T
HE rapid increase in the use of high-speed Diesel engines for automotive and locomotive as well as for stationary power has led to a large increase in the demand fur suitable fuels. Between 1032 and 1936 the annual sale of Diesel horsepower increased from 130,000 to 1,830,000. There has also been a tendency toward Diesel engines of lower horsepower, for it has been estimated that 90 per cent of the installations in 1936 C. G. DRYER, J. A. CHENICEK,GUSTAV EGLOFF. were engines of less than 100 horseAND J. c. MORRELL power (6). Universal Oil Products Company,Riverside, Ill. These smaller high-speed engines will require approximately an additional 7,500,000 barrels of fuel on the basis of 4.5 barrels per horsepower per year. Most of this additional fuel will necessarily be of Solvent extraction of cracked Diesel fuels with sulfur higher ignition quality than that dioxide and furfural produced raffinates with improved igniused for the large slow-speed ention quality without materially affecting other physical gines because the high-speed engines operate under widely varying conproperties. The improvement depended upon the solvent, ditions of load and speed. It is the percentage removed, and the method of extraction. The under such conditions that results rdffinates showed unchanged susceptibility to a pour-point accompanying the use of unsatisfacdepressant. The extracted portions had high-octane blendtory fuels are most easily noticed. ing values and low pour points. Under light load conditions, rough running, exhaust fumes, and inExtraction of straight-run fuels caused less improvement creased fuel consumption are more in ignition quality than solvent treatment of cracked fuels. easily observed. Less evident but Acid treatment resulted in a negligible increase in ignition more important effects following quality. Hydrogenation of fuels of low sulfur content prothe use of unsatisfactory fuels are duced fuels of high ignition quality. c a r b o n deDosits o n t h e sDrav nozzles, exhaust valve deposit;, an& piston-ring gum formations. A survey of twenty-five manufachigh ignition and low pour point by this method is limited. turers of high-speed Diesel engines showed that fuels with a Woods' data follow: minimum cetane number of 45 are recommended for highspeed engines (IS). Fuels of similar ignition quality are 100 40 60 80 20 Straight-run fuel, % ' 0 also recommended for use in Diesel electrolocomotives. The Cracked fuel, 100 80 60 40 20 0 71 53 58 66 Diesel index No. 43 48 Army Air Corps and the Bureau of Aeronautics of the Navy 15 30 50 -5 -40 -25 Pour point, F. specify fuels for aircraft purposes with a minimum Diesel index of 60 and a minimum aniline point of 150"F. As previously stated, the need for Diesel fuels of high igniIt is essential that large quantities of fuels of high ignition tion quality and low pour point necessitates the use of special quality be available in order to ensure the continued developrefining processes. The following are possible methods of ment of the Diesel engine. Straight-run fuels produced from treatment : solvent extraction, hydrogenation, acid treatvarious crudes appear to have a sufficiently high cetane ment, mild cracking, dewaxing, addition of dopes, addition number to operate most high-speed engines. Cracked fuels, of pour-point depressants, blending of straight-run and derived from crudes of various fieIds, however, have an cracked fuels, and polymerization. average cetane number of 35 which is well below that recThe work included here is concerned primarily with the ommended by the engine manufacturers (12). A fuel of suffirst of these methods-namely, solvent extraction-which ficiently high ignition quality for most purposes probably is considered most important. Results obtained by the use can be made by blending straight-run and cracked fuels in of some of the other methods are included for comparison. the proper proportions. However, fuels for aircraft engines, Various other means of producing high-quality Diesel fuels besides having high ignition quality, must have other are possible, but before discussing these methods it would be desirable properties such as low cloud and pour points. desirable to show the relation of physical properties and Straight-run fuels which have high ignition quality unfortuchemical composition to ignition quality. nately have high pour points and consequently cannot be used a t low temperatures such as those accompanying use in Ignition Quality aircraft. Cracked fuels, on the other hand, have low pour The ignition quality of a fuel is determined by the ignition points but also low ignition quality. Woods (23) found that delay or the amount of time elapsing between injection and it is possible to prepare fuels of good ignition quality and combustion of the fuel. Paraffins have the shortest and lowered pour point by blending cracked and straight-run aromatics have the longest ignition delay; consequently, they oils. However, it is necessary to use a straight-run fuel of have, respectively, the best and poorest ignition quality. very high ignition quality. The quantity of such fuels is Naphthenes, isoparaffins, and olefins fall between these two limited, and therefore the possibility of producing fuels of
