factor” due to velocity u 0 (at some distance from the sphere). Experiments
convective dispersion. It will be noticed that the enhancement were performed to test this result over a wide range of tem-
factor is independent of Sc for high values of this parameter peratures, as described in the next section.
and dependent on Sc for Sc ⱕ 550. This is because mass-
transfer rates around the sphere depend strongly on D T and the Experiment
value of D T /D⬘m is independent of Sc only for Sc ⬎ 550, as Experiments were performed on the dissolution of single-
shown recently by Delgado and Guedes de Carvalho (2001) in spheres of both benzoic acid and 2-naphthol, buried in beds of
a detailed study on dispersion in liquids. sand (0.219 mm or 0.496 mm average particle diameter)
A convenient way of presenting the result given by Eqs. 13 through which water was steadily forced down at temperatures
and 14 is then in the range 293 K to 373 K.
Figure 4 shows the experimental rig (not to scale). The
Sh⬘ Sh⬘md distilled water in the feed reservoir was initially dearated under
⫽ 䡠 ␩, (16) vacuum to avoid freeing gas bubbles in the rig at high temper-
⑀ ⑀
ature. The test column was a stainless steel tube (100 mm ID
and 500 mm long), and it was immersed in a silicone oil bath
with kept at the desired operating temperature by means of a ther-
mosetting bath head (not represented in the figure). The copper

␩⫽ 1⫹ 冋 Pe⬘p
9 冉 冊冉 冊
⫺
Sc
1500
4.8
4
3
Pe⬘p 册
4.83⫺1.3log10 共Sc兲 1/ 2 tubing feeding the pressurized water to the column at a constant
metered rate was partly immersed in a preheater, and it had a
significant length immersed in the same thermosetting bath as
for Sc ⱕ 550 (17a) the test column; the copper tubing leaving the test column was

68 January 2004 Vol. 50, No. 1 AIChE Journal

 Figure 4. Experimental setup.

Figure 5. Dependence of Shⴕ/å on Peⴕ for DT ⴝ DL ⴝ Dⴕm

(throughout).
immersed in a chiller to cool the outlet stream before reaching
the UV analyzer.
In order to replace the sphere buried in the sand, valves B1
and B2 were closed, the lid of the test column was removed, Results and Discussion
and a source of distilled water was connected near N to direct Accurate values of c* and D⬘m are needed in the treatment of
water “backwards” to the test column. Valve B2 was then the mass-transfer data obtained in our experiments. Values of
opened gently to reach incipient fluidization of the sand; the those parameters at different temperatures are given (and dis-
soluble sphere was then replaced, the water flow was immedi- cussed) in Appendix B for the systems used.
ately stopped, and the lid of the test column was placed back in In order to test Eq. 15, it was necessary to perform some
position. The downflow of water through the test column was experiments at very low Pe⬘p . A significant number of data
then initiated at a very low flow rate, while the column was points were obtained with both benzoic acid and 2-naphthol
immersed in the silicone oil bath and the temperature was and, in Figure 5, they are shown to be in excellent agreement
allowed to rise (slowly) to the value intended for the experi- with Eq. 15. The agreement is so good that in Appendix B we
ment. The flow rate of water was then adjusted to the required suggest that measuring the rate of dissolution of a buried sphere
value, v, and the concentration of solute in the outlet stream at low Pe⬘p may be an accurate method for determining the
was continuously monitored by means of a UV/VIS Spectro- molecular diffusion coefficient of slightly soluble solutes. At
photometer (set at 274 nm for 2-naphthol, and at 226 nm for higher values of Pe⬘p , it may be seen, from Eq. 13, that Sh⬘/ ⑀
benzoic acid). The instant rate of dissolution of the sphere was is expected to be a function of d/d 1 , Sc, and Pe⬘, since Pe⬘p ⫽
calculated from the steady-state concentration of solute at the Pe⬘(d/d 1 ).
exit, c e , as n ⫽ vc e . Our experimental points were therefore organized in three
The spheres of benzoic acid and 2-naphthol were prepared plots of Sh⬘/ ⑀ vs. Pe⬘ (one plot for each value of d/d 1 ), as
from p.a. grade material, which was molten and then poured shown in Figures 6, 7, and 8. These figures show that the
into moulds made of silicone rubber to give spheres with agreement between experiment and Eqs. 13 and 14 is very
approximate diameters of 11, 15, 19, and 25 mm. If any slight good, and this is strong support for our theory.
imperfections showed on the surface of the spheres, they were In each of the plots of Figures 6 – 8, the solid line represents
easily removed by rubbing with fine sandpaper. Vernier calli- the equation for gas flow obtained and tested by Guedes de
pers were used to make three measurements of the diameter of Carvalho and Alves (1999). It can be seen that the data ob-
each sphere, along three perpendicular directions. The spheres tained in the present work virtually bridge the gap between the
would be discarded if the difference between any two mea- extremes of gas flow (Sc ⬵ 1) and cold-water flow (Sc ⫽
surements exceeded 1 mm; otherwise, they would be kept for 980). Also, the agreement between theory and experiment is
the experiments and taken to have a diameter equal to the excellent over the entire range; and since D T /D m is indepen-
average of the three measurements. dent of Sc for Sc ⱖ 550 [see Delgado and Guedes de Carvalho
The silica sand used in the experiments was previously (2001)], it can be stated that Eqs. 13 and 14 (or, alternatively,
washed, dried, and sieved in closely sized batches. Table 1 Eqs. 16 and 17a,b) accurately predict the coefficient of mass
gives the corresponding size distribution, as determined by a transfer between a sphere buried in a packed bed of inerts and
laser diffraction technique. the fluid flowing around it, in parallel uniform flow (far from

