Chapter 21 Lubricant contamination 21.1
At present, the operating life of rolling bearings can be calculated by formulae published by bearing manufacturers which are also generally supported by international and national standards; e.g. ISO, ANSI, DIN, etc. The standards specify methods of calculating basic dynamic load rating and rating life of rolling bearings well-manufactured from good quality steel, and basically of conventional design with regard to the shape of rolling contact surfaces. The methods contained in the standards are based on the work of G. Lundberg and A. Palmgren conducted during the late 1930’s and early 1940’s [Lundberg 1947, Lundberg 19521. The ball and roller bearing load ratings and life calculation methods were, and remain, representative of the manufacturing methods, materials, lubrication methods and testing methods then available. Moreover, all experiments used by Lundberg and Palmgren were performed under the cleanest laboratory conditions that could be attained at that time in order to give the correct parameter values in the life calculation formulae for rolling element bearings. Thus it was assumed that there was no influence from dirt particles on the bearing lives. However, the filtration level for the lubricants in those experiments, whenever filters were used, was no better than 10 pm absolute, thus some of the failures were probably caused by dirt particles and asperity interaction, leading to surface-initiated failure. How well “well-lubricated” bearings were lubricated was not defined until elastohydrodynamic theory became available in the 1960’s [Grubin 1949, Petrusevich 1951, Dowson 19591. Then the oil film thickness to surface roughness ratio A = 2 to A = 4 was defined as good lubrication. The ball bearings in Lundberg’s and Palmgren’s investigations had fine enough surfaces to obtain A-values of this order, but the old roller bearings had coarser surfaces, thus the major part of the old tests performed with roller bearings seems to have been carried out at A-values around 0.5. Today we know that such a low A-value decreases the life of the bearing and also causes wear particles to be produced in the bearing. Through the years, the load carrying capacity, found in laboratory experiments with rolling element bearings, has increased. This has mainly been assumed to depend on the cleaner steels and the finer surfaces of the bearings, but at the same time the filtration of the lubricants used in the tests has become better and better, so some of the contribution to longer lives has come from cleaner lubricants. Today we know that hard particles larger than the oil film thickness are usually detrimental 341
CHAPTER 21. LUBRICANT CONTAMlNATlON
to bearing life. In rolling element bearings the oil film thickness is normally of the order 0.1 to 3 pm, thus filtration should be kept at a similar level to obtain maximum bearing lives. Even dirt particles much smaller than the mean film thickness give wear if they are hard and in sufficiently large concentrations. If they are large enough to penetrate the oil film, they give local stresses at the surfaces and thereby shorten the life of the bearing considerably. In all normal endurance tests, lubricants are virtually free from water. In outdoor applications and when the machines contain water or steam, the lubricant is very easily contaminated with water. It is also well known that free water in a lubricating oil decreases the life of rolling element bearings ten to more than a hundred times, depending on the water content. How water can decrease bearing life as much as it does is not well understood, but a small concentration of water of 0.01 per cent is enough to decrease the bearing life to half its original value. The reduction of bearing life with water concentration is steepest when the water can be dissolved in the oil, but even at high concentrations more water gives shorter life. This means that it always pays to keep the dirt and water concentration in the lubricant as low as possible. In connection with the above, the ongoing research into contamination is reviewed. Here 3 - D mapping of surface areas is used in which dents generated by debris are present. These mapped areas are then used in a numerical model to calculate the contact pressure between the dented raceways and the rolling elements. From this pressure distribution the sub-surface state of stress within the bearing raceway is finally calculated. The stress information, which reflects the pressure of dents, is used in conjunction with a new fatigue model to predict the reduction in the fatigue life of the bearing. In addition to this prediction of life reduction, debris deformation studies are reported which help us to understand how particle hardness and particle size affect the surface damage. Finally, examples of predicted life reductions resulting from contamination are given together with examples of bearing operation in real applications [Ioannides 19891.
