Rolling-element bearing life predictions are based on statistical calculations of the fatigue life of the various bearing components. An assortment of parameters, which include load, rpm, lubrication type, viscosity, bearing material and even lubricant cleanliness enter into the calcualtion. The predicted life is expressed as the number of hours that a specified percentage of bearings will survive. For most industrail machinery, that percentage is typically 90% and you will hear the term "L-10" bearing life as a point at which there will be 10% failures.
Bearing catalogs have a dynamic load rating which is defined as the point at which 90% of a large statistical sampling of identical bearings under near perfect conditions will survive 1 x 10^6 cycles. That isn't long, just the short side of 300 hours for a 3600 RPM motor.
If you examine the equations for estimatating bearing life, it is apparent that the life of a bearing is a nonlinear function of applied load. For a ball bearing, it is related to load to the 3.00 power and in roller and needle bearings, it is the 3.33 power. Obviously that means that a relatively small increase in bearing load can cause a significant reduction in the calculated bearing life. A good source for the rating equation is the SKF site. You can also plug in differing values for the variables and see the effect. To get a statistical 5% failure rate (so called L-5 bearing life) for instance, the dynamic load rating has to be about 1.67 higher for a ball bearing. To achieve a 2% failure rate (called L-2 life) requires a dynamic load rating 3X higher. The motor in a plant is likely designed with the L-10 life in mind, whereas an aeroderivative gas turbine for an aircraft is more likely designed to an L-2 criteria, or even tighter.
In the real world many different factors that are beyond the basic rating equation come into play in the design of a machine. Lubricant type, viscosity and cleaniliness are very important factors in the life calcualtion and relatively minor changes may seriously impact the calculated life.
If you want to play a bit with the numbers so you have a better understanding of the effect of changes, I suggest the SKF website. You will also notice that three "lives" are predicted; one is the original based on fatigue life studies done I believe in the 1940's, another method commonly used by machinery manufacturers in the design stage, and the last life which is used by non-technical sales and marketing people. In that order, each gets higher and higher in predicted L-10 life.
Posts: 2743 | Location: Exton PA | Registered: 22 February 2005
If you examine the equations for estimatating bearing life, it is apparent that the life of a bearing is a nonlinear function of applied load. For a ball bearing, it is related to load to the 3.00 power and in roller and needle bearings, it is the 3.33 power. Obviously that means that a relatively small increase in bearing load can cause a significant reduction in the calculated bearing life.
So, doesn't that mean if the load applied to the bearings is increased even slightly, no amount of increasing lubrication can prolong the life of the bearings?
What is the main function of lubrication? Is it not to reduce the coefficient of friction?
Posts: 4285 | Location: Borneo | Registered: 13 February 2005
The basic dynamic load rating C is defined in the ISO specification 281:2007. Again, the basic dynamic load rating C is based on one million revolutions. A parameter for lubrication related variables is not part of the basic calculation.
ISO 281:1990/Amd 2:2000 also makes provisions for individual bearing manufacturers to recommend a suitable method for calculating the life modification factor to be applied to a bearing, based on operating conditions, which may include lubricant conditions and a "contaminant" factor. The various factors incorporated by SKF include lubricant conditions (viscosity ratio κ) and a contaminant factor ηc. These factors in theory would take into account lubricant effects based on actual operating conditions. You can play with the numbers but none will have a major effect like load, which again changes life by the third or 10/3 power.