September 1950
INDUSTRIAL AND ENGINEERING CHEMISTRY LITERATURE CITED
CONCLUSIONS
1. Increasing the ratio of holdup-to-charge in a series of batch distillations gave a sharper separation, a less sharp separation, or made no difference in the sharpness of separation. The result obtained depended on the reflux ratio. 2. As the reflux ratio was increased, with all other variables held constant, increased holdup became more and more detrimental to sharpness of separation. A critical reflux ratio was noted at which holdup had no effect on the sharpness of separation. As the reflux ratio was increased above this critical value the effect of increased holdup became increasingly detrimental to sharpness of separation. At reflux ratios below this critical value the effect of increased holdup was beneficial. 3. Varying reflux ratio had little effect on the shape of distillation curves for large holdup. However, the effect of increasing reflux ratio, when the ratio of holdup to charge was small, was to improve the sharpness of separation markedly. As the reflux ratio was increased the sharpness of separation for small holdup became sharper until approximately superimposed on the curve for large holdup. This point is referred to as the critical reflux ratio. If the reflux ratio is increased beyond this pzint, sharpness of separation for large holdup still is not improved appreciably, but the sharpness of separation for small holdup increases markedly, and holdup becomes detrimental to sharpness of separation. 4. The critical reflux ratio was different for columns with different numbers of theoretical plates. For column 2 (20plates) the critical reflux ratio was 9:1, whereas under the same conditions the critical reflux ratio for column 1 (40 plates) was more than 24:l.
1879
(1) Bowman, J.
R., and Briant, R. C., IND.ENQ.CHEM.,39, 745
(1947). (2) Bowman, J. R.,and Cichelli, M. T., Ibid., 41,1985 (1949). (3) Colburn, A. P.,and Steams, R. F., Trans. Am. Inst. Chem. Engrs., 37,291-309 (1941). (4) Edgeworth-Johnstone, R., IND.ENQ. CWEM.,36, 1068-70 (1944). (6)Houston, Reagan, M.S. thesis, Pennsylvania State College (1947). (6)Pfeiffer, Carl, M.S. thesis, Pennsylvania State College (1947). (7) Pigford, R. L., Tepe, J. B., and Garrahan, C. J., presented at Third Meeting-in-Miniature, Philadelphia Section, AMERIC4~ CHEMICAL SOCIETY (January 1949). (8)Quiggle, Dorothy, and Fenske, M. R., J . Am. C h a . Soe., 59, 1829-32 (1937). (9) Rose,Arthur, IND. ENG.CHEM.,32, 675 (1940). (IO) Rose, Arthur, Bailey, C. R.,and Bertram, L. L., presented before the Division of Petroleum Chemistry at the 102nd Meeting, AMERICAN CHEMICAL SOCIETY, Atlantic City, N. J. (11) Rose, Arthur, and Houston, Reagan, presented before the Division of Industrial and Engineering Chemistry at the 113th Meeting of tho AMERICANCHEMICAL SOCIETY, Chicago, 111. (12) Rose, Arthur, Johnson, R. C., and Williams, T. J., IND.ENO. CREM.,in press. (13) Rose, Arthur, and Welshans, L. M., Zbid., 32,668 (1940). (14)Rose, Arthur, Welshans, L. M., and Long, H. H., Ib$d., 32, 673-5 (1940). RECEIVED October 3, 1949. Presented before the Division of Industrial and SOCIETY, Engineering Chemistry ut the 116th Meeting, AMERICANCHEMICAL Atlantic City, N. J.
Hydrodesulfurization of Heavy Petroleum Oils E. C. HUGHES, H. M. STINE, AND R. B. FARIS The Standard Oil Company (Ohio), Cleveland, Ohio
T h e sulfur contents of several high-sulfur gas oils and a high-sulfur reduced crude were reduced to those of corresponding fractions from a low-sulfur Illinois crude by relatively mild hydrodesulfurization. The heavy oils were desulfurized over a cobalt oxide-molybdenum oxide-aluminum oxide catalyst at 750' F., lo00 cubic feet of hydrogen per barrel of hydrocarbon, and 300 pounds per square inch pressure. In some cases a portion of the charge was in the liquid phase. The distribution of sulfur between the paraffinic-naphthenic fraction and aromatic fractions before and after desulfurization of the gas oil was determined by silica gel fractionation and sulfur analyses. The reduced crude was distilled and asphalts prepared from the bottoms.