Solvent Extraction of
Diesel Fuels
O
813
INDUSTRIAL AND ENGINEERIKG CHEMISTRY
814
extremes. The effect of a paraffinic side chain on the ignition quality of a Diesel fuel is much less than it is on the antiknock properties of gasoline. The ignitibility of cetane is scarcely effected by.the introduction of a side chain, and no appreciable effect is apparent unless there is a large number of side chains (8). Kreulin (16) was able to show that the chemical composition of a Diesel fuel and its cetane number are closely related. He was able to calculate the cetane number to within five units by the use of the following expression: cetane value = 0.2 A 0.1 N 0.85 P where A = aromatic ring content N = naphthenic ring content P = paraffinic side-chain content as determined by the ring analysis method (21)
+
+
The standard method of evaluating ignition quality is the determination of the cetane number in a standard test engine operated under specified conditions. By cetane number is meant the percentage of cetane in a reference fuel composed of t h a t hydrocarbon and a-methylnaphthalene which has the same ignition characteristics as the fuel being investigated. This test is probably the best available method of evaluating Diesel fuels but involves considerable expensive equipment. For this reason numerous ignition quality expressions have been advanced which are based on readily determined physical properties of the fuel such as aniline point, specific gravity, etc. Certain Diesel engine manufacturers and fuel consumers prefer to specify ignition quality in terms of the Diesel index number proposed by Becker and Fischer (2): Diesel index =
aniline point
(O
F.) X gravity at 60' F. ( O A. P. I.) 100
This expression is used more widely than any other since it is readily calculated and can be correlated with the cetane number as determined in an engine. The viscosity-gravity number, which depends on the specific gravity and kinematic viscosity, has been suggested by Moore and Kaye (17). Their formula is a modification of one used by Hill and Coats in connection with lubricating oils (10) :
+
G = 1.082 A = 0.0887 (0.776 - 0.72A) X loglog (KV where G = specific gravity a t 60" F. A = viscosity-gravity constant KV = kinematic viscosity at 100' F., millistokes
- 4)
The boiling point-gravity number was proposed b y Jackson (14): G=A (68 - 0.703A) log BP where G = gravity, O A. P. I. A = boiling point-gravitz constant B P = 50% boiling point, 0 .
+
The Universal Oil Products characterization factor was suggested by Watson and Xelson (26): K = -Ta' '3
S
where Tb = molal av. boiling point,
O
Rankin
I n the case of Diesel fuels the 50 per cent distillation point in degrees Rankin is used instead of the molal average boiling point. The use of the parachor, a function of the surface tension which was first suggested by Sugden (20)as a method of determining molecular constitution, was introduced by Heinoe and Marder (7) : specific parachor =
where
CY
=
surface tension
0.25
sp. gr.
VOL. 30, NO. 7
The same authors recently used the Siedekennzifler (boiling characteristic) in conjunction with the parachor to determine ignition quality (9). They believe that, while the physical methods of determining ignition quality give an approximate value, the method ,would be more accurate and more useful if the chemical composition of the oil were taken into consideration. The ignitibility of a member of a series of hydrocarbons varies with the molecular weight. Therefore, if a property of the fuel which is related t o the molecular weight is used in a formula for expressing ignition quality, the formula should become more accurate. Heinze and Marder attempted to do this by the use of the Siedekennzifler. This property is determined by distilling the fuel in an ordinary Engler-Ubbelohde apparatus and noting the temperature a t which 5, 15, 25, etc., up to 95 per cent of the fuel by volume distills over. The sum of these recorded temperatures is then divided by 10. The value thus obtained is roughly proportional to the average molecular weight of the oil. By the combined use of this value and the specific parachor Heinze and Marder claim to be able to calculate cetane numbers which do not vary more than 1.9-2.1 from values determined on an engine. The ignition quality number (11) also attempts to take into account the nature of the molecules composing the oil by combining the 50 per cent distillation point of the fuel with the Diesel index number : *
G X A X B P 100,000 ignition quality number where Q A = aniline point, O F. G = gravity at 60" F., O A. P. I. BP 50 per cent distillation point, =
5
5-
O
F.
