Generalized Correlation for Loading in

Packed Towers with Countercurrent

Gas-liquid Flow
J . E. HOWKINS and J. F. DAVIDSON
Cambridge University, Cambridge, England

Experiments in which a liquid film runs over a vertical string of spheres surrounded
by a concentric tube through which air is blown upward have shown that loading in a
packed tower is due to the formation of standing waves on the liquid film. In the ball-and-
tube system a wave is formed just below the equator of each ball, owing to the pressure
gradient within the air stream as it accelerates through the narrowing gap between the
ball and the tube. Interfacial shear and surface tension are of secondary importance. The
similarity between the characteristics of the ball-and-tube system and those of the randomly
packed tower suggests that loading in the latter system is also due to wave formation.
With this concept of loading, a correlation has been d8rived.

The importance of loading and flooding I

.A

in the design of packed towers for counter- 1

current gas-fiquid flow is well known, and Fig. 2. Sectional scale drawing of the ball- I
numerous correlations for the prediction and-tube apparatus.
4
of these phenomena have been proposed
(1 to 4), but none of them are based 0
bi----iJ
1 2 3
SCALE (IN.)
on a clear understanding of the detailed
mechanism of loading and flooding. I n
a previous paper (5)the authors suggested
that loading is due to the formation of
standing waves on the liquid film within
the packing. The present paper gives
further support to this idea, which has
been used as the basis for a correlation t o
predict the loading point with counter-
current gas-liquid flow.
The proposed mechanism involves a
balance between the forces shown in
Figure 1, which depicts an element of
liquid running down the packing in a
region where the gas is accelerating as it
moves upward. The acceleration is due
to contraction of the flow area available
to the gas, and it follows from Bernoulli's
theorem that there must be a pressure
gradient within the gas stream and that
consequently the pressure a t P is greater
than at &, giving a net upward force on

WALL SHEAR. {; GRAVITY ACCELERATING

\ \ '\
pRopoRlloNAL~p GAS
TO I
t PRESSURE \, A B
GRADIENT C
~ i 1. ~F~~~~~
. on an element of the liquid Fig. 3. Liquid paraffin (p== 300 centipoises) flowing over l.~-in.-diarneter balls inside
film of thickness 1. 1.75-in.-diameter tube.
(A) Loading
- (. B, ) (C) Floodinn I

J. E. Howkins is with Standard Oil Company,

Liquid rate cc./min. = 6 6 6
Sarnia, Ontario, Canada. Air rate cu. ft./min. = 4.31 4.87 6.06

Page 324 A.1.Ch.E. Journal September, 1958

 LL
0 0.06-
u
f

a
-4 AIRFLOW G [LB./(HR.) (sQ.FT.)]

Fig. 6. Variation of holdup with air flow. A, B,

C: 1.49-in. balls in 1.75-in. tube with water rates
AXIAL DISTANCE FROM CENTER OF BALL (IN.) of 117, 860, and 3,410 lb./(hr.) (sq. ft.); D (22):
0.5-in. balls (random) in 2.89-in. column with
Fig. 4. The pressure distribution along the water rate of 3,170 lb./(hr.) (sq. ft.) ; E, F (23):
tube beside three successive balls: first, rn ; 0.541. rings in 10-in. column with water rates
second, X ; third, +. of 1,000 and 3,500 lb./(hr.)(sq. ft.)

