(28 McMullen A. K., Miyauchi, T., Vermeulen, T., Univ. (43) Sleicher, C. A,, Jr., A.Z.Ch.E. J . 5 , 145 (1959).

dalifornia Radiation Lab., Rept. UCRL-3911-Suppl. (1958). (44) Thornton, J. D., Chem. Eng. Prog~.Symp. Ser., No. 13, 50,
(29) Matsuyama, T., Chem. Eng. (Japan) 14, 249 (1950). 179 (1954).
(30) Miyauchi, T., Univ. California Radiation Lab., Rept. (45 V‘an Deemter, J. J., Zuiderweg, F. J., Klinkenberg, A,,
UCRL-3911 (AuPust 1957). dhem. Eng. Sci. 5 , 271 (1956).
(31) Miyauchi,‘ T.,-unpubliihed manuscript. (46) Vermeulen, T., Lane, A. L., Lehman, H. R., Rubin, B.,
(32) Vyauchi, T., McMullen, A. K., Vermeulen, T., Univ. Univ. California Radiation Lab., Repts. UCRL-2983, UCRL-
California Radiation Lab., Rept. UCRL-9112 (1960). 718 (revised), in press.
(33) Nagata, S., Eguchi, W.. Kasai, H., Morino, I.,’ Chem. Eng. (47) Vogt, H. G., Geankoplis, C. J., Znd. Eng. Chem. 45, 2119
(Japan) 21, 784 (1957). (1953) : 46. 1763 119541.
(34) Oldshue, J. Y . , Rushton, J. H., Chem. Eng. Progr. 48, 297 (48) Wkhner,’J. F., ’Wilhklm, R. H., Chem. Eng. Sci. 6 , 89 (1956).
(1952). (49) Wicke, E., Travinski, H., Chem. Zng. Tech. 25, 114 (1953).
(35) Otake, T . , Kunugida, E., Chem. Eng. (Japan) 22, 144 (1958). (50) Yagi, S..Miyauchi, T., Chem. Eng. (Japan) 17, 382 (1953);
(36) Otake, T., Kunugida, E., Kawabe, A , , Ihid., 23, 81 (1959). 19, 507 (1955).
(37) Prausnitz, J. M., A.Z.Ch.E. J . 4, 14M (1958). (51) Young. E. F., Chem. Eng. 64, 241 (February 1957).
(38) Reman, G. H., “Joint Symposium on Scaling-Up Problems,”
p. 35, Inst. Chem. Eng.: London, 1957. RECEIVED
for review March 8, 1960
(39) Reman, G. H.: Olney, R. B., Chem. Eng. Progr. 51, 141 ACCEPTED February 2 5 , 1963
(1958).
(40) Scheibel: E. G., A.I.Ch.E. J . 2, 74 (1956). Studies conducted in Lawrence Radiation Laboratory under
(41) Sege, G., LYoodfield, F. W., Chem. Eng. Progr. Symp. Ser., asupices of the U. S. Atomic Energy Commission. Part of the
No. 13, 50, 39 (1954). work of T. Miyauchi was done under a fellowship from the
(42) Sherwood, T. K., Pigford, R. L., “Absorption and Extrac- Japan Atomic Energy Commission. Computations made in the
tion,” 2nd ed., Chap. 5 . McGraw-Hill, New York, 1952. Universitv of California Computer Center.

LIQUID-PHASE MASS TRANSFER

A T LOW REYNOLDS NUMBERS
J. E. W I L L I A M S O N , K . E . B A Z A I R E , A N D C. J . G E A N K O P L I S
The Ohio State University, Columbus 10, Ohio

Liquid-phase mass transfer coefficients were obtained for packed beds of benzoic acid spheres and water
in the Reynolds number ( N ” R ~range
) of 0.08 to 120. All previous data were obtained with granular solids
or modified spheres. The previous correlation for J’ of Gaffney and Drew extended down to N ” R of ~ 0.8,
with considerable scatter of the data and uncertainty of the slope of the line in the range of N ” R below
~ 50.
The present data confirm the exponent on the Schmidt number in J’ of 0.58 for liquids. The data check
those of Goffney and Drew and McCune and Wilhelm above a N ” R of ~ 0.8. Equations are proposed for
the entire liquid range.