Table 1. Properties of the Sand Beds

Particle Diameter (⬍ ␮m)
Nominal Size 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(mm) ⑀ ␶ vol. vol. vol. vol. vol. vol. vol. vol. vol. vol.
0.219 0.40 1.41 161 180 195 210 225 241 259 281 313 476
0.496 0.34 1.41 277 340 389 434 480 530 588 660 756 900

AIChE Journal January 2004 Vol. 50, No. 1 69

Figure 6. Comparison between experimental points (this Figure 8. Comparison between experimental points (this
work) for d/d1 ⴝ 0.02 and the theory repre- work) for d/d1 ⴝ 0.045 and the theory repre-
sented by Eqs. 13 and 14. sented by Eqs. 13 and 14.

the sphere). The approximate conditions of validity of those valid for any value of Sc. The numerical solution of the
equations are equation representing the theory gives the “exact” values of
Sh⬘/ ⑀ , and these are well represented by Eqs. 13 and 14 (or,
Re p ⬍ 25 (18a) alternatively, by Eqs. 16 and 17a,b).
The large number of experimental points reported, covering
d 1/d ⬎ 10 (18b) a wide range of the relevant parameters, provide strong support
for the theory developed.
Pe⬘p ⬍ 1300 (18c) In Figure 9 predictions of the theory presented in this work
are compared with those given by equations proposed by other
authors (see Table 2). Figure 9 refers only to one value of d/d 1
and these are easily met in most situations of practical interest. (similar curves are obtained for other values of d/d 1 ), and the
The limitation on Re p ensures that Darcy’s law is observed two “extreme” values of Sc ⫽ 1 and Sc ⫽ 980, corresponding
with good approximation, d 1 /d ⬎ 10 is an approximate con- approximately to gas flow and cold water flow, respectively,
dition establishing that the active particles are large compared are considered. With hot water (or other low-viscosity liquids)
to the inerts, and the limitation on Pe⬘p is related to the disper- or supercritical fluids, the intermediate range of Sc is covered
sion data available (see Figure 1); it should be emphasized that and the lines giving the different correlations are situated
near the surface of the sphere the fluid velocity varies up to between the two extremes.
1.5u 0 . The large number of experimental values reported in the
present work, together with the many available for gas flow
Conclusions [see Guedes de Carvalho and Alves (1999)], confirms the
The present work shows that a theory for mass transfer
between a sphere buried in a packed bed of inerts and the fluid
flowing past it may be derived from first principles, which is

Figure 9. Comparison of equations available for the pre-

Figure 7. Comparison between experimental points (this diction of mass transfer from a buried sphere
work) for d/d1 ⴝ 0.026 and the theory repre- (the experimental points for Sc ⴝ 1 and Sc ⴝ
sented by Eqs. 13 and 14. 980 are a small sample of the available data).

