LUBRICATING OILS SOLVENT EXTRACTION IN A REFLUX EXTRACTION UNIT M.R. CANNON AND M.R. FENS= The Pennsylvania State College, State College, Pa
inoleetilar size aiul is operable imly bel&, tlir cracking t&perat,itre; on the other hand, sulreirt pro( played for separation according to eitlitiler type arid are operablc over the ooi,ipli:te rarigc of tiyilrocarbons that occur in petroleitrii. The efiificiency i d a solverit process is ciepeiidetrt upmi two iiiajor factors---iiu,iiely, the selectivity of tlie solvelit and the type of process employed. Tlie first of tliesn frotors has mceived a great deai of study in many iahoratorics and resulted in tlie prxctieal application of different solvents. The srcoird Itnotor has receivcd less attention and its possibilities I i a v ~not heerr fully apprcciateii. 11 a i d Fcnske (4)sliowed the ttm)rt:tical reIxtiwis that clraract,erize the various types of proc Of , rcfltix extraotion is iirriong tlie niost eflicicrrt. It is ~inssiihto obtain the same end products, with a relatively pour solvent in a reflux process, BS can iic obtained with a very selective solvetrt in it process that does not ernploy roflux. This fact materially increases the iiurnber of solvents that ciin be etnployed for a given purpose; Ern altl~oiiglireflux itiems greater energy coiisuniption, many solvents ciiti be
A solvent extraction unit employing reflux is described. With this unit it is possible t o increase the efficiency of the process of solvent extraction t o such an extent that solvents formerly regarded as relatively poor can produce the same end products as solvents known to be very selective. This materially increases the number of solvents that may be employed for specific purposes. Quantitative data are presented on the operating characteristics of this type of extractor, and comparisons are made with other types showing the reflux extractor t o be very efficient.
tho oil. The solution r,f 0i1 in rulvrot p r o ~ . e tdt,he ~ d ~ t o p of ti;e tower whero i t strikes w. s t ~ l t ~coil n which vspm The vaporized solvent passes u p ~ a r dt,llrough % steam-l~?ated numihid and is c0ndenst.d Ili,ove i,ire liquid level in the tow?? so that ii,is fed bauk intu the Imw crf t,hr 1.0wr by gravity. Tile rate rri solvent flow is therefore governd Ity t,lie liquid driving head ernph,.vrd, arid can be varied simply hv adding or remnving solvent,from tbe system. In the disengaging section the temperature is much biglm than that, in t,he tower section, and consrquent,ly more oil will dissolve in tile s c d w x l t in this locality dhan in any u t l w pwt oi the unit. As more and more oil is brought into this srrt.ion by the rising solvent, l l i e quantity of oil that the fixed amount of solvent which is maintained there can dissolve is woii exceeded, causing precipitation. The oil prcseii>itaI,esaut
unt,il eqiiilihrium conditiorw itw
I
reaelied. whereupon t,lie entire
T h e t.ower is again'placed in opemiioil until equilibi.iurn conditions m c reachrd, and then a, second cui, is withdrawn. The ent,ire tower section is jacketed so tliat i.lie temperitture of operat,ion may be controlled. The solvent flow is much greater in quarrtit,y than t,he flow of oil that i s refluxed; consequrntly the first f ~ w inchev o f tlie top of i.lie foivrr section m e all that is affeckd by the near-by sf,eittn coil. The ton-ex seotion of this unit i s 10 feet (3 meters) long and 1 inch (2.5 em.) in diameter, the disengaging section is 1.5 feet (45.7 ern.) long and 2 inches 1035
INDUSTRIAL AND ENGINEERING CHEMISTRY
1036
TABLEI.
PROPERTIES O F PRODUCTS OBTaINED BY SOLVENT WITH
Original Oil 94.5
Designation Kinematic viscositv indexD Viacosity : Centistokes a t 100' F. (37.8' C.) Saybolt sec. a t 100' F. Centistokes a t 210" F. (98.9' C . ) Saybolt Bec. at 210' F. Sp. gr 20/20° C. R e f & i v e indEx at 20' C., ng Towertemp C. Solvent flow,'liters/hr. Extn. time, hr. yie&y#rams
VOL. 28, NO. 9
EXTRACTION O F A LUBRICATIXG OIL ACFTONE
Run 1, Charge 430 Grams First ext. Second ext. Raffinate -33.4 6.7 110.0
Run 2 , Charge 430 Grams First ext. Raffinate -12.3 107.2
REFLUXEXTRLCTOR
IN .4
R u n 3, Charge 438 Grams First ext. Holdup Raffinate -18 1 91.0 122.0
151.7 109.7 32.1 119.2 32.9 128.5 39.8 700 505 150.2 562 154 592 185.4 9.23 8.41 5.30 8.45 5.33 8.69 5.74 55.6 52.8 42.8 53.0 42.9 53.8 44.2 0.9585 0.9351 0.8756 0.9412 0.8629 0.9455 0.8795 1.4853 1.5370 1.5257 1.4733 1.5289 1.4760 1.5318 1.4872 .... 9.0 9.0 9.0 23 23 12 .... .... 5.4 5.4 ... 4.6 .... 4.9 .... .... 2.7 3 .... 3 .... 6 .... 75 15 300 73 325 81 77 17 18 17 4 70 17 75 a Calculated according to data of Hersh et al. ( I ) ; Saybolt viscosities obtained from centistokes by conversion of McCluer and Fenske ( 2 ) . 38.1 177.4 5.64 43.9 0.8746
.... ....
