--
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT
Fractionating Efficiency of Coke Packings FUMITAKE YOSHIDA AND TETSUSHI KOYANAGI Deparfmenf of Chemical b g i n e e r i n g , Kyofo University, Kyofo, l o p a n
P
ACKED distillation columns are simpler in construction than are plate columns. However, it does not necessarily follow that packed columns are less expensive, especially in the case of large industrial columns. The cost of packing is often a considerable fraction of the total equipment cost, even when conventional packings such as Raschig rings or Berl saddles are used. The cost per theoretical plate of efficient packings such as Mchlahon or Stedman is several times higher than that of conventional packings. The work described in this paper was planned to study the distillation performance of coke packing, which is considered to be the most inexpensive packing available; it costs less than one tenth as much as Raschig rings or Berl saddles. It was anticipated that the coke packing might be efficient in distillation columns, for the same reason that wire gauze or protruded packings are efficient-Le., because of its high wettability. Experiments were carried out a t atmospheric pressure using two experimental setups. The 15- and 25-mm. coke packings were tested in 100- and 150-mm. columns, respectively, using an ethyl alcohol-water mixture. Smaller size coke packings were tested in a 1-inch column with benzene-ethylene dichloride. Coke packing performance tests include operation of columns with ethyl alcohol-water and benzene-ethylene dichloride
..
-
The general setup of the experimental apparatus was the same as the one used in a previous study (IS), except that the columns were 100 and 150 mm. in inside diameters, and mere packed to a height of 1 meter. Runs were made only a t total reflux. The reflux rate was determined by means of a graduated vessel which was placed in the liquid return line leading from the column bottom to the reboiler. The coke packings used were 15 and 25 mm. in average diameter with maximum deviation of 2 mm. They were prepared by carefully screening metallurgical coke from an iron works. The bulk density of such screened coke was about 0.56 gram per cc. The 15-mm. coke packing and, for the sake of comparison, 15-mm. ceramic Raschig rings were tested in the 100-mm. column. The 25-mm. coke packing was tested in the 150-mm. column. To obtain reproducible data, it was necessary either to preflood the packing or to operate the apparatus for 15 to 20 hours at the start of a series of runs. This was also true with the smaller apparatus. It required 3 to 5 hours for each of the runs to reach steady conditions. The liquid samples of the ethyl alcohol-water mixture were taken from the column top and from the reflux rate measuring vessel a t the column bottom, They were analyzed by density measurement. The equilibrium data used were those of Kirschbaum and Gerstner (7). The number of theoretical plates was determined from a plot of theoretical plate versus liquid composition. The pressure drop through the packing was measured by means of the manometers attached to the top and the bottom of the column. In order to eliminate the effect of vapor density, which varies considerably with composition in the case of ethyl alcoholwater mixture, the pressure drop test was made by distilling pure water a t total reflux.
April 1955
Metallurgical Coke as Column Packing
. . . compares well with Raschig rings in H.E.T.P. and pressure drop
. . . has a cost advantage which may easily outweigh disadvantage of higher holdup values
Small-Sized Coke Packings. The smaller equipment, in which smaller coke packings, single-turn helices, and McMahon packings were tested, was made entirely of glass. The column was 1.0 inch in inside diameter and 44.1 inches in packed height. It was enclosed in two concentric glass jackets. The inner jacket was wound with Nichrome heating wire and the heating was controlled by a variable voltage transformer. The air temperatures a t the top and the bottom of the annular space between the column and the inner jacket were so controlled that they agreed within 0.5' C. with the vapor temperatures a t the top and the bottom of the column, respectively. By adjusting the insulation on the outer jacket, it was possible to obtain proper temperature difference between the top and the bottom of the air jacket. The still pot had a capacity of 1000 ml. and was heated by a heating mantle. The reflux was supplied by a coil-type head condenser. Two eizes of coke packing were tested in the small column. One had itn average diameter of 4.4 mm. and was of very uniform size-Le., passed through the 4.5-mm. screen and was retained on the 4.3-mm. screen. Another passed through the 6.5-mm. screen and was retained on the4.5-mm. screen-i.e., had an average diameter of 5.5 mm. For comparison, l/rinch wire gauze McMahon saddles and 2.5-mm. stainless steel single-turn helices were also tested in the same column. The packings were supported by a stainless steel wire spiral a t the column bottom. For each run, 4 to 8 hours was required to attain steady conditions with this apparatus. The liquid samples were taken from the reflux line a t the column top and from the reflux rate measuring bulb a t the column bottom. Analyses of the benzeneethylene dichloride mixture were made by measuring the refractive index with a Pulfrich refractometer. Both benzene and ethylene dichloride were prepared by repeated fractionation with a column 1 meter high and packed with McMahon saddles. The pure samples of benzene and ethylene dichloride had refractive indices of 1.50075 and 1.44465, respectively, a t 20' C. The composition of liquid and the number of theoretical plates were obtained from the values of the refractive index using the data of Bragg (11. Holdup Determination. Table I presents the experimentally determined values of the void space as well as the static liquid
INDUSTRIAL AND ENGINEERING CHEMISTRY
711
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT COLUMN T O T A L REFLUX
DIA.