ANY D A T Ahave been reported for mass transfer of gases surface area by the indirect method of measuring the perme-
in packed beds of solids. In recent years, mass transfer ability of the bed.
with liquids in lobv velocity regions has assumed a greater In the present work, mass transfer coefficients of benzoic
importance because of the use of ion exchange and chroma- acid spheres to water were obtained in packed beds. Data
tography in packed beds. Relatively few d a t a have been obtained in the ‘V‘‘R~region of 0.08 to 120 were correlated and
obtained for liquids. particularly in the low Reynolds number compared with data from the literature. Equations are
regions, and no data for liquids have been obtained using beds presented for calculating the J ’ factors.
of true spheres.
Gaffney and Drew ( 4 ) made a complete study of the effect
Experimental
of Schmidt number and varied it from 150 to 1300 for liquids.
They found the exponent of A’sc in the J’ factor to be 0.58 T h e process flow diagram is shown in Figure 1. All previous
and not 2 / 3 as for gases. using modified spheres in the . V ” R ~ investigators (2, 4>8) used modified spheres made by pelletizing
range of 0.8 to 1500. Below a .Y”R~of 50, there was consider- powder. T h e pellets consisted of a short cylindrical section
able uncertainty of J’ because of scatter of the data. They with hemispherical caps. I n this rvork? true spheres were
stated that the data of Hobson and Thodos (6) for liquids made by casting.
should not be considered because of uncertain extrapolations. Figure 2 shows the two halves of the copper mold for casting
McCune and Wilhelm (8) used modified spheres of 2-naph- the spheres. T h e sphere mold was made by cutting approxi-
tho1 and obtained data for liquids for a S’’R~ range of 40 to mate hemispheres about 0.115 inch deep in each of the two
5000. Dryden, Strang, and PVithrow (2) used modified matching plates. Steel ball bearings 0.250 inch in diameter
spheres of 2-naphthol and benzoic acid in water in a ,V’‘R~ \yere inserted in the holes of one plate, and the two plates were
range of 0.004 to 7. Different lines were obtained for the two clamped together.
solutes using the 0.58 exponent for N s c . T h e plates were pressed slowly together several times using a
Evans and Gerald ( 3 ) obtained data for liquids with granules pressure of 200,000 p.s.i. to form 0.250-inch holes. While
of benzoic acid. However, it was necessary to obtain the still clamped together. holes were drilled for steel taper pins

126 I&EC FUNDAMENTALS

to allow precise alignment of the plates when reassembled.
After separating the pla.tes, channels lvere drilled to and from
the holes for pouring molten benzoic acid, and then the plates
were nickel plated.
H a r d , glossy white 'benzoic acid spheres were produced?
as shown in Figure 3. T h e diameter of 20 spheres for series I
runs was measured in 10 different directions. T h e maximum
of any one of the 200 measurements was 0.2533 inch, and the
minimum was 0.2452 inch. T h e average sphere diameter
was 0.2484 inch. with the maximum average diameter of a
single sphere being 0.2496 inch and the minimum 0.2468
inch. The average diameter of the spheres after a series of mass
transfer runs was 0.236 inch, or a decrease of 5% in diameter.
For calculations, the avlerage diameter of 0.242 inch was used.
For series I1 runs! the average was 0.2505 inch before the runs
and 0.2444 inch after th.e runs, or a decrease of 2.4y0 in diam-
eter. For calculations. the average of 0.2475 inch was used.
T h e average void fr,action of the bed was determined as
0.431 for series I and 0.441 for series I1 by counting the number
of spheres in a 2.00-inch packed height and calculating the
void fraction.
I n making R run, the flow was downflow and the outlet fluid
rate was determined by timing several weighed samples. T h e Figure 1. Process flow diagram
total outlet flow was coniposited and analyzed by titration with
A. Storage of distilled water
0.01O.V N a O H solution. All samples were collected under 6 . Throttle valve
nitrogen and titrated under nitrogen to prevent contamination C. Overflow constant-head tank
by ( 2 0 1 . .4 mixture of phenol red and bromothymol blue was D. Flow control valve
E , F, G. Rotameters
used as the indicator. This gave a sharp end point in the 7.2 H. Glass tower (2.625-inch inner diometer)
to 7.4 p H range. with packed bed
K. Packed bed of benzoic acid spheres 2.00
inches high
Results and Calculationls I. Bed of 6-mm. glass beads 2.0 inches high
above and below bed of benzoic acid
All of the mass transfrr runs were made with the inlet water spheres
concmtration of benzoic acid a t zero. T h e mass transfer
coefficient can be calcuhted by Equation 1 :
T h e J factor is defined as:
L(C2 - C,) = kLa,X(AC)I, (1)

If C1 = 0 and b = CJC,, then

T h e J' factor of Gaffney and Drew ( J ) is:

J' = 2
k
L
(5)
0.53
(4)