All of these expressions for ignition quality give results which can be correlated with cetane numbers %s determined in the standard engine. However, the usefulness of most of the formulas is limited since they cannot successfully be applied to doped fuels, fuels of origin other than petroleum, or the standard reference mixture of a-methylnaphthalene and cetane. They are sufficiently valuable, nevertheless, to warrant their use as a means of following the efficiency of a solvent extraction. It will be shown later that the increase in ignition quality is a function of the amount of fuel extracted by a given solvent. Previous Work Those constituents of a lubricating oil which impair its value are the same as those that decrease the ignition quality of a Diesel fuel. It is reasonable to assume that the same types of solvents can be used to extract Diesel oils as are used t o treat lubricating oils. I n both cases the object is to remove undesirable constituents-i. e., aromatics and olefinsand to increase the paraffinicity of the oil. However, the need for Diesel fuels of high ignition quality is relatively recent, and consequently little is available on this subject in the literature. Edeleanu (4) has several patents for the production of fuels of high ignition quality by extraction with liquid sulfur dioxide. Two methods of obtaining such fuels are outlined: (a) an extraction of up to 75 per cent by volume of the constituents soluble in liquid sulfur dioxide or (b) an amount not exceeding 30 per cent of a raffinate produced by liquid sulfur dioxide extraction is added to a Diesel oil. Pyzel (18) discussed the use of sulfur dioxide to extract gas oils to yield raffinates suitable for compression ignition engines and extracts which give, on cracking, gasolines of high antiknock value. More recently Marder (16) improved brown-coal Diesel fuels by extraction with a sufficient quantity of a
INDUSTRIAL AND ENGINEERING CHEMISTRY
JULY, 1938
Woods (23) also investigated the possibilities of other methods of producing fuels of high ignition quality and low pour point. Mild cracking carried out in an iron bomb a t 70G800" F. for 2, 4,or 8 hours caused a lowering of the pour point but there was a considerable cracking loss and decrease in ignition quality. I n one case the pour point was lowered from 35" to -20" F., but there was a cracking loss of 19.1 per cent and the Diesel index number fell from 62 to 50. Pour point depressants such as wool fat or Paraflow, when added in amounts of 0.1 and 0.3 per cent, were most effective on those fuels which already had low pour points so that their use is limited. Other workers (3) also found that fuels of high pour point are not greatly affected by the addition of Paraflow alone but that compounding is necessary to produce fuels of low pour point. Woods (23) also determined the effect of solvent dewaxing on the ignition quality of Diesel fuels. Dewaxing losses were large but it was possible to obtain a fuel of low pour point from a straight-run oil of high ignition quality. A stock with an original Diesel index number of 73 and a pour point of 55 O F. had a Diesel index of 58 and a pour point of - 30 " F. after naphtha dewaxing treatment. Woods also carried out acid treatment of fuels with 98 per cent sulfuric acid a t room temperature. Treating losses were large and there was relatively small gain in ignition quality. Woods concluded that of all the methods used, solvent extraction offered the greatest possibilities.
selective solvent. The raffinates thus produced were stable and could not be distinguished either by appearance or behavior from the best petroleum Diesel oils. Glacial acetic acid was also suggested as a solvent for the extraction of the Diesel oil fraction of petroleum (1). Feigin, Obleukhova, and Prorokov ( 5 ) treated a gas oil from Balakhany topped crude with furfural. They were able to increase the Diesel index by 12 units by extracting 38 per cent of the fuel a t 20" C.; this is not a very large increase, considering the amount removed, but the extract could be converted into aviation fuel by hydrogenation : Volumes Furfural 0
Sp. Gr. a t 150 c . 0,898 0.884 0.874 0.870
Raffinate,
%
73:O 64.0 62.0
1 2 3
Diesel Index No. 62.4 68.4 72.2 74.3
Woods (23) extracted several Diesel fuels a t 25" F. with liquid sulfur dioxide. Batch procedure was used, and the sulfur dioxide was allowed to boil off a t atmospheric pressure. Last traces were removed by caustic and steaming in an open vessel for several minutes. These experiments shorn that cracked fuels can be extracted to produce raffinates of substantially improved ignition quality but that treating losses are relatively large. I n order to make the solvent extraction of Diesel fuels commercially feasible, some use for the extracted portions must be found. Solvent extraction has the advantage of increasing ignition quality without causing a large rise in the pour point of the fuel. A raffinate of 60 Diesel index was obtained with a pour point of -35" F., whereas a straight-run fuel with the same Diesel index has a pour point of about 15" F. The use of more than two volumes of solvent is questionable, for the results indicate that the increase in Diesel index is too small to compensatefor the treating loss and rise in pour point. Table I, taken from Woods, also shows that the low-grade straight-run f'uels do not respond to solvent treatment as well as the cracked fuels. Steffen and Saegebarth (19) also recently investigated the action of liquid sulfur dioxide a t 30" F. on straight-run paraffinic, straight-run nonparaffinic, and cracked gas oils. They found that the cracked gas oils were especially susceptible to treatment and that the straight-run paraffinic fuels were least improved. The results they obtained are as follows: Son used, vol. % Straight-run paraffinic oil: Raffinate, weight % Diesel index No. Straight-run nonparaffinic oil: Ra-ffinate, weight % Diesel index No. Cracked gas oil: Raffinste, weight % Diesel index No.