The earliest papers on contaminated lubricants for rolling element bearings treated water contamination. In 1968 and 1969 Schatzberg and Felsen published two papers on the effect of dissolved water on rolling element fatigue life [Schatzberg 1968, Schatzberg 19691. The nominal stress level on the four-ball machine was varied between 6.4 GPa and 9.0 GPa, but of course there was a great deal of plastic deformation in the balls at these high nominal stress levels. Under these high stress conditions and in a laboratory environment, the presence of 100 ppm (0.01 per cent) water dissolved in the squalane lubricant decreased the fatigue life 32 to 48 per cent. Schatzberg and Felsen proposed that water collected in micro-cracks in the ball surface and that this caused corrosion and hydrogen embrittlement, giving a reduction in fatigue life. Later the same year, Yardley and Crump [Yardley 19691 showed some results from grease lubricated rolling element bearings where they stated that dirt introduced into the bearing by the grease packing procedure or through the grease channels on re-greasing could rapidly destroy the bearing. In 1972 Felsen, McQuaid and Marzani [Felsen 1972) studied the effect of sea water on floodlubricated angular contact ball bearings. The lubricants tested were of different physical and chemical classes and included five mineral oils, one triaryl phosphate and two mineral oil-base
21.2. LITERATURE REVlEW
seawater emulsifying fluids of different viscosities. The fatigue life behaviour of the different lubricants was very similar. When the lubriction was good and A > 1.5, the life of the bearings 0.5. For both lubrication regimes X 1.5 and was about ten times higher than when X X 0.5, the introduction of 0.1 to 1 per cent seawater decreased the L ~ life o by about a factor of five and the Llo life to half its value. In the worst case, fatigue life decreased by as much as 80 per cent. Depending on the solubility of water in the oil, the influence on bearing life was different. As long as the oil could dissolve all water or keep it floating in the form of microscopic droplets, the endurance life of bearings decreased rapidly with increasing water content. As soon as the water separated and fell to the bottom of the lubricant container, the influence of increased water content was much less pronounced. Another consequence of the addition of sea water was a general increase in Weibull slope with increasing amounts of dissolved water. This was also accompanied by a change of failure mode, from ball failures to race failures, when the water content increased. In two papers Fitzsimmons together with Cave [Fitzsimmons 19751 and together with Clevenger [Fitzsimmons 19771 investigated the wear of taper roller bearings as a function of the amount of contaminant in the lubricant. They found that :
1. The wear was proportional to the amount of contaminant.
2. Tapered roller bearings will continue to wear as long as the particle size of the contaminant is larger than the lubricant film thickness between the bearing surfaces.
3. For bearings to wear significantly, the contaminant particle hardness has to be larger than or equal to the hardness of the bearing material.
In their paper, they also refer to an earlier paper by Okamoto et al. [Okamoto 19721 where they found life reductions of 80 - 90 per cent when bearings were contaminated with different amounts of particles of various hardness. In their own papers they do not give any absolute values for the life reductions. In (Fitzsimmons 19771 they claim that the wear rate changes with time so that when the dirt particles have been worn down, the wear rate decreases until new sharp debris is added to the oil. In a paper from 1976, Tallian describes an experimental investigation of rolling contact fatigue life for contaminated deep groove ball bearings [Tallian 19761. Depending on the lubrication system and the type of lubricant and contamination, he obtained different lives for the bearings. He also showed the influence of different types of steel. With vacuum arc remelted, carburized 8620 steel, the Llo lives were about seven times the values for vacuum degassed 52100 steel when they were both oil lubricated. For grease lubrication with mineral oil based lithium soap grease, the difference was much less. The 8620 steel lasted about 50 per cent longer than the 52100. The lives of 52100 steel bearings in an open housing with daily regreasing were 15 times longer than those in a closed housing with no re-greasing, but even that was inferior to lubrication with a well-filtered oil which gave a life 50 times longer. The importance of cleanliness for bearing life is demonstrated by three groups of experiments, where the first group is ultrasonically washed, greased and sealed. The second group is kerosene washed, greased and open. The third group is sump lubricated with oil artificially
C H A P T E R 21. L U B R I C A N T CONTAMINATION
