INDUSTRIAL A N D ENGINEERING CHEMISTRY
February, 1927
259
Prediction of Flash Point of Blends of Lubricating Oils' By E. W. Thiele
*
S T A N D A R D OIL C O M P A N Y ( I N D I A N A ) , W H I T I N G , I N D .
0 FAR as is known, no method has been published for calculating the flash point of a mixture of lubricating oils when the flash point of the individual components is known. Such a method would be a valuable guide in the blending of oils, such as is carried out extensively in the manufacture of finished lubricants. This paper presents a method for making the calculation; it is believed to be simple in application and - sufficiently accurate for most practical purposes.
S
and B (Figure I), which are calculated on the basis of the foregoing equations. The use of the tables is indicated on the figure. The tables may also be used for mixtures of more than two components, by considering the various ingredients as added one a t a time. Physical Basis of the Method
The method is based on the idea that the flash point is the Nature of the Method temperature a t which the vapor pressure of an oil reaches a The minciple of the method may be expressed thus: certain fixed value (which is somewhere in the vicinity of 10 the recibrocai of the antilogarithm of-one onelhundredth of millimeters of mercury). the flasi point ( " F.) is anI t is well known that the additive property of the oil vapor pressure of any oil is (on a volume basis). related to the temperature A method is presented for calculating the flash point In symbols this principle by the equation of any ordinary mixture of lubricating oils when the may be stated as follows: A flash point of the constituents is given. The method l o g p = -B Let F, F', F" = flash points of the constituents (in' F.) (it is assumed that there are three constituents) X,XI,. X u = volume fractions (=0.01 of the volume per cents) of the three constituents F m = flash point of the mixture Then
F, = -1OOlog
-&)I
4 )+ X"(10
10
Example of Working of Method
+
where A and R are (substantially) constants, and T is the absolute temperature.
Over a limited range this may be replaced by the approximate equation:
+
log p = M T K', where M and K'are new constants Also, if t is the temperature
in
[x( -A)+ X'( 10
T
is rapid and simple in application. While it is probably not extremely accurate, the experimental evidence presented indicates that, when applied to a wide variety of commercial lubricants, it gives correct results to about the degree of accuracy attained in flash point determinations. The theoretical basis for the method is also discussed briefly.
log p = 1Mt f K' where K is a new constant.
O
F. this may be written:
+ 460 M
=
Mt
+K
It happens that for oils in the range in which lubricating oil flash points lie, M = 0.01 nearly.2 K evidently varies with the kind of oil. Let
P
=
constant vapor pressure at which flashing begins
Suppose it is required to find the flash point of a mixture p = pressure of any oil a t a fixed arbitrary temperature, here taken as 400" F. of 75 per cent by volume of red oil (flash point 380" F.) Consider a binary mixture, the flash points being denoted by and 25 per cent of Pennsylvania mineral seal (flash point F', and the fractions of the constituents by x, 2'. 260' F.). One one-hundredth of 380 is 3.80. The number F, Then for the first constituent whose logarithm is 3.80 is 6300, the reciprocal of which is lop P = 0.01 . F +K 0.000159. Similarly the number whose logarithm is 2.60 K = -0.01 @ T log P (0.01 X 400) K = 4 - 0.01 F log P is 398, and its reciprocal is 0.002515. Each of the quantities Also log p p = = 104 + log P - 0.01F = (104 + log P ) (10-0.01F) 0.000159 and 0.002515 can be found by one setting of a slide rule. Similarly for the other constituents:
-
p'
25 per cent of 0,002515 = 0.000628 75 per cent of 0.000159 = 0.000119 TOTAL0.000747
Use of Tables
Use of the slide rule or logarithm tables may be entirely avoided for blends of only two oils by the use of Tables A
* Received August 28, 1926. Presented before the Division of Petroleum Chemistry a t t h e 72nd Meeting of the American Chemical Society, Philadelphia, Pa , September 5 to 1 1 , 192G
= (104
+
10s P )
(1o-O'OlF')
At 400" the vapor pressure of the mixture, p,, is
+ x'p'
xp
The reciprocal of this total is 1339, the logarithm of which is 3.13. Hence the flash point of the mixture will be 313' F. The flash point can be found from the quantity 0.000747 by one setting of a slide rule. With a slide rule, therefore, the whole operation may be performed very simply and rapidly.