70 January 2004 Vol. 50, No. 1 AIChE Journal

 Table 2. Equations Available in Literature for Mass Transfer from Large Active Particles in Beds of Small Inerts
Reference Model Equation

La Nauze et al. (1984)

Sh⬘
⑀
⫽ 2␶ ⫹ 冉 冊
4␶
␲⑀
1/2
共Pe⬘兲1/2 19
(gas flow)
Sh⬘
Prins et al. (1985) ⫽ K共Pe⬘兲1⫺m 20
⑀

K⫽
⑀ 冉 冊冉 冊冋
␶m 1 ⫺ ⑀ m d1 m
⑀ d
0.105 ⫹ 1.505
d1
d 冉冊 册⫺1.05
Scm⫺2/3 20a

冉冊
m ⫽ 0.35 ⫹ 0.29
d 0.5
d1
20b

Coelho and G. Carvalho (1988b) Sh⬘

⑀ 冋
⫽ 4 ⫹ 0.576共Pe⬘兲0.78 ⫹ 1.28共Pe⬘兲 ⫹ 0.141
d
d1
共Pe⬘兲2 冉冊 册 1/2
21

Agarwal et al. (1988) Sh⬘

⑀
⫽ Kp冉 冊
␶ CD ⑀ a 1/3
⑀ 8
共Pe⬘兲1/3 Sc⫺1/3
22

关1 ⫹ 共⑀/␶兲Pe⬘兴1/3 ⫺ 1 22a
Kp ⫽ 0.69
关共⑀/␶兲Pe⬘兴1/3

CD ⑀ a ⫽
⑀Pe⬘ 冋
24␶2 Sc 2Z共1 ⫺ ⑀兲
␶
⫹ 10␺ ⫺ 1 册 22b

冉 冊
⑀ 䡠 Pe⬘ 0.369
冉 冊
⑀ 䡠 Pe⬘ 0.431
0.124

再 冉 冊冎
␺ ⫽ 0.261 ⫺ 0.105 ⫺ 22c
␶ 䡠 Sc ␶ 䡠 Sc ⑀ 䡠 Pe⬘ 2
1 ⫹ Log
␶ 䡠 Sc
5.3
Z⫽ 2
⑀
␶ 1⫺⑀ 冉 冊 0.3
22d

冋 册冋 冉冊 册
1/2 1/2
G. Carvalho and Alves (1999) Sh⬘ 4 4 1 d 23
⫽ 4 ⫹ 共Pe⬘兲2/3 ⫹ Pe⬘ 1⫹ Pe⬘ 共gasflow兲
⑀ 5 ␲ 9 d1

validity of Eqs. 13 and 14 over the entire range of Sc. A small u r , u ␪ ⫽ components of fluid interstitial velocity
sample of data points for naphthalene-air and 2-naphthol–water v ⫽ volumetric flow rate
at ambient temperature are represented in Figure 9. It can be
seen that the other equations give very inaccurate predictions Greek letters
over a wide range of values of Pe⬘p .
⑀ ⫽ bed voidage
⌽ ⫽ dimensionless potential function (as defined in Eq. A4)
Notation ␾ ⫽ potential function (defined in Eq. 1)
␩ ⫽ enhancement factor due to convective dispersion (defined in
A ⫽ area Eq. 16)
c ⫽ solute concentration ␮ ⫽ dynamic viscosity
c0 ⫽ bulk concentration of solute ␪ ⫽ spherical angular coordinate
c* ⫽ saturation concentration of solute ␳ ⫽ density
ce ⫽ concentration in the outlet stream ␶ ⫽ tortuosity
C ⫽ dimensionless solute concentration (as defined in Eq. A1) ␻ ⫽ cyclindrical radial coordinate (distance to the axis)
d ⫽ diameter of inert particles ⌿ ⫽ dimensionless stream function (as defined in Eq. A5)
d1 ⫽ diameter of active sphere ␺ ⫽ stream function (defined in Eq. 2)
DL ⫽ longitudinal dispersion coefficient
Dm ⫽ molecular diffusion coefficient
D⬘m ⫽ effective molecular diffusion coefficient (⫽ D m / ␶ )
DT ⫽ transverse (radial) dispersion coefficient Dimensionless groups
K ⫽ permeability in Darcy’s law Pe⬘ ⫽ Peclet number based on diameter of active sphere (⫽ u 0 d 1 /
k ⫽ average mass-transfer coefficient D⬘m )
n ⫽ mass-transfer rate Pe⬘p ⫽ Peclet number based on diameter of inert particles (⫽ u 0 d/D⬘m )
p ⫽ pressure Re p ⫽ Reynolds number based on diameter of inert particles (⫽
Pe L (⬁) ⫽ asymptotic value of Pe L when Re p 3 ⬁ ␳ ud/ ␮ )
Pe T (⬁) ⫽ asymptotic value of Pe T when Re p 3 ⬁ Sc ⫽ Schmidt number (⫽ ␮ / ␳ D m )
R ⫽ radius of the sphere Sh⬘ ⫽ Sherwood number (⫽ kd 1 /d⬘m )
ᑬ ⫽ dimensionless spherical radial coordinate (⫽ r/R) Sh⬘md ⫽ Sherwood number when D T ⫽ D L ⫽ D⬘m (i.e., Pe⬘p ⬍ 0.1)
r ⫽ spherical radial coordinate (distance to the center of the soluble
sphere)
U ⫽ dimensionless interstitial velocity (⫽ u/u 0 ) Subscripts and superscripts
u ⫽ absolute value of interstitial velocity
u ⫽ interstitial velocity (vector) i, j ⫽ grid node indices (see Figure A1)
u0 ⫽ absolute value of interstitial velocity far from the active sphere n ⫽ iteration number

