Sunteți pe pagina 1din 10

ENGINEERING. DESIGN.

AND PROCESS DEVELOPMENT

Liquid-Liquid Extraction Spray Columns


Drop Formation and Interfacial Transfer Area
FREDERICK W. KEITH, JR.~, AND A. NORMAN HIXSON
Universifpof Pennsylvania, Philadelphia, Pa.

ASSAGE of droplets of a liquid through another part,ially or


totally immiscible liquid due to a difference in density is a
familiar phenomenon. The process is used in both industrial
and laboratory spray columns for countercurrent liquid-liquid
extraction. Design of such equipment requires a knowledge of
flow rates, concentrations, and the rate of material t,ransfer for
the system. The last factor depends on the diffusion and convection conditions of the system and on the interfacial area available for transfer; it is a t present necessary to determine these factors by scaling up 1aborat.ory data or by making a rough estimate
on the basis of a very similar system operating on a large scale.
Such approximations are moderately successful only for small
differences in diffusional Characteristics (26). Few quantitative
data are available to show the effect on transfer area of changes
in flow rates, number and size of nozzles, and physical characteristics of the solvents and the solute.
The ultimate effect of changing these physical variables is a
change in average drop size of the dispersed phase. The transfer
area depends on holdup, rate of drop formation, and the average
surface area per drop. These factors, in t a n , are functions of
drop size, rate of rise of drops, and the distortion and extent of
coalescence of drops. It has also been shown t h a t transfer is
dependent on convection as well as diffusional forces and that the
proportion due to convection is affected by drop size (28, 40).
Unusually large material transfer has been found a t the dispersed phase inlet (40) or outlet ( 1 4 ) ; these end effects are also
partially dependent on drop size and number.
Very little information is available on the formation of drops,
their sizes and shapes, and the conditions associated with their
mot,ion through the ambient fluid in a liquid-liquid system. As
a basis for considering these data, it is necessary to outline the
visible changes t h a t occur as the flow rate through a nozzle is increased. At low flows, drops form individually a t the nozzle tip
and grow in size until the buoyance force overcomes interfacial
tension and the drop is released. At increased flow rate, a point
is reached where a very short continuous neck of liquid exists
between the nozzle tip and the point of drop detachment; this
velocity will be called the jetting point. Further increases
rapidly lengthen the jet, which appears as a smooth column of
liquid with occasional transient lumps. Finally, the jet takes on
a ruffled appearance a t its outer end and the drops formed are
less uniform than in the earlier stages; this occurs a t or near
the maximum length of the jet and will be termed the critical
velocity. Increasing the flow rate further decreases the jet
length and increases drop nonuniformity until the jet breakup
point retreats to the nozzle tip and a nonuniform spray of rather
small drops results; this last point will be called the disruptive
velocity. Most of the recorded work on jets has dealt wit,h factors affecting jet length ( 3 , 7 , 10, 11, 38,@) for liquid-into-air jets,
both upward and downward. Smith and Moss ( 4 2 ) showed t h a t
the critical velocity occurred when eddy turbulence overcame the
restraining forces of interfacial tension. Merrington and
Richardson (52) referred to the range of velocities between the
1

Present address, The Sharples Corp., Philadelphia, Pa.

258

jetting point and the critical velocity as the varicose region because of the occasional lumps in the jet; between the critical
and disruptive velocities lay the sinuous region, as identified by
irregular weaving of the jet. These characteristics are not always obvious in liquid-into-liquid jets, but the terms will be retained for identification.
A number of investigators have studied drop size resulting
from the breakup of liquid-into-air jets ( 1 1 , 2 2 ,52, S S , S 7 ) and gasinto-liquid jets (15, 27, $1) Some papers on extraction spray
towers have made qualitatiye note of drop sizes ( 4 , 26,40). Hayworth and Treybal(20) studied drop size in liquid-liquid systems
for a static continuous phase a t flow rates mostly below the jetting point. A correlation for drop diameter was developed for
these systems The recommended range of rates for this correlation, however, is too low for practical application in most spray
towers. Photographs of their flow range were presented and
also some data for the largest drops at higher flow rates, but the
results are too scattered to fix a peak condition. Several studies

t-L- Gapillary
air leak

4
f
* 6
-tc

t o vacuum

Capillary tubing:
7.39 mm. O.D.
2.1 mm. I.D.

Stopcock

6.5 cm.
~

?.

t-L- 14.835 cm3

L1.0mm. bore
glass copillary
glass vent tube

35cm.

~~/-Thin-walled
I/z in.
Tyg on tubing

End
rough

6. 75 mm.

DETAIL OF
DROPPING TIP

I%Dropping
Figure 1 .

tip

Apparatus for measuring interfacial tension

INDUSTRIAL AND ENGINEERING CHEMISTRY

Vol. 47, No. 2

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT

Table 1.
Solvent
Isoamyl alcohol
Benzene
%-Butylalcohol
n-Butyl chloride
Cyclohexane
Diisopropyl ketone
Ethyl acetate

Sources and Boiling Ranges of Solvents


Source

Paragon
Paragon
Baker
Paragon
Paragon
Eastman
Phillips & Jacobs

Quality
Purification
Technical
Distilled
Thiophene-free
None
Analyzed
None
Reagent
None
Reagent
None
Practical
Distilled
Commercial
Distilled

Baker
Phillips & Jacobs

Commercial
Technical
Petroleum
Practical
Reagent
Analyzed
Commercial

ketone

Table II.

Distilled
None
Distilled
Distilled
None
None
Distilled

Boiling 9 n t
Range, C.
131.3-132.3
80.1
116 -118
77 - 78
80.6- 80.9
123 -124.5
Azeotropic
at 70.4
35 - 36.5
144.2-147.1
66.2- 69.2
114.9-116.0
67 - 69
110 -111
139.3-141.3

from laboratory extraction tower results


and to permit more accurate estimation of the effect on transfer efficiency
of various changes
- in the Dhvsical characteristics of extraction columns and
their flow systems. The investigation
was limited to the use of organic solvents lighter than mater as the dispersed
phase and water as the continuous
phase. Solutes were used in mut,ually
saturated systems only for their effect
on the phy.sica1 properties and not for
mass transfer area; this is similar to
t h e a p p r o a c h of H a y w o r t h a n d
Treybal (20).
-

Physical Properties of Phases from Tower Runs


Physical properties are determined
for 12 water-organic solvent systems

Solvent Phase

Dynes/Cm.