PACKED HEIGHT
100
CsH, -CzH,CLz T O T A L REFLUX
MM
1000 M M 3 .__
2
25-MU
COKE CCOLUMN
DIA
I 5 0 MM
>
RMCHIC
a 5 I-
I
W
I
+
I
COKE
15-HM.
W
COLUMN
DIA.
PACKED
HEIGHT 4 4 I I N
I
0
2
VAPOR
Figure 1. H.E.T.P.
VELOCITY
4
3
,
2
I
0
REFLUX
FT/SEC.
values for medium-size coke packings and Raschig rings
holdup expressed in percentage of the total packed volume for the various packings employed in the present work. The static holdup was determined by weighing the packings which were dipped in benzene and were drained for 2 hours a t 20' C. The operating holdup was determined in the same columns used in the distillation tests but without vapor flow, since it was considered that the vapor flow has little effect on the liquid holdup. The packings were irrigated for a time by benzene a t 20' C. at various flow rates. Then, the stopcock in the liquid feed line and the two-way cock in the liquid exit line a t the column bottom were simultaneously turned off. The amount of benzene received in a vessel ~7asmeasured after 2 hours to give the operating holdup.
Table 1.
~ _ _
~
I
O
Figure 2.
RATE
4
1
,
LITER5/HR.
H.E.T.P. values for small-size coke packings and conventional packings
H.E.T.P. for the 15-mm. coke packing in the 100-mm. column is about 5 inches and is considerably lower than that for the Raschig rings of the same size tested in the same column. However, as shown by the sharp rises of the curves a t increased vapor velocity, the limiting vapor velocity is lower for the coke packing than for the Raschig rings. The data of Kirschbaum and David (6) on the distillation of ethyl alcohol-water a t atmospheric pressure with 8-mm. Raschig rings in a column of the same dimensions used in the present work (100 mm. in diameter and
Void Space and Static Holdup of Packings
Diam., Mm. Void Space, % ' Coke 4.4 Coke 5.5 Coke 15 Coke 25 Single-turn helices 2.5 McMahon saddles . 2 5b Raschig rings 15 a After 2 hours draining (benzene a t 20" C.). b Inch. Packing
Static Holdupa,
Yo
12.5 11.7 6.2
4.7 5.2 2.1 0.9
H.E.T.P. values for coke compare favorably with those for conventional packings
The theoretical basis for the use of H.E.T.P. (height of equivalent theoretical plate) seems to be weaker than for the use of H.T.U. (height of transfer unit) in distillation correlations. However, it cannot be denied that the use of H.E.T.P. is more convenient in practice. The use of H.E.T.P. and the usual practice of correlating H.E.T.P. with the throughput may be justified a t least at total reflux, where H.E.T.P. values usually approximate the values of over-all H.T.U.'s, especially when the average slope of the vapor-liquid equilibrium curve, m, is near unity. The effects of vapor and reflux rates on H.E.T.P. cannot be separated a t total reflux. The problem of the way in which the H.E.T.P. values are affected by the reflux ratio or other factors is beyond the scope of the present work. Detailed data, including H.E.T.P. values, for operation of the columns with each of the packings are available from the American Documentation Institute. Figure 1 shows the H.E.T.P. values for the packings tested in the larger columns plotted against the superficial vapor velocity relative to the gross cross-sectional area of the columns. The 712
0.2
4
VAPOR
Figure 3.
6
8
1.0
VELOCITY,
2
4
6
8
FT./SEC.