~~
TABLF 11.
hr.
Extn. time, hr. Weight, grama Yield, % a
Original Oil 94.5
38.1
177.4 5.64
Run 1, Charge 417 Grams Ext. Raffinate -47
159 5 736
9.31
43.9
55.9 12
.... .... .,..
4.6 3 69 17
....
....
1.4712 ....
....
.... 280 64
~
PROPERTIE8 O F PRODUCTS OBTAINED BY SOLVENT E X T R ~ C T IOO FN .i LUBRICATIXG OIL IN WITH ACETONEBY SEMI-CONTISUOUS OPERATIOS
Designation Kinematic viscosity indexa Viscosity: Centistokes a t loOD F. (37.8' C.) Saybolt sec. a t looo F. Centistokes a t 210' F. (98.9' C.) Saybolt sec. a t 210" F. Tower temp. ' C. solvent flow', liters/
29.5 138.7 5.19 42.5 0.8541
119 30.3 142.1
Run 2, Charge 417 Grams Ext. Raffinate -27
140.0 646
5.22 42.6
.... .... ....
265 64
8.95
116
30.8
144.2 5.22
Run 3, Charge 417 Grams Ext. Raffinate -9 128,5 593
8.86
113
Run 4, Charge 440 Grams Eut. Raffinate -32
31.1 145.9 5.23
150.2 694 9.20
114
31.1 146.0
5.24
54.7
42.6
....
54.4 12
42.6
55.5 11
4.3 3 90
.... ....
4.4 3 95 23
.... ....
3.1 .... 3.25 . . . . 88 363 20 82
12
22
314 75
....
338 81
42.7
....
Run 5, Charge 440 Grams Ext. Raffinate -26
141.0
651
9.01
54.9 12 3.1 3.25
107 24
116 31.0
145.2 5.26
A
REFLUXEXTRiCTOR
Run 6. Charge 440 Grams Ext. Raffinate -29
142.3 657
8.99
114
31.1 145.9
5.25
Run 7 Charge 440 'Grams Ext. Raffinate -34
114
149.4 690
9.14
31 0 145 4 5.23
42.7
42.8
....
54.8 12
42.7
55.3 11
....
2.9
....
2.9 .... 3.25 . . . . 92 346 23 79
....
315 71
3.25 92 21
.... ....
364
83
....
Calculated according to data of Hersh et al. ( 1 ) ; Saybolt viscosities obtained from centistokes by conversion of McCluer and Fenske ( 2 )
(5.1 cm.) in diameter. There are several contacting sections t h a t can be attached t o the tower. The one described here is 4 feet (1.2 meters) long and 1.5 inches (3.8 cm.) in diameter with a capacity of approximately 1 liter. The unit is operated entirely
by exhaust steam and needs no supervision.
Discussion of Results Table I contains the results of processing a light oil by acetone in the reflux extraction unit shown in Figure 1. It is apparent that an excellent separation was obtained. The viscosity index of each extract cut was decidedly negative; that of the raffinate was increased almost one point per per cent of extract yield. Furthermore the sharpness of separation is pronounced; for example, the first cut of run 1 had a viscosity index of -33 and a 100" F. (37.8' C.) viscosity of 700 Saybolt seconds, whereas the second cut had a viscosity index of f6.7 with a 100' F. viscosity of 505 Saybolt seconds. Comparison of runs 1and 2 indicates that a low temperature is advantageous inasmuch as the extract was of lower viscosity index a t 9" C. than it was a t 23' C. In runs 1and 2 the same amount of solvent was recycled and the same yield of extract was obtained, but a t 9" C. the viscosity index of the extract was -33 with a 100' F. viscosity of 700 Saybolt seconds, while a t 23" C. the viscosity index was -12 with a 100" F. viscosity of 562 Saybolt seconds. The raffinates of runs 1and 2 were obtained by completely draining the entire tower a t the end of each run. Since the tower section is filled with acetone which contains oil of quality somewhat inferior to that in the oil phase, it is obvious that such a procedure is not a true indication of the maximum efficiency of this unit. Accordingly, in run 3 the oil in the acetone phase was recovered separately and its properties are listed under the term "holdup." With this technic the viscosity index of the raffinate was 122, but the yield was only 64 per cent. It is obvious that the tower holdup has an important influence on the type of end products obtained. A true indica-
tion of the maximum efficiency obtainable in this unit could be determined either by making the oil charge so large that the holdup would be unimportant or by operating the unit in a semi-continuous manner without removing the solvent phase from the tower but simply removing the oil phase and recharging with the same stock until equilibrium conditions were reached. The latter method was used. A series of runs was made in the manner described above; i. e., the acetone solution in the tower proper was not removed but the oil phase only was withdrawn and a new charge of oil was added for each run. The data from these experiments are presented in Table 11. After run 3 the extract and raffinate yields were equal to the charge, indicating that the holdup quantity had become a constant, and for the following runs a yield of approximately 80 per cent of raffinate with a viscosity index of 114 was obtained. Comparison with Table I shows that operation in the