I N D U S T R I A L A N D ENGINEERING CHEMISTRY
August, 1924
~~
789
~
T h e Viscosity of Oils a t High Temperatures' By A. R. Fortsch and Robert E. Wilson STANDARD OIL Co.,WHITING,IND.
tation of viscosity data on the N THE modern oil reThe viscosity of oils at high temperatures is very important in lighter hydrocarbons up to finery, calculations inrefinery engineering calculations involving the flow of fluids and the slightly above octane. The volving the flow of heat flow of heat, but there are practically no published data on this subject fluidity (l/viscosity in poises) and the flow of fluids are was plotted against temperaat temperatures above 2 I2 " F. Suggested methods of extrapolation ture in Centigrade degrees. becoming more and more are not at all reliable above 300" F., while the data are often badly The relation was almost linear important in permitting the needed at temperatures between 400' and 600 O F. with the very lightest hydrodesign of more efficient stills, This paper presents data and CUYVCS covering viscosity determinacarbons, but showed an aph e a t e x c h a n g e r s , conp r e c i a b l e curvature with tions on a wide variety of oils from Midcontinent and similar crudes octane. densers, coolers, and similar u p to a temperature of 495" F. The curves are discussed and a A third method of presenequipment. If such calcumethod of interpolation is outlined by which it should be possible ' tation was that of Bahlke, lations are to be reliable, it to obtain to within a few per cent the viscosity of practically any oil Waldron, and Wilson,* who is essential to hav%reasonfound that for the lighter derived from Midcontinent or other intermediate base crude at any hydrocarbons pIotting the log ably accurate data on the temperature up to 550" F., given a single reliable determination of its of the viscosity against the viscosity of the fluid a t the viscosity at ordinary temperatures. reciprocal of the absolute temtemperature a t which the perature gave approximateIy flow-of heat or of oil takes a straight line. None of the three methods gives straight lines for all oils. place. In refinery practice these temperatures are frequently far above the range a t which viscosity measurements are ordi- All of them were applied t o the data obtained by the writers, although none gave -a 'linear relation for most of the oil. narily made. Furthermore, the fluid is not a pure compound Herschel's method (log viscosity us. log Fahrenheit temperature) but a mixture of unknown compounds about whose properties approached this much nearer than the other methods. It was therefore decided to use this method in the presentation of the a t high temperatures little is known. A careful survey of the literature indicated that with the results. exception of octane the determinations did not include EXPERIMENTAL temperatures much above 260" F., while the large majority With the exception of a few lower temperatures on two of the were below 212" F. No matter what method of plotting the results is used, the uncertainty of extrapolating to tempera- oils, the measurements were made with a modified type of tures of 400" or 500" F. is obvious. It was therefore consid- Ostwald viscometer such as that described by Bingham. ered desirable to make viscosity determinations on various The slower instrument was calibrated using water a t various temperatures, and the more rapid one by comparison with the oils a t elevated temperatures. first on various oils. After the constants of the instruments had been determined, the various oils were run a t the temperaPltEVIOUs W O R K AND METHODS O F PLOTTING tures desired.
I
A large amount of work on the viscosity of various crude oils and vegetable oils up to 210" F. has been done by Herschel.2 He has found it desirable to plot log viscosity against log tems plots gave approximately perature ( " F.),because in all c a ~ e such straight lines between room temperature and 212' F. Herschel found that if these lines were extended in the direction of higher temperatures they met approximately a t a point. This point of intersection was different for oils from different types of crude (paraffin or asphalt base). He suggests that for an unknown oil of a given base the viscosity may be determined with a fair degree of accuracy by using the intersection point and one other point, such as that deter; mined by the viscosity a t 100 F. Obviously, this line could not be safely used much beyond the range of the experimental work, since the viscosity of all oils certainly does not become the same a t temperatures around 400' F. as would be predicted by this method of plotting. A second method of plotting A was used by Herschel3for presenB
It
-!r
C
Presented before the Division of Petroleum Chemistry a t the 67th Meeting of the American Chemical Society, Washington, D. C., April 21 to 26, 1924. 1
2
THISJOURNAL, 14, 715 (1922).
a PYOC. A m . SOC.Testing
22, 677 (1923).