Pressure drop for medium-size coke packings and Raschig rings C = Coke R = Raschig
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 47, No. 4
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT 1 meter in packed height) are also plotted in Figure 1. The curve for the Kirschbaum-David data is quite similar in shape to that for the present data. The differences in H.E.T.P. and in the limiting vapor velocity between 8- and 15-mm. rings are probably attributable to the difference in the packing size. Furthermore, the H.E.T.P. values for 15-mm. Raschig rings from the Kirschbaum data (6) with ethyl alcohol-water using a 400mm. column 1 meter in packed height are in approximate agreement with the present data for 15-mm. Raschig rings. With the 150-mm. column, in which the 25-mm. coke packing was tested, i t was impossible t o increase the throughput to the limiting velocity because the reboiler capacity was insufficient. Figure 2 shows the H.E.T.P. values for the small coke packings as well as for 2.5-mm. single-turn helices and '/d-inch McMahon saddles tested in the 1-inch column. The H.E.T.P. values increase in general with increasing throughput. Small coke packings are slightly more efficient than 2.5-mm. single-turn helices, but are less efficient than '/l-inch McMahon saddles. Again the limiting reflux rate is lower for coke packings. The H.E.T.P. values for single-turn helices are 2 to 3 inches and roughly agree with the data of Fenske and coworkers ( 2 ) for a 2-inch column and also with the data of TValsh and coworkers (11) for 1- and 2inch columns. However, these values are higher than those of Tongberg and coworkers (10)for a 0.79-inch column and lower than the values of Weedman and Dodge (19) for a 1.94-inch column. The H.E.T.P. values for McMahon packing range from 1 to 2 inches, in agreement with the data of McMahon (8) for a 2.88-inch column and with the data of Struck and Kinney (9)for a 0.75-inch column. These values are lower than the
61
0.2
I
I
4
6
REFLUX
I
4
1.0
RATE,
I
I
l
i
2
4
s
e
LITERS/HR.
Figure 4. Pressure drop for small-size coke packings and conventional packings C = Coke M.M. = McMahon S.T.H. = Single-turn helices
April 1955
4
0
8 100
2
LIQUID
Figure 5.
4
RATE
,
B
8 1000
2
4
LB./FT!HR.
Operating liquid holdup as function of liquid rate C = Coke M.M. = McMahon R = Raschig S.T.H. = Single-turn helices
values obtained by Forsythe and coworkers ( 4 ) with a 6-inch column and those obtained by Fisher and Bowen ( 3 )with a 4-inch column. Pressure Drop. Figure 3 shows the results of the pressure drop tests for 15- and 25-mm. coke packings and 15-mm. Raschig rings obtained with pure water a t total reflux. I n order to estimate the pressure drop for other vapors, it is necessary to multiply the pressure drop obtained from Figure 3 by the ratio of the average density of the vapor to that of saturated steam at atmospheric pressure. The curves for 15-mm. coke and 15-mm. Raschig rings (Figure 3) show sharp increases in pressure drop corresponding to the limiting vapor velocities. However, with the 25-mm. coke packing, which was tested in the 150-mm. column, i t was impossible t o obtain high vapor velocities. AH expected, the coke packing shows higher pressure drop than t h e Raschig rings of the same size. However, the pressure drop per theoretical plate is roughly equal for both types of packing. The pressure drop through the smaller packings tested in the 1-inch column is shown in Figure 4. The data were obtained during the distillation runs a t total reflux with benzene-ethylene dichloride. The coke packing exhibits the highest pressure drop, while McMahon saddles show the lowest pressure drop. Liquid Holdup. Figure 5 shows the operating holdup for all the packings tested as a function of the flow rate of benzene a t 20" C. in pounds per (hour) (square foot). The operating holdup of the coke packing is not so large, although its static holdup is relatively large. Large coke packing exhibits about twice as large operating holdup as the Raschig rings of the same size, but the operating holdup of small coke packings is almost comparable to t h a t of McMahon saddles. It is expected that the holdup would vary, depending on the physical properties of liquid. Conclusions
When used in distillation columns, screened coke packings show considerably lower H.E.T.P. and nearly equal pressure drop per theoretical plate as such conventional packings as Raschig rings of the same size. The H.E.T.P. values for small coke packings are comparable to those for some efficient packinga used in laboratory fractionating columns. The fact that the
INDUSTRIAL AND ENGINEERING CHEMISTRY
713
ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT coke packing has relatively large static holdup might be undesirable in some batch distillations. However, it would not be disadvantageous in the case of continuous industrial distillation, for which the coke packing may find i h major use. The main advantage of the coke packing-its unparalleled low cost-will outweigh its defects. Acknowledgment
The authors wish to express their appreciation to the Hirohata Iron Works for preparing and donating the samples of the coke packing and to several senior students for their assistance in the experimental work. literature cited
(1) Bragg, L. B., Ind. Eng. Chem., Anal. Ed., 11, 283 (1939). (2) Fenske, M. R., Lawroski, S., and Tongberg, C. o., IND. ENG.