Isoamyl alcohol
%-Butylalcohol
n-Butyl chloride
Cyclohexane
Diisopropyl ketone
Ethyl acetate
Ethyl ether
Ethylhexanediol
%-Hexane
Methyl isobutyl ketone
Isopropyl ether
Toluene

5.1
1.8
24.2
40.7
16.4
4.5
10.5
4.1
40.6
10.7
18.3
35.6

p0

0.8242
0.8427
0.8784
0.7723
0.8031
0,8988
0.7103
0.9445
0.6746
0.8006
0.7225
0.8606

pW

0.9929
0.9865
0.9957
0.9968
0.9963
0,9922
0.9891
0.9967
0.9971
0.9950
0.9949
0.9969

fig

3.566
2.740
0.428
0,877
0.616
0.476
0.235
79.23
0.320
0.576
0.324
0.549

fiw

0.991
1.191
0.924
0.895
0,909
1.203
1.021
1.046
0.891
0.933
0.930
0.894

For solvents that are not available commercially in reasonably


pure condition, a technical or commercial grade was used. I n
order to avoid, as far as possible, the interference of any surface
active impurities, these solvents were water-washed and then
distilled t o remove low and high boiling cuts. I n no case was
more than about 5% of the original solvent eliminated. For
solvents that are commercially available in satisfactory purity a t a
reasonable price, the reagent grade was generally used. Laboratory distilled water was used throughout the investigation.
There were two exceptions to this procedure. Ethylhexanediol
was not distilled although a techniral grade was used. To obtain
the high value of interfacial tension desired, it was necessary to
use a reagent grade cyclohexane.
The solvents tested are listed in Table I with the source, grade,
and boiling range. The physical properties of these solvents in
mutually saturated water-solvent systems are listed in Table 11.
The solutes used were C.P. 99.5% glacial acetic acid Baker analyzed, reagent benzoic acid (Mallinckrodt), sucrose (Paragon), and
Alkaterge C (Commercial Solvents Corp). Solutions of Alkaterge C were prepared on an approximate volumetric basis ae
the surface activity of such a material is probably not strictly
reproducible. The physical properties of the toluene-water system containing solutes are shown in Table 111.
The drop volume method wag used to measure interfacial tension because i t seemed comparable to other methods in accuracy
when prolonged aging of the surface was not under study ( 1 , 8,
24, 36). The method of calculation and the table of correction
factors of Harkins and Brown ( 1 7 )were used.
The data of Addison on migration times ( 2 ) caused some concern since they indicated uncertainty as to the correct value of the
interfacial tension that should be used when the jet varied in
length with flow rate. These data, however, also showed that
the time required to reach surface equilibrium for pure solvente
was extremely small, so that static interfacial tension as determined by the drop volume method would seem to apply to jet

on extraction a t very low flow rates have determined drop sizes


(13,50,40, 44).
A number of theoretical and experimental results in liquidliquid systems are available for rates of drop rise and the corrections required for the usual laws of terminal velocity of particles
(6, 8, 9, 10, 12, 15, SO, 44). The breakup of drops exceeding a
critical size at a related critical rate of drop rise has also been
treated experimentally ( 3 2 ) and theoretically ( 2 1 ) . Several investigators have noted the flattening effect on drops during
motion (16, 21, do), while qualitative observations have also
been made on coalescence in extraction towers ( 4 , 6 , 26). Farmer
( 1 9 ) made precise eccentricity measurements on a few drops, but
the data are too limited for average results. Johnson and Bliss
( 2 6 ) found that spray tower efficiency was increased by avoiding
nonuniform drop sizes in the dispersed phase. Plate-type distributors tend to give irregular drop sizes, as noted by Blanding
and Elgin (6); a picture by Laddha and Smith ( 2 9 ) shows the
nonuniformity of drop sizes from a hemispherical distributor a t
low throughput.
Sherwood, Evans, and Longcor (40) gave the first quantitative
data on calculated interfacial areas and their effect on transfer
coefficients for single drop transfer. They found that application of a calculated area term resulted in a coefficient showing an
increase in transfer with drop size, as would be expected because
of the increased internal convection.
Other data (15) show little effect of
drop size on the coefficient or show the
Table 111. Physical Properties of Toluene-Water System with Solutes
coefficient becoming constant above a
certain drop size.
Concentration,
Interfacial
Grams/c~.
Density
Viscosity,
The investigation described in this
Wt. %
Tension
CP.
Solute
Toluene
Water
Dynes/CA.
po
pw
l"0
fiW
paper
was undertaken to determine
_
Sucrose
Negligible
40.4
34.2
0.8592
1.1784
0.547
5.55
quantitatively the factors affecting
Sucrose
Negligible
17.2
36.3
0.8594
1.0686
0.546
1,550
26"
Negligible
9.2
0.8631
drop size in liquid-liquid systems and
Alkaterge C
0.9971
0.555
0.918
Alkaterge C
6.5"
Negligible
15.4
0.8620
0.9970
0.552
0.910
to obtain data on other factors contribAlkaterge C
l.05
Negligible
25.3
0.8610
0.9970
0.550
0.902
Acetic acid
0.52
9.36
21.7
0.8614
1.0133
0.549
1.101
uting to the calculation of interfacial
Benzoic acid
4.00
0.21
24.3
0.8689
0.9970
0.576
0.890
transfer area. The results can serve
Benzoic acid
6.75
0.29
21.8
0.8760
0.9974
0.607
0.898

two purposes-to allow calculation of


more fundamental transfer coefficients
February 1955

Drops/IOO co. of toluene.

INDUSTRIAL AND ENGINEERING CHEMISTRY

259

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT


breakup a t any length. The case for systems containing solutes
is different and wiIl be treated in a subsequent paper.
T h e interfacial tension apparatus was adapted from t h a t of
Powney and Addison (36)and is shown in Figure 1. I n operation
the stopper was inserted in a 250-cc. wide-mouthed extraction
flask almost filled with equal quantities of t h e phases of t h e liquidliquid system being tested. T h e whole apparatus was immersed
to the stopcock in a constant temperature bath. With the Tygon
tubing acting as a water tight flexible bellows, the calibrated
bulbs could be filled as often as necessary without removing the
apparatus from t h e bath. A run consisted of counting the number of drops formed during emptying of a calibrated bulb. The
rate of drainage was controlled by t h e size of the capillary air
leak. The liquid preferentially wetting the glass tip should be
used as the drop-forming phase.

Figure 2.

Apparafus

for drop size measurements

9 preliminary study on the benzene-water system showed that


a drop formation time of 3 minutes minimum gave reproducible
results t h a t checked well with t h e literature. For solvents of
lower interfacial tension, 2 minutes was sufficient, while as long
as 6 minutes was needed for higher tensions. Except for oc. . . ,
,
casional check runs the smaller bulb was usea tnrougnout t h e
study. The dropping tip was prepared to meet the requirements
of Harlrins and Brown. They recommended a 9- to 11-mm. tip
diameter for liquid-liquid systems, but a 6.375-mm. tip was used
t o give a greater number of drops and thus reduce the error a t the
start and finish of a run Values of interfacial tensions for unknown systems may frequently be estimated from the literature
for systems without solutes ( 5 , 18, 26, 35, 39, 4 1 ) and with
solutes ( I S , 19, $4).
Viscosities n-ere measured in standard Ostwald-Cannon-Fenslre
viscosimeters calibrated on water, acetone, carbon tetrachloride,
benzene, and glycerol. Densities of the phases were determined
in capped pycnometers. All measurements of physical properties
.>

260

1,

.1

were made on mutually saturated phase mixtures, including


solutes, at 25.0 i 0.1" C. I t is estimated t h a t the interfacial
tensions were accurate to k 0 . 2 dyne per em., t h e densities to
10.0003 gram per cc., and the viscosities t o about 1 0 . 0 0 4 cp
Drop size measurements are made on
dispersed solvent phase from single nozzle