Table 1. Experimental Mass Transfer Data

Run Tpm,b., L> C'Z,
KO. c. Lb./Sq. Ft.-Sec. "Re N"R~ J J' Grams/ Liter
1-1 22.2 1.382 43.6 101.4 0.216 0.117 0.150
1-2 22.2 0.512 16.1 37.4 0.357 0.194 0.244
1-3 22.3 0.869 27.4 63.7 0.274 0.149 0.190
1-4 22.3 0.712 22.6 52.3 0.326 0.177 0,225
1-5 22.3 0.187 5.90 13.7 0.770 0.418 0.501
1-6 22.4 0.0940 2.98 6.88 1.115 0.605 0.718
1-7 24.6 0 . C814 2.71 6.28 1.200 0.663 0.904
1-8 24.6 1.581 52.7 122.5 0.205 0.113 0.175
1-9 22.5 0.440 14.00 32.4 0,455 0.248 0,330
1-10 22.5 0,564 17.9 41.5 0,350 0.191 0,256
1-11 23.0 0,232 7.65 17.8 0.592 0.325 0.437
1-12 22.9 0.165 5.30 12.3 0.690 0.378 0.501
1-1 3 22,8 0.138 4.44 10.3 1.103 0,605 0.755
11-1 24.3 0.00595 0.201 0,453 7.35 4.02 2.805
11-2 25.2 0.00183 0.0629 0.143 11.70 6.41 3,244
11-3 25.4 0,00439 0.152 0.344 10.13 5.56 3.208
11-4 23.2 0.00305 0.1004 0,227 11.40 6.22 3,006
11-5 23.7 0.00107 0,0357 0.081 23.28 12.91 3.256
11-6 23.3 0.0730 2.41 5.46 1.551 0.843 1.029
11-7 23.3 0.0126 0.415 0.941 4.89 2.67 2,267
11-8 23.5 0,0371 1.230 2.79 2.29 1.25 1.418
11-9 23.5 0.0186 0.615 1.40 3.66 2.01 1.950
11-10 23 :4 0.0239 0.791 1.79 3.00 1.64 1.705
11-11 22 , '7 0.1264 4.11 9.31 0.980 0.531 0.672
~ ~~~ ~~~~~

VOL 2 NO. 2 M A Y 1 9 6 3 127

 Figure 2. Metal mold for costing
-Figure 3.
~~

Benzoic acid sphere

suheres

T h e solubility data for benzoic acid given by Steele and T h e equation relating the variables in the N'n. range of
Geankoplis ( 9 ) and the Schmidt number data of others were 0.035 to 55 is:
used (7,9). T h e final calculated results are given in Table I. J = 2.50 ( J V ' ~ ~ ) - O . ~ ~ (5)
T h e average void fractions used in recalculating the d a t a of
other investigators were 0.36 for McCune and Wilhelm (8), This means k, is proportional to Lo." in this region.
0.51 for Gaffney and Drew (4, 0.36 for Dryden, Strang, and T h e d a t a of McCune and Wilhelm (8)correspond closely
Withrow ( Z ) , and 0.39 for Hougen and coworkers (5, 70). to those of this work. T h e line of Haugen and coworkers for
gases is 10 to 25% above the data for liquids.
In several runs a t N t n , of 0.1 to 0.5, the values of J were
Discussion below the line shown in Figure 4 and a n apparent peak was
I n Figure 4 the J factor data of this work are plotted us. found. Later it was found that the peak was erroneous, and
Nfn,. For the straight portion of the line, the mean deviation
of the data poinrs Of this work is 15.8%.

Figure 4. Plot of J vs. N ' R ~

1 . Dato of this work, liquids
1 . Doto of this work, liquids 2. Line of Goffney and Drew I41 and McCune and Wilhelm (81. liquids
2. Dato of McCvne and Wilhelm (81. liquids 3. ~~~ ~ ~
, ~ ~_. .
.. Line of Druden. Strano. and Withrow (21.
~ . . . liauidr:
, A. Benzoic acid
3. Line of Bar-ilon and Rernick (11. gases downflow. B. 2-naphthol dawnflow. C. 2-naphthol vptlow
4. Line of Hovgen and coworkers (5, 101, gores 4. Line of Hougen and coworkers ( 5 , 101. gores

128 l&EC FUNDAMENTALS

 ----\3
- ’ ’ ’ 4 Nomenclature
a,,, = sq. ft. surface a r e a i c u . ft. volume of bed
b = C,/Cs = C‘2/C’s
C’1 = inlet concentration, grams/liter
CI = inlet concentration, lb./lb.
C’? = outlet concentration, gramsiliter
CZ = outlet concentration, lb./lb.
C’s = saturation solubility, gramsiliter
Cs = saturation solubility, lb./lb.
D = molecular diffusivity, sq. ft./sec.
D, = Darticle diameter. ft.
JP = ik,/L)(:vsr)2’3
J’ = (k,/L)(,V),,OJ*
001
01
I.
7-;--, Lk r
05
mass transfer coefficient, lb. solute,!sa ft.-sec.-AC
I I I , , , , , ,
liquid flow. lb./sq. ft empty column cross section-sec.
5 IO 50 100
I , , , , ,