T
H E removal of sulfur from heavy petroleum fractions, such as gas oils and reduced crudes, by hydrogenation was undertaken only after it was clear that less expensive methods such tu caustic washing and cracking were unsatisfactory. Since the bulk of the sulfur in such fractions is usually of an unreactive cyclic type, an easy solution to the problem would not be expected. Hendricks, Huffman, Parker, and Stirton ( 9 ) have reported the hydrodesulfurization of two Santa Maria 'aa oils. One waa a 33" A.P.I. Santa Maria virgin gas oil co:itsining 2.3% sulfur and the other a 20" A.P.I. coker gas oil fro i Santa Maria crude containing 3.6% sulfur. A cobalt oxidc -molybdenum oxidealuminum oxide catalyst was used at a temperature of
760' F., 250 pounds per square inch total pressure, and a hydrogen circulation rate of 3000 to 0000 cubic feet per barrel of charge. Brown, Voorhies, and Smith (I) described the use of high pressure hydrogenation for the combined hydrogenation and desulfurization of gas oil at 750 to 3000 pounds per square inch over an undisclosed catalyst. The desulfurization of gas oils described below differs from the experimental work described by Hendricks et al. ( 9 ) in that the gas oils which were processed were of lower sulfur content, and the hydrogen circulation rates were lower. The objective was reduction of the sulfur contents to thoge of gas oils from sweet crudes. Conventional refinery equipment would handle such stocks. The hydrodesulfurizations were carried out a t mild operating conditions which would accomplish this objective. MATERIALS AND APPARATUS
CATALYST.The catalysts used in this experimental work were cobalt oxide-molybdenum oxide-aluminum oxide compositions. The preparation of this type of catalyst has been reported by Byrns, Bradley, and Lee (2). In the initial work, samples of this catalyst were obtained from the Union Oil Company. Later, such catalysts were prepared in the authors' laboratory. The catalysts were used as '/cinch pellets or 10- to 20-mesh granules. APPARATUS.The catalytic units used were of conventional fixed-bed design. The feed stock was delivered by a reciprocating pump to the inlet manifold at the top of the reactor where it was mixed with either hydrogen or recycle gas. The reactor was stainless steel and held 200 ml. of catalyst. The oil and hy-
INDUSTRIAL AND ENGINEERING CHEMISTRY
1880
-
I
Figure 1. Laboratory Hydruclesulfurization Unit
CONDlTlONS OVER COBALT MOLIBDATE O N ALUMINA AT
75m.L 3w,ur, .L
0.4-
$023
i
L
HYDRODESULFURIZATION OF G A S OILS
N
I
I
When recycle gas was used, the gas from the product receiver was directed to a caustic scrubber where hydrogen sulfide waa removed as shown in Figure 1. The effluent from the caustic scrubber went to the suction side of the recycle unit. The recycle unit comprised a laboratory aspirator through which ethylene glycol was pumped by a gear pump. The glycol and recycle gas from the aspirator went t.0 a gas separator. By using a conventional laboratory water aspirator it was possible
24
I-
1. The system was operated with gas recycle until recycle gas was of constant hydrogen content. 2. The system was isolated by closing make-up hydrogen inlet and product gas outlet, and it was operated 12 to 24 hours without product removal. The pressure was observed at the end of the petiod. The size of the receiver was chosen such that the decreasr in gas volume brought about by liquid roduct accumulation in the receiver maintained approximately t i e oprrating pressure. 3. The receiver was drained and the final pressure noted. The draining of the receiver lowered the pressure to 50 to 100 pounds er square inch below the operating ressure. 4. TEe hydrogen consumption was calcukted from the pressure differelice, the known volume of the system, and the hydrogen content of the recycle gas.
Regeneration of the catalyst was carried out by passing an airnitrogen stream through the catalyst. In some instances the catalyst was first extracted with benzene to remove tarry materials. Silica gel separations of gas oils before and after desulfurization were carried out as described by Mair, Sweetman, and Rossini (4). These separations combined with sulfur analyses determined the sulfur distribution between paraffins-naphthenes and aromatics. The absorbant was fresh silica gel with a particle size distribution of 60% between 40 and 80 mesh. The desorbing liquid was ethyl alcohol. The inert gas pressure applied to the column was such that the rate of flow of liquid was 10 to 20 ml. per hour.
