I S D C S T R I A L A.VD ESGIATI1’EERISG L’HEMISTRY
April, 1924
347
T h e Mechanism of Lubrication‘ 111-The Effect of Oiliness on the Behavior of Journal Bearings By D. P.Barnard, H. M. Myers, and H.0.Forrest ~IASSACHCSETTS INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MASS.
This paper describes the results obtained f r o m a series of expericarriedona contin~ousfluid merits conductedfor the purpose of determining the e#ect of oiliness film of the lubricant. Tests this series were given, of the lubricant on the carrying power of a conuentional type of on bearings operating under respectively, the results of preliminary investigajournal bearing. These results indicate that the carrying power such conditions have shown is somewhat greater when oiliness, as measured by fhe coeficient fairly definitely that vistions of the property of cosity is the O l l b property of static friction, is increased. This increase i n carrying power oiliness of lubricants, and is small f o r oariations among commercially practicable lubricants. of the lubricant which influa review of the available ences the friction. Oiliness literature on experimental exerts no detectable effect. mork on journla and bearing friction It was pointed out that, while some work (largely of The bearing is therefore said to be running in the region of a speculative nature) had been done on the rather intangible “fluid film” lubrication. property of oiliness, practically no reliable information was On the other hand, when the value of z n / p becomes arai]ab]e as to the effects of this property on the behavior sufficiently small, it is impossible for a fluid film to be mainof jouriial bearings. The object of this paper is to show tained. Lubrication is effected largely by the ability of the to mhat extent oiliness may be expected to influence the lubricant to adhere to the surfaces. The bearing in this case is said to be operating under conditions of “adsorbed behavior of lubricated bearings. film” lubrication, and the effects of friction are many times PRELIMIKARY DISC~;SSION larger than normally occur under fluid film lubrication.
I
3 Parts I2 and 113of
I n Part I it was pointed out that f , the coefficient of frittion of a journal bearing, is a function of at least nine variables--i, n, p , c, d , I , S,M , and 0, where 2: =
IZ =
p
=
c =
d =
I =
S
=
M 0
= =
viscosity of the lubricant a t the temperature of the lubricating film, centipoises speed of rotation of the journal, r. p. m. . nominal pressure on the bearing, pounds per square inch of projected area diametrical clearance between journal and bearing, inches diameter of journal, inches length of bearing, inches surface conditions-composition of metal, smoothness, etc. method of supplying the lubricant oiliness factor of the lubricant
There is, therefore, for a given combination of bearing and journal, some value of x n / p where a transition occurs from fluid film to adsorbed film lubrication. This is illustrated qualitatively in Fig. 1. TOthe right of the transition point, A , is the region of fluid film, and to the left is that of adsorbed film lubrication.
’>
It was also shown that f is a factor without dimensions. Therefore, in accordance with dimensional analysis, f must be a fuiiction of these variables combined in such a manner that all dimensions cancel out. Such a function may be expressed by For anv given combination of iournal and bearing, therefore. variat’iom’s in f are defined by” the modulus, z n7p, and the oiliness factor, 0. For such a combination in which the lubricant is not changed (and t’herefore the factor, 0, kept constant;), changes in f are determined entirely by the modulus x n/p. Identical values of f are obtained for the same value of x n / p , regardless of changes in theindividual variables. The soundness of this line of reasoning has been well verified by the experimenta.1 data available in the literature. When a bearing is operating a t a comparatively high value of x n/p-i. e., high speeds or light loads-the journal is 1 Presented under the subtitle before the Division of Petroleum Chemistry a t the 66th Meeting of the American Chemical Society, Milwaukee, Wis., September 10 to 14. 1923. * Wi1i:on and Barnard, J. SOC.nulomotive Eng., 11, 49 (1922); abstracted in THISJOURNAL, 14, 682 (1922). 8 Wilson and Barnard, THIS JOURNAL, 14, 683 (1922); J. SOC.Automotive Eng., 11, 143 (1922). It should be noted t h a t z n / $ is actually a dimensional term, However, z ? I / $ is used because of its convenient engineering units, and is directly proportional to the theoretical modulus, which is without dimensions.
*
1
R
1 ,
Z77 - fWSCGJ/7?u -C&~,vmrf~)(APfl,J PHZJSURZ /h’LES PLR JQ/N)
-_
D ,
FIG I - ~ R A S S I T I O N
FROM
FLUIDFILMTO ADSORBEDF I L M
LUBRICATION
It has been found more satisfactory to designate these two regions of lubrication as stable and unstable, respectively. The characteristics of the two regions are more accurately defined by this nomenclature. In the range to the right of the transition point, friction produces effects (heating) which tend to lower the value of z n / p by decreasing the viscosity and to lessen the magnitude of the friction. This region is, therefore, one of natural stability. This is not true of that range to the left of the transition point, as the action of friction in producing effects tending to lower x n / p causes a rise in friction. This action is cumulative, and the tendency of a bearing operating in this range is to heat rapidly and eventually to seize. It is at the transition point that the effects of oiliness of the lubricant may be expected to show themselves and, as the value of z n / p is still further decreased, t o become more and more marked.