contaminated with powdered hardening scale. The relative lives of the three groups are 100 per cent, 23 per cent and 2 per cent respectively. The sump lubrication in a differential gear oil with 5 mg of powdered heat treating scale up to 600 pm platelet diameter thus gave a reduction of life of 1:50 or more. On the other hand, Tallian shows that extreme cleanliness pays off, giving 15-30 times longer life than expected from normal calculations. Under laboratory conditions, the standard lives of bearings in 1976 were about 10 times the ISO-Llo lives as compared to the 30 times achieved in aircraft turbine bearings. This meant that under so-called laboratory conditions, the dirt made the standard test damage surface-dominated and the measured lives were probably at least three times lower than they could have been with clean oils. Before this was published, Dalal et al. [Dalal 19741 made some preliminary tests under ultraclean conditions where the lubrication oil was filtered through a 3 pm Millipore filter. The five bearings in the test ran more than 40 times their theoretical Llo lives without failures. Under standard test conditions, those bearings were known to have 4-5 times their theoretical Llo life. The significant increase in life observed demonstrated the beneficial effect of lubricant cleanliness on bearing life. To investigate the influence of a few hard particles in the oil, Dalal and Senholzi [Dalal 19771 performed experiments with deep groove ball bearings and tapered roller bearings, in which they had made Vickers hardness indentations in the inner ring surface. The diagonal dimension of the indentations was 150-153 pm as compared to the Hertzian contact width of 760 pm and 230 pm respectively for ball bearings and roller bearings. Seven of the fourteen ball bearings failed at the Vickers indentation, five failed elsewhere and two were suspended. None of the inner ring failures for the tapered roller bearings occurred at the Vickers indentation. The life of the indented ball bearings was about 7 times the Llo as compared to more than 40 times the Llo life for the non-indented bearings in the earlier investigation. For the indented tapered roller bearings, the Llo life was 11.4 Mrev. instead of the theoretical 10 Mrev. Part of this difference in life for the ball and roller bearings can be explained by the higher A-value for the ball bearing. In an interesting paper from 1979 Loewenthal and Moyer describe laboratory experiments with deep groove ball bearings lubricated with an uncontaminated and an artificially contaminated lubricant [Loewenthal 19791. The lubricant type was neopentylpolyol(tetra)ester which is normally used as a gas turbine engine lubricant. The artificial lubricant contaminant contained 88 per cent carbon dust, 11 per cent Arizona test dust and 1 per cent stainless steel particles. In the tests with uncontaminated lubricant, the lubricant was circulated through a 49 pm absolute filter. This means that a lot of particles produced within the system were circulated through the bearings. Despite this, the 10 per cent life for the test bearings was 672 hours whereas the AFBMA catalogue 10 per cent life was 47 hours. Even if the adjustment for good lubrication is taken into account, the observed life was 2.7 times longer than expected. When test dirt was added to the oil at a rate of 125 mg per bearing per hour, the Llo lives were not changed very much for oils filtered through 3 pm and 30 pm filters, but the Lso lives were decreased. As soon as dirt particles were introduced into the lubrication system, the Weibull slope increased considerably and the surfaces of the balls and races started to show progressive surface damage. Using the coarsest filters in the investigation, 49 pm and 105 pm absolute, the bearing surfaces showed extensive micro-cracking and wear. In fact, with the 105 pm filter the wear was so extensive that the test could not be continued to fatigue failure. The same year, Perrotto, Riano and Murray published an investigation of the effect of abra-
21.2. LITERATURE REVIEW
sive contamination on the performance of ball bearings [Perrotto 19791. They did not study the life of the bearings but how hard contaminants of different particle size affected the oil film separation of the bearing surfaces. The particles were of two materials and four sizes: Silicon dioxide ( S O 2) particles 50 pm in diameter, and aluminium oxide (A1203) particles with diameters of 3, 0.3 and 0.06 pm. The calculated oil film thickness for the deep groove ball bearings tested was 0.33 pm between the balls and the inner race, and 0.38 pm between the balls and the outer race, when they were running at 1700 rpm. This means that it should be possible to have the two finest grades of aluminium oxide particles in the oil film without causing damage to the surfaces. When the tests started, the bearings were run in for 70 to 80 hours with clean oil until all electrical contact through the oil film disappeared. Then new test oil with contaminants was injected into the circulating system and the bearings were run with the dirty oil for one hour, whereafter the oil was drained out and a fresh charge of clean oil was put into the system. The test was then continued to determine if the bearing could run in again. The purpose of this step was to obtain some qualitative measure of how badly the bearing surfaces had been damaged by the abrasive particles. Perrotto et al. found that bearings exposed to particles larger than the EHL film thickness (3 pm and upwards) showed no signs of running back in. When the dirty oil was replaced with clean lubricant, there was no change in the percentage of electrical contact through the oil film. Even after more than two days of running with clean oil, there was no indication of run-in. Only when particles smaller than the EHL film (0.3 and 0.06 pm) were