the element of liquid. As the gravity

force is balanced by the sum of the
pressure gradient and wall shear forces
and since the latter force is inversely
proportional to the film thickness t, the
accelerating gas flow must increase t .
It can easily be seen that there must be a
I I 1 ~ 1 1 1 1 I I I I I I I I maximum gas flow beyond which the film
I L cannot run down, because a t a certain
gas flow the pressure-gradient force will
balance the gravity force and the film
thickness must become very large if the
viscous force is to be small. This kind of
instability is thought to give an explana-
L I tion of the phenomenon of loading and
was investigated by the authors (b),
using an apparatus specially designed for
studying the instability rather than for
simulating the behavior of packed towers.
This paper is a detailed study of what
is thought to be a good, though simplified,
model of a packed tower. The model
consists of a vertical string of spheres
covered by a liquid film and surrounded
by a concentric tube through which gas is
blown upward. At a certain gas flow a
small wave is formed on the liquid film
just below the equator of each sphere.
The forces in Figure 1 account for the
appearance of the waves, which form,
as would be expected, near the point of
maximum gas acceleration and pressure
iI gradient. Measurements of pressure varia-

2! 0.01
100
I , ,
500
I

1000
I I ,,,I I I 1

5000
1.1 I., ,I
tion up the tube, together with theoretical
calculations, make it clear that the wave
formation is due to the pressure gradient
balancing the gravity forces.
AIRFLOW C [LB./(HR) (SQ.FT.)]
That the wave formation corresponds
to loading in a packed tower is clear from
Fig. 5. Variation of mean pressure gradient a study of the holdup and pressure-drop
with air flow. 1.49-in. balls in 1.75-in. tube:
A = no liquid, B = water rate of 117 lb./ characteristics and a comparison of them
(hr.)(sq. ft.); 8.66-in. column with 1-in. with the corresponding characteristics
ceramic rings (9): C = no liquid, D = water for packed towers. This comparison of
rate of 1,250 Ib./(hr.)(sq. ft.), E = water characteristics shows that the ball-and-
rate of 12,500 lb./(hr.)(sq. ft.). tube system is a good model of a packed

