Abstract

A hierarchical Bayesian approach to analysing fatigue test data including reinserted specimens is proposed. It is found that the inference model is capable of fitting the fatigue damage model to the observed data well. After the addition of rerun specimen data, the results show a significant change in the predictive SN curves. For the analysed 40CrMo8 fatigue data sets, the deviation of fatigue limit is the primary explanatory mechanism for the observed fatigue life scatter. The change of predictive fatigue limit distribution after the addition of rerun data is compared to the change due to additional (simulated) virgin specimen tests.