I
JAMES R. WHITE and ALBERT T. FELLOWS
. Socony Mobil Oil Co., Inc., Paulsboro, N. J.
Design of Ports a n d Manifolds in
...
Thermal Diffusion Columns Efficiency of heat utilization in continuous-flow thermal diffusion columns can be improved by properly designing inlet and outlet ports
B m x u s E large amounts of heat are expended, continuous flow thermal diffusion columns for separating liquids must utilize effectively the largest possible fraction of the heat. Such columns have been designed with uniform spacing and temperature gradients between the hot and cold surfaces. But for columns to be thermally efficient, proper provisions must be made for continuously introducing mixtures to the column and for removing the separated streams. Evidence is presented demonstrating the need for uniform lateral distribution of inlet and outlet streams in thermal diffusion columns-a need reasoned to be important a priori. This evidence was obtained by deliberately creating experimental conditions of nonuniformity in columns and making observations. When these were judged sufficient to establish that nonuniformity was deleterious to column operation, the study was discontinued, since little of significance seemed realizable from a more extensive study of poor apparatus design. Three kinds of transport can be identified in vertical thermal diffusion columns. First there is convective flow in the vertical or 2 direction. The temperature gradient imposed by the flow of heat from the hot to the cold walls causes diffusive flow of the constituent molecules perpendicular to these walls, or in the X direction. There is also a lateral direction, Y. I n the absence of thermal or concentration irregularities there is no force which serves to transport material in the side-to-side, or Y direction. But this lateral transport must be provided if the entire working surfaces are to be efficiently used. The lack of this lateral, side-to-side
transport has been observed in columns made of glass, both in the form of concentric cylinders and parallel plates. In these experiments the diffusion space between the heated and cooled glass surfaces is filled with a colorless “white oil.” When a little of the same white oil, dyed by addition of a brightly colored dye, is injected into the oilfilled diffusion space, a colored, threadlike column of oil rises to the top, returns to the bottom, and then rises again to the point of entry. This convective loop proceeds over and over
PARALLEL PLATE
c-2 I
I
I I
I
I I
I HOT SURFACE
A When the column is tilted slightly from the vertical, the individuality of each cycle within the diffusion space becomes apparent
b Dyed oil injected into the diffusion space shows that this port arrangement, for the parallel plate and concentric cylinder, results in low thermal efficiency
I
i I
I I I
I \ \
CONCENTRIC CYLINDER VOL. 52, NO. 5
MAY 1960
389
Stain pattern left on the working surface of the hot plate shows that only about one third of the total volume between the plates was acting as an efficient separating column A. Outlet B. Feed port
ONE OF TWO SURFACES D E F I N I N G THE SEPARATION SPACE
with practically no lateral displacement or fanning out. If the column is tilted slightly from the vertical, each convective cycle is displaced from its previous path, and the individuality of each cycle is apparent. If this port were the sole means of entry for a stream to be separated, and similar ports above and below were the means of withdrawal of the separated products, the whole effective column would be similar to the narrow thread of dyed oil in the experiments just described. The rest of the column will operate essentially in a steady-state, dissipating heat throughout the column. But only that portion of the heat flow which is across the limited lateral distance shown by the dyed oil experiments will be effective in leading to a net separation. The net result of such a port arrangement is low thermal efficiency.
Both Flow Rate and Total Separation Improve When Full Width of the Column Is Used Pumping R ~ ~Product ~ , Via. at 37.7' C., C S . Cc./ Min. TOP Bottom Entire Lateral Dimension 4
140 148
8 1/3
4 8
390
271 252
Lateral Dimension 149 163
A further experiment demonstrating the importance of lateral distribution uniformity has been carried out with a parallel brass plate column with entry and take-off manifolds (Figure 1, B ) , extending across the full width of the column. The normal operation of this column-Le., efficient operation using manifolds the full width of the plateshas been compared with otherwise similar experiments made with the same apparatus using only one third of the manifold, conveniently arranged by closing off the other two thirds of the manifold with metal filler. An unrefined oil having a viscosity of 190 cs. at 37.7' C. was pumped into the column and divided equally into top and bottom streams. Results shown in the table were obtained at a mean hot wall temperature of 173' C., a mean cold wall temperature of 90" C. and a spacing of 0.0838 cm. between plates. The diffusion space was 58.4 cm. high and 22.9 cm. wide. Using viscosity to measure separation, the tabulation shows the improvement, whether in terms of total separation or as a flow rate, when the full width of the column was used over that obtained when only one third of the column width was used,
255
230
TUBE PERSPECTIVE
PROFILE
ONE OF TWO SURFACES D E F I N I N G
as is the case where the manifolds were plugged. The oil used in this experiment was suffciently unstable thermally to result in a stain on the working surface of the hot plate after the experiments where two thirds of the manifolds were blocked off. The feed port is near the bottom, and, because the" left two thirds is plugged with metal, all the feed is from the one third at the right. The top product left the column through the top manifold, the left two thirds of which was blocked off by the neoprene gasket. The bottom product was similarly withdrawn by an identical slot and manifold at the bottom of the cold plate which is not shown, but the bottom of the neoprene gasket shown covered the left two thirds of the slot also. The oil entering from the feed port thus rises in the convective upstream outlined by the visible stain pattern and leaves by the top and bottom manifolds. The stain pattern outlined shows that only slightly over one third of the total volume between the plates was effectively acting as a separating column, the remaining slightly less than two thirds of the space between the plates being essentially inoperative for separation but nevertheless transferring heat from the hot to the cold plate as though the entire three thirds of the plate were effective in the separation. A small lateral bowing of the flow lines is evident in the stain pattern. This bowing results from a small flowfringing of the oil which was pumped from the bottom to the top of the column. Such flow-fringing is absent in batch columns, but makes a small contribution to lateral transport in continuous-flow columns. The desired uniform lateral distribution in the column can be obtained by introducing the mixture and withdrawing the products through manifolds of the type shown in Figure 1 (7), which, of course, extend across the full lateral dimension. Columns using the ports typified in Figure l,B have been operated with thermal efficiencies which approach thermodynamic limits ( 4 ) . Jones ( 3 ) has proposed a manifold comprised of a series of closely spaced holes opening into a flow-equalization channel and has suggested a porous manifold similar to Figure 1, A to minimize turbulence (2). Literature Cited
TUBE PERSPECTIVE
PROFILE
Figure 1. Uniform lateral distribution can b e obtained b y using manifolds such as these for introducing the mixture and withdrawing the products A. Porous manifold minimizes turbulence 8. Brass plate column with entry and take-off manifolds extending full width of the column. The brass plate contains a slot
INDUSTRIAL AND ENGINEERINGCHEMISTRY
(1) Fellows, A. T., White, J. R., U. S. Patent 2,834,464 (May 13, 1958). (2) Jones, A. L., U. S. Patent 2,720,976 (Oct. 18, 1955); British Patent 725,753 (March 9, 1955). (3) Jones, A. L., U. S. Patent 2,720,977 (Oct. 18, 1955). (4) Lt'hite, J. R., Fellows, A. T., IND.END. CHEM.49, 1409 (1957). RECEIVED for review May 22, 1959 ACCEPTEDFebruary 12, 1960