Vol. 4, No. 3 A.1.Ch.E. Journal Page 325

tower, The coincidence of the wave measured by recirculating the liquid and formation in the converging parts of the
formation with the loading point for the fixing all the volumes in the system except packing.
ball-and-tube system, as determined from the volume of a graduated tank into which
the pressure-drop and holdup charac- the liquid flowed from the column. Changes Holdup
in level within the tank gave changes in It has been found (10, 1 1 ) for packed
teristics, suggests strongly that loading holdup, and the total holdup was obtained
in a packed tower is due to wave forma- by adding the initial holdup with zero towers that holdup is a function pri-
tion within the interstices. This being the airflow, obtained from data of Cullen (8), marily of the liquid rate up to the loading
case, the dimensionless groups used in who used a weighing technique. point and that above the loading point
the theory for predicting wave formation the holdup increases sharply with gas
are applicable to the correlation of loading Behavior flow. The ball-and-tube system behaves
points. The resulting correlation is not The general behavior of the system is similarly, as shown in Figure 6, where
entirely satisfactory in that i t is semi- shown in Figure 3. Low air rates have a holdup is plotted as a function of air
empirical and for a given packing the small effect on the liquid film and merely flow at constant liquid rates. The point
loading points have to be determined increase the amplitude of moving ripples, at which standing waves are formed is
experimentally at two liquid rates. which are present at all but very low liquid marked by an X . The increase in holdup
This concept of loading due to pressure- rates. At a critical air rate the stationary after loading is sharper for the ball-and-
gradient forces rather than to interfacial waves shown in Figure 3A are formed, the tube system than for the randomly
initial formatiop being just below the packed tower, as is consistent with the
shear accords with the calculations of equator of each ball. With water the wave
Chilton and Colburn (6, 7), who showed breaks into spray as soon as it is formed; idea that loading is due to wave formation
that with single-phase flow in a packed with liquid paraffin the wave is stable and within the interstices. I n the randomly
tower the over-all pressure drop is is blown up above the equator, as in Figure packed tower, wave formation would be
due largely to repeated accelerations and 3A. A t higher air rates the motion shown in expected first in the most serious restric-
that wall friction is of secondary impor- Figures 3B and C is induced, the tube a8 tions, and holdup would increase in these
tance. well as the balls being covered with liquid. regions (in the manner of Figure 3B)
Ultimately flooding is reached, when liquid before wave formation had begun in the
will not run out of the base of the apparatus. other parts of the tower. The loading
APPARATUS AND PROCEDURE process would not be complete until
The ball-and-tube apparatus, shown in RESULTS: LOADING DUE TO wave formation had begun in all parts
Figure 2, was designed to fit into the air WAVE FORMATION of the packing. With the ball-and-tube
distributor and liquid separator described system, loading occurs all at once through-
in the previous paper ( 5 ) . A string of table- Pressure Variation over a Single Ea11 out the tower and the increase in holdup
tennis balls of 1.49-in. diameter was held Figure 4 shows the results from the is therefore sharper.
concentriddly within a Lucite tube A forty-six tappings in the wall of the
of 1.75in. bore. Water was fed onto the 1.75-in.-diameter tube, with 1.49-in.
top ball through the annular gap between THEORETICAL DEVELOPMENT OF THE
the end of the conical nozzle B and the balls and no liquid flow. The results from
GENERALIZED CORRELATION
W-in.-diameter rod carrying the balls. the three successive balls have been made
Adjacent to the liquid inlet at the top of to agree at point A , and their similarity The basic assumptions for the general-
the apparatus was a small outlet tube, at other points shows that end effects are ized correlation are as follows. (See
normally blanked off, which was used to small. I n the accelerating region the Figure 7.)
eliminate air. The water ran uniformly in a results agree, as would be expected, with
thin film over the outside of the spheres and simple theory based on Bernoulli’s
was removed at the bottom by the %-in.- theorem, uniform flow across each hori- n
diameter rod which passed through the
centers of the spheres. Air after being zontal section being assumed. Down-
uniformly distributed passed through the stream there is a small pressure recovery,
Lucite orifice C, up the tube between the followed by a constant-pressure wake in
inner wall and the balls, and out to atmos- the same way as with an infinite stream
phere at the top. flowing past a sphere.
The pressure variation was measured
over the whole column and across three Over-all Pmssum Drop
balls in the middle of a string of twenty.
This number was enough to counteract end Figure 5 shows the mean pressure
effects, as shown by the similar pressure gradient as a function of air flow, both in
profiles across each of the three middle balls the dry ball-and-tube system and with
(Figure 4). The pressure tappings were a constant liquid rate, and the results
formed by forty-six holes of equal diameter, from randomly packed towers given by
drilled at 0.1-in. pitch in a straight line Sarchet (9) are also shown. The following
down the tube. Because of the sharp points of similarity may be noted:
variations in pressure, the final tapping was
0.012 in. in diameter. The holes were
normally plugged, but, when a measure-
ment of the pressure was required, a plug
1. With no liquid flow, In AP is a
linear function of In G, the slope being
about 1.9 for the ball-and-tube system
f t U,t t
Fig. 7. Ball and tube with liquid film.
was replaced by the tapping D, which was and for the randomly packed tower, where
connected to a Chattock gauge measuring AP is the mean pressure gradient and G
the pressure to &l%.I n Figure 2 the top is the superficial gas rate.
tapping is ready for use and the forty-five 2. With liquid flowing, the relation 1. The viscosity of the air is neglected
holes below are plugged. between AP and G is similar in both and its velocity across any horizontal
Balls and tubes of other diameters systems. The loading point is followed section is assumed uniform. This assump-
(without pressure tappings) were used to in each case by a line of increased slope. tion is justified by the agreement between
investigate the effect of size and voidage on For the ball-and-tube system i t is known theory and experiment in Figure 4. From
the loading velocities. Orifice C was changed, continuity
but the general layout of the apparatus from visual observation that the loading
was the same ae in Figure 2. Liquid paraffin point X is coincident with the wave 7 7