used did there seem to be a true recovery. The bearing contaminated with 0.3 pm particles returned to its well run-in state in less than 24 hours, and for the bearing contaminated with 0.06 pm particles, the recovery time was about 15 hours. One observation made by the authors was that the particles seemed to be crushed as soon as they came into the EHL contact, thus the damage from dirty oil stopped as soon as the particles were crushed down to a small enough size. Obviously this size was much smaller than the oil film thickness as even the 0.06 pm particles showed the same behaviour despite the six times larger oil film thickness. Independent of particle concentration, all particles larger than the oil film thickness caused permanent damage to the bearing surfaces before they were crushed, and the 3 pm A1203 particles, or fragments of them, became embedded in the steel surfaces. In [Loewenthal 19791 and [Perrotto 19791 the influence of filtration level was studied for one type of bearing steel and one type of heat treatment each. Under clean conditions, the life of bearings normally increases with the hardness of the steel. In a paper from 1981 Sugiura et al. (Sugiura 19821 state that the heat treatment which decreases the crack sensitivity of a bearing steel can increase the rolling contact fatigue life significantly, especially when using debris-contaminated lubricants but also under non-contaminated lubrication. For lubrication with 20 ml 56 cSt turbine oil and 0.1 g foreign particles, the optimum heat treatment gave 5-6 times longer ,510 and L ~ lives. o In a continuation paper to their 1979 work, Loewenthal, Moyer and Needelman studied the effect of ultraclean and centrifugal filtration on rolling element bearing life [Loewenthal 19821. This time they used the 3 pm absolute filter in the circulation system, also when no artificial contamination was added. By doing so they increased the Llo life of the bearings by a factor of two as compared to the contaminated case with 3 pm filtration, and the Llo life increased by 64 per cent relative to the baseline tests with a laboratory clean lubrication system. They also investigated the wear rate as a function of the filtration. The non-failed bearings in the baseline tests with clean oil and 49 pm absolute filtration had 3.9 times the weight loss of the
CHAPTER 21. LUBRlCANT CONTAMINATION
bearings from the ultraclean tests while the bearings from the centrifugal filter, 3, 30, 49 and 105 pm filter tests with contaminated oil had 6.6, 7.4, 12.5, 16.2 and 344.6 times respectively the weight loss of the ultraclean test bearings. As the wear rate is so high when the bearings are lubricated with contaminated oil, it is not only the filter rating that is important but also how big a flow of oil per time unit the filter can accept. The filter must at least be able to catch as many particles per hour as the amount produced by wear in the bearing. In a paper from 1982 Sayles and Macpherson found that the early failures determining the Llo lives for contaminated bearings were mostly surface-initiated. They experimented with standard 25 mm bore, extra light series, single row cylindrical roller bearings with a brass cage. The contamination of the lubricant was produced by a helicopter gear box lubricated with the same oil as the test bearings. The oil containing the wear debris from the gear box was pumped through different filters to the bearing test rig. The cartridge-type filters were of absolute ratings 40, 25, 8, 3 and 1 pm. The amount of particles in the oil was strongly influenced by the filter rating giving 100 particles larger than 2.5 pm per millilitre of oil in 4 hours for the 40 pm filter and in 24 hours for the 3 pm filter. Sayles and Macpherson found that oil film thickness had a minor influence on bearing life, but that the filter rating had a much bigger influence. The finer filters always gave longer bearing lives with typically 6 times longer Llo life with a 3 pm filter than with a 40 pm filter. The most striking of their results is that surface damage by dirt particles in the oil does not recover if the oil is cleaned after a short while. They made pre-runs with 40 pm filter for 30 minutes and then the rest of the test was run with 3 pm filtration. Despite the fine filtration during the main part of the test, the bearings failed just as early as if they had been lubricated with the dirty oil all the time. The damage done to the bearing surface by the dirt particles during the first half-hour was enough to cause early failure. These early failures were in the form of transverse cracks perpendicular to the bearing surface and with a depth of 0.2 to 0.75 mm. This indicates that the mechanism of failure is associated with a change in surface contact properties rather than with the continued presence of a third body. In their experiments, the finest filtration level, 1 pm, did not give longer lives than the 3 pm filtration. This means that the oil film thickness and surface roughness were such that no large, extra stresses were introduced by the particles passing the 3 pm filter and therefore the life was the same. Sayles and Macpherson found a relationship between life of the bearings (L)and the absolute filter rating ( p ) . For the plate-formed wear debris produced by the gear box, the life of the bearings was proportional to the filter rating raised to the power minus two-thirds.
