Using Weibull regression to estimate the load-life relationship for rolling bearings.

Statistical techniques have recently been developed based on the method of maximum likelihood, for forming exact interval estimates of the parameters of a power function relation between a Weibull scale parameter and an independent variable known generically as 'stress'. This article illustrates the application of these techniques to point and interval estimation of the load life relationship in rolling bearings given the observed lives in identically censored endurance tests run at two or more loads. Interval estimates are also given for the Weibull shape parameter (assumed invariant with load), the tenth percentile life at the stresses at which the testing was done, and the extrapolated tenth percentile life at other (lower) stresses. The article includes a method for testing whether the power law model fits the observed data. Tables of the numerical values needed for implementing the various computations are given for a range of practical sample sizes. (A)