A Comparative Study of the Various Mean Stress-Based Fatigue Life Prediction Models for 880-Grade Indian Rail Steel in Presence of Ratcheting Phenomenon

In this work, the efficacy of various existing mean stress-based prediction models to estimate fatigue life in the presence of ratcheting phenomena is compared with the experimentally observed life for the 880-grade rail steel. A set of 14 existing stress-based life prediction models have been considered in this investigation. This investigation reveals that all linear models provide conservative predictions, and the Half-slope model has the better efficacy with a stress deviation of 0.0364. On the contrary, the fatigue life estimated by exponential models (i.e. ASME, Bagci, and Gerber) are on the non-conservative side, and the prediction by the ASME model has a stress deviation of 0.0430. Among all the models, the Dietman model predicts ratcheting life with the 2nd best estimate having a stress deviation of 0.0289. However, the prediction efficacy of the Walker model, which has merely a stress deviation of 0.0090, is the best compared to all the other methods considered here. This is because the prediction by the Walker model depends on the adjustable material parameter, gamma, which is derived by multiple linear regression of the data collected from ratcheting experiments under asymmetric cyclic stress. Nevertheless, this dependency may make this efficient model an unfavorable option for many designers, as obtaining the c requires several complex and tedious experimentations.