For other forms of dirt and wear debris, the relation can be different. Another influence of wear is the redistribution of forces and stresses in the bearing. Lorosch [Lorosch 19831 showed that experiments with worn bearings always exhibited pitting failure in the non-worn parts as the contact pressure there was higher. He also showed that calculated bearing life was always shorter than the real life of the bearings, the only exception being when dirt particles entering with the oil made sharp indentations in the bearing surfaces. In that case, the real life and the catalogue life of the bearings were similar. Under ideal lubricating and cleanliness conditions, the bearing life increased with decreasing load to a much greater extent than obtained by standard calculations. He also found that the influence of contaminants is less for large bearings than for small ones. Consequently, small bearings have
21.2. LITERATURE REVlEW
a larger capacity than calculated if they are reasonably clean. To investigate the influence of dirt, Lorosch made small indentations in the rolling track with hard balls of different sizes. The steeper the edges of the indentations, the shorter the life of the bearing. When a 0.4 mm ball was pressed into the surface to a depth of 13 pm, the life of the bearing decreased by 98.5 per cent to only 1.5 per cent of the life without indentation. His experiments showed that the sharp indentations never disappeared but stayed and gave high stress concentrations for the rest of the bearing life. When it comes to wear, the situation can be slightly different. Skorynin and Michenya showed in their paper [Skorynin 19841 that the wear rate was very much dependent on the size of hard particles. When they added 10 ppm of electrocorundum particles, 5 pm in size, to the lubricating oil, the balls wore more than twice as much as the inner race, the outer race and retainer together because these particles were embedded in the plastic retainer surface. This wear rate decreased when the whole bearing, including the retainer, was cleaned of the particles. With larger particles, diameter 12 pm, wear of the rolling tracks also increased rapidly and for high loads the wear rate increased and was concentated on certain bands around the balls and tracks. If one of the rolling element surfaces is much harder than the other, the result will be similar if the A-value is low. Ishibashi, Hoyashita and Sonoda [Ishibashi 19851 studied the influence of different hardnesses for the two surfaces working together. The surface roughness of the two roller surfaces was 3 pm R,,, and the calculated oil film thickness was about 1 pm for the load range 1.5 GPa to 2.8 GPa. When a 520 HB roller was combined with a case hardened 800 HV roller at 1.5 GPa pressure, they never wore in but metallic contact continued for 800.000 revolutions, after which the surfaces exhibited pitting failures. In contrast to this, when the 525 HB rollers with the same roughness of 3 pm R,, were used in equal hardness combinations, the surfaces wore in so that 90 per cent of the time there was no metallic contact and they could be rotated 6.5 x lo6 revolutions at a contact pressure of 2 GPa without failure. For rollers with a mirror-like finish, 0.2 pm &,, and equal hardness, the metallic contact stopped after lo6 revolutions and no pitting occurred within lo’ revolutions, even at 2.4 GPa pressure. Their observation of metallic contact through the oil film for the first million revolutions is very interesting as the A-value for the mirror-like rollers was about 15. This means that it does not matter how fine the surfaces are ground, run-in can improve them. This also means that wear particles are always produced in new machinery and should be transported away from the elastohydrodynamic contacts. When Ishibashi et al. used one case hardened roller together with a 500 HB roller, both with mirror finish, at 2.3 GPa pressure, the life was less than lo6 revolutions before severe pitting occurred. This means that the occasional asperity contacts, even at A = 15, between a soft and a hard steel surface will decrease the life considerably compared to when the surfaces are of similar hardness. If they are of similar hardness they will wear in and the surfaces will be so fine that metallic contact stops. In the experiments of Ishibashi et al., a pitting limit of p,,,=0.45 HB was found, which is considerably higher than the conventional pitting limit of pmn,=(0.2-0.3) HB. From the above review it is obvious that the water and dirt content in the rolling element bearing lubricant should always be as low as possible. Even minute amounts of water of the order per mille and dirt particles of the order parts per million are enough to considerably decrease the life of bearings. If, on the other hand, it is not possible to reach those low contamination levels, life is always increased if the dirt level is lowered. In two papers
CHAPTER 21. LUBRICANT CONTAMINATION
Scales in pm
Figure 21.1 Mapped dents. dealing with sealed rollneck bearings for steel rolling mills, the sealing is not very successful [Yamamoto 1985, Kusiner 19861. In the lubricating grease, the typical water content level reached is about 10 per cent and the hard particle content is 1 per cent. This is still so much better than the old value of water content in the grease of 40 per cent that the number of catastrophic failures has decreased from several per year to zero between the yearly inspections.