semi-continuous manner materially increases the efficiency, for in simple batch operation the removal of 20 per cent resulted in a raffinate viscosity index of 110 whereas in the second method of operation the viscosity index was 114. A second important method for evaluation of the efficiency of this reflux extractor is t o compare it with simple batch extraction. The same oil was treated with acetone using a 2-to-1 solvent-oil ratio and stirring in a flask a t 25" C. An extract yield of 15 per cent was obtained with a viscosity index of 48, and the raffinate had a viscosity index of 103. When a series of twelve flasks was employed in countercurrent operation and a 2-to-1 solvent-oil ratio, the viscosity index of the extract phase for 15 per cent removed was 5 while the raffinate was 106. In view of the fact that the unit has been proved to be efficient, it is interesting to compare it when operating with acetone to a more selective solvent such as Chlorex (p, p'dichloroethyl ether), Page, Buchler, and Diggs (3) showed that Chlorex, when applied to an oil of viscosity index 77
SEPTEMBER, 1936
INDUSTRIAL AND ENGINEERING CHEMISTRY
with a 2-to-1 solvent-oil ratio a t 25’ C., gave an extract yield of 25 per cent oil with a viscosity index of 5 and a raffinate of 104 viscosity index. They characterize the effectiveness of a solvent by the term “selectivity” which is expressed as the viscosity index of the raffinate minus the viscosity index of the extract. Chlorex had, accordingly, a selectivity of 98, and their data show acetone to have a selectivity of 40. However, when acetone is employed in a reflux extractor at 9” C., it has a selectivity of about 145. It is now evident that, in an attempt t o classify and describe the operation characteristics of a solvent process, the method of operation employed is just as important as the solvent used. A better combination for a solvent process would be a highly selective solvent of good solvent power operating in a highly efficient unit. Since the quantity of low-viscosityindex material in a given oil is fixed, the end products may be the same if a relatively poor or very good solvent is employed. However, a selective solvent should require much less reflux and consequently less energy consumption. There are many other factors to be considered; Consequently the quedion of
IIT H E
1037
which combination to employ becomes a problem in economic balance. In a subsequeqt paper the application of this extraction unit to the study qf the composition of lubricating oil will be discussed. ‘, . !,
Acknowledgment The authors are indebted to R. H. McCormick and J. N. Bresnowitz for their aid in securing many of the data. This paper is published by permission of the Graduate School of The Pennsylvania State College.
Literature Cited (1) Hersh et al., IND. ENG.CHEM.,27, 1441 (1935). (2) McCluer and Fenske, I b i d . , 27, 82 (1935). (3) Page, J. M., Buchler, C. C., and Diggs, S. H., Ibid., 25, 418
(1933).
K. A,, and Fenske, M. R., t o be published. RECEIVED June 16, 1936. Part of a thesis submitted by M. R. Cannon in (4) Varteressian,
partial fulfillment of the requirements for the degree of doctor of philosophy in chemical engineering a t The Pennsylvania State College.
DISSOLUTION
BY J a m e s Gillray
O I t is to be expected that James Gillray (1757-1815), the famous English caricaturist,
would utilize the ideas of Teniers, van Ostade, Steen, and other famous European painters, as shown in their works which have appeared in the Berolzheimer Series of Alchemical and Historical Reproductions. Many of Gillray’s twelve thousand drawings are political in nature, and “The Dissolution” or “An Alchemist Producing an -4etherial Representation,” No. 69 in our series, is no exception. .41so, it is particularly applicable to present-day conditions. “Many amalgam did I make, Wenying to fix these to grett avayle And there to sulfur dyd I take Tartar, eggs whyts and the oyle of snayle. But ever of my purpose dyd I fayle, For what for the more and what for the lesse, Evermore something wanting there was.” “The Compound of Alchymie,” by George Ripley, 1475.
-4 detailed list of the first sixty reproductions, together with full particulars for ohtainin photographic copies of the originals, appearef in onr issue for January, 1936,.page 129, where also will be found Reproduction No. 61. Reproduction No. 62 appears on page 241 of our February issue, No. 63 on page 280 of March, No. 64 on age 413 of April, No. 65 on page 572 of May, 0.66 on page 677 of June No.67 on page 788 of July, and No. 68 on page 9i4 of August.
5