Treat,
0
50
100
300
100 68
92 73
90 75
84 79
100 47
82 48
77 61
55 67
100 40
65 62
56 73
49 81
815
Experimental Procedure The object of the present investigation was to determine the effect of extracting various amounts of fuels with several solvents, on the properties of representative straight-run and cracked fuels. The relative efficiency of two solvents, furfural and liquid sulfur dioxide, was determined, and the results of extraction with these solvents were compared with results obtained by other methods. No attempt to recommend specifications for Diesel fuels to be used in high-speed engines was made. The solvent extractions with furfural were carried out in a continuous countercurrent extraction tower: The tower consisted of a glass tube 120 cm. (47.2 inches) long and 4 cm. (1.6 inches) in diameter fitted with inlets and outlets for the fuel and solvent. The fuel was pumped into the tower through an inlet tube fitted with a sintered-glass disk which projected in 16 cm. (6.3 inches) from the bottom of the tower. The sintered-glass disk was made from 40-mesh ground glass and served to break the entering fuel into small droplets. The solvent was pumped into the top of the tower through a similar tube which projected in an equal distance from the top of the tower. The space between the inlet and the outlet tubes was filled with glass Raschig rings of uniform size. The treated fuel or raffinate was removed at the top of the tower, and spent solvent was removed at the base of the column. The spaces between the inlet
TABLE I. EXTRACTION OF DIESEL FUELSWITH SULFUR DIOXIDE --loo7 - - z x loo---3 x 1007
Yo
Fuel cracked a t 395-650' F.: Yield, 70. Gravity, A. P. I. Aniline point, O F. Diesel index No. Pour point, O F. Fuel, cracked a t 600-640° F.: Yield, % I. Gravit.y, O,A. Aniline point, F . Diesel index No. Pour point, F . Straight-run fuel (low-grade): Gravity, A. P. I. Aniline point F. Diesel index No. Pour point, F.
SO2 as Solvent
7
7
Pennsylvania Cracked Diesel Fuel Ratio solvent/solute Extrabtio?, % Gravity, A. P. I. Aniline. point, O F. Diesel index No. Blending val!e Flash point, F. Pour oint ".F. C!ou$poiLt, F. Viscosity a t 100' F., sec. Total sulfur, 70 Conradson carbon, % Distillation, F.: Initial b. p. 107 over 50% over 9 0 7 over E n 8 point
0 0 30.1 131.7 39.6
R a t i o solvent/solute Extrahtion, % Gravity, ',A. $. I. Aniline, point, F. Diesel index No. Blending v a l t e Flash point, F. Pour oint, O F. ClouBpoint 0 F. Viscosity at'1000 F., sec. Total sulfur, % Conradson carbon, 7 0 Distillation ' F.: I n i t i d b.'p. 107 over 50% over 90% over E n d point
0 0 28.1 98.2 27.7
...
180 5 20 37.2 0.05 0.02 338 476 537 616 660
0.25 8 14.0 -6 -0.8 79
1 .o 22 13.0 -1 -0.1 82 240 < -35 -20 38.0 0.12 0.05
0.5 16 12.9 3 0.4 82 230 < -35 -20 42.0 0.15 0.02
... .. .. .. ... ...
... ... ... ... ... ...
1.5 28 13.3 3 0.4 80 240 < -35 -15 40.0 0.11 0.04
$60
456 501 533 607 660
0.3 19 14.5 -20 -2.9 79
0.9 27 13.3 -36 -4.8 84 195 < -635 . . . . . . . . .Too dark. 38.4 38.2 0.12 0.10 0.18 0.12
... ... ... ...
... ... ... ... ... ... ...
474 516 542 609 661
000 530 594 640
0.6 25 13.4 4 0.5 85 195 < -35
412 478 528 616 664
1.2 29 13.2 -2 -0.3 86 190 < -35
.. . . . . . . . .
38.0
...
0.12
410 475 528 612 664
378 468 528 ~~. 616 644
Mid-continent Cracked Diesel Fuel
'
i82
< -35
-5 35.4 0.23 , 0.14 398 451 484 548 610
0.25 11 17.2 -4 -0.7 78
... ... ...
... ... ... ... ... ...
455 474 499 545 588
364 467 498 543 589
414 473 498 545 593
... ...
420 458 494 582 622
...
... ...
410 459 494 583 624
413 459 495 576 628
410 458 495 579 637
Mixed E a s t and West Texas Cracked Diesel Fuel 0 0
30.3 118.8 36.0 Blending value Flash point, ' F. Pour point, ",F. Cloud point F. viscosity at'1000 F., sec. Total sulfur, % Conradson carbon, % Distillation, O F.: Initial b. p. 107 over 5_0% over 9 0 7 over E n 8 point
...
16: a
Too dark
36.0 0.60 0.10 350 $423 516 639 687
0.3 12 11.6 8 0.9 80
0.5 16 11.4 8 0.9 79 230 < -35 10 42.0 1.78 0.09
... ...
... ... .. .. .. ... ... ... ... ...
1.0 25 10.9 8 0.9 81 240