Composite probability distribution for fatigue life prediction of API X65 steel via Vickers hardness

Abstract

Accurate prediction of the fatigue life of API X65 steel is crucial in various applications. However, the traditional bootstrap method has inherent limitations, such as a tendency to deviate from the true distribution with insufficient sample sizes, difficulty in identifying extreme statistics, and an inability to generate distributions closer to the original sample. These deficiencies lead to overly conservative S-N curve designs and pose challenges in data collection, particularly for small samples. To address these issues, we propose an improved bootstrap method using a composite probability distribution. This method enhances the sampling range and improves prediction accuracy for parameter uncertainty ranges by considering both small samples and extended virtual samples’ probability distribution. Comparative analysis through Monte Carlo simulation demonstrates the superior parameter estimation of our method for small samples. Our case analysis further explores the relationships between Vickers hardness, tensile strength, surface roughness factor, and intercept constant. The findings led to a novel method for estimating the S-N curve confidence interval of API X65 steel from Vickers hardness. Analysis of fatigue life test data for API X65 steel yielded favorable results, confirming the effectiveness and feasibility of our improved method.