Residual fatigue life prediction based on a novel improved Manson-Halford model considering loading interaction effect

Abstract

The classical nonlinear fatigue cumulative damage model Manson-Halford is widely used in fatigue life analysis and prediction, but this model lacks consideration for the effects of load interaction. At present, some studies have proposed new correction models to overcome the shortcomings of classical models. However, the problem with these models is that the parameters are difficult to determine, resulting in a cumbersome application process or the correction process is based only on a single parameter. Insufficient correction leads to large fluctuations in life prediction errors. To overcome the above problems, this article proposes a new correction method and establishes a new improved model. Unlike existing models, the newly improved model is based solely on S-N curve parameters and does not require additional material parameters. It fully considers the relationship between adjacent loads and the difference in fatigue life under different stress states, and dynamically modifies the classical model based on larger influence weights. By conducting multi-level fatigue tests on 300M steel, a new improved model was used to accurately predict its remaining fatigue life. Compared with the classical model, the prediction error was reduced by 11.75 %. Utilized fatigue test data from various other materials to predict the fatigue life of the improved model under different levels of stress loading. The results indicate that in the vast majority of cases, the improved model has the highest prediction accuracy among all compared models, with a minimum relative prediction error of only 3.79 % and small fluctuations in prediction error. Compared with the classical model, the maximum reduction in relative prediction error after model improvement reached 28.73 %, indicating the effectiveness and accuracy of the new improved model.