The layout and flow diagram for the apparatus used for the
measurement of drop sizes is shown in Figure 2.
The column itself, E, was a glass tube 3.75 cni. (1.44 inches) in
inside diameter. For the majority of the runs the active distance
between stoppers was 87.0 em. (20.2 inches) or 118.4 cm. (46.6
inches), but there was no apparent effect of column length on the
drop measurements. The bottom was a neoprene stopper with
the dispersed phase inlet a t its center and the continuous phase
outlet to the side. The top was a cork stopper through which
passed the dispersed phase overflow and the continuous phase
inlet; the latter consisted of five 3/82-inchholes in 9-mm. glass
tubing bent a t right angles t o provide horizontal motion of the
inlet water phase. Starting 10 cm. above the longest nozzle, a
50-cm. length of column was marked off in IO-cm. lengths for
use in counting drop frequency. Various nozzles were connected
by short pieces of neoprene tubing to the solvent inlet above the
bottom stopper; a wire spider near the end of the nozzle ensured
centering of the jet. Neoprene tubing in contact vrith the solvents
was extracted with acetone and benzene prior to use.
Solvent, stored in a 2-liter flask, D, was fed to the column by
controlled air pressure in the flask. A pinch clamp on a short
piece of neoprene tubing provided a shutoff from the tower, but
it was not used for flow control and was wide open during operation.
Compressed air was metered to the system through a globe and
a needle valve in series and passed through a trap, A . A 1-gallon
bottle acted as a pressure reservoir and surge tank. The pressure
to the feed flask, D ,was controlled by an adjustable leg of glass
tubing in about a 2-foot column of acetylene tetrabromide; the
lower end of the glass leg was drawn to a fine nozzle t o give a
more even flow of air. The top of this pressure leg column was
closed and a saran tube led the excess air to a scrubbing flask
filled with the same solvent as the dispersed phase. From here
it was led to the bottom of the 2-liter flask, K , used for colIecting
the dispersed phase overflow. As the solvent level in the feed
flask, D,was lowered during operation, the increasing head in
flask K provided most of the necessary increase in back pressure
to maintain a constant flow to the nozzle. This setup worked
very well and gave constant flow rates with excellent control,
A wide-mouthed half-gallon bottle, G, served as a reservoir for
the continuous phase. Liquid was pumped, H , to the I-liter
flask, J , which served as an adjustable-level constant head feed
tank. The flow was from this tank to the top of the column,
while the continuous phase was down through the column,
through the interface level control, and bark to the reservoir, G .
A cooling coil in the reservoir was adequate t o hold the temperature a t 25 f 2" C. A two-way 120" stopcock placed after the
interface-level control permitted determination of the continuous
phase flow rate.
A short mercury manometer, F , open to the atmosphere, indicated the static pressure head a t the bottom of column E , A
longer water manometer approximately indicated the pressure
drop across the nozzle plus a small length of inlet tube (the latter
was negligible).
Photographs were taken of the tower during operation with
an Eastman Kodak Model B. S , with portrait attachment So.
8, as shonn a t L. The camera was on a level with the center of
the column and a t a distance of 33 inches. A strip of nonreflecting white paper pasted t o a board the length and width of the
column and placed 1/4 inch behind the column was a satisfactory
background. This could be adjusted with respect t o the lighte
and camera t o give a dark crescent on both sides of each drop.
Nozzles. The nozzles were made from pieces of capillary or
thin-walled glass tubing. The tip was cut! a t right angles to the
axis and ground, when necessary, to remove nicks and chips.
The inlet end was sealed to a 1-inch length of 7-mm. thin-walled
tubing to fit the standard neoprene connection a t the bottom
stopper. For most of t h e runs the length of t h e main part of the
nozzle was 3 inches. Nozzle A was also made in lengths of 2, 1,
and 0.15 inches; nozzle G was also made in a 1-inch length. The

INDUSTRIAL AND E N G I N E E R I N G C H E M I S T R Y

Vol. 47, No. 2

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT


diameters of the nozzles were measured with a calibrated microscope scale. D a t a on t h e nozzles are given in Table IV. Nozzles
are referred t o by letter and length in inches-that is, a 3-inch
length of 2.77-mm. (inside diameter) capillary is nozzle A-3.

Table IV.
Nozzle
Designation

A
B
C
D
F
G

Measurements of Glass Nozzles


Inside Diameter
Inch

Mm.
2.77
2.12
1.15
0.71
4.33
6.66

Outside

Diameter, Mm.

0.109
0.0835
0.0453
0.0280
0.170
0.262

8.94
7.39
6.32
5.39
6.58

..

NOZZLE SIZE ON

.-

D w e r h e d Phose Flow

R o t e , Uy ( c m ' / s a e )

Figure 3

Experimental Procedure. Prior to use, solvent and water were


kept in mutual contact with frequent shaking a t or near 25" C.
for about 48 hours. The only preparation required for the column
was t o insert the proper nozzle. I n starting up, the aqueous
continuous phase was circulated, the rate adjusted, and the
temperature regulated at 25 f 2' C.by t h e coil in t h e reservoir.
The dispersed solvent phase was quickly passed through the
tower once to bring it to t h e temperature of t h e continuous phase;
it was then returned to the feed flask and the run waa started.
The first run of a series was made a t the jetting point, the second
below the maximum jet length, t h e third near or a t the maximum,
the fourth above t h e maximum, and the fifth a t a very high rate.
When the jet length was set and the interface had reached equilibrium just below the continuous phase inlet, the flow rates were
checked, a picture was taken, t h e jet length was checked, and the
uniformity of drop size was noted. The rate of drop rise for
5 t o 20 drops of representative sizes was determined over a 40or 50-om. length of column and t h e temperatures were checked.
The photographic film was developed but not printed as it was
found that i t was easier t o count the drops by transmitted light.
For most runs the number of drops over a 40- or 50-cm. length of
column was counted directly from the negative with the aid of a
jeweler's eyeglass. For more complete data the negative was
projected on a screen; horizontal and vertical diameters of all the
drops in a representative section of column were measured and
an enlargement factor was calculated from a marked length of
column on the screen. The results of the projected counts agreed
well with the direct counts and the error of this measurement was
probably about 3%
The average drop volume was calculated from the average
rate of drop rise, the number of drops, and the dispersed phase
flow rate. By assuming a spherical drop, the drop diameter,
drop surface area, surface area per unit time, and surface area
per unit volume of dispersed phase could be calculated. The
full results of the 570 runs made are on file with A. N. Hixson
at the University of Pennsylvania.
February 1955

Each system exhibits maximum interfacial transfer


area at a particular flow rate of dispersed phase

The most easily observed effect of increasing the flow rate


through a nozzle was t h e varying length of the jet. It was very
small at the jetting point, increased to a maximum, and then became shorter. The length of the jet was also affected by the flow
rate of t h e continuous phase; it was longest with no flow and
became shorter with increases in rate. This pattern was found
for all t h e liquids tried and for all nozzle sizes. Results for a
range of nozzle sizes for the toluene-water system are shown in
Figure 3 for a continuous phase flow rate of about 100 cc per
minute. The almost linear variation below the maximum and
the rather sharp maximum seemed t o be characteristic of t h e
phenomenon except for the largest nozzle.
The average drop volume passed through a rather sharp minimum when plotted against the dispersed phase flow rate. T h e
form of this curve was nearly a V, as shown by thorough tests a t
the start of the investigation. I n later runs where the number
of tests was not so large, i t often occurred that t h e flow rate giving
the exact minimum was bracketed but not hit. I n this case the
minimurn was readily determined with only a small uncertainty
by graphical interpolation of t h e available data. Corresponding
to this minimum drop volume was a minimum drop diameter
based on the assumption of a spherical drop. These two minima
did not correspond in any regular manner with flow rate
maximum jet length but generally occurred near the latter.