+
+
p,
= (104
10s P )
+
(approximately) __
(x[10-0.01R]
+-XI
[10-O.OlF'
+
I>
log p m = 4 log P log (z[10-0.0~~] 4-x' [10-0.01F' Let Fm = flash point of the mixed oil K , = value of K for this mixture log P = 0.01 F,,, K, K , = -0.01 F, log P = (0.01 X 400) K, = 4 - O.OIFm log P log P ,
++
+
+
Equating the two expressions for log p,:
4
+ log P + log ( x [
lO-O'o'"]
I>
+
XI[
10-O.O'F'])
+
=4 - O.OIFm log P
2 See the vapor pressure chart of Wilson and Bahlke, THIS JOURNAL, 16, 116 (1924).
260
INDUSTRIAL A N D E,VGI.NEERI,VG CHEMISTRY
VOl. 19, No. 2
INDUSTRIAL AND ENGI,VEERING CHEMISTRY
February, 1927 Hence F, =
- 100 log
(*[
+ [10-".o'F'I>
10-O.OIF] x i
T a b l e 111"
-
FLASH MINERAL OIL
which is the equation already given, on which the tables are based. Flash
Point o f
x Observed
Oil M i x t u r e s r ; ]
Lirie d r a w n t h r o u g h c a l c u l a t e d points-
I
I
Calculated
Observed
Per cent O F. F. Petroleum Lubricating Oil and Neat's-Fool Oil 452 0 ioi 406 25 387 384 50 373 370 75 ... 100 362 Petroleum Lubricating Oil and S p e r m Oil 0 538 ... 25 416 420 50 388 391 75 376 374 100 362 Petroleum Lubrrcating Oil and Lard Oil 0 421 25 392 50 374 75 100 Petroleum Lubricating Oil and Cottonseed Oil 0 602 25 420 422 50 390 392 374 75 378 100 362 ...
Points
Upper Line: Penn. S t e u m Ref'ld C y l i n d e r Stock - L t q h t M t d Continent Lubricant
261
5oo
... ...
... ...
1, 16 Data of Sherman, Gray, and Hammerschlag, THIS JOURNAL, (1909).
T a b l e IVa
Figure 2
FLASH
It will be seen that the derivation of the equation rests
MINERALOIL
on a considerable number of approximations. Thus, the vapor pressure curve is assumed to be linear in T instead of 1/T, and vapor pressures are assumed to be additive with respect to volume concentrations, instead of with respect to molecular concentrations. P is also a rather indefinite quantity. It appears, however, that the various errors more or less cancel each other, for the results obtained in use on the ordinary range of lubricating oils are usually correct within the limits of accuracy of a flash point determination. The method is not applicable, however, to low-boiling oils, such as kerosenes.
Pe7 cent
FLASH Observed
-
T a b l e 11-Heavy
a n d L i g h t P e t r o l e u m Oila
FLASH
HEAVSOIL Observed
Per cent 100 90 80 70 60 50 40 30 20 10 0
F. 430 405 390 375 375 370 360 360 355 345 340
-
Calculated
F.
...
407 392 381 372 365 357 353 34s 344
.
...
Data of Sherman, Gray, and Hammerschlag, THIS JOURNAL, 1, 16
d ionoi ,I""",.
' F. 369 376 396 435 490 520
...
36i.5 369 376 396 434 488 517.5
... ...
100 98 95 80
359 350 321 292 275 268 267
50
'0
2 0 a
Calculated
3si.5
100 98 95 80 50 20 5 2 0
Calculated
Per cent F. F. Light Midcontinent Lubricating Oil and Red Of1 100 380 ... 75 360 36 1 50 340 348 25 340 335 ... 0 330 Penns3lzania Mineral Seal and R e d Ozl 100 380 ... 310 312 75 25 265 271.5 0 260 ... Y o r t k Louisiana Lubricatina Oil and Pennsylvania Mineral Seal 100 432 ... 75 315 318 50 295 289 270 272 25 260 0 ...
F.
100 98 95 80 50 20 5 2 0
Table I HEAVY OIL
dbserved
...
Data of Keesler and Mathiason, THISJOURXAL, 3, 71 (1911)
Experimental Data
Five oils were used in testing the accuracy of the method. Their flash points were: O
Pennsylvania mineral seal Light midcontinent lubricating oil Red oil Xorth Louisiana lubricating oil (close cut) Pennsylvania steam-refined cylinder stock
P.
260 330 380 432 605
Five different pairs of these oils were studied, mixtures in the proportions 25:75, 50:50, and 7 5 : 2 5 being made up. The A. S. T. M. Cleveland open tester was used. Only one determination of the flash point was made for each mixture, The results are shown in Table I and in Figure 2, which is given to show the general character of the curves obtained. It illustrates the very rapid falling off in flash point which occurs when a small amount of oil of low flash point is added to an oil of high flash point. It will be seen that the greatest difference found between calculated and observed values was loo, and this in cases where the flash points differed by 275"
1 5
c
J 262
INDUSTRIAL A N D ENGINEERING CHEXISTRY
or more. Considering the usual accuracy of a determination the agreement may be regarded as good. Data from the Literature
Some values for the flash points of mixtures were found recorded in the literature. These are compared with the
VOl. 19, No. 2
calculated values in Tables 11, 111, and IV. Most of these mixtures have a vegetable oil as one constituent, and it is remarkable that the agreement is as good as it is. In the single case, Table 11, where two mineral oils are mixed, the greatest deviation is 7".