I N D L'STRIAL A X D E,\'CINEERING
348
It may be well at tlis time to call attention to the fact that, in designating the two regions of lubrication as stable and unstdle, the t.ransition point is defined as that part of the friction curve whcre the slope becomes zero, and is identical with the point of minimum friction.
FIO.2 .Bsnni~i:FRICTION TESTINO M I C I ~ L
In Part 11 was described a met.hod of measuring oiliness in a comparative sense by the determination of the coefficient of stat,ic frictjon existing between two metallic surfaces wet with the lubricant under investigation. In the present work friction coefficients obtained by this method are used as indications of the oiliness of the luhricants. The writers believe that this method represents the best single measure of t.he property of oiliness. The results of investigations available in the literature offer no data that permit the forming of any definite conclusions as to the effects of oiliness on the behavior of commercial types of bearings. The purpose of the work reported herein is to determine the influence of oiliness, as indicated by the coefficient of static friction, on hearing performance, particularly as this property affects the position of the transition point from stable to unstable lubrication. To t.tiis end a number of carefully conducted tests have been made, using journals and bearings approaching as closely as possible tliose of commercial machinery. Every effort has been made to control carefully each of the numerous variables involved. EXPERIMENT.4L
hIETHon
The experimental procedure employed consisted, in principle, of the nieasurement of the torque imparted by a rotating journal to a cylindrical bushing "floating" on it. The diita include approximately forty complete tests, in eacb of which the value of tlie modulus z n / p was varied to provide a number of observations in the two regions of lubrication. By this means the position of tlle transition point was fairly accurately determined. All the variahlesespecially the clearance between bearing and journal-were controlled as carefully as possible. The testing machine (Pig. 2 ) was constructed from a lGinch engine lathe which had been stripped of carriage and other accessories and provided with supporting hearings for the test journal. These bearings could be moved to any position along the lathe bed in order to accommodate shafts of diffemnt lengths. The supporting bearings were bronze lined and were of very rigid construction. The lathe spindle wss carefully balanced to Dermit operation at speeds up to 1500 r. p. m. The machine was driven by a shunt motor, which, with the speed regulation provided in the driving sys-
CHE.MISTRY
Vol. 16. T o 4
tem of the lathe, made readily attainable any s p e d fro111 10 r. p. m. up to 1500 r. p. m. The test bushings, which were Aoated on the central, or test, section of the journal, were mounted in a yoke, to which the load N&Sapplied by means of a lever acting through a parallel linkage system. This system, while dowing the application of any desired load to tho bearing, permitted the bearing to move freely about the shaft. The torque imparted to the bearing by friction was measured by means of a balance reading to 0.01 gram. The test journals were made of carefully ground steel and the test portion was finished by precision Iapping methods. The diameters of the test sections were measured by means of interference precision gages and were reported to the nearest 0.00001 inch. The test sections presented smooth, mirror-like surfaces. The test bearings were made of a commercial bearing bronze, machined from solid stock into bushings 2 inches long by 1 inch internal diameter. The outside diameter was 2'1, inches. After reaming, the surface of the bore was finished hy broaching with hardened steel plugs until all scratches were removed and a polished surface was ubtained. The diameters of the holes were measured by means of a special set of plug gages, to the nearest O.ooOo5 inch Fhch hearing was provided R-ith an oil-distributing groove running parallel to the axis and terminating a/ig inch from each end. 1)iamctrieally opposite were fitted two thermocouples so mounted that the junction came almost flush with the bearing surface. The test bushings were mounted in the machine with the thermocouples and oil grooves in line with the direction of application of the load, the thermocouples being placed a t t.he point of greater pressure. The potentiometer used gave tlie temperature to the nearest 0.1" C. The test shafts and bushings are illustrated iu Fig. 3.