IO W H OF 27'ARI PAoAFFlN DISTILLATE (I,?!@
a
42, No. 9
to build up a pressure of ahout 10 pounds above that of the system. The gas flow from the separator, whirh was controlled by a needle valve, was then metered through a rotameter back to the catalyst. The system was found to be very satisfactory for recycling a small amount of gas under pressure. Since it was an entircly closed system with no mechanical gas pumps, it was particularly advantageous in the estimation of hydrogen consumption on a small scale where even very small leaks in the system ran ruin the experiment. Hydrogen consumption data were obtained in the following manner :
MARE-UP WDROGEN
t
VOI.
CHARGE ST(XK- SO&
2ol
W T E X A S REO CRUDE
The sulfur contents of the stocks processed were reduced successfully to those of gas oils of the same boiling range derived from sweet crudes by hydrodesulfurization a t relatively low hydrogen circulation rates of 1000 cubic feet per barrel. A 36' A.P.I. Illinois crude oil containing 0.19% sulfur was taken as representative of sweet crudes. The effect of the hydrogen circulation rate upon the hydrodesulfurization of a 27' A.P.I. paraffin distillate from Mississippi-Louisiana crudes is shown in Figure 2. A circulation rate of 1000 cubic feet per barrel reduced the sulfw- content from 1.2 to about 0.25%. Higher circulation rates did not show a marked advantage and lower rates were noticeably poorer. At this hydrogen circulation rate 1500
I
I,O
K
1
CLl:
BOTTOMS AT e50mF.(CORR ATM
PRESSURE:
ac IPRODUCT
.-
1
HOURS
Figure 3.
ON
STREAM
Hydrodesulfurization of Reduced Crude
1
1
BOTTOMS 1
1
September 1950
INDUSTRIAL AND ENGINEERING CHEMISTRY
-I-
TABLE I. HIGHBOILINGOVERHEAD STOCKS HYDRODESULFURIZED (Conditions: 1.0 vol./vol./hr. of stock listed below over cobalt oxide-mokybdenum OXide-alUminUm oxide at 300 lb./sq. inch, 750' F.,and 1000 cu. ft. hydrogen/barrel of hydrocarbon) -Sulfur Conten*Fraction fiom Hydrogen sweet Consumption, Stock Before After crude" Cu. Ft./Bbl. 38O A.P.I. light Hawken gas 0.14% ... oil 0.55% 0.10% 36' A.P.I. light West Texas 0.16 0.16 80 gaa oil Shell Oil Company) 1.1 32' A.P.i. wide cut Hawken 0.24 1.4 0.20 gas oil 29 A.P.I. wide cut W e 4 Texas gaa oil (Shell Oil 0.30 1.1 0.30 Corn any) I. wide cut West 27ke!i'ga oil (Standard 0.33 0.25 2.0 ... Oil Corn any of Indiana) 27O A hlissiasi pi0.25 0.41 120 Louisia% 'araffin distilyate 1.2 25O A.P.I. $est Texas fluid CatalytlC gas oil (Shell oil 0.36 0.35 170 Company) 1.6 22O A.P.I. MissisaippiLouisiana light cylinder 0.45 150 0.50 1.5 stock
...
...
31
4 Undeaulfuriaed fraction of corresponding boiling range from 36O A.P.I. Illinois crude (0.19% sulfur on total crude).
TABLE 11. PROPERTIES OF HIGHBOILING OVERHEAD STOCKS Engler, O F. _______ ~Aniline Pt.,
Stock 38" A.P.I. light Hawken gas oil 35O A.P.I. light West Texas gas o!l 32' A.P.I. wide cut Hawken gas 011 29O A.P.I. wide out West Texas gas oil 27O A.P.I. wide cut West Texas gas oil 27O A.P.I. Mississippi-Louisiana garaffin distillate 25 A.P.1. West Texas fluid Catalytic gas oil 22O A.P.I. light cylinder stock
10% 30% 60% 90% 458 486 616 612 460 502 638 623 440 520 622 720
O F .
160 154 158
548
624
674
784
569
610
651
751
162
635
682
714
744
187
811 838
643 858
658 872
687 940
161 200
175
volumes of feed per volume of catalyst were processed without noticeable decrease in activity. The effect of other procase variables such as hydrocarbon flow rate were found to be as reported by Hcndrich et al. (3). A list of the high boiling overhead petroleum cuts which were hydrodesulfurized , a t relatively mild conditions (750' F., 300 pounds per square inch, IO00 cubic feet per barrel) is given in Table I. These stocks ranged from a light gas oil to a light cylinder stock with a fluid catalytic cycle stock included. The properties of these stocks are given in Table 11. In all cases the sulfur contents of the products were no greater than the sulfur contents of the same boiling range material from undesulfurized Illinois crude. Hydrogen consumption varied from 80 cubic feet per barrel for the light West Texas gas oil to 170 cubic feet per barrel for the catalytic cycle stock.
