Fatigue model of bearing steel based on stressed volume

Abstract

Fatigue is one of the most important life limiting factors for bearings. A number of different kinds of tests are used to assess the behaviour of bearing steels when subjected to repeated loading. The work presented here attempts to rationalise these test methods on the basis that they should all be related by the volume of material that is stressed. The analysis focuses on one particular steel, ‘AISI 52100’, which is the most common alloy used in the manufacture of bearings. It is demonstrated that a physically reasonable model can be produced using the neural network technique, in spite of some quite dramatic differences in sample geometry and loading conditions, thus establishing that simple tests can, in principle and practice, be used to estimate the performance during the fatigue experiments.