was used in place of water t o find the effect formation shown in Figure 3 A ; it there- U = uo
d2 (1)
of higher viscosity.
Liquid holdup on the 1.49411. balls was
fore seems likely that loading in a ran-
domly packed tower is due t o wave
1 -D COS’ e - 4dt
o” cos e
Page 326 A.1.Ch.E. Journal September, 1958
From Bernoulli's theorem TABLE1. LOADING
CORRELATION
[Equation (9)
LY=
Reference A B 1.195A
2. The liquid-film thickness is calcu- (1 - €)"2(3€ - l)'l'
lated from a balance between the forces
in Figure 1. The velocity is assumed to be Ball and tube
a parabolic function of distance measured Ball Tube
normal to the surface, and since the diam., diam.,
surface of the ball is nearly vertical in. in.
where the wave is formed, cos 8 1. 1.49 1.75 0.032 26 0.51 0.27
The film thickness is therefore calculated 1 .oo 1.22 0.058 14 0.55 0.31
by neglecting the kinetic energy of the 0.50 0.63 0.064 5.9 0.58 0.25
liquid in the same way as in reference 5 : 1.96 2.50 0.066 21 0.60 0.23
1.49 2.00 0.11 21 0.64 0.26
1.49 2.50 0.16 21 0.77 0.20
Quartz, 2 in. 0.015 63 0.46 0.27
Serrated grids
From Equation (3) it is clear that 2 in. 0.16 77 0.83 0.18
- d p / d z cannot exceed p L g , and in the 1.5in. 0.10 41 0.89 0.10
previous paper (5) it was shown that Stacked rings
wave formation occurs when 3-in.stoneware 0.058 31 0.76 0.075
Random rings
0.5-in.stoneware 0.024 16 0.61 0.075
-dp = apLg (4) 1-in. carbon 0.018 23 0.66 0.039
dz 1-in.carbon 0.032 22 0.70 0.055
1 .Sin. stoneware 0.028 28 0.72 0.044
where a is less than unity and varies 1 .@in. stoneware 0.032 25 0.73 0.048
from system to system. a can be predicted 1 .@in. stoneware 0.030 19 0.80 0.035
theoretically by relaxation methods (5), 0.Sin. metal 0.024 13 0.87 0.024
but in the present paper it will be left as 2.0-in.metal 0.033 31 0.92 0.034
an unknown parameter.
To give dp/dz a t the wave crest,
Equation (1) is differentiated, with the
simplification that dt/dz = 0 a t the wave This equation can be generalized by this occurs it is necessary to put
crest. In Equation (1)it is also permissible writing d and D in terms of the voidage, dp/dzz = 0. Using Equations (l), (2),
to put t C O S ~8 2 t.since the wave is c, and a, the wetted area per unit volume: and (6), with t = 0 and 22 = -d Sin 8,
formed near the equator, and using the gives the angle for maximum pressure
resulting expression for dU/dz with gradient:
Equations (2) and (4)gives the condition
for loading: 3e - 1
sin2 6 = 15(1 - e) (7)

D ~ ( I-
4U02dsin 0
aCOs a
Dda e --
a = D4d2 J This equation gives the angle for maxi-
mum pressure gradient in the absence of
D2
4dt>" The wave will be formed where -dp/dz the liquid film and will therefore be
is maximum, and to find the angle where applicable only a t very low liquid rates.
--
- (5)
ffpLg Eliminating t, d p / d z , 8, d, and D from
PO
Equations (3) to (7) gives the final
condition for loading:
-
ORDINATE 0.19

=I----- 5

This may be written in the form

0 06
where A and B are functions which are
given in temis of the packing dimensions

q+
004
by Equation (8) for the ball-and-tube
system but have to be determined experi-
6.02
mentally for other packings. The area a
0 0
0
002
002
.0 0 4
.0 0 4
006
006
0.00
0.00
010
010
01'2
01'2
014
014
016
016
. 0.18
has not been included in the constants
A and B to keep them dimensionless.
($)+
($)i
APPLICATION OF THE CORRELATION
Fig. 8. Determination of A and B, Equation (Q), for the ball-and-tube systems. In each
circle the upper and lower numbers are the ball and tube diameters. The left-hand symbol To get A and B in Equation (9),
is for air-water; the other is for air-liquid paraffin. Gpa/pOpLg a t loading is plotted as a