Matevials,
D E F G
H K L
M
Note.-The
constants are C and CIin the equation
z
= cpt
C1P -t
where Z = absolute viscosity in centipoises p = pressure in grams per square centimeter t = time of flow in seconds p = density in grams per cubic cenL timeter a t the temperature of the fluid when determination was made
For the two viscometers the values were:
No. I
l
1 2
l
CYLINDER NITROGEN GAS BUNSEN PRESSURE REGULATOR MANOMETER P R E S S U R E RESERVOIR DRYING BOTTLE THREE-WAY STOPCOCK VISCOMETER THERMOMETER VAPOR B A T H (LAGGING NOT SHOWN) REFLUX CONDENSER BOILING LIQUID
FIG.~-APPARATWSFOR VISCOSITYDETBRMINATIONS
c
CI
0.001055 7.4 0.00001104 11.5
The temperatures up to 250" F. were obtained by immersing the viscometer in an oil bath provided wit+ close thermostatic regulation and with glass windows so that the observations could be taken t h r o u g h t h e bath liquid. Various liquids were tried, and it was found that up to this temperature mineral seal oil would not darken ap4
6
p. 76.
Laboratory report. "Fluidity and Plasticity," 1922,
I N D USTRIAL;AND ENGINEERING CHEMISTRY
790
preciably and an observer had no trouble in taking the necessary readings. Beyond 250" F. the oil used for the bath gradually became dark and was in general very disagreeable to work with. Therefore, for these higher temperatures a special vapor bath was devised, the details of w@ch are shown in Fig. 1. The viscometer was cemented into a brass cap (not shown in sketch) which screwed into a brass gland at the top of the tube constituting the vapor chamber. The slight leak in this gland made it possible to keep the level of condensing vapor within an inch of the cap, insuring that the oilin the viscometer was completely surrounded by the condensing vapor. The liquids used for the vapor bath were: Liquid Xylene Aniline Naphthalene Diphenyl ether
Boiling Point O F. (Approx.) 280 to 287 363 to 364 424 495
Vol. 16, No. 8
The data on the various oils are given in Table I. I n Fig. 2 the data have been plotted according to the method of Herschel.2 An auxiliary scale has been added so that the viscosities and the temperatures may be read directly. This group of curves covers data on the various oils over a temperature range from 90" to 495" F. DISCUSSION AND USE OF CURVES The logarithmic viscosity temperature curves for seventeen of the nineteen oils on which data were obtained are plotted on a large scale in Fig. 2. It will be noted that with one or two exceptions the determined points fall on the smooth curves and, what is more important, thatall the curves appear to be members of a single family. The curve for the light paraffin oil (79 seconds Saybolt) appears to be substantially straight throughout and the curves for the oils on &her side curve in and tend to become parallel to this line a t very high temperatures. It will also be noted that in general these graphs show considerable curvature, whereas those of Herschel were approximately straight lines. It should be remembered, however: that Hersehel's resulta lay in rather narrow .temperature range between 70" and 212" F. and did not cover the very light or very heavy oils, which show the most curvature. In this region the lines determined by the writers are not far f
A total immersion thermometer was suspended in the vapor bath in the same manner as the viscometer, the bulb being a little below the bulbs of the viscometer. The pressure tending to force the oil through the viscometer was maintained constant a t approximately 50 grams per square centimeter by means of an ordinary Bunsen pressure regulator. The Source of pressure was a cylinder of nitrogen.
