Statistical Cumulative Damage Theory for Fatigue Life Prediction

A statistical cumulative damage theory is developed with the purpose of prediction of mean, standard deviation and probability density of fatigue lifetime of randomly variable specimens subjected to the same deterministic cyclic loading program. The theory requires availability of a deterministic cumulative damage theory for ideal nonvariable specimens, called clones. Detailed analysis is given for two-stage cyclic loading based on a previously developed deterministic cumulative damage theory and log-normal distribution of S—N curve lifetimes. Results indicate that the usual interpretation of deterministic cumulative damage theory in terms of means of lifetimes is not valid for significant scatter. Preliminary experimental results for two-stage loading are compared with analytical predictions.