Fro. 3 - T n s ~ J o u n n n ~A N D I l s n ~ i ~ a
Measurements of the shaft and bearing diameters as outlined above permitted the determination of the clearance with a probable error of less than 10 per cent. The clearance was changed in different tests by the use of journals of varying diameters. The extent to which the individual variables were investigated is outlined 5s follows:
Spindle oil Spindle oil = 2 per cent oleic acid Laid oil Glycerol-watei
125
0.sm
123 20"
0.sxz
40
0 887
($7
95
1.200
48
12
0.170 O.12h
0.ias Above U.300
.. M*IERI*L
Bronze Soft steel ciesrance: 0.00s inch (dit = 2001 LO o.oou53 Load: SO to 130 pounds per square inch Speed: 20 to 1000 r. D. m. BearinLE iournsi
% ? % t ~ " ~ ~ ~ ~
iilCll ( d / C
= IS2UI
.... e. 0
I1 I S ,
. ..
I N D UXTRIAL A N D ENGINEERING CHEXISTR Y
April, 1924
PRoc'EDuRE-The machine was started under a comparatively light load (50 pounds per square inch) and run at a speed of 400 to 600 r. p. m. until the flow of oil to the bearing had become steady. The desired load was then applied and simultaneous readings were taken of speed, temperature, and friction torque. Without changing the load or speed the machine was reversed and another set of readings taken. The fonvard reading was then repeated, and if a check on the first was obtained, the mean of the forward and reverse readings was taken as the true torque. This method of taking observations was repeated at successively lower values of x n / p (obtained by decreasing the speed) until several readings below the transition point were obtained. These observations were checked by another series made a t successively increasing speeds. As a general rule but one set of check runs was made, in order to avoid undue wear on the test parts. The procedure of averaging the readings taken in the two directions of rotation corrects for any lack of balance of the parts or slight shifting of the linkage system.
.
349
While the maximum differences observed are not so large as might be anticipated, yet they are definite and fall clearly in the order expected. I n this case the order of oiliness happens to be also the order of the viscosities of the lubricants. However, a series of tests in which the glycerol solution used had the highest viscosity gave the transition points falling in the same order as in Fig. 5.
RESULTS The character of the results obtained by operating the machine in the manner outlined above is illustrated by Fig. 4. It will be noticed that the observations in the fluid film range fall very nearly on the straight line drawn through them. The intercept with the vertical axis is much smaller than that reported in the majority of friction investigations available in the literature. I n all the tests made with this machine the intercept varied between f = 0.0008 and f = 0.0011, as against an average of approximately f = 0.002 as previously reported. The accuracy of the results is shown by the fact that for three tests identical with that illustrated in Fig, 4,except that different lubricants were used, the intercepts checked exactly and the slopes were 2.21, 2.24, and 2.20 X low6,respectively. The general shape of the curve a t the transition point varied somewhat, the rise frequently being slightly more abrupt than that of Fig. 4.
1'113.
OBSERVED FRICTION CURVE (RUN 22) Journal. , , .Steel Bronze Bearinn.. Diameier Clearance . * * . 1 1 5 0 Oil. .Velocite B spindle oil
4-TYPICAL
. ... .
.... . ...
In Figs. 5, 6, and 7 are shown curves for comparable tests using different lubricants. The horizontal scale has been expanded somewhat over that in Fig. 4 in order to illustrate better the nature of the curves a t the transition points. I n Fig. 5 is given a comparison between a solution of glycerol and water, a straight mineral Iubricant, and a fixed oil. The first possesses practically no oiliness, while the last, lard oil, is recognized as being oily to a very high degree.
FIQ.5-EFFECT
OF OILINESS O N VALUE OF Z n/P A T TRANSITION POINT .Steel Journal. .Bronze Bearina . Diameter Clearance . * . * 3 5 9 Load . I 5 1 pounds per square inch A .Glycerol-water solution (Run 20) -t ..Vetocite B spindle oil (Run 15) , Lard oil ( R u n 18) 0
. . .. ....
. . . .. ...... .. .... .. . ........... .. ...
