A Fatigue Strength Prediction Model Based on True Stress-Strain Response and Its Validation

Abstract

Rapid estimation of fatigue strength is important for the preliminary durability assessment of critical load-bearing components. The widespread use of conventional fatigue testing is nevertheless limited by its high cost and low efficiency, making it poorly suited to rapid material screening and engineering-oriented evaluation. Here, 304 stainless steel (SS 304) was adopted as a model material, and different rolling reductions were applied to produce distinct microstructural states and corresponding strength levels. On this basis, tensile and high-cycle fatigue tests were conducted to systematically examine the relationships among microstructural evolution, true stress–Vstrain behavior, and fatigue strength. As the rolling reduction increased, the material showed pronounced grain refinement accompanied by enhanced work hardening. Accordingly, the true ultimate tensile strength, true yield strength, and fatigue strength increased, whereas the post-yield hardening reserve progressively decreased. To account for these trends, a constrained semi-empirical fