INDUSTRIAL AND ENGINEERING CHEMISTRY
1676
Vol. 45, No. 8
Acknowledgment
which did not give the same type correlations as the cation exchange resins also showed more discoloration within the resin, disintegration, and irregular s~vellingthan the cation exchangers. Figures 4,5 , and 6 show the effect of flow rate on the utilization of the anion exchange resin capacity. The results obtained in this xork indicate the following ronelusions:
This work was done in partial fulfillment of the requirements for a doctoral degree in Chemical Engineering at the University of Tennessee, and under a fellowship granted by the Oak Ridge Institute of Nuclear Studies, Oak Ridge, Tenn. The authors also acknowledge the assistance of E. I. du Pont de Xemours & Co., Inc., which also granted a fellomhip for this project.
Ion exchange will take place a t a n appreciable rate between most nonaqueous solvents and the water-wet form of exchange resins. The best resins for use with organic solvents are those which are not changed by the solvents and retain their aqueous volume when in contact with the nonaqueous solutions. The exchange mechanism in this case may differ from the aqueous one primarily by the additional step of solute transfer from one liquid phase to another a t the surface of the particles, and may be almost independent of the properties of the solution. Regardless of mechanism, this is a significant consideration since it shows the very durable, solvent resistant, commercial resins to be suitable for use in many organic solvents. The influence of the solvent on the resin may be specific, 80 extrapolation of results to other solvents may lead to only rough approximations. An estimation of the feasibility of the use of cation exchange for a process solution may be made by considering the volume change of the resin in the solution and the viscosity and electrical conductivity of the solution. High electrical conductivity, low fluid viscosity, and lack of volume change of the water-wet resin when equilibrated with solvent are conducive to fast rates of ion exchange. Anion removal may take place from solutions that have no measurable electrical conductivity and so, presumably, only negligible ionization.
literature Cited
Doc. Inst., Doc. 3985, L i b r a r y of Congress, Washington 25, D. C. (2) Bhatnagar, S.S., Kapur. A. N., and Puri, A I . L., J . Indian Chem. Soc., 13,679 (1936). (3) Icressman, T. R. E., and Kitchener, J. A , J . Chem. Soc., 1949,
(1) Am.
p. 1211. (4) Kunin, Robert, and Myers, R. J., “Ion Exchange Resins,” C h a p .
5 , New Pork, John Wiley B: Sons, 1950. ( 5 ) Myers, F. J., IND.EXG.CHEM.,35,858 (1943). (6) Robinson, D. A., and Mills, G. F.,I b i d . , 41,2221 (1949). (7) Wiegner, G., and Jenny, H., Kolloid-Z., 42,268 (1927). RECEIVEDfor review November 4, 1952. ACCEPTED M a y 4, 1953 Presented as part of the Symposium on Nonaqueous ilpplications of Ion Exchange before the Division of Colloid Chemistry a t the 122nd Meeting of the AMERICAN CHEMICAL SOCIETY, Atlantic City, S . J. Material supplementary to this article has been deposited as Document KO.3956 with the AD1 Auxiliary Publications Project, Photoduplication Service, Library of Congress, Washington 2 5 , D . C. h copy may be secured by citing the document number and by remitting 81.25 for photoprints, or $1.25 for 36mm. microfilm. Advance payment is required. Make checks or money orders payable t o Chief, Photoduplication Service, Library of Congress.
0
cking Design Data
0 0 0
L. B. BRAGG Packed Column Corp., 3 0 Church St., New York, N. Y .
0
0
ovei the entire range of sizes fiom 3/*-inch diameter to 11-feet diagonal of a hexagon.
PERATING characteristics of Stedman packing have been described previously by Stedman (6, 7 , 8, 9) and Bragg ( 1 , 2 , 3, 4). The Stedman packed column with the largest cross section that has been built to date, a hexagon 11 feet across the corners, was described by Bragg and Morton ( 5 ) . This paper describes an interesting correlation of the number of theoretical plates per foot that are obtainable with Stedman packed columns,
For the purposes of this correlation selected points have been taken from the data published by Bragg ( 1 , 2 , 3, 4 ) . The data
a-
-
Figure 1.
NO
REFLUX RATE GALLONS PER HOUR Correlation of Plates per Foot and Reflux Rate
129
36 IN
HEXAGONAL
INDUSTRIAL AND ENGINEERING CHEMISTRY
August 1953
Table 1. Packing Type NO. Conical 105
104
Round triangular pyramid
Diameter Inches Feet
H.E. T.P., Inches
0,031
150 180 200 225
0.040 0.048 0.053 0.059
28.2 24.9 22.1 19.5
0.42 0.48 0.54 0.61
0.750
0.062
200 300 400 500 600
0.053 0.079 0.11 0.13 0.16
24.1 21.3 18.3 16.8 15.3
0.50 0.56 0.65 0.71 0.78
200 400 600 800 1000
0,053 0.11 0.16 0.21 0.26
21.9 17.4 15.5 14.1 12.3
0.55 0.69 0.77 0.85 0.98
0.75 1.1 1.5 2.0 2.5
12.2 10.3 9.1a 8.4 7.4
0.98 1.17 1.32 1.43 1.62
10.0 8.3 7.8 7.4 7.1 6.7 6.5
1.19 1.45 1.55 1.62 1.70 1.78 1.86
0.984
0.082
128
2.080
0.173
Correlation Data
Theoretical Plates/ Ft.
0.375
112
107
Reflux Rate MI./ Gal./ hr. hr.
1677
points were selected to cover a range of reflux rates from approximately that corresponding to the highest number of theoretical plates for each size of packing to a reflux rate a t approximately the loading point. I n one case a datum point was read from the curve drawn through the actual data. The data for the 12-inch diagonal hexagonal-shaped column and the smaller columns were obtained with binary test mixtures of benzene and ethylene dichloride. The data for the 36-inch diagonal hexagonal-shaped column, which have not been published previously, were obtained by using mixtures of 2,2,4-trimethylpentane and methylcyclohexane. To provide a proper correlation with benzene-ethylene dichloride data, the reflux rates determined were reduced 25% following the principles set forth by Bragg (4). The data for the 132-inch diagonal hexagonal-shaped column, which also have not been published previously, were obtained by using a n isomeric heptane mixture. These data were adjusted by reducing the reflux rates 22%. The dat,a used in the correlation are presented in Table I and are plotted in Figure 1. These data may be represented satisfactorily by the equation
where P is the number of theoretical plates per foot of height of the column, and R is the reflux rate expressed as gallons per hour at operating temperatures. The curve in Figure 1 was drawn through points calculated from this equation. While the data indicate that the number of theoretical plates per foot of packed height is substantially independent of the column diameter for any given reflux rate, this is the case only within the limits of suitable loading of the column. The reflux rate cannot exceed the loading point rate for the column in question and there is doubtless a limiting lower reflux rate for any particular column, which will vary with the operating conditions. Stated differently, the data given in this paper should not be used to determine column cross section but rather to determine the theoretical plates per foot of packed height, with the cross
Packing Type No.
Triangularshaped triangular pyramid
116
Diameter Inches
12
Feet
Theoretical Plates/ Ft.
H.E. T.P., Inches
10.2 7.9 7.4 6.7 6.3 5.8
1.18 1.52 1.61 1.78 1.91 2.08
15 30 40 50 60 7n 80 90
8.3 6.6 6.0 5.8 5.6 5.4 5.0 4.8
1.44 1.82 1.98 2.05 2.12 2.22 2.38 2.46
351 44 1 468 675 702
4.1 4.1 4.0 3.8 4.1
2.90 2.90 3.02 3.15 2.92
2.1 2.5 2.6 2.7 2.7 3.2
5.7 4.8 4.6 4.4 4.5 3.7
Reflux Rate hr.
1.0
hr.
."
129
36
3.0
From curve drawn through actual data.
section being determined by the usual allowable vapor velocity and liquid loading methods.
Discussion The equation indicates 2.8 theoretical plates per foot of packed height for a column of infinite reflux rate, but additional data on larger columns will be needed before this value may be assumed to be correct. The data obtained on the 132-inch diagonal hexagonal-shaped column slopes in a reverse manner to all of the rest of the datathe number of theoretical plates per foot of packed height increasing with increasing reflux rate. This column was the first large column that was built and the method of reflux distribution used was crude as compared to later designs. It is believed t h a t imperfect reflux distribution, particularly a t the lov,-er reflux rates, caused the reduced number of theoretical plates and the reverse slope of the data. The data presented were obtained a t total reflux but there are unpublished data which indicate that the value of P would not be appreciably effected by reduced reflux ratio until the reduction in the reflux ratio resulted in significant reduction in the liquid loading of the column. A reflux ratio of 3 t o 1 would conservatively fall in the region where no appreciable effect would be expected. literature Cited (1) Bragg, L. B., IND.ENG.CHEM.,ANAL.ED.,11, 283-7 (1939). (2) Bragg, L. B., IND.ENG.CHEM.,33, 279-82 (1941). (3) Bragg, L. B., Refiner Natural Gasoline M f r . , 18, 295-8 (1939). (4)Bragg, L. B., Trans. Am. Inst. Chem. Engrs., 37, 19-50 (1941). (5) Bragg, L. B., and Morton, F., Proc. Am. Petroleum Inst., 22
(III), 38-44 (1941). (6) Stedman, D. F., Can. Chem. Met., 21, 214-16 (1937). (7) Stedman, D. F.7 Can. J . Research, 15B7 383-400 (1937). (8) Stedman, D. F., Natl. Petroleum News, 24, No. 34, R 125-6, 128 (1937). (9) Stedman, D. F., Trans. Am. Inst. Chem. Engrs., 33, 153-61 (1937). RECEIVED for review October 16, 1952.
ACCEPTEDApril 14, 1953.