500 1000 5000 10000

=
=
%E iV’Re = D,L/p
lV“xc = D,L/ep
Figure 6. Plot of J’e vs. ”’Re
-vsc = P PD
I . This work X = packed bed length, ft.
2. Gaffney and Drew ( 4 ) e = void fraction
3. McCune and Wilhelm (8) P = viscosity. lb./ft.-sec.
4. Dryden, Strong, and Withrow (2): A. Benzoic acid downflow. 6. p = density, lb./cu. ft.
2-naphthol downflow. C. 2-naphtha1 upflow

literrrture Cited
sufficient time to reach steady state had not been used. Such a
(1) Bar-Ilan, hfoshe, Resnick, \Villiam. Ind. Eng. C em. 49, 313
peak had been found ( I ) for gases for upflow (Figure 4 ) . (1957).
This peak ivould be expected for upflow and is probably- due to (2) Dryden, C. E., Strang, D. A,, IVithrow, A. E., Chem. Eng.
Proqr. 49, 191 (1953).
natural convection effects in laminar flow, as discussed in
(3) Evans, G. C.. Grrald. C. F., Ibid.. 49, 135 (1953).
detail by others (7: 2). Dryden and co\zorkers (2) found (4) Gaffney, B. .J., DreLv. T. B., Znd. En?. Chem. 42, 1120 (1950).
different effects of natural convection on up- and doivnflo\v (5) Gamson, B. L$T.> Thodos, George, Houyen, 0. A , Trans.
A.1.Ch.E. 39, 1 (1943).
runs a t ver)- lo\\-Reynolds numbers. ( 6 ) Hobson, Mark, Thodos: George, Chem. Eng. Prop. 47, 370
I n Figure 5: the J ’ factors using -Vs,0.5* are plotted versus (1951).
the - V ” R ~values. T h e data of this work check very closely IV, H.. Sherwood, T. K.. Ibzd., 46, 258 (1950).
(7)’ Linion,
(8) McCune. L. I(..\\’ilhelm, R. H., Ind. En?. Chem. 41, 1124
those of Gaffney and Drew ( 4 ) and McCune and \t‘ilhelm /I040)
”i’
\ A ’

(8). T h e data of this work also confirm the line to be used at (9) Steele, I,. R.. Geankoplis, C. J . . A.1.Ch.E. J . 5 , 178 (1959).
(10) IVilkr, C . R., Hougrn. 0. A , , Trans. ‘4.I.Ch.E. 41, 445
values of *Y’’ne below 50> since the data of Gaffney and Dreiv (1 345).
( 4 ) scattered considerabl). in this region. T h e line for gases is
about 1 0 0 n higher than that for liquids. Hence, it appears RECEIVEDfor review August 31, 1962
A C C E P T E D February 25, 1963
that using the 0.58 power on the Schmidt number corre!ateq
only the liquid data. T h e data for Dryden and coworkers (2) Work supported by Procter and Gamble Co. and Esso Research
in Figure 5 are about 407; above the line of Gaffney and Drew and Enqineering Co.
(4).
T h e recommended equations to use for liquids for the entire
Reynolds number range are Equation 6, based on data of this
ivork primarily: and Eqiuation 7 > based on data of Gaffne)-
and Dreiv (4) and h)fcCune and \Vilhelm (8):
Correction
J’ = 2.40 (.V”n,)-0.86 N ” R 0.08
~ to 125 (6)
J’ = 0.442 ( - V ’ ’ ~ ~ ) - 0 . 3 ~ I Y ’ ’ R125
~ to 5000 (7) MASS TRANSFER AND CHEMICAL REAC-
T I O N I N A TURBULENT BOUNDARY LAYER
For the values of , V S R e below 0.08, the data are inconclusive
and extrapolations are hai:ardous. More experimental data are I n this article b>-\V. R . Vieth, J. H. Porter, and T. K.
needed in this region. I t is expected that a series of lines would Sherivood [IND.ENG.CHEM.FUSD.~MEST.ALS 2 , 1 (1963)], thrre
be obtained for different geometries, solubilities, and Grashof are two typographical errors in the equations on page 2.
numbers. Equation 3 should read (final!!.)
I n Figure 6: the data are plotted as J‘e us. ‘V’’R~. This
method of correlation \vas used by Dryden and coworkers (2)
and brings their data closer to those of Gaffney and Drew ( J ) .
Holvever. the data of thir; work and of McCune and Wilhelm Equation 12 should read (hnally)
(8) are separated from those of Gaffney and Drew by 15 to - Dbl
,p = ~

30y0. Hence, it is recommended that Figure 5 be used. Yo

VOL. 2 NO. 2 MAY 1963 129