Modeling metallic fatigue data using the Birnbaum--Saunders distribution

This work employs the Birnbaum--Saunders distribution to model metallic fatigue and compares its performance to fatigue-limit models based on the normal distribution. First, we fit data for 85 fatigue experiments with constant amplitude cyclic loading applied to unnotched sheet specimens of 75S-T6 aluminum alloys. The fit obtained by the Birnbaum--Saunders distribution is noticeably better than the normal distribution. Then, we define new equivalent stress for two fatigue experiment types: tension-compression and tension-tension. With the new equivalent stress, the statistical fit improves for both distributions, with a slight preference for the Birnbaum--Saunders distribution. In addition, we analyze a dataset of rotating-bending fatigue experiments applied to 101 round bar specimens of 75S-T6 aluminum. Finally, we consider a well-known dataset of bending tests applied to 125 specimens of carbon laminate. Overall, the Birnbaum--Saunders distribution provides better fit results under fatigue-limit models with various experimental setups.