' --y/\ -'- I


I

EFFECT OF
NOZZLE SIZE ON

l -

The surface area per unit volume of dispersed phase flow, A s ,


depends only on average drop size; thus, by the geometry of a
sphere, A s = ~ / D D .Therefore, t h e interfacial contact area
per unit volume of flow, assumed to be t h e same as transfer area,
depends only on drop diameter in an inverse ratio. Consequently, when the drop diameter went through a minimum, the
transfer area went through a maximum a t the same flow rate.
Transfer area plotted against dispersed flow rate, UT,gave an
inverted-V type of curve with varying degrees of flatness; this
type of plot allowed determination of an interrun maximum
point, as was the case with minimum drop volume. Such plots
are shown in Figure 4 for the toluene-water system with several
nozzle sizes.
The only data available in the literature for comparison with
these area results are calculated values from the correlation of
Hayworth and Treybal (20). This correlation applies strictly
only for nozzle velocities below 10 em. per second; for rates of 10
t o 30 cm. per second the equation correlates only the largest drops
observed, and a t higher rates the data were considered too erratic for correlation. Nozzle rates in the present study were
greater than 10 em. per second for the common solvents except
in the largest nozzle and many of the maximum area values occurred a t rates greater than 30 em. per second. I n Figure 4

INDUSTRIAL AND ENGINEERING CHEMISTRY

261

ENGINEERING. DESIGN. AND PROCESS DEVELOPMENT


are shown the area curves for toluene in water for nozzles C, A,
and F as computed from the Hayworth and Treybal correlation.
Nozzle velocities and drop sizes for these curves are as follows:

Uv cc./sec.
UN: cm./sec.

Do (by correlation of HayTq-orth


and Treybal, ZO), cm.
DO (observed), om.

C
0.28- 0 . 4 3
2 7 . 0 -41.4

Nozzle
A
0.90- 1.80
14.9 - 2 9 . 9

0.34- 0 . 2 5
0.37- 0 . 3 3

0 67- 0 . 5 1
0.74- 0 . 5 5

F
170- 4.0
1 1 . 5 -27.2
0.90- 0 . 6 8
0 82- 0 . 7 7

Agreement is reasonably good considering t h a t all the values


are beyond the recommended range of the correlation where the
latter shows an ever-decreasing drop size and that the correlation
is based on a static continuous phase
The rate of formation of new transfer area is given by AT
as square em. per second and is found from the product of A s and
Uv. The maximum value of As and its corresponding UV and
their product, A T , reprebent the optimum operating conditions
for the dispersion nozzles of a spray tower and, therefore, are the
basis for most of the correlations obtained in this investigation
These three functions are defignated by the additional subscript
M ; thus A ~ . w U
, V M ,and AT.M indicate values corresponding to
the peak data of A s while B s , Uv and AT are used for any other
point.
Correlations are based on peak values
that represent optimum operating conditions

I n view of the importance of A m and its corresponding data,


since every solvent tested shoned such a maximum, the values
found for these peak data are given in Table V. Some of the
solvents of higher viscosity and lower interfacial tension would
not form individual drops from the larger nozzles. When the
dispersed phase flop rate was reduced below the point of a medium
length jet, the jet collapsed and retreated back inside the nozzle.
For isoamyl alcohol, n-but3 1 alcohol, and ethylhexanediol, the
flow giving 4831 v a s the loaest flow that could be run; this was
considerably below maximum jet length. It is obvious that
operation of a tower a t such a n unstable point nTould not be
feasible, but these data were considered t o be peak data in order
to be consistent with the other results
Inasmuch as peak values depend on a number of factors, including a graphical interpolation, their reproducibilitv was tested.
This is shown in Table VI for repeat runs from Table V with the
same nozzles for the toluene-water and diisopropyl ketone-water
systems. These check runs weie interspersed with other runs
both in operation and in calculation so a s to avoid wishful
control a s much as possible. The reproducibility for A S Mand
UVMis good. Both show an average mean deviation of about
2.6% and a maximum of about 470
The results of repeat iuns in Table Tr also indicate that the
velocity of a moving continuous phase has no effect. Apparently,
drop formation and the method of jet breakup are independent of
the increased turbulence of the higher flow rates. Since the
length of the jet is affected by the continuous phase flow rate, it
is apparent that diop formation is not a function of jet length.
It should be noted, however, that a static continuous phase gives
somewhat different results for the peak data than does a phase
moving in the range of 15 to 81 cubic feet per (hour) (square foot).
The static values for A ~ . vand UVMare generally higher than the
moving phase values. The differences, hoxTever, are sufficiently
small that the data of Tables V and VI are probably good for
continuous phase flow rates well below the tested minimum of 15
cubic feet per (hour) (square foot).
The peak values of the dispersed phase flow late, UVM,are
plotted against nozzle diameter, DN,in Figure 5. For some solvents this curve appears to be smooth, but for others the line for
nozzles G, P, and A seems to show a break with the curve for
nozzles B, C, and D.
The effect of nozzle length on peak data wm not sufficiently
tested to tell its exact nature and magnitude. One set of toluene
262

Table V.
Nozzle
Size

Maximum Surface Area and Related Data for


Solvents and Water
Av. LV
4sx,
U v .w ,
ATM,
Cc,/&,

Sq. cm./Cc.

Cc./Sec.

Sq. Cm./Sec.

-4-3
F-3

c-3

162
144
135

Isoamyl Alcohol
21.0
16.7
18.1

c-3
B-3
4-3
F-3
G-3

98
107
114
171
103

%-Butyl Alcohol
25.9
29.2
17.7
20.6
26.0b

0,050
0.08
0 . 38Qa
0.549
1.0b

1.30
2.37
6.89
11.3
26.0

c-3
B-3
A-3
F-3
G-3

130
112
125
120
120

n-Butyl Chloride
21.4
11.4
13.2
9.47
5.23

0.313
0,932
1,524
2.25
4.02

6.68
10.62
16.4
21.4
20.2

D-3
c-3
B-3
.4-3
F-3
G-3

112
138
141
173
I57
112

Cyclohexane
32.6
18.3
14.4
11.8
7.20
6.89

0.350
0.820
2.00
2.04
4.13
5.97

11.4
15.0
28.8
24.1
29.7
41.2

c-3
B-3
B-3
-4-3
A-3
A- 1
F-3
F-3
G-3

82
113

103
106
95
120
100
117

c-3
A-3
F-3

141
121
153

E t h y l Acetate
24.0
12.0
10.5

0.153
0.545
0.680

3.67
6.54
7.11

c-3
A-3
F-3

178
119
118

Ethyl E t h e r
25.5
18.8
17.6

0.510
0.527
0.551

13.0
9.91
9.68

c-3
4-3
F-3

95
110
165

Ethylhexanediol
20.6
17.1
19.2

0.0595

1.23
1.54
1.94

c-3
B-3
A-3
F-3
G-3

128
110
152
119
120

n-Hexane
21.4
15.2
8.90
8.10
7.63

C-3
B-3
A-3
F-3
G-3

99

105
140
167
102

D-3
c-3
B-3
A-3
F-3
G-3

91
141
88
108
105
104

111

0.235
0,2025
0 , 235a

4.94
3.38
4.25

Diisopropyl Ketone
22.0
0 . 408
15.6
0,860
16.9
0.855
14.1
1.08
15.2
1.08
14.3
1.17
11.8
1.50
11.7
1.63
11.1
2.45