AIChE Journal January 2004 Vol. 50, No. 1 71

Literature Cited c ⫺ c0
C⫽ (A1)
Agarwal, P. K., W. J. Mitchell, and R. D. La Nauze, “Transport Phenom- c* ⫺ c 0
ena in Multi-Particle Systems—III. Active Particle Mass Transfer in
Fluidized Beds of Inert Particles,” Chem. Eng. Sci., 43, 2511 (1988).
Coelho, M. N., and J. R. Guedes de Carvalho, “Transverse Dispersion in u 共u r2 ⫹ u ␪2兲 1/ 2
Granular Beds: Part I—Mass Transfer from a Wall and the Dispersion U⫽ ⫽ (A2)
u0 u0
Coefficient in Packed Beds,” Chem. Eng. Res. Des., 66, 165 (1988a).
Coelho, M. A. N., and J. R. F. Guedes de Carvalho, “Transverse Dispersion
in Granular Beds: Part II—Mass Transfer from Large Spheres Immersed r
in Fixed or Fluidised Beds of Small Inert Particles,” Chem. Eng. Res. ᑬ⫽ (A3)
Des., 66, 178 (1988b). R
Currie, I. G., Fundamental Mechanics of Fluids, McGraw-Hill, New York,
p. 150 (1993).
Delgado, J. M. P. Q., and J. R. F. Guedes de Carvalho, “Measurement of 4 ␾
the Coefficient of Transverse Dispersion in Packed Beds Over a Range ⌽⫽ (A4)
3 u 0d 1
of Values of Schmidt Number (50 –1000),” Transport Porous Media,
44(1), 165 (2001).
Dunker, C., “Benzoic Acid,” Kirk-Othmer Encyclopedia of Chemical ␺
Technology, Vol. 3, 2nd ed., Interscience, New York, p. 422 (1964). ⌿⫽ . (A5)
Ferziger, J. H., and M. Peric, Computational Methods for Fluid Dynamics, u 0d 12
Springer-Verlag, Berlin, p. 42 (1996).
Gaskell, P. H., and A. K. C. Lau, “Curvature Compensated Convective
Transport: SMART, A New Boundedness Preserving Transport Algo- Equation 5 can be rearranged to
rithm,” Int. J. Numer. Methods Fluids, 8, 617 (1988).
Ghosh, U. K., S. Kumar, and S. N. Upadhyay, “Diffusion Coefficient in
Aqueous Polymer Solutions,” J. Chem. Eng. Data, 36, 413 (1991).
Guedes de Carvalho, J. R. F., M. A. Alves, and J. M. P. Q. Delgado,
⭸C
⫽
⭸⌽ ⭸⌽
⭸
A
⭸C
⭸⌽ 冋 册
⫹
⭸
⭸⌿
B
⭸C
⭸⌿
, 冋 册 (A6)
“Determination of Molecular Diffusion Coefficient Through Measure-
ment of Mass Transfer Around a Cylinder Exposed to Liquid Flow in a
Packed Bed,” Int. J. Heat Technol., 19(2), 95 (2001). with
Guedes de Carvalho, J. R. F., and M. A. M. Alves, “Mass Transfer and
Dispersion Around Active Sphere Buried in a Packed Bed,” AIChE J.,
45, 2495 (1999). 4 DL
A⫽ (A7)
Guedes de Carvalho, J. R. F., and J. M. P. Q. Delgado, “Mass Transfer 3Pe⬘ D⬘m
from a Large Sphere Buried in a Packed Bed Along Which Liquid
Flows,” Chem. Eng. Sci., 50, 1121 (1999).
Guedes de Carvalho, J. R. F., and J. M. P. Q. Delgado, “Lateral Dispersion 3 ᑬ 2sin2 ␪ DT
in Liquid Flow Through Packed Beds at Pe m ⬍ 1400,” AIChE J., 46(5), B⫽ , (A8)
1089 (2000).
16 Pe⬘ D⬘m
La Nauze, R. D., K. Jung, and J. Kastl, “Mass Transfer to Large Particles
in Fluidised Beds of Smaller Particles,” Chem. Eng. Sci., 39, 1623 and the appropriate boundary conditions are
(1984).
Lozar, J., C. Laguerie, and J. P. Couderc, “Diffusivité de l’Acide Benzo-
ique dans l’Eau: Influence de la Température,” Can. J. Chem. Eng., 53, ⌽ 3 ⫺⬁, ⌿ ⱖ 0 C 3 0 (A9)
200 (1975).
McCune, L. K., and R. H. Wilhelm, “Mass and Momentum Transfer in
Solid-Liquid System,” Ind. Eng. Chem., 41(6), 1124 (1949). ⌽ 3 ⫹⬁, ⌿ ⱖ 0 C 3 0 (A10)
Moyle, M. P., and M. Tyner, “Solubility and Diffusivity of 2-Naphthol in

