SEDL / STP / STP1419-EB / STP10850S

Statistical Prediction of the Maximum Inclusion Size in Bearing Steels

Shi, G
Research Associate, University of Sheffield, Sheffield,
Section Leader, TWI Limited, Abington, Cambridge

Atkinson, HV
Reader, University of Sheffield, Sheffield,

Sellars, CM
Professor, University of Sheffield, Sheffield,

Anderson, CW
Professor, University of Sheffield, Sheffield

Yates, JR
Professor, University of Sheffield, Sheffield,

Pages: 13    Published: Jan 2002

Download this paper for $25 PDF (252K)          View License Agreement
Abstract

Large and brittle oxide inclusions may initiate fatigue failure in bearing steels. The size of the maximum inclusion in a large volume must be predicted by statistical analysis because only small samples can directly be analysed and there are limitations on non-destructive testing methods. A new method based on the Generalized Pareto distribution (GPD) was recently proposed by the Sheffield group. This allows data on inclusion sizes in small samples to be used to predict the maximum inclusion size in a large volume of steel. The method has advantages over other statistics of extremes methods. The number of sources of failure and the failure rate of practical bearings can be estimated from the predictions of the GPD and the stress distribution in the bearing. Here the GPD method is compared with the Statistics of extreme values (SEV) method developed by Murakami and co-workers. The application of predictions from the GPD in the safe design of bearings will be illustrated.