Determination of Paraffin Wax in Crude Wax' By L. M. Henderson and S. W. Ferris THEATLANTIC REFINING Co., PHILADELPHIA, PA.
RESENT methods for the analysis of oil-wax mixtures has certain decided advantages over other methods and one are not altogether satisfactory. Some give results of of the most important of these is a working temperature of doubtful accuracy and others are long and tedious. 15.6" C. (60" F.), instead of the low temperatures employed The several methods are of three types-namely, those by earlier investigators. involving (1) a pressing process, (2) a sweating process, and One inherent fault in all these solvent methods is that (3) a separation of oil and wax by means of selective solvents. they require one or more transfers of the wax or wax soluThe press method2 does not give satisfactory quantitative tions. The acetone method, for example, involves two transresults. The sweating process3 is tedious and of doubtful fers and a filtration. There is difficulty in maintaining a c c u r a c y . Selective- Solconstant temperature while vents have been used by filtering and washing, and if several investigators. the temperature rises some A method for determining wax in oil-wax intermediHolde4 employed a 50:50 wax, goes into the filtrate. ates has been developed which obviates objectionable mixture of absolute alcohol T r a a e r s of wax are obof previous solvent methods. Nitrobenzene features and ether and precipitated j e c t i o n a bl e because the is used as the selective solvent. the wax from solution by physical properties of wax An apparatus has been devised which makes it poschilling to -20" C. This are such that it is practisible to carry out the entire procedure of separating cally impossible to transfer method necessitates mainand weighing the wax without transferring the sample. taining a constant temperait repeatedly n-ithout loss. The working temperature is never below 32" C. The To avoid this loss it has ture of -20°C. throughout melting point may also be determined in the same the various steps in the probeen the usual practice to apparatus. cedure. It is neither easy weigh the oil and determine nor convenient to maintain t h e wax b y difference. this low temperature conHowever. if there have been stant during t'he successive filtrations and washings required. one or more transfers or filtrations in the procedure, an exMethylethyl ketones has been suggested, but its use also re- cess of solvent has been used for washing purposes and some quires cooling to -20 " C. Wilson and Wilkinse have described of the wax has surely been dissolved. The result is a lorn a solvent-index of refraction method of determining oil in wax value for the wax content of the original sample. It is the purpose of the present investigation to modify which makes use of ethylene dichloride. Here also the filtration and washing are carried out in a Holde apparatus at - 18" the method of Wyant and Marsh so as to eliminate the transC. (0" F.), However, instead of weighing the residual oil ob- fers and filtration and to develop a dependable and rapid tained after evaporating the solvent from the filtrate, it is method of analysis. Acetone was found to be unsatisfactory dissolved in a special mixture of mineral seal and ligature oils when tried out under the conditions here imposed. Thereand the index of refraction of the solution determined a t fore, nitrobenzene was tried, for it is well known that it may 25" C . (77" F.). The amount of oil present in the solution be used for separating oil and wax. is then ascertained from a reference curve which relates index of refraction with the oil content of standard solutions. Solubility of Wax and Oil in Acetone and Nitrobenzene The wax is obtained by difference. The authors state that To establish the suitability of nitrobenzene for the purpose, the method is not so simple as might be desired. Wyant and Marsh' conducted an extended investigation the solubilities of both wax and oil were first determined in each of the two solvents. The oil was pressed heavy paraffin on the analysis of oil-wax mixtures by the solvent method. distillate, and the wax a finished plant product from paraffin Among the solvents studied by them were acetone, a very distillate. The method of determining the solubilities was volatile gasoline, benzene, and ethyl acetate. They conas follows: cluded that acetone was especially suitable for use in the separation of oil and wax. The method which they evolved A weighed amount of the solute was put into solution in the
P
1 Received August 28, 1926. Presented under the title "Determination of Wax in Intermediates" before the Division of Petroleum Chemistry at the 72nd Meeting of the American Chemical Society, Philadelphia, Pa., September 5 to 11, 1926. 2 Day, A Handbook of the Petroleum Industry, Vol. I, p. 724, John Wiley 81 Sons, Inc., 1922. 8 Holde-Mueller, "Hydrocarbon Oils and Saponifiable Fats and Waxes," 2nd ed., p. 249 (1922). 4 Holde-Mueller, loa c i t . , p. 108. 6 I b i d . , p. 110. 4 THIS JOURNAL, 16, 9 (1924). Bur. Mines. Tech. Paper 368 (1926)
solvent, by heating in the case of nitrobenzene, and by means of a reflux distillation in the case of acetone. The first solution prepared was of high concentration. The dilute solutions were prepared by mixing a portion of the stock solution with a definite amount of the fresh solvent.
The cloud point of each solution was then determined, first by slowly cooling until the cloud could not be eliminated by shaking, and then by gently heating until the cloud just disappeared. The points so determined were found to check closely.