I
INDUSTRIAL AND ENGINEERING CHEMISTRY
August, 1924
from straight, and if only these intermediate data were available one might be justified in drawing straight lines. Any extrapohtion of these straight lines to higher temperatures would, however, give very erroneous results for the heavier oils. For example, the extrapolated viscosity of cylinder stock at 500" F. would be only about half of the actual viscosity, whereas that for the lighter oils would be considerably higher than the actual figures. It must be emphasized that all but two of these oils were derived from Midcontinent crude, and that these two were from a north Louisiana crude which appears to be similar in its viscosity-temperature relationship. From data a t 100' and 210' F. it appears that the temperature coefficient of viscosity for oils from Pennsylvania crude is somewhat lower than that for Midcontinent oils, while that for Gulf Coast crudes is still higher. The difference is relatively small except for the heavier oils, and prqbably decreases with increasing temperature. The only oils that do not fit in with the family of curves are the road oil, made by the reduction of pressure tar from the Burton process, and Midcontinent crude. The slight variations in the latter are undoubtedly due to'the loss of some of the light ends at the higher temperatures, but in the case of the road oil the viscosities a t low temperatures are considerably too high to fit in with the family. This is probably a colloidal phenomenon and may indeed be due largely to the adsorption of a film on the walls of the viscometer capillary, which seems to take place a t the lower temperatures. The curves for these two oils are shown in Fig. 3. If the results shown in Fig. 2 were limited to these oils alone, the expenditure of labor necessary to obtain them would z
1' 0
17.
32 50 68 86
104 12% 140
41 50 77 9 !j 113 131
~
-,
-3
Centipoises O F. Octane ilandolt-Bornstein) 158 0.706 176. 0.616 194 0.542 212 0.483 230 0.433 248 0.391 0.355 Gasolines (Hersche1)s ,-----Z Centipoises No. 1 No. 2 0.770 0.715 0.603 0.662 0.518 0.589 0.472 0.515 0.426 0.468 0.382 0.412
z
Centipoises 0.324
,0 .297 0.273 0.252 0.234 0.216
-------. No. 3 0.775 0.641 0,540 0.493 0.440
z Density Density T Z . ' F. Centipoises G./Cc. G./Cc. Centipoises Kerosene (455 seconds Saybolt Tlzermo) 2,422 0.823 194 0.7140.777 61 1,632 0.812 203 0.672 0.769 91 1 , 2 2 5 0.801 250 0.515 0.750 121 160 0.891 0.785 306 0.395 0.730 Midcontinent crude (45 seconds) Mineral seal (44 secondsIa 4 7720.835 47 19.068 0 853 95 2 932. 0.822 95 5.828 0 838 133 1.3010.792 132 3.698 0 825 212 422 0 402 0.715 212< 1.812 0 800 0 299 0.689 495 Pressed distillate (57 seconds) 7,850 0.861 205 2.102 0.823 100 125 5.368 0.852 231 1.665 0.813 3.859 0 , 8 4 3 269 1.262 0.798 148.5 177.5 2.721 0.834 297 1.048 0,788 Light paraffin oil (79 seconds) 16.41 0.873 282 1.431 0 804 89 0 776 7.138 0.857 362 0.853 131 0 754 163 4.500 0.846 4240.628 201 2.869 0.833 495.5 0.456 0.729 241 1,952 0.819 Whit.e oil (113 seconds) 3.966 0.811 198 0.855 38.35 74.6 .' 2.801 0.799 231 0.846 20.96 97.0 259 2.199 0.790 0.837 12.04 123.5 1.518 0.773 304.5 7.777 0.829 148 172 5.475 0.820 T F.
(I
scarcely be justified. However, the curves in Fig. 2 appear to belong to a single family. It is possible to start with a single point, such as the viscosity at 100' F. or a t 210' F. of a given oil, and construct a curve which fits in with the family. For most purposes it is not necessary actually to draw this curve, since the value a t any temperature can be determined approximately by inspection.
00
is Loa Tc-mueroiure (Drqrees fohrenheif)
FIG:3-TsMPERATUR&-VrSCOSITY
CURVES
For oils from Midcontinent or other intermediate base crudes, it is believed that this will give results within 2 or 3 per cent, even if carried over a range of 250" F., provided, of course, that there is no loss of volatile constituent.