1881
Z I SAMPLE A ( 2 . 0 x S )
--c
x 0.8
0
I
40
I I I I20 I80 200 HOURS IN STREAM
0
I 240
I
320
3 0
Figure 5. Hydrodesulfurisation of T h e r m a l Cracking Tars
0.5% sulfur which is the sulfur conbent of the 0.5 vol./vol./hr. product during the initial portion of the run. At all other conditions the desulfurization waa not aa complete. The hydrogen consumption was approximately 225 cubic feet per barrel of reduced crude. A product from the 0.5 vol./vol./hr. run and the ohsrge were vacuum fractionated into four overhead cuts and the bottoms. The cut point waa approximately 850' F. when corrected to atmospheric pressure. The properties of the materials distilled are shown in Table I11 and the sulfur distribution in the fractions is shown in Figure 4. The desulfurization of the gas oil portion of the sample distilled was comparable to that obtained if the gas oil were processed in the absence of the bottoms materials. The desulfurization of the heavy portion of the reduced crude, although it must have passed through the unit unvaporized, was nearly as well desulfurized as the lighter vaporizable portions. As indicated in Table 111a viscosity breaking occurred. The catalyst lost desulfurizing activity more rapidly in processing the reduced crude than in the gas oil desulfurization. The catalyst was regenerated satisfactorily by air burning. Poisoning of the catalyst by salt deposition was not observed and no change in salt content of the reduced crude during desulfurization was noted.
TABLE111. HYDRODESULFURIZATION OF 500' F. REDVCEU WESTTEXAS CRUDE (Conditions:
0.6 vol./vol./hr., 400 lb./sq. inch, and 2000 cu. ft. hydrogen barrel Of charge) Charge Product
HYDRODESULFURIZATION OF REDUCED CRUDE
Instead of processing the gas oil, it is possible to hydrodeaulfurize the entire crude or a reduced crude. Removal of sulfur from the reduced crude would reduce the over-all refinery corrosion but would be more expensive since larger amounts of stock would have to be processed and the catalyst would probably deteriorate more rapidly. The course of the desulfuriaation at 300 and 400 pounds per square inch of a 500' F. West Texas reduced crude obtained from the Shell Oil Company is shown in Figure 3. A considerable portion of the charge was in the liquid phase. The reduced crude was desulfurized at two conditions as indicated. The catalyst lost desulfuriring activity at about the same rate a t both 0.5 and 1.0 volume per volume per hour (vol./vol./hr.) even though only one half as much stock was processed per unit of catalyst rtt 0.5 vol./vol./hr. A sweet crude reduced to the same extent would contain about 0.4 to
TABLEIV. ASPHALTS
DESULFURIZED WEST TEXAS CRUDES
FROM
Stock Preparation
Undesulfurieed Vacuum reduction to 254 penetration and air oxidation Yield, % on reduced crude 29 Penetration, Ductility 77O F. 73
77O F. 60° F. Soft point, rinnand ball, Penetration 77O F. Penetration 3 P k'. sulfur
F.
loo+ loo+ 118
Desulfurized Vacuum reduction to 200 peneVacuum tration and air reduction oxidation 16 19 80 90
loo+ loo+
3.4
116 4.2
3.1
1.6
92 42 114 3.6 1.8
INDUSTRIAL AND ENGINEERING CHEMISTRY
1882
The bottoms of the reduced crudes were flashed in order to prepare vacuum-distilled asphalts and bottoms for oxidation to asphalt. As listed in Table IV one vacuum-reduced and one oxidized asphalt were prepared from the desulfurized stock and one oxidized asphalt was prepared from the undesulfurized. The asphalts prepared from the desulfurized and undesulfurized bottoms were essentially the same except that the ductility of the oxidized desulfurized bottoms was less than 100 em. The temperature susceptibility as measured by the ratio of penetration at 77" F. to penetration at 32" F. was poorer than those of mid-continent asphalts. Probably more hydrogenation is needed to improve the temperature susceptibility.