In addition to the experimental work described above, theoretical calculations have been published in recent years in which an attempt has been made to predict the life reduction caused by contamination. Previous experimental work [Sayles 19821 indicates that, under certain conditions, dents, formed by overrolling of debris, act as sites of stress concentrations and as such are responsible for the initiation of surface fatigue associated with a life reduction of the bearings. This was particularly true in [Sayles 19821 where a gear machine produced the debris predominantly in the form of ductile particles which, in turn, produced smooth dents, see figure 21.1. With the advent of the New Life Theory [Ioannides 19851 it was possible to utilise stress information in the locality of the dents to predict the reduction of the fatigue life of the rolling bearings. In 1985 Webster, Ioannides and Sayles [Webster 19851 published such calculations in which a number of dents from the tests described in [Sayles 19821 were selected and mapped (3-D), see figure 21.1. The information contained in these topography measurements was subsequently used in conjunction with a numerical contact model, coupled to a sub-surface finite element analysis, to evaluate the stresses in the bearing raceways in the presence of the dents. Then, as indicated above, these stress fields were used to calculate life reductions resulting from the presence of the dents. In the event, the predicted life reduction was in reasonable agreement with the one observed
21.3. THEORETICAL RESULTS
I 1> i:
1 0 N L
Figure 21.2a The numerical solution for a frictionless elastic/plastic contact of a smooth steel roller against a hypothetical stepped flat surface simulating the presence of debris. The top diagram shows the pressure distribution whilst the two lower ones show the original configuration and the resulting contact geometry. in the tests described in [Sayles 19821. Since then, additional work has been published. Sayles and Ioannides [Sayles 19881 examined the effect of debris type and geometry in relation to the formation of dents. The concept of elastic conformity around entrained debris was studied and it was shown that a critical debris aspect ratio may exist beyond which no damage on the contacting surfaces occurs. This is because the surfaces close elastically around the debris and further loading is taken up elastically by them, see figure 21.2. In a sequel to this work, Hamer, Sayles and Ioannides [Hamer 19871 studied the debris and raceway deformation modelled as an extrusion process. The effects of the film thickness, the raceway and particle hardness and the particle size were introduced into the analysis and combinations of these parameters define a boundary between debris that can damage the contact surfaces and debris which cannot. An example of these regions for a particular configuration is given in figure 21.3. This work was experimentally verified by the same authors [Hamer 1988a] when particles of different hardness were pressed between two anvils and the measured dents were compared to the predictions of [Hamer 19871. In 1988 KO and Ioannides [KO19881 made a finite element calculation of the debris denting and obtained as a result the residual stresses left in the steel after the indentation. They could therefore calculate the decrease in fatigue life caused by the indentation. The FEM
CHAPTER 21. LUBRICANT CONTAMINATION
Figure 21.2b As figure 21.2a, but with an approximate doubling of the normal load.
(Initial panicle diamcwY(fdm(shim) Ihickncss)
Figure 21.3 Mapping of safe/damaging regimes of debris size and hardness combinations.
calculations also gave as a result the actual form of the indentation and the height of the ridge around the dent. Under this ridge is a region of tensile residual stress which not only gives a lower endurance strength but the ridge itself causes high stress variations when it is overrolled. This induces a large reduction in life. In 1988 Hamer, Lubrecht, Ioannides and Sayles [Hamer 1988a] made a similar calculation using slip line field analysis which can be applied to deep transverse indentations, where the assumptions of plane plastic strain can be justified. The most striking conclusion is that where failure is initiated through surface indentation and associated residual stresses, the expected lives may not increase with decreasing load as quickly as would be predicted by conventional models. The reduction in expected lives is very sensitive to the size of the dent in relation to the roller radius, and this may well have important consequences in terms of critical particle size and safe and unsafe levels of filtration. Ioannides, Jacobson and Tripp [Ioannides 19881 introduced the decrease of endurance strength with hydrostatic tensile stresses in the steel caused by residual stresses or hoop stresses from the mounting of the bearing. The calculations show a slight increase in bearing life for clean and undented conditions without hoop stresses, but a sharp decrease in life when tensile stresses and hoop stresses are introduced into the calculations. For a hoop stress of 325 MPa, the life decreases to only 1 per cent of the life for a bearing with the same load but without hoop stresses. The indentations in bearings caused by overrolling of wear debris also induce tensile residual stresses close to the raceway surface and therefore give a large reduction in life, not only by causing high stress variations but also by decreasing the local endurance strength of the bearing steel.