TABLE I-VISCOSITY DATAOF T
791
VARIOUS OILS
z
T
Density T Z Density G./Cc. F. Centipoises G/Cc. Light p a r a f i n oil (140 seconds) 87 36.89 0.882 252 1.993 0.812 121 16.04 0.870 362 1.126 0.783 162 7.738 0.855 424 0.832 0.761 201 4.436 0.841 496.5 0.574 0.736 252 2.560 0.823 Distillate from north Louisiana heavy crude (173 seconds) 100 32.76 0.903 424 0.790 0.784 15.34 130 0.592 494 0.605 0.758 4.660 0.863 210 Light motor oil (219 seconds) 0.893 241 3.566 75 90.62 0.834 0.884 282 2.410 40.52 101.1 0.819 0.873 362 1.314 19.27 131.0 0.790 0.859 423 0.934 9.580 170 0.768 6.183 0.847 496 0.658 202 0.742 Rcd oil (485 seconds) ' 0,900 282 3.627 84 163.4 0.828 0.890 362 1.864 59.41 0.799 115 0.873 424 1.251 21.14 0.776 158 0.858 495 0.848 10.02 0.751 200 5.1Q6 0.840 250 Road oil (67 seconds Saybolt Furol at 1.22' P.) 401.9 1.036 422.0 1.467 0.920 100 110.4 1.026 494 0.974 0.895 130.5 13.91 0.999 203,5 North Louisiana heavy crude (89 seconds at 210' F.) 239.3 0.928 423 1.681 0.819 100 88.47 0,918 494 1.097 130 0.795 15.33 0.891 210 Heavy motor oil (121 seconds at 210' F.) 549.7 0.903 282 7,423 0.832 91 184 2 0.891 361 3.368 0.804 121 48.40 0.875 424 2.124 0.781 168 24.39 0.862 495 1.377 0.755 201 11.14 0.844 250 Cylinder stock (156 seconds at 210' F.) 0.806 0.877 361 4.162 73.64 167 0.783 0.864 424 2.660 32.86 204 0.757 0.850 494 1.630 16.35 244 9.629 0.835 282 14.5 grawity Midcontinenl resid_uum (384 seconds at 210' F.) 100 5638.5 0.956 422 4.321 0.849 132.5 857.9 0.945 494 2.515 0.825 202.5 89.80 0.922
F.
Centipoises
.
The figures in parentheses refer to the Saybolt Universal viscosity of the oils a t 100'
F.unless otherwise specified.
INDUSTRIAL A N D ENGINEERING CHEMISTRY
792
For oils from other types of crude more work will have to be done, but even in this case it is believed that the error involved in using these curves will be much less than for any other existing method of extrapolating to high temperatures. Since the curves of Fig. 2 appear t o be nearly straight in the neighborhood of 500’ F., it is probable that they could be extended to 600” F. without the introduction of any serious error. The ordinary Saybolt determinations are always accessible,
Vol. 16, No. 8
so there is no particular need for extrapolation to lower temperatures. Except for a small range, such extrapolation is likely to be in error due to the arbitrary nature of the scale, since the temperature scale extends to negative infinity as the temperature approaches 0” F. These is also a tendency for the introduction of larger errors in the viscosity determination with the heavier oils as they approach their cold test temperature. For these reasons extrapolation to lower temperatures is not recommended.