Vol. 4, No. 3 A.1.Ch.E. Journal Page 327

 I I I function of (Lu2p/pL2g)”3 and -4 and B
are determined from the best straight
21 BALL & TUBE. 50; line through the points. Figure 8 shows
,
ALL
y BALL TUBE 5% thr results for the ball-and-tubc system,
$4 DIA.- DIA.- %4 and similar graphs were plotted by
:$ IN. IN. $ 2
STACKED R I N G S means of the data for various packed
A 1.49 1.75 (14) 3 in STONEWARE towers of Shulman, Cllrich, and Wells
v 1.00 1.22 RANDOM RINGS (13) and Morris antl ,Jackson (14).
+ 0.50 0.63 (13) 0 5 in STONEWARE
Table 1 shows the resulting values of d
T 1.96 2.50 (14) I in CARBON
/’ 1.49 200 (13) I in CARBON and B , which m-ere used to replot all the
I 1.49 2.50 (13) 1.5in STONEWARE results in Figure 9. The points from
QUARTZ 2in. (13) 1.0 in STONEWARE referencLes 13 and 1G
. were selerted from
SERRATED GRIDS (14) I.O in STONEWARE
2in. O 5 i n METAL smoothcd experimental data as the
2.0in METAL original data were not available. I t is not
certain whether the original data would
give a better or worse scatter than is
shown in Figure 9.

Validity of the Correlation

In Figure 8 the agreement between the

results with water and liquid parafin
for the same system is satisfactory except
in the case of the 0.50-in. balls, where the
poor correlation is thought t o be the
effect of surface tension, which was not
allowed for in the derivation of Equation
(9) and is magnified with smallcr packing.
The agreement between experiment
and theory in Figure 9 indicates that the
niechanism of loading in the ball-and-
tube system is similar to thc mechanism
in the other packings.
A further check on the validity of the
theory is given in Figure 10, which is
similar to Figure 9, but with different
values of A and B for each liquid. To get
I I 1 I I I I I I
\ Figure 10, Figure 8 was redrawn, the
0.1 0.2 03 0.4 0-5 0.6 07 0.8 09 i.0 behavior of each ball-and-tube system
bring represented by two straight lines,
one for water and one for liquid paraffin.
Essentially, Figure 10 is a plot of G2
against L“3, since A , B, a, p L , pot and p
Fig. 9. Generalized loading correlation.
are fixed for any one sgstclrn, and i t
shows very good agreement with theory.
---
The fact that Figure 10 is a better corre-
lation than Figure 9 indicates that the
constants A and B are not entirely
independent of liquid viscosity for a

-1
0 3111STONEWARE particular system. This is further indi-
RANDOM RINGS cated in Figure 8, where for any ball-and-
> 0 5111 STONEWARE
tube system the results are somewhat
0 I i n CARBON
A I In CARBON scattered around the fitted straight line;
V I Sin STONEWARE
Figures 8 and 9, nevertheless, correlate
< I Oln STONEWARE
8 I. O m STONEWARE the results reasonably well and justify
6 0 5!n METAL
the assumption that, to a first approxi-
@ 2 0 i n METAL
~

mation, A and B are indrpendent of

liquid viscosity for each packing.
The theory can also be checked by
calculating a, which is a meamre of the
maximum pressure gradient within the
- IQ. 0.31 packing, in terms of liquid head. It
should be somewhat less than unity, and
its significance will be discussed later,
with reference to interfacial shear. a can
*%\ be calculated for the ball-and-tube
a, \><
0\ system, because in this case Equations
\+\ 0 +
(8) and (9) should be identical. A and B
L-!
0.1 0.2 0.3 04 0.5
I
0.6 0.7
I
oa
I I
0.9
u
1.0
d
1.2
1.1
can therefore be related to a :tnd E, and,