0 . 0Q03a

0.101a
0.515

1.30
1.76
2.80
9.16

Methyl Isobutyl Ketone


25.3
0.262
18.4
0.548
14.7
0.655
10.8
1.03
13.8
1.50
Isopropyl E t h e r
35.6
25.3
14.1
12.9
10.9
9.92

0.150
0.265
0.620
0,840
1.25

1,63

Toluene
23.6
0.490
269
c-3
24.1
0.488
89
c-3
22.0
0.490
133
c-3
22.8
0.466
457
c-3
15.5
1,28
111
B-3
14.7
1.18
128
B-3
12.3
1.R5
96
A-3
1
2
.
8
1.39
133
A-3
12.3
1.56
253
A-3
12.4
1.57
417
A-3
12.3
1.47
130
A-2
12.3
1.32
137
A- 1
12.4
1.23
125
A-0.15
8.35
2.95
86
B-3
8.67
2.73
121
F-3
8.10
2.92
247
F-3
7.97
2.70
436
F-3
6.63
5.60
94
G-3
a Showed peak at lowest dispersed rate attainable.
b Estimated; too m a n y drops in column for good count.

INDUSTRIAL AND E N G I N E E R I N G C H E M I S T R Y

8.99
13.5
14.4
15.3

16.5
16.7
17.6

19,1

27.3

11.0
19,5

15.7
22.7
69.9
6.63
10.1
9.62
11.1

20.7

5.33
6.70
8.74
10.8
13.6
16.1
11.5

11.8

10.8
10.6
19.7
17.3
19.1
17.7
19.2
19.5
18.1

16.2
15.2
24.7
23.7
23.6
21.6
37.2

Vol. 41,No. 2

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT

,26 Nozzle G-3

2.0

.IO

- A-3
-

.00

= e-3

.06

e= .I2

.oe -

02

.IO

U"M

(cs.hsc.)

1.0

Figure 5

data shows a deviation of UVMgreater than would be expected


from the average error and also shows a small but distinct increase
in velocity, b u t not area, with nozzle length. A set of data for
diisobutyl ketone indicates the opposite trend with velocity but
again no effect on area. It would appear, then, that ABMis
independent of nozzle length and that UVMis either independent
or shows only a very slight increase with increasing nozzle length.
The length of the shortest nozzle, 0.15 inch, approximates
thickness of a heavy plate distributor, indicating that the correlations of this investigation might be used for plate distributors
or sieve plates operating with the appropriate flows and no surface agglomeration.

water system. The only reason for. selecting


this system was that the relative refractive
indices of toluene and water produced a very
sharp outline of the drops so that measurements were facilitated. It was necessary to
obtain an equivalent spherical diameter from
the horizontal and vertical diameters measured
from the projected photograph. From a consideration of the limited accuracy of the measurements and the large number of drops involved, it was decided that an arbitrary diameter calculated as the mean of the two
measured diameters was satisfactory. Size distribution on a percentage basis was plotted
on arithmetic-probability paper against the
mean equivalent diameter, D'D, as measured
directly from the projections since all the latter
had the same enlargement factor.
Typical results for these counts are shown in
Figure 6 for nozzle F-3. The lowest flow rate,
IO
1.50 cc. per second, was just above the jetting
point and shows good uniformity of drop size.
The next flow rate, 2.54 cc. per second, was a
little below the critical velocity and shows only
a slight decrease in uniformity but a marked
decrease in drop size, as would be expected when UVMis approached. The next flow rate, 3.10 cc. per second, was above
UVMand shows the effect of a number of large drops. The
highest flow rate indicates a corresponding increase in the propor-

Uniformity of drops i s related to flow


rate regions defined for jet length

I n general, the uniformity of the drop sizes obtained on jet


breakup could be related to the flow rate regions as defined for
jet length. Below the jetting point the drops formed were very
uniform, and in the varicose region they showed almost as much
uniformity. In the region of the critical flow, a decrease in uniformity was noted and in the sinuous region a n increasingly wide
range of drop sizes was obtained. As the disruptive stage was
reached, the sizes became somewhat more uniform and usually
tended to be fairly small. When operation was extended into the
spray region, which occurred a t a fairly low velocity for the larger
nozzles, the drops remained small and sometimes decreased further in size.
Drop size distribution was studied thoroughly for the toluene-

c
E

.,
.Pa

Percanlage of Tolel Drape Smaller

Table VI.

Reproducibility of Peak Data for Solvent-Water


Systems
Surface Area (Cm.a/Cm.a)
AY. deviation,

Nozzle
Size

Av. A B M

e-3
B-3
A-3
F-3

23.1
15.1
12.4
8.27

B-3
A-3
F-3

16.3
14.6
11.7

February 1955

Toluene
3.12
2.65
1.40
2.90

Flow Rate, Cc./Seo.


A v . deviation,

AY. U V M

0.483
1.23
1.52
2.82

1.82
4.06
4.15
3.90

Diisopropyl Ketone
3.84
0.86
3.76 ,
1.08
1.56
0.21

0.30
0.00
4.15

in

S i r e Than

Figure 6

tion of large drops with a decrease in uniformity. It is the sudden


appearance of larger drops near the critical velocity that terminates the constant increase of surface area found in the varicose region. As the flow rate is increased further, the frequency
of occurrence of these large drops increases a t a faster rate and
thereby accounts for the sudden drop in surface area beyond the
maximum value.
The results of a uniformity analysis, as measured by the standard deviation for a normal probability curve, are given in Figure
7 for a number of operating conditions. Uniformity is definitely

INDUSTRIAL AND ENGINEERING CHEMISTRY

263

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT


8
6

a function of the flow rate for all the nozzles. Uniformity is better
for smaller nozzles but is also good for lower flow rates with larger
nozzles. Uniformity starts to decrease as UVM is approached
and it falls off very sharply for flows only somewhat larger than
UVM,which is indicated by the arrow8 (Figure 7 ) . A decrease
in nozzle length, as for nozzle A-I, decreases uniformity in the
lower flow range when compared to a longer nozzle such as
nozzle A-3. It seems that a decrease in interfacial tension alone,
as shown by the nozzle F-3 runs containing Alkaterge C, is insufficient t o cause a change in the uniformity.
Rates of drop rise are lower than predicted by
Newtons law or drag coefficient for spheres

Both qualitative and quantitative data were obtained o n drop


behavior during free rise in the column after formation. Rates
of droT rise, relative to the continuous phase, were measured
approximately in determining drop volumes. All but the smallest
drops encountered were in the Reynolds number range covered
by Newtons law. Figure 8 shovs this law for toluene drops in
water and also shows the rates predicted by the drag coefficient
for rigid spheres. The expelimental points for drops over 0.4
cm. in diameter fall well below both curves. These results are
in line with the observation of Hinze ( 6 1 ) that the deformation
and vibration of liquid drops in a fluid will increase the resistance
to flow appreciably above that for rigid globules. Hughes and
Gilliland ( d 3 ) studied the factors affecting deformation of liquid
drops in air and gave relations for predicting the distortion
based on terminal velocity. However, application of their results to liquid-liquid drop systems is uncertain because of the
continuous oscillation and fluttering tendency of such systems.
Qualitatively, however, in agreement with the results of Bond and
Newton (8, 9), it appears that the drag coefficient for drops may
be less than that for solid spheres in the smaller drop sizes primarily because of the finite viscosity of the drop liquid; a t larger
drop sizes, distortion should increase with a resultant increase
in the drag coefficient to a value greater than that for spheres.
The transition diameter toluene drops in water should lie in the
range of 0.15 to 1.5 cm. as appears to be the case in Figure 8.
Farmer ( 1 3 ) and West and coworkers (44) reported drop rise
curves showing similar deviations. It seems, therefore, that
experimental rates of rise are necessary for extraction area calculations.
Figure 9 shows the observed rates for all the pure solvents
tested and Figure 10 s h o m the results for the toluene-water system with solutes. These curves represent the best lines through
the data and have an average error between 2 and 3% for each of
the solvents.
Drop measurements in sufficient numbers to give flattening
264

data were available for only a feff solvents; these diameters were
corrected for the knovm enlargement of the projections. In
addition, the horizontal diameters were divided by an experimentally determined factor of 1 2 5 to correct for magnification
in the round column. The vertical diameter was corrected for
drop rise during the */ip-second camera exposure. The best
curves for these corrected data are shown in Figure 11; the average deviation of the points is 5 to 10%. A comparison of the
graph with the physical properties of the systems indicates that
flattening is proportional to drop size and density difference
and inversely proportional t9 interfacial tension. I t should be
noted that the drop shapes were constantly varying and that the
data of Figure 11 therefore represent the mean shape based on
measurements of a large number of drops for each system Using
photographs of several individual drops of carbon tetrachloride
in water, Farmer ( 1 3 ) found eccentricities of 1.10 to 1.86 for
drops with equivalent spherical diameters of 0.26 to 0.52 cm.,
respectively. These data fall along the curve for ethyl acetate,
a s expected, since these systems have almost identical ratios of
density difference to interfacial tension.
The flattening of a drop of any given volume has an effect on
surface area. I f it is assumed that the flattened shape is that
of a n oblate spheroid, the mag4tude of the effect is readily calculated from the equations for the volume and surface area of a
sphere and of a spheroid. For drops of the same volume, eccentricities of 1.2, 1.6, and 2.0 increase the surface area of the sphere
by 0.6, 4.1,and 9.6%, respectively.
Observations on coalescence in the column were based entirely
on watching drops rise the full height of the tower and recording
whether any of these drops coalesced with any others. A distinction was made between coalescence of a few or of most of the
drops. These results are recorded in Table YII.

RATE OF DROP RISE

RELATIVE TO CONTINUOUS

>

---

-.I5

.2

.4

.3

1 j

Observed rate
Newtons law f o r spheres

Drag coefficient for

.5

.6

.7 .S .9 1.0

DD (em.)

Figure 8

Some of the solvents used seemed t o form drops no larger than


a certain size. This was particularly true of ethyl acetate and
n-butyl alcohol which coalesced readily to form largc drops that
immediately split into two or more smaller drops. Other solvents showing some coalescence also shoa-ed a distinct upper
limit to the size of drops found among the larger nozzles. For
toluene, several nozzles larger than nozzle G were tested for
single drop formation, and during the course of this m-ork a few

INDUSTRIAL AND ENGINEERING CHEMISTRY

Vol. 41, No. 2

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT


The peak data obtained with these solutes'
Comparison of these results with those for pure
toluene shows little effect for the Alkaterge C
but a marked change for the other solutes.
These results will be discussed more fully in a
subsequent paper dealing with correlation of
the data.

in toluene-water are given in Table IX.

limits of suitable operating range


for dispersion nozzles are defined by
jetting point and onset of nonuniformity

RATE OF DROP RISE


RELATIVE TO
CONTINUOUS PHASE
01

Volumetric

Drop Dlameler,

10

I1

Do (cm.)

Figure 9

of the largest drops broke up while rising. As a result of these


observations, approximate estimates of critical drop size may be
made for a few solvents. These data are shown in Table VIIT as
ranges within which the critical drop size probably falls.
These sizes are probably sensitive to any marked difference in
the turbulence conditions found in the column.

As a result of the observations on jets, it is


possible to define a most suitable range of
operation for the dispersion nozzles of liquidI2
13
liquid spray towers. The lower end of the
range is set by the jetting point velocity since
this combines the lowest economical flow rate
with high uniformity. This velocity for the common solvents is 50 to 80% of UVM. The upper
limit of the suitable range is determined by the onset of appreciable nonuniformity of drop sizes because this factor gives rise to a
reduction of transfer area, loss of fine droplets by entrainment,
and excessive coalescence. The rate a t this point is frequently
only a few per cent higher than UVM,seldom more than 20y0.
--

Third components differ markedly in


effect on peak data

Table VIII.

The number of solutes tested was not large, but their physical
characteristics varied widely a s a means of high-spotting the effects of solutes on tower operation. All solutes were used in the
toluene-water system so t h a t comparison of results would be
facilitated. Alkaterge C was a strongly surface active material
soluble in toluene but negligibly so in water; it markedly changed
the interfacial tension but, had almost no effect on the other
physical properties. On the other hand, sucrose was soluble
in the water phase, but negligibly so in the toluene phase; it
markedly changed the density and viscosity of the water but
had no effect on the toluene phase and little effect on interfacial
tension. Acetic acid and benzoic acid changed the physical
properties of both phases. but the former favored the water
phase and the latter the toluene phase, both by roughly 20 to 1
on a weight percentage basis.

Table VII.
Solvent
Phase
Toluene

Dispersed Phase
Toluene
Ethyl acetate
Isopropyl ether
Diisopropyl ketone
Methyl isobutyl ketone
n-Butyl alcohol

Table IX.

l.On
l.Oa
6.5a
6.5"
265
26a

Solute
Concn.,
Solute

...

sucrose' '
40.4
Sucrose
17.2
Acetic acid
0.52
Benzoic acid
4.00
Benzoic acid
6.75
Alkaterge C
1.0"
Alkaterge C
6.5a
Alkaterge C 2 P

..

..
..
..
..
..
......

C
NC
NC

NC
NC
NC
NC
NC
NC
NC NC

Nozzle Size
B
A
F
G
NC NC N C N C
.. N C N C
. . NC . . . .
NC . .
. . . . N C ..
NC
. . . . NC . .
. . . . NC
. . . . NC . .
NC N C NC NC
SC MC MC ..
MC MC MC MC

..

... .
. . . .

.....
.....
.....
.....

...
.*.
...
...
...
...

..
..
..
..

..

..
..

negl
negl.
negl.
negl.
negl.
negl.

Nozzle
Size

A8M.
. UVM,
Sq. Cm./
Cc./

cc.

.Set.

ATM
Sq. Cm./

Sec.

C-3
F-3
C-3
F-3
C-3
F-3

0.545
2.69
0.520
3.06
0.412
2.97

12.6
23.7
11.7
27.0
9.85
26.2

Acetic Acid
118
21.5
112
11.5

0.375
1.65

8.06
18.9

Benzoic Acid
111
22.2
118
10.1
128
22.5
125
10.5

0.370
1.96
0.326
1.78

8.21
19.7
7.32
18.7

1.22
0.360
1.02
1.65

16.5
7.87
15.4
20.7

9.36
9.36

C-3
F-3

4.00
4.00
6.75
6.75

0.21
0.21
0.29
0.29

C-3

17.2
40.4
40.4
40.4

L17,

Cc /
Min.

Alkaterge C
128
23.2
122
8.80
107
22.5
116
8.84
125
23.9
123
8.80

0.52
0.52

negl.
negl.
negl.
negl.

Cm.
1 . 3 (approx.)
0.7-0.8
0.7-0,s
0.7-0.8
0,6-0.7
0.3-0.4

Maximum Surface Area and Related Data for


Toluene-Water with Solutes

Concentration,
Wt. %
Toluene
Water

Coalescence of Solvent Phase Drops

Cyclohexane
.....
...
n-Butyl alcohol
.....
...
n-Butyl chloride
.....
...
.. SC
Diisopropyl
ketone
.....
NC NC N C N C
.....
...
,. NC , . SC MC
Ethyl acetate
Methyl isobutyl
ketone
NC N C SC MC
.....
...
.. NC . . N C N C
Isoamyl alcohol
. . NC
N C N C
Ethylhexanediol
Ethyl ether
sc
MC MC
.....
n-Hexane
NC NC N C N C
Isopropyl ether
N C SC SC N C MC
Drops/100 ml. toluene.
N C = no coalescenoe
SC = some coalescence
MC = much coalescence

:February 1955

Estimated Critical Drop Sizes for


Solvent-Water Systems
Solvent Phase as
Critical Dn,

F-3
C-3
F-3

A-3
C-3
A-3
F-3

Sucrose
102
113
123
123

13.5
21.8
15.1
12.6

Drops of Alkaterge C/100 ml. of toluene.

MC

..

SC

..
..

..

NC
MC

Within the range defined above lies the flow rate, UVM,as
determined by the point of maximum transfer area. Such a critical point has not previously been reported and is important in
a spray extraction column. The maximum of the area is sharp
with a rapid decrease in area on either side of the peak. For almost all of the solvents and nozzles tested, this maximum area in
terms of the area per unit volume of dispersed phase flow, A ~ M ,
is 10 to 25% greater than t h a t at the jetting point and is 3 to

INDUSTRIAL AND ENGINEERING CHEMISTRY

265

ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT

Table X.

DX,
Inch

UN,

Cm./Sec.
12 -87

0,120
0.095
0.042
0.250

I ,

.3

.4

.5

.l

.6

V ~ l u r n e t iCrop
~ ~ Oiomeler,

.B
1.0
D g (crn.)

II

IZ

13

Figure 10
Concn., W t .
Solute
Acetic ocid
Benzoic acid
Benzoic acid
Alkaterge C
Alkaterge C
Alkoterge C
Sucrose
Sucrose

Toluene
0.52

4.00

%
Water

9.36
0.21

6.75

0.29

0.01
0.065
0.26

Negligible
Negligible
Negligible
17.2

Negligible
Negligible

40.4

Alkaterge C conc. i n drops/cc. of toluene; phases mutually saturated with


respect to all components

15y0higher than t h a t a t the nonuniformity flow rate. The peak


measurements appear to be reproducible with less than 5% error.
For the best operation of a sprag tower, several factors appear
to be important. Flow of the dispersed phase must be even and
steady; each of the dispersion nozzles must receive an equal
portion of the total feed; and the nozzle length should be 2
inches or greater. A dispersion head of the type used by Sherwood, Evans, and Longcor (40)seems particularly suitable.
The experimental data are insufficient to show conclusively
whether or not there is a real discontinuity in the curve of UVM
versus nozzle diameter in the region of a 0.1-inch nozzle. Cyclohexane and the alcohols show a break well beyond the average
error of the measurements; most of the solvents not indicating a
break between nozzles B and A do show a marked curvature.
The exact cause of the change in behavior in this nozzle range
is not known, but it is suggested that the effect may mark the region where the jet begins t o taper after leaving the nozzle so
that D N is not strictly equal to DJ. This tapering was very
marked in the high viscosity solvents but wa8 also apparent with
most solvents in the larger nozzles. Data on one toluene jet
from nozzle A-3 showed a 22% reduction in diameter from the
nozzle to the breakup point. No particular value of the dimensionless correlation groups seemed to be involved in the transition from nozzles B t o il.
I n view of the observation by Johnson and Bliss ( 2 6 )that poor
results with spray towers for high rates of flow might be due
to excessive nonuniformity, it is interesting t o compare the
suitable operating range, a s defined above, with the rates actually used by several investigators. This is done in Table X
for methyl isobutyl ketone using the linear nozzle velocity, U N .
Sherwood and coworkers ( 4 0 ) operated at very high velocities
and the nonuniformity of drops may well have been responsible
for the decrease found in the mass transfer coefficient a t higher
rates. Johnson and Bliss ( 2 6 ) changed the number of nozzles in
their distributor so that their flow rates probably remained in the
range shown; their maximum rate agrees well with that found
in the present work and it is of interest. that their capacity coeffi266

Nozzle Velocities Used


Phase

for Methyl Isobutyl Ketone


Suitable Range, Cm./Sec.
Jet
Nonpoint
L 7 ~ . % 4 uniformity

Reference
(40)
($6)
(6)

4.5-21
38 (rnax.)
61 (max.)

9
12

i)

..

(6)

15
20
38
4

..
..

, .

cients did not show a maximum with dispersed flow rate. Blanding and Elgin ( 6 ) apparently s p t their maximum rates to avoid
the disruptive range but, for the larger nozzle, were well into the
nonuniform region.
The results of several investigators ( I S , SO, 40, 44) show up to
40% of the extraction in a tower occurring a t the dispersed phase
inlet. All of these tests were made below the jetting point;
with a jet there is no expansion time and a very short formation
time, EO that no extra transfer would be expected a t this point
for a tower under normal operating conditions, as reported by
Geankoplis and Hixson (14).
A study of the solvents shocr-ing appreciable coalescence, a s
listed in Table VII, indicates several strong trends. Coalescence
is promoted by low interfacial tension and by increasing nozzle
size and therefore by increasing drop size; the last also generally
implies a n increase with higher rates of drop rise. These trends
are opposite to those noted by Appel and Elgin ( 4 ) The conditions found to accompany coalescence in the present investigation seem reasonable because increasing drop size and decreasing
interfacial tension both promote oscillation and vibration of the
drops and thus increase the chances of contact between drops

I 71

.2

.4

/~

'

.8

.6

I RISE IN COLUMN I 1

10

1.2

Drop Diameter f o r Sphere of Same Volume os Spheroid, 0,

Figure

1.4

(cm)

11

One criticism of large spray towers is that in long columns the


drops of the dispersed phase may coalesce so much that transfer
area is greatly reduced. It is noteworthy that the solvent
characteristics promoting coalescence are apparently the same
ones, in general, that lead to a relatively small critical drop size
a t the velocities achieved in a liquid-liquid system. Table VI11
shows that the critical drop sixes for the lower interfacial tension
systems are often smaller, even in large nozzles, than the drop
sizes attained in medium nozzles with higher tension system?.
For strongly coalescing solvents in long towers, the transfer area
should be based on the critical drop size and not on the initial
drop size.