The Acceleration of Vulcanization’ I-Influence
of the Acetone-Soluble Constituents of Rubber on the Physical and Chemical Properties of Accelerated Rubber Stocks By L. B. Sebrell and W. W. Vogt GOODYEAR TIRE& RUBBERC o . , AKRON,OHIO
A
All accelerators require soluble zinc in order to produce the best ~CCeleratorson the physical amount Of data has physical properties, also in stocks containing the same accelerator Properties of vulcanized rubber, have recorded some been Published from the chemicalcure is no index of the physical properties. time to time during the last The production of maximum physical properties seems to be deinteresting data On the effect of the resins on the ten O r twelve Years On the pendent upon the formation of a zinc-accelerator compound. ,!hch a compound has not been isolafed in any case, except with the merrate of cure. These workers influence Of the resins and capfo accelerators, where it is believed to be [he zinc salt of the acselected two samples of Proteins on the VulcaniZation of rubber. This has celerator. pale crepe rubber, extracted been done to determine Some accelerators. such as diphenylguanidine and dimethylthem with acetone for 36 aminodimethyldithiocarbamate,are able to react directly with zinc hours, and vulcanized them their effect both upon the oxide in order to form the necessary zinc-accelerator compound, with suitable controls for chemical and Physical Propwhile others are dependent upon the presence of a rubber-soluble equal lengths of time in a erties of the rubber. Rezinc salt, either formed by the reaction of the resin acids on zinc oxrubber-sulfur mix* A desuits have been Obtained termination of the coeffifor Pure gum mixes, and ide or added directly to the mix. for those containing certain cients of vulcanization inorganic accelerators, such s h o w e d t h a t one exas litharge and magnesia, but in only a very few cases for tracted sample contained more combined sulfur than the unextracted control, while in the other case there was no compounds containing organic accelerators. Seid12 in 1911 and Beadle and Stevens3 in 1912 made a difference. They further found that, if magnesia was used study of the nature of the resins contained in rubber and in the mix under the same conditions as just previously detheir relation to the quality of the vulcanized product. scribed, the rate of cure was greatly reduced if extracted rubThe latter workers used a simple compound of rubber with 5 ber was used, and that this was equally true of both samples. per cent of sulfur and in some cases with the addition of zinc Magnesia would then seem to function in exactly the same oxide. They found that the removal of the acetone-soluble way as litharge. If, however, a small amount of an organic material retarded the cure and that the tensile properties of accelerator (termed “Accelerator A”) was added to each of the rubber were greatly reduced. At about t,he same time the extracted samples and their controls and cured as before, Weber4 announced the results of work that he had done on then the resin-free rubber again showed an increased rate of the action of the resins, He had used a litharge stock and cure in both cases. The reason for the increased rate of cure with extracted rubber had obtained such poor results that he in this last case was not considered, nor has it since been stated that no vulcanization had taken place. Stevens5 re- satisfactorily explained or further investigated. The effect peated Weber’s work and found that in a litharge stock pre- of the resins on the action of organic accelerators has also been pared with acetone-extracted rubber the physical properties mentioned by Maximoff.’ He was unable to obtain any were indeed poor, that the rubber was more susceptible to accelerating effect with the zinc salt of dimethyldithiocarbaging, but that it did contain some combined sulfur. There- amic acid unless zinc oxide was present or an excessive fore some vulcanization had taken place. I n a 90 : 10 rubber- amount of the accelerator was used. This fact he attributed sulfur mix he found that the removal of the resins retarded the to the acid nature of the resins, which he thought retarded the rate of cure, as shown by the reduced coefficient of vulcaniza- rate of vulcanization. Zinc oxide by neutralizing this acidity allowed the vulcanization to proceed a t the normal rate. We tion and lower physical properties. Kratz and Flower,B in a study of the effect of certain now know that this explanation is not the correct one, since a more plausible explanation of the inactivity of this acceleraPresented under the tit,e ,,The Acceleration of Vulcanization I--The Influence of the Nonrubber Constituents on the Physical and Chemical tor in the absence of zinc oxide has been set forth by Bedford Properties of Rubber Stocks” before the Division of Rubber Chemistry a t and Gray.8 the 66th Meeting of the American Chemical Society, Milwaukee, Wis., Stevensg in some later work has also found certain rubbers September 10 to 14, 1923. which, after removal of the resins, vulcanized a t a faster rate * Gumini-Ztg : 88, 710, 748 (1911).
C 0 N S I DE RAB L E
8th Mfernal. Cong. A g p l . Chcm., 1912, pp. 25, 581. Ibid., 1912, pp. 9, 95. I J . SOC.Chcm. I n d . , $6, $74 (1916). 6 T ~ I JOUR,NAL, S $2, 971 (1920). 8
4
7 8
0
Rubber A g e , 10, 53 (1921). THISJOURNAL, 15, 720 (1923). J . SOC.Chem. I n d . , 41, 3261’ (1922).