CHEM.,30, 297 (1938).
(3) Fisher, A . W., Jr and Bowen, R. J., Chem. Eng. P V O L ~ ~45, . , 359 (1949). (4) Forsythe, W. L., Jr., Stack, T G., Wolf, J. E., and Conn, A. IND. ENG.CHEM..3 9 , 7 1 4 (1947)
L.,
(5) Kirschbaum, E., "Destillier- und Rektifiziertechnik," 2. Aufl., Springer, Berlin, 1950. (61 Kirschbaum, E., and David, A., Chem.-Ing.-Tech., 25, 592 (1953). (7) Kirschbaum, E., and Gerstner, F., Verfahrenstechnik, Beih. 2. Ver. deut. Ing., 1, 1 0 (1939). ( 8 ) McMahon, H. O., IND. ENG.CHEM., 3 9 , 7 1 2 (1947). (9) Struck, R. T., and Kinney, C. R., Ibid., 4 2 , 7 7 (1950). (10) Tongberg, C. O., Lawroski, S., and Fenske, M. R., Ibid., 29. 957 (1937). (11) Walsh, T. J., Sugimura, G. H., and Reynolds, T. W., Ibid., 45, 2629 (1953). (12) Weedman, J. A., and Dodge, B. F., Ibid., 3 9 , 7 3 2 (1947). (13) Yoshida, F., Koyanagi, T., Katayama, T., and Sasai, H., Ibid., 46, 1756 (1954). RECEIVED for review September 13, 1954. ACCEPTED November 5 , 1954. Material supplementary to this article has been deposited as Document No. 4501 with the AD1 Auxiliary Publications Project, Photoduplication Service, Library of Congress, Washington 25. D. C . A copy may be secured by citing the document number and by remitting $1.25 for photoprints or $1.25 for 35-mm. microfilm. Advance payment is required. Make checks or money orders payable to Chief, Photoduplication Service, Library of Congress.
Fractionation of Mixed Aromatics from Udex Extraction E. R. FENSKE
AND
D. B. B R O U G H T O N
Universal O i l Producfr Co., Dee Plainer, Ill.
U
D E X process was developed to assist in meeting the greatly expanded demand for benzene, toluene, and xylene of aufficiently high purity for chemical use. The National Production Authority has estimated t h a t requirements for aromatics will increme 30% from 1952 to 1955, even under a normal peacetime economy. Since the potential increase in supply from coke oven operation is small, the additional requirements are expected to be met largely from petroleum sources. Udex has been directed chiefly a t utilizing this source by recovering pure aromatics from reformed naphthas. The Udex process w w originated by Dow Chemical Co. and engineered and licensed by Universal Oil Products Co. Recovery of pure aromatics from mixtures with nonaromatic hydrocarbons is complicated by the occurrence of minimum boiling azeotropes. Benzene, for instance, will form azeotropes with nonaromatics boiling within about 30" F. of itself and toluene with those boiling within 15' F. This behavior prevents production of pure aromatics b y simple fractionation. T o illustrate, a batch column with infinite trays and infinite reflux operated to produce narrow cuts t o contain, respectively, all the benzene and all the toluene from an average reformate, would give products containing only about 40% benzene and 75% toluene. I n actual continuous operation these concentrations would be still lower. By utilizing differences in chemical type rather than in relative volatility, the Udex process simultaneously extracts all aromatic types from the feed stock, t o produce an extract of benzene, toluene, and xylenes containing less than 0.2% nonaromatics. Postfractionation of the aromatic mixture to produce t h e individual aromatics is relatively easy, because of the wide differences in volatility between benzene and toluene and between toluene and the xylenes. The necessity of precise prefractionation, to produce narrow cuts containing each aromatic singly, and of then treating each cut separately to produce high purity individual aromatics is thus avoided. 714
The main component of the Udex solvent is diethylene glycol, chosen because of its favorable solubility-selectivity relation for hydrocarbons, low solubility in hydrocarbons, stability, and noncorrosive nature. I t s low volatility is advantageous in eliminating the heat load that would be required to vaporize a volatile solvent. Heat requirements are further reduced by operating t h e extractor and stripper at substantially the same temperature and thus avoiding sensible heat loads on the circulating solvent. Water is used as a secondary component of the solvent, to adjust solubility and to provide a stripping medium. The boiling range specifications of