INDUSTRIAL AND ENGINEERING CHEMISTRY

Vol. 47, No. 2

ENGINEERING. DESIGN, AND PROCESS DEVELOPMENT


Inasmuch as a point of maximum area was found for each of the
nozzles and solvents tested, these peak values were considered
to be unique functions for each system. An applicable equation
was developed by dimensional analysis and correlations were obtained for both maximum area and its corresponding velocity.
These results permit calculation of the interfacial area per unit
volume of extractor and will be presented in a subsequent paper.
Nomenclature

The metric system of measurement was used throughout this


investigation, except ae noted on graphs. The units given are,
therefore, the ones in which the factors were measured or calculated. If a factor was not measured or calculated, no units are
indicated.

As = surface area per unit volume of dispersed phase flowing,


sq. cm./cc.

AT = surface area formed per unit time, sq. cm./sec.


DD =i diameter of spherical drop, cm.
DD = ( D E
DE')/^ = equivalent drop diameter from pro-

jected photographs, mm.


= horizontal diameter 01 flattened drop, mm.
= vertical diameter of flattened drop, mm.
= diameter of jet a t breakup point
= inside diameter of nozzle, cm.
= DE/DE = eccentricity of flattened drop
= length of jet to breakup point, cm.
= volumetric flow rate of continuous phase, cc./min.
= linear flow rate of dispersed phase in nozzle, cm./sec.
= velocity of drop rise relative to continuous phase, cm./sec.
UT. = volumetric flow rate of dispersed phase, cc./sec.
TD = average drop volume of dispersed phase, cc.
U
,
viscosity, cp.
p
= density, grams/cc.
u p = standard deviation of normal probability distribution, cm.
=i

Subscripts
C refers to continuous phase

D refers to dispersed phase


M refers to peak values corresponding to maximum area

0 refers t o organic phase


W refers to aqueous phase
literature Cited
(1) Adam, N. K., Physics and Chemistry of Surfaces, 3rd ed.,
pp. 387-8, 424, Oxford Univ. Press, New York, 1941.
(2) Addison, C. C., J . Chem. SOC.,1943, p. 535; 1944, pp. 252, 477;

1945, p. 98, 354; 1946, p. 579; 1948, p. 963.


(3) Addison, C. C., Phil. Mag., 36, 73 (1945).
(4) Appel, F. J., and Elgin, J. C., IND.
ENG.CHEM.,29, 451 (1937).
(5) Bartell, F. E., Case, L. O., and Brown, H., J . Am. Chem. Soc.,
55, 2419 (1933).
( 6 ) Blanding, F. H., and Elgin, J. C., Trans. Am. Inst. Chem. Engrs.,
38, 305 (1942).
(7) Bohr, N., Trans. Roy. SOC.(London), 209A, 281 (1909).
(8) Bond, W. N., Phil. Mag., Ser. 7, 4, 899 (1927).
(9) Bond, W. N., and Newton, D. A., Ibid., 5, 794 (1928).

February 1955

(10) Chemical Engineers Handbook (J. H. Perry, editor), 3rd ed.,

Section 11 (1950).
(11) De Juhasz, K. J., Zahn, 0. F., and Schweitzer, P. H., Penna.

State College, Bull. No. 40, 24 (1932).


(12) Elgin, J. C., and Browning, F. M., Trans. A m . Inst. Chem.
Engrs., 31, 639 (1935); 32, 105 (1936).
(13) Farmer, W. S., Oak Ridge Natl. Laboratories, Unclassified
Rept. 635, 1950.
ENG.CHEM.,42, 1141
(14) Geankoplis, C. J., and Hixson, A. N., IND.
(1950).
(15) Guyer, A., and Peterhans, E., Helv, Chim. Acta, 26, 1107 (1943).
(16) Hadamard, J., Compt. rend., 152, 1735 (1911).
(17) Harkins, W. D., and Brown, F. E., J . Am. Chem. SOC.,41, 499
(1919).
(18) Harkins, W. D., and Cheng, Y . C., Ibid., 43, 35 (1921).
(19) Harkins, W. D., and Humphrey, E. C., Ibid., 38, 228 (1916).
ENGI.
CHEM.,42,1174
(20) Hayworth, C. B., and Treybal, R. E., IND.
(1950).
(21) Hinze, J. O., Appl. Sci. Research, A-I, 263, 273 (1949).
(22) Holroyd, H. B., J. Franklin Inst., 215, 93 (1933).
(23) Hughes, R. R., and Gilliland, E. R., Chem. Eng. Progr., 48, 497
(1952).
(24) Hutchinson, E., J. Colloid Sci., 3, 219, 235 (1948).
(25) International Critical Tables, IV, 435-7; ,McGraw-HilI, New
York, 1930.
(26) Johnson, H. F., Jr., and Bliss, H., Trans. Am. Inst. Chem.
Engrs., 42, 331 (1946).
(27) Krevelen, D. W. van, and Hoftijzer, P. J., Cliem. Eng. Progr.,
46, 29 (1950).
(28) Kronig, R., and Brink, J. C., A p p l . Sci. Research, A2, 142
(1950).
(29) Laddha, G. S., and Smith, J. XI., Chem. Eng. Progr., 46, 195
(1950).
(30) Licht, W., Jr., and Conway, J. B., IND.
ENG.CHEM.,42, 1151
(1950).
(31) Maier, C. G., and Ralston, 0. C., U. S. Bur. Mines, Bull. 260,
1927.
(32) Merrington, A. C., and Richardson, E. G., Proc. Phys. SOC.
(London), 59, 1 (1947).
(33) Neumann, H., and Seeliger, R., 2. Physik, 114, 571 (1939).
(34) Ohnesorge, Q., 2. angew. math. Mech., 16, 355 (1936).
(35) Pound, J. R., J . Phys. Chem., 30, 791 (1926).
(36) Powney, J., and Addison, C. C., Trans. Faraday SOC.,33, 1243
(1937).
(37) Prausnitz, P. H., Kolloid-Z., 76, 227 (1936).
(38) Rayleigh, Proc. Roy. SOC.(London), 29, 71 (1879); 34, 130
(1882).
(39) Satterly, J., and Collingswood, L. H., Trans. Roy. SOC.Can.,
111,25, 205 (1931).
(40) Sherwood, T. K., Evans, J. E., and Longcor, J. V., IND.ENQ.
CHEM.,31, 1144 (1939).
(41) Silbereisen, X., 2. physik. Chem., 143A, 157 (1929).
(42) Smith, S. W. J., and Moss, H., Proc. Roy. Soc. (London), A93,
373 (1917).
(43) Tyler, E., and Watkin, F., Phil. Mag., 14, 849 (1932).
(44) West, F. B., Robinson, A. P., Morgenthaler, A. C., Jr., Beck,
T. R., and McGregor, D. K., IND.
ENG.CHEM.,43,234 (1951).
R ~ C E I V Efor
D review April 2, 1952.

INDUSTRIAL AND ENGINEERING CHEMISTRY

ACCEPTED
September 9, 1954.

267

S-ar putea să vă placă și