When dirt particles in lubricating oils are considered, it is obvious that the cleaner the oil the better it is. Even dirt particles much smaller than the mean film thickness cause wear if they are hard. If they are large enough to penetrate the oil film thickness, they cause local stresses at the surface and thereby shorten the life of the bearing considerably. This was shown already in 1973 when Leibensperger and Brittain [Leibensperger 19731 measured the stresses below asperities in Hertzian contacts using photoelasticity. Such stresses can now be modelled theoretically and, with the use of the New Life Theory, predictions of life reduction in the presence of contaminants can be provided. Finally, the reason why water can decrease bearing life as much as it does is not well understood, but as small a concentration as 0.01 per cent is enough to decrease the bearing life to half its original value. The reduction of bearing life with water concentration is steepest when the water can be dissolved in the oil, but even at high concentrations, more water gives shorter life. This means that it always pays to keep the dirt and water content in the lubricant as low as possible.
Bibliography [Dalal 19741
Dalal, H.M., et al., “Progression of surface damage in rolling contact fatigue”, U.S. Navy Office of Naval Research, Final report on contract No. N00014-73-C-0464, SKF report AL75T002, AD780453.
Dalal, H.M., and Senholzi, P., “Characteristics of wear particles generated during failure progression of rolling bearings”, ASLE Trans., Vol. 20, 3, 1977, pp. 233-243.
Dowson, D., and Higginson, G.R., “A numerical solution to the elastohydrodynamic problem”, J. Mech. Eng. Sci., 1(1), 1959, pp. 7-15.
Felsen, I.M., McQuaid, R.W., and Marzani, J.A., “Effect of sea water on the fatigue life and failure distribution of flood-lubricated angular contact ball bearings”, ASLE Trans., Vo1.15, 1, 1972, pp. 8-17.
Fitzsimmons, B., and Cave, B.J., “Lubricant contaminants and their effect on bearing performance”, Society of Automotive Engineers, 750583.
Fitzsimmons, B., and Clevenger, H.D., “Contaminated lubricants and tapered roller bearing wear”, ASLE Trans., V01.20, 2, 1977, pp. 97-107.
Grubin, A.N., and Vinogradova, I.E., “Investigation of the contact of machine components”, Central Scientific Research Institute for Technology and Mechanical Engineering, No. 30, Moscow.
Hamer, J.C., Sayles, R.S. and Ioannides, E., “Deformation mechanisms and stresses created by third body debris contacts and their effects on rolling bearing fatigue”, Proc. of the 14th Leeds-Lyon symposium on Tribology, Lyon, September 1987, Interface Dynamics, Tribology Series, 12, Elsevier, pp.201-208.
Hamer, J.C., Sayles, R.S., and Ioannides, E., “Particle deformation and counterface damage when relatively soft particles are squashed between hard anvils”, Paper presented at ASME/STLE Tribology Conference, Baltimore, USA, October 1988.
Hamer, C., Lubrecht, A.A., Ioannides, E., and Sayles, R.S., “Surface damage on rolling elements and its susequent effect on performance and life”, Proc. of the 15th Leeds-Lyon Symposium on Tribology, Leeds, September 1988. 353
Ioannides, E., and Harris, T.A., “A new fatigue life model for rolling element bearings”, ASME Journal of Lubrication Technology, Vol. 107, 1985, pp. 367-378.
Ioannides, E., Jacobson, B., and Tripp, J., “Prediction of rolling bearing life under practical operating conditions”, Proc. of the 15th Leeds-Lyon Symposium on Tribology, Leeds, September 1988.
Ioannides, E., Jacobson, B., and Sayles, R.S., “Contamination in Lubricants - Reduction in Bearing Life”, Proc. of the 5th Eurotrib Conference, 1989, pp. 14-19.
Ishibashi, A., Hoyashita, S. and Sonoda, K., “Remarkable increase in pitting limit of through hardened steel rollers under rolling with sliding conditions”, Proc. of the JSLE International Tribology Conference, July 1985, Tokyo, Japan, pp. 929-934.
KO, C.N., and Ioannides, E., “Debris denting - the associated residual stresses and their effect on the fatigue life of rolling bearings: An analysis.”, Proc. of the 15th Leeds-Lyon Symposium on Tribology, Leeds, September 1988.