再
Water,” Ind. Eng. Chem., 45(8), 1794 (1953).
Prins, W., T. P. Casteleijn, W. Draijer, and W. P. M. Van Swaaij, “Mass ⫺1 ⱕ ⌽ ⱕ 1 C ⫽ 1 共A11a兲
Transfer from a Freely Moving Single Sphere to the Dense Phase of a
⌿ ⫽ 0 兩⌽兩 ⬎ 1 ⭸C
Gas Fluidised Bed of Inert Particles,” Chem. Eng. Sci., 40, 481 (1985). ⫽0 共A11b兲
Sahay, H., S. Kumar, S. N. Upadhyay, and Y. Upadhyay, “Solubility of ⭸⌿
Benzoic Acid in Aqueous Polymeric Solutions,” J. Chem. Eng. Data, 26,
181 (1981).
Seidell, A., Solubilities of Organic Compounds, 3rd ed., Van Nostrand, ⌿ 3 ⫹⬁, all ⌽ C 3 0 (A12)
New York (1941).
Steele, L. R., and C. J. Geankoplis, “Mass Transfer from a Solid Sphere to
Water in Highly Turbulent Flow,” AIChE J., 5, 178 (1959). Discretization
Vanadurongwan, V., C. Laguerie, and J. P. Couderc, “Diffusivité Moy- Equation A6 was solved numerically using a finite difference
enne de l’Acide Benzoique dans l’Eau Entre la Dilution Infinie et la
Saturation—Influence de la Température,” Can. J. Chem. Eng., 54,
method similar to that adopted by Guedes de Carvalho and
460 (1976). Alves (1999). A second-order central differencing scheme
Wilhelm, R. H., “Progress Towards the a Priori Design of Chemical (CDS) in a general nonuniform grid was adopted for the
Reactors,” Pure Appl. Chem., 5, 403 (1962). discretization of the diffusive terms (Ferziger and Peric, 1996)
Wilke, C. R., and P. Chang, “Correlation of Diffusion Coefficients in that appear on the righthand side of Eq. A6. The convection
Dilute Solutions,” AIChE J., 1(2), 264 (1955).
term that appears on the lefthand side of Eq. A6 was discretized
with the SMART high-resolution scheme of Gaskell and Lau
(1988), which preserves boundedness even for convective
Appendix dominated flows.
In order to integrate Eq. 5 with auxiliary Eqs. 1 and 2, it is The discretized equation resulting from the finite difference
convenient to define the dimensionless variables approximation of Eq. A6 reads