B (9): the latter being known, two 1-alues of a

are calculable, one from A antl one from
B. I n the derivation of Equations (7) and
Fig. 10. Nongeneralized loading correlation. (B), however, it was assumed that the

Page 328 A.1.Ch.E. Journal September, 1958

liquid-film thickness t was negligible. It 4. Voidage E. The effect of increasing NOTATION
would therefore be expected that the voidage is to move the point of wave
constant A , which depends upon the A = packing constant
formation. Table 2 shows the angle below
loading-gas rate a t L and t being equal the equator for maximum pressure
to zero, would give reasonable values of a. gradient, calculated from Equation (7).
B, which gives a measure of the variation It was difficult to observe the position of for the ball-and-tube system
in the loading-gas rate due to finite the initial waves on the balls, because as a = interfacial area, sq. ft./cu. ft.
liquid-film thicknesses, would not be soon as the waves were formed they were B = packing constant
expected to give a reasonable estimate of pushed above the equator as in Figure 3A.
a. These aqsumptions are confirmed by
calculation; Table 1 shows values of a‘
calculated from the experimental values
At very high voidagcs, however, a wave
formed on the lowest ball of the series a t
a point well below the equator, as would
-
-
[”I
(I - a)f 3E - 1
for the ball-and-tube system
of A , but values of a calculated from B be expected from Table 2; a detailed D = tube internal diameter, ft.
are nearly unity and in one case negative d = ball diameter, f t .
and have riot been included. G = superficialgasrate, lb./(hr.)(sq.ft.)
g = acceleration of gravity
TABLE
2. ANGLEFOR WAVE FORMATION
Factors Affecting loading L = superficial liquid rate, lb./(hr.)
1. Liquid Viscosity. The viscosity of [Equation (7)] (sq. ft.)
the liquid paraffin was about 300 times AP = mean pressure gradient, in. water/
E 0.333 0.4 0 . 5 0 . 6 0 . 7 0 . 8
that of water, and the agreement obtained edegrees 0 9 15 21 30 43 ft. height
between the two sets of results for the p = pressure, lb./(ft.)(hr.z)
ball-and-tube systems indicates that the t = liquid-film thickness, ft.
final correlation, Equation (9), could be U = gas velocity a t any section, ft./hr.
used for an approximate prediction of treatment of this special case is given UO = gas velocity in the empty tube,
the behavior of randomly packed towers elsewhere (15). It is interesting to note ft./hr.
with any liquid when the air-water results that in the present experiments the x = axial distanre, ft.
are used. This method could not be 1.49411. balls in the 2.50-in. tube loaded
Greek letters
applied to small packings. by wave formation on the bottom ball
2. Interfacial Shear. The approximate with air and water, and that therefore the a = fraction defined by Equation (4)
dependence of A and B on E and the resulting points on Figure 9 have a large E = voidage
relative constancy of LY (Table 1) indicate deviation from Equation (9). 0 = angle indicated in Figure 7
that the neglect of interfacial shear in p = liquid viscosity, lb,/(ft.)(hr.)
deriving Equation (8) is justified. As pG = gas density, lb./cu. ft.