I n Fig. 6 are given the results of another series of tests on a bearing of somewhat different characteristics, as will be noted by the difference in magnitudes of the transition points. In this case the load was 71.28 pounds per square inch, as against 151.28 pounds in the preceding series. The clearance was also about 20 per cent smaller. These results may be taken as indicative of the maximum differences to be expected among the various types of commercial lubricants. Fig. 7 represents a third set of tests, in which lard oil is compared with the petroleum lubricant. The differences between the transition points are not so marked in this as in the two former cases, and indicate again that the effects due t o variations in oiliness observed in practice are very small in the case of well-lubricated journals and bearings. All the foregoing tests, with the exception of that on the glycerol-water solution, were made in the order of increasing oiliness. This procedure is necessitated by the fact that oily films will remain adsorbed on the metal surfaces for a considerable length of time, even though the original lubricant is replaced by one inferior in oiliness. The glycerolwater solution was tested subsequent to the oils in order to avoid risk of damage to the surfaces. It is probable that this precaution was not entirely necessary. The differences between the glycerol solution and the oils would probably have been greater had the solution been tested first. I n the remaining tests duplicates were made with one lubricant before changing to one of greater oiliness. I n all this work only as many duplicate tests were made as were absolutely necessary in order to avoid undue alteration of the parts. A number of combinations of bearings and journals were investigated, with the result that several differences, as yet unexplainable, between individual bearings were noted. It is believed, however, that the results from any one bearing are in themselves fairly comparable. I n order to make certain that the differences in the transition points observed between different lubricants could not be due to changes in the clearance by abrasion, several series
350
IXDUSTRIAL AhTDENGINEERISG CHEMISTRY
*
of tests were made in which the clearance was purposely varied by the use of journals of different diameters. The results of one such series are given in Fig. 8, in which the value of z n / p at the transition point is plotted against the ratio d/c. It will be noted that the slope of the curve is
FIG.6-EFFECT
O F OILINESS ON VALUE O F 2
n/P
AT
TRANSITION POINT
Journal. . . . . .Steel Bearing.. .Bronze Diameter .I150 Clearance .71 pounds per square inch Load.. T . . ........ ..Velocite B spindle oil (Run 2 2 ) 0. .Lard oil (Run 23) A.. ..Velocite B+2'% oleic acid (Run 24)
... -. ... ...... ..........
.........
very small for a considerable range either side of the value of d / c = 1000. As this figure represents approximately the clearance a t which the tests on oiliness were conducted, it seems improbable that the very small amount of wear occurring could have produced any appreciable variation in the position of the transition point. The tendency of the observed curve to rise a t values of d / c greater than 1000 (for the 1-inch diameter bearing this corresponds to a clearance of 0.001 inch) indicates clearly the reason for seldom fitting continuously loaded bearings tighter than this figure. In fact, the large majority of complete bearings carrying uni-directional loads are fitted so that d / c is rarely greater than 500. These figures do not, of course, apply to bearings in which the direction and magnitude of the load are constantly changing, as in the crankshaft and connecting rod bearings of reciprocating engines.
Vol. 16, No. 4
stable lubrication for a complete cylindrical bearing is altered by variations in oiliness of the lubricant. 2-The transition point occurs a t a lower value of the modulus, z n / p , when oiliness, as indicated by the coefficient of static friction, is increased.
FIG
7-EFFECT
O F O I L I N E S S ON VALUE O F 8
Journal.. Bearing. Diameter Clearance
. . . . ,Steel
% / aA T TRANSITION POINT
..... .Bronze .... ,1110 + ......... .Velocite B spindle oil (Run 12) 0 . ......... .Lard oil (Run 13) .I
3-The variation of the transition point to be expected with commercial lubricants is small, certainly not over 10 per cent. 4-While it is possible to detect changes in oiliness by means of a suitable journal-testing machine, the experimental difficulties are such as to preclude the use of such a machine for measurement of this property. It is believed that the coefficient of static friction offers the most convenient single measurement of oiliness. 5-The position of the transition point is strongly affected by variations in clearance.
ACKNOWLEDGMENT The writers wish to express their indebtedness to the General Motors Research Corporation and the Standard Oil Company of New Jersey for permission to publish this paper; to the Godfred Cabot Fund a t the Massachusetts Institute of Technology for funds from which to construct the machine; to Charles Mills, of the Saco-Lowell Shops, for his coijperation and advice in planning the experimental procedure; to F. W. Perkins, of the Mechanical Engineering Department of the Institute, and to H. Van Keuren, of the Van Keuren Instrument Company, for the care and precision with which they constructed, respectively, the testing machine and the test parts and gages.
Gasoline Statistics
F I O . 8-EFFECT
OF CHANGING C L E A R A N C E ON V A L U E OF Z ? t / P AT
TRANSITION POINT
CONCLUSIONS The results of the foregoing work justify the following conclusions : 1-The position of the transition point from stable to un-
Gasoline production in the United States in January amounted to 695,322,500 gallons, and established a new high record of monthly production, according to figures compiled by the Bureau of Mines. The month of December had also established a new high record production mark. The daily average of gasoline production in January amounted to a n increase of 5.5 per cent over the December daily average and a n increase of 11.5 per cent over the rate for January a year ago. Imports of gasoline during January were 19,309,197gallons, a decrease of approximately 6,000,000 gallons from the December imports. The new gasoline supply (production plus imports) was 714,631,697 gallons, an increase of 12.1 per cent over last year's figures. In comparison with the gasoline statistics for January, 1923, the figures for January, 1924, show an increase of 12.1 per cent in the new supply, of 13.2 per cent in the total demand, and of 19.9 per cent in refinery stocks.