72 January 2004 Vol. 50, No. 1 AIChE Journal

 C i⫺1/ 2, j ⫽ C i⫺2, j ⫹ Ĉ i⫺1/ 2, j共C i, j ⫺ C i⫺2, j兲, (A17)
with

Ĉ i⫺1/ 2, j ⫽ max 冋 Ci⫺1, j ⫺ Ci⫺2, j

Ci, j ⫺ Ci⫺2, j
, min 3 冉
Ci⫺1, j ⫺ Ci⫺2, j
Ci, j ⫺ Ci⫺2, j
,

3 3 Ci⫺1, j ⫺ Ci⫺2, j
⫹
8 4 Ci, j ⫺ Ci⫺2, j
,1 冊册 (A18)

Substitution of Eqs. A14 and A17 into Eq. A13 leads to the
final form of the discretized equation, which can be cast in
compact form as
Figure A1. Computational grid. C i, j ⫽ 共FC i⫺2, j ⫹ GC i⫺1, j ⫹ HC i⫹1, j ⫹ IC i, j⫺1 ⫹ JC i, j⫹1兲/E.
(A19)
The resulting system of equations (Eq. A19) was solved
C i⫹1, j ⫺ C i, j C i, j ⫺ C i⫺1, j iteratively using the successive overrelaxation (SOR) method
A i⫹1/ 2 ⫺ A i⫺1/ 2
C i⫹1/ 2, j ⫺ C i⫺1/ 2, j ⌽ i⫹1 ⫺ ⌽ i ⌽ i ⫺ ⌽ i⫺1 (Ferziger and Peric, 1996), and the implementation of the
⫽ boundary conditions was done in the same way as described in
⌽ i⫹1/ 2 ⫺ ⌽ i⫺1/ 2 ⌽ i⫹1/ 2 ⫺ ⌽ i⫺1/ 2
our previous work (Guedes de Carvalho and Alves, 1999).
C i, j⫹1 ⫺ C i, j C i, j ⫺ C i, j⫺1 It should be noted that we always started our calculations with
B j⫹1/ 2 ⫺ B j⫺1/ 2
⌿ j⫹1 ⫺ ⌿ j ⌿ j ⫺ ⌿ j⫺1 a zero concentration field on a coarse grid. A converged solution
⫹ , (A13) could be obtained very quickly (O(10 s) in a desktop PC with a 1.4
⌿ j⫹1/ 2 ⫺ ⌿ j⫺1/ 2
GHz AMD processor), and then we proceeded to a finer grid
where the values of the ⌽, ⌿, A, and B coefficients are easily (doubling the number of grid points in each direction). Instead of
computed using their definitions (Eqs. A4, A5, A7, and A8). restarting the calculations with a zero concentration field in this
Please note that these coefficients only have to be computed new finer grid, we simply interpolated the solution obtained in the
once, since they are dependent on a priori known quantities, coarse-level grid, leading to a significant decrease in the CPU time
and are not influenced by the unknown concentration field. required to attain convergence. This fully automated procedure
The C i⫹1/ 2, j and C i⫺1/ 2, j values have to be conveniently was repeated until the finest mesh calculations were performed.
interpolated from the known grid node values (represented as The use of Richardson’s extrapolation to the limit allowed us to
circles in Figure A1) using the SMART high-resolution scheme obtain very accurate solutions (with errors in the computed Sh⬘
to ensure numerical stability and good precision: value below 0.1%). A more elaborate multigrid technique could
have been implemented to further increase the convergence rate,
but we found that this simple technique was sufficient to obtain
C i⫹1/ 2, j ⫽ C i⫺1, j ⫹ Ĉ i⫹1/ 2, j共C i⫹1, j ⫺ C i⫺1, j兲. (A14) mesh-independent solutions in affordable CPU times.
The converged solution obtained yields values of Ci,j, from
The limiter function Ĉ i⫹1/ 2, j used in this work is expressed which the overall mass-transfer rate from the sphere, n, could be
as (Gaskell and Lau, 1988) calculated and expressed by means of an average Sherwood
number
Ĉ i⫹1/ 2, j kd 1
Sh⬘ ⫽ ⫽ 关n/共 ␲ d 12兲共c* ⫺ c 0兲兴d 1/D⬘m.
冋 冋
(A20)

冦
C i, j ⫺ C i⫺1, j Ci, j ⫺ Ci⫺1, j 1 D⬘m
3 if 僆 0, - The value of n was evaluated by numerically integrating the
C i⫹1, j ⫺ C i⫺1, j Ci⫹1, j ⫺ Ci⫺1, j 6