was explained above, a may be thought CONCLUSIONS pL = liquid density, lb./cu. ft.
of as the maximum pressure gradient
within the packing in feet of liquid per 1. Loading in a packed tower is caused LITERATURE CITED
foot of height and cannot exceed unity. by the formation of standing waves on
the liquid film a t points within the packing 1. Sherwood, T. K., G. H. Shipley,
In reference 5 a was about 0.6 and the and F. -4.L. Holloway, Ind. Eng.
difference from unity was attributed where the pressure gradient, due to
acceleration of the gas, is maximum. In Chem., 30, 765 (1938).
partly to interfacial shear and partly to 2. Bain, W. A., and 0. A. Hougen,
interaction between the water film and a randomly packed tower loading is
Trans. Am. Inst. Chem. Engrs.,
. . 40,.
the air stream. I n Table 1, a is about thought to be a gradual process, the 29 (1944).
wave formation occurring over a small 3. Dell. F. R.. and H. R. C. Pratt. J.
0.25 for the ball-and-tube system,
and simple boundary-layer calculations range of air flows, although the evidence A p p i . Chem.’ (London),2, 429 (1952).
showed that the lower value is due to for this is not direct. 4. Hoffing, E. H., and F. J. Lockhart,
interfacial shear. . 2. -4t largc voidages loading will start Chem. Eng. Progr., 50, 94 (1954).
a t the bottom, and the practice of putting 5. Davidson, J. F., and J. E. Howkins,
Strictly, it is not possible to calculate Proe. Roy. SOC. (London), 240A, 29
a for the other packings, since its relation
larger, and possibly graded, packing a t
the bottom is a good one. (1957).
to A is not known. However, the values 6. Chilton, T. H., and A. P. Colburn,
in Table 1, calculated from the ball-and-
3. Interfacial shear has a secondary
Ind. Eng. Chem., 23, 913 (1931).
tube formula, indicate interesting trends. effect on loading and is most important 7. - , Trans. Am. Inst. Chem. Engrs.,
The serrated grids show, as might be a t large voidages. 26, 178 (1931).
4. Surface tension has a secondary 8. Cullen, E. J., Ph.D. thesis, Univ.
expected, the greatest similarity to the
effect on loading, although it is quite Cambridge, England (1956).
ball-and-tube system, For the random
important for small sizes of packing. 9. Sarchet, B. R., Trans. Am. Inst.
paekings it is not clear whether the low Chem. Engrs., 38, 283 (1942).
5. Equation (9), which is a linear rela-
values of a are due to the packing geom- 10. Leva, Max, “Tower Packings and
tion between the dimensionless velocity
etry or to interfacial shear; however, the Packed Tower Design,” p. 35, United
head in the gas and the dimensionless
success of Equation (9) in correlating all States Stoneware Co., Akron, Ohio
liquid-film thickness, is suitable for corre-
the results indicates that shear is of (1953).
lating loading points. The correlation is 11. Piret, E. L., C. A. Mann, and Thomas
secondary importance in randomly packed
semiempirical, because constants A and Wall, Ind. Eng. Chem., 32, 861 (1940).
towers.
3. Surface Tension. I n Figures 8 and 9, B have to be determined for each packing 12. Elgin, J. C., and F. B. Weiss, Ind.
by measuring the loading point a t two Eng. Chem., 31, 435 (1939).
the results for the 0.50-in. balls are
liquid rates. The loading point a t other 13. Shulman, H. L., C. F. Ullrich and
extremely scattered, probably because of N. Wells, A.I.Ch.E. Journal, 1, 247
liquid rates, and for different gases and
surface tension, which would affect the (1955).
liquids, can then be predicted.
loading point by stabilizing the wave and 14. Morris, G. A., and J. Jackson, “Ab-
by altering the shape of the interstitial sorption Towers,” pp. 24, 25, 38, 39,
meniscus. Both these effects would be Butterworths, London, (1953).
magnified for smaller packings and would ACKNOWLEDGMENT 15. Howkins, J. E., and J. F. Davidson,
be different for water and liquid paraffin, J. E. Howkins wishes to acknowledge the Chem. Eng. Sci., 7, 235 (1958).
the surface tension of the latter being support given by a Gas Council Scholarship Manuscript received Aun. 1 4 , 1957; revised Jan.
approximately 30 dynes/cm. to enable him to carry out this work. $0, 1958; accepted Feb. 6 , 1958.

Vol. 4, No. 3 A.1.Ch.E. Journal Page 329