Kusiner, W.J., Morinaga, N., Mauel, H.M., and Akiyama, M., “Sealedclean rollneck bearings”, Iron and Steel Engineer, June 1986, pp. 52-56.
[Leibensperger 19731 Leibensperger, R.L., and Brittain, T.M., “Shear stresses below asperities in Hertzian contacts as measured by photoelasticity”, ASME Trans., Journal of Lubrication Technology, July 1973. [Loewenthal 19791
Loewenthal, S.H., and Moyer, D.W., “Filtration effects on ball bearing life and condition in a contaminated lubricant”, ASME Trans., Journal of Lubrication Technology, Vol. 101, April 1979, pp. 171-176.
Loewenthal, S.H., Moyer, D.W., and Needelman, W.M., “Effect of ultraclean and centrifugal filtration on rolling element bearing life”, ASME Trans., Journal of Lubrication Technology, Vol. 104, July 1982, pp. 283292.
Lorosch, H-K, “Research on longer life for rolling element bearings”, ASLE Lubrication Engineering, Vol. 41, 1, 1983, pp. 37-43.
Lundberg, G., and Palmgren, A., “Dynamic capacity of rolling bearings”, Acta Polytechnica, Mech. Eng. Vol. 1, No. 3, 1947.
Lundberg, G., and Palmgren, A., “Dynamic capacity of rolling bearings”, Acta Polytechnica, Mech. Eng. Vol. 2, No. 4, 1952.
Okamoto, J., Fujita, K., and Yoshioka, T., “Effects of solid particles in oil on the life of ball bearings”, J. of Mech. Eng. Lab., Tokyo, Japan, Vol. 26, No. 5, 1972.
Perrotto, J.A., Riano, R.R., and Murray, S.F., “Effect of abrasive contamination on ball bearing performance”, ASLE Lubrication Engineering, Vol. 35, 12, 1979, pp. 698-705.
Petrusevich, A.I., “Fundamental conclusions from the contacthydrodynamic theory of lubrication”, IZV. Akad. Nauk. SSSR (OTN), p. 209.
Sayles, R.S., and Macpherson, P.B., “Influence of wear debris on rolling contact fatigue”, Rolling Contact Fatigue Testing of Bearing Steels, ASTM STP 771, J.J.C. Hoo, Ed., American Society for Testing of Materials, 1982, pp. 255-274.
Sayles, R.S., and Ioannides, E., “Debris damage in rolling bearings and its effect on fatigue life”, ASME Journal of Lubrication Technology, Vol. 110, 1988, pp. 26-31.
Schatzberg, P., and Felsen, I.M., “Effects of water and oxygen during rolling contact lubrication”, WEAR, 12, 1968, pp. 331-342.
Schatzberg, P., and Felsen, I.M., “Influence of water on fatigue failure location and surface alteration during rolling contact lubrication”, ASME Trans., Journal of Lubrication Technology, Vol. 91, 2, 1969, pp. 301-307.
Skorynin, Y.V., and Minchenya, N.T., “Rolling bearing wear model”, Trenie i Iznos, Vo1.5, No.2, 1984, pp. 266-272.
Sugiura, I., Kato, O., Tsushima, N., and Muro, H., “Improvement of rolling bearing fatigue life under debris-contaminated lubrication by decreasing the crack sensitivity of the material”, ASLE Trans., Vol. 25, 2, 1982, pp. 213-220.
Tallian, T., “Prediction of rolling contact fatigue life in contaminated lubricant: Part I1 - Experimental”, ASME Trans., Journal of Lubrication Technology, 1976, pp. 384-392.
Webster, M.N., Ioannides, E., and Sayles, R.S., “The effect of topographical defects on the contact stress and fatigue life in rolling element bearing”, Proc. of the 12th Leeds-Lyon Symposium on Tribology, Lyon, September 1985. Published by Butterworths in “Mechanics of Surface Distress”.
Yamamoto, K., Yamazaki, M., Akiyama, M., and Furumura, K., “Introducing of sealed bearings for work roll necks in rolling mills”, Proc. of the JSLE International Tribology conference, July 1985, Tokyo, Japan, pp. 609-614.
356 [Yardley 19691
BIBLIOGRAPHY Yardley, E.D., and Crump, W.J.J., “Some failures of grease lubricated rolling element bearings”, Proc. Inst. Mech. Engs., Vol. 184, 1969, pp. 63-73.