Vol. 42, No. 9
coal tar is shown. I t was possible to desulfurize one of the samples to approximately 1% sulfur for an extended period. However, the other one was extremely difficult to desulfurize and soon blocked up the catalyst chamber with coke. Analyses made on t.he two samples are shown in Table V. The only appreciable difference noted was the resin content as determined by Hoiberg analysis. It seemed likely that this resinous material was fouling the catalyst and thermal polymerization of this material blocked the catalyst chamber.
TABLE V. ANALYSES OF THERMAL TARSAMPLES Sample Source
+Zi$fo
F. iscosity, Furol Pentane inaolubles Resins by Hoiberg Conradson carbon
A
€3
Miss.
Miss.-La
4.0
2.0
49 6.6 16 14
10.4 2.2 85 6.1 34 12
HYDRODESULFURIZATION OF HYDROCARBON CLASSES AS SHOWN BY SILICA GEL ADSORPTION
4 3.O 1
t
3.0
DESULFURIZED CATALYTIC CYCLE STOCK (035'l.S)
A 29" A.P.I. West Texas virgin gas oil obtained from the Wood River Refinery of the Shell Oil Company was subjected to silica gel fractionation. Five fractions were separated, one of which contained paraffins and naphthenes and the remaining four contained the aromatics. The latter were taken by elution from the column so that they represent increasing adsorbability from left to right in Figure 6, above. These data show that most of the sulfur in the charge is concentrated in the aromatic portions. The more difficult the elution from the gel, the higher the sulfur content in the fraction. The amount of paraffinic-naphthenic fraction was increased about 5y0 during the desulfurization. A similar silica gel fractionation was made of cycle gas oil from a fluid unit running on West Texas gas oil. This cycle stock was obtained from tbe Wood River Refinery of the Shell Oil Company. The data summarized in Figure 6, below, show that t,he paraffinic-naphthenic portion was increased about 10% in the processing. The catalytic cracking had efficiently desulfurized the paraffinic-naphthenic portion but had left the sulfur in the aromatic fractions. The hydrodesulfurization removed 83% of the sulfur in the aromatic fractions from the cycle gas oil but as shown in Figure 6, above, removed only 70% of the sulfur in the aromatic fractions from the virgin gas oil.
l----lJ-7
I
p'
SUMMARY
I
I L-
Figure 6. Sulfur Distribution of West Texas Virgin Gas Oil (above) and Cata-
Several high sulfur gas oils were reduced in sulfur contents to those of gas oils from sweet crudes by hydrodesulfurization over cobalt oxide-molybdenum oxide-aluminum oxide a t 750" F., 300 pounds per square inch pressure, 1.0 vol./vol./hr., and IO00 cubic feet of hydrogen per barrel of charge. A similar desulfurization of a 500" F. reduced West Texas crude was effected at 750" F.,400 pounds per square inch pressure, 0.5 vol./vol./hr., and 2000 cubic feet per barrel of charge.
HYDRODESULFURIZATION OF PETROLEUM TARS
LIqERATURE CITED
4
I
12 . 2 b ' 4 0 'A THROUGH COLUMN
* L
Iro
' I & '
lytic Cycle Stock (below) before and after Desulfurization
TARS. In Ohio a considerable amount of heavy fuel oil is supplied to steel mills. Some steel companies have set specifications of 0.75% sulfur on the heavy fuel used in their open hearth furnace operations. The tar obtained from thermal cracking operations is one of the components of the heavy fuel oil. If the tar has no more than 1% sulfur, it can be disposed of satisfactorily by blending with low sulfur components. Some tars obtained from Mississippi and Louisiana crude processing have 1.5 to 2.5% sulfur. A partial desulfurization would allow their use in heavy fuel. In Figure 5 the hydrodesulfurization of two samples of cracking-
(1) Brown, C. L., Voorhies, Alexis, and Smith, W. M., Irw. ENG. CIIEM.,38, 136 (1946). (2) Byrns, A. C., Bradley, W. E., and Lee, M. W., Ibid., 35, 1160 (1943). (3) Hendricks, G. W., Huffman, H. C., Parker, R. L., Jr., and Stir-
(4)
ton, R. I., paper presented before the Division of Petroleum Chemistry at the 109th Meeting of the AM. CAEM. SOC., Atlantic City, N. J. Mair, B. J., Sweetman, A. J., and Rossini, F. D., IND.ENG. CREM.,41, 2224-30 (1949).
RECEIVED March 26, 1949. Presented before the Division of Petroleum Chemistry a t the 115th Meeting of the AMERICANCXWICALSOCIETT,Ssn Francisco, Calif.
