On the Use of Cox Regression for Statistical Analysis of Fatigue Life Results

The presence of scatter is inherent to fatigue life studies of materials and structures; a wide variety of statistical tools exists for analysis. Among them, Cox (proportional hazards) regression is widely encountered in survival studies of medical sciences, but has not found frequent use in fatigue of materials. Of interest is the ability to treat scatter in both applied load and fatigue life as well as censored measurements of the latter. In the present work, this statistical approach is adapted for fatigue life studies. For ease of its application, it is numerically implemented as a Python module and provided for permissive free use. The module allows us to carry out Cox regression on raw fatigue life results. Survival estimates, which provide a relationship between the probability of failure and fatigue duration, can be additionally plotted with the module. Finally, the module allows us to plot probabilistic Wohler (S-N) curves, which leads to a more accurate representation of scatter in fatigue life results.