冋 册
3 3 Ci, j ⫺ Ci⫺1, j Ci, j ⫺ Ci⫺1, j 1 5 diffusive/dispersive flux of solute perpendicular to the sphere
⫹ if 僆 , along its surface (Eq. 6)
8 4 Ci⫹1, j ⫺ Ci⫺1, j Ci⫹1, j ⫺ Ci⫺1, j 6 6

册 册
⫽
1 if
Ci, j ⫺ Ci⫺1, j
Ci⫹1, j ⫺ Ci⫺1, j
5
僆 ,1
6
n ⫽ ⫺⑀ 冘 D Rsin共␪ 兲 23 u sin共␪ 兲
i
T i 0 i

Ci, j ⫺ Ci⫺1, j Ci, j ⫺ Ci⫺1, j

Ci⫹1, j ⫺ Ci⫺1, j
if
Ci⫹1, j ⫺ Ci⫺1, j
ⰻ关0, 1兴,

(A15)
⫻ 冉 冊 ⭸c
⭸␺ i,1
2␲R2 冋 2 册
cos共␪i⫺1 兲 ⫺ cos共␪i⫹1 兲
, (A21)

or, in compact form

which in dimensionless discretized form reads

Ĉ i⫹1/ 2, j ⫽ max 冋 Ci, j ⫺ Ci⫺1, j

Ci⫹1, j ⫺ Ci⫺1, j
, min 3 冉
Ci, j ⫺ Ci⫺1, j
Ci⫹1, j ⫺ Ci⫺1, j
, Sh⬘
⫽⫺
3
冘 冉D⬘D 冊 sin 共␪ 兲
T 2
⑀
冊册
i
8 m i,1
3 3 Ci, j ⫺ Ci⫺1, j i
⫹
冉 冊冋 册
,1 (A16)
4 8 Ci⫹1, j ⫺ Ci⫺1, j ⭸C cos共␪i⫺1 兲 ⫺ cos共␪i⫹1 兲
⫻ . (A22)
Similarly, for the left face relative to node (i, j), one obtains ⭸⌿ i,1
2

AIChE Journal January 2004 Vol. 50, No. 1 73

Appendix B: Relevant Properties of the
2-Naphthol–Water and Benzoic Acid–Water
Systems
In the interpretation and correlation of the experimental data
reported, it is important to know the solubility and the coeffi-
cient of molecular diffusion of 2-naphthol in water and of
benzoic acid in water, in the range 293 K to 373 K.
Data from a number of sources are plotted in Figures B1 to
B2 for the 2-naphthol–water system. The majority of data
points are from McCune and Wilhelm (1949) and Moyle and
Tyner (1953); the latter also provide the fitted functions

c* ⫽ 5.93 ⫻ 10 ⫺6exp共0.0394T兲 (B1)

and
D m ⫽ 9.95 ⫻ 10 ⫺7exp共⫺2012/T兲, (B2) Figure B3. Solubility of benzoic acid in water.

and these are drawn in the figures (c* in kg/m3, D m in m2/s,

and T in K).

Figure B4. Coefficient of molecular diffusion for benzoic

acid in water.

Figure B1. Solubility of 2-naphthol in water.

We performed some measurements of c*, by saturating
water with 2-naphthol up to 363 K, and obtained the points in
Figure B1, in excellent agreement with the line proposed by
Moyle and Tyner (1953). From our measurements of the mass-
transfer rate around buried spheres at very low Pe⬘p , values of
D m were also obtained, making use of Eq. 10. The correspond-
ing points are also shown in Figure B2, and they clearly
confirm the validity of Eq. B2.
For the benzoic acid–water system, a very large number of data
points are available in the literature, as shown in Figures B3 and
B4. The lines on those figures are given by the equations
c* ⫽ 1.44 ⫻ 10 ⫺5exp共0.0414T兲 (B3)
and
D m ⫽ 2.20 ⫻ 10 ⫺6exp共⫺2295/T兲 (B4)
that were fitted to the data points shown.
Figure B2. Coefficient of molecular diffusion for 2-naph-
Manuscript received Apr. 10, 2002, revision received May 24, 2003, and final version
thol in water. received Aug. 11, 2003.

74 January 2004 Vol. 50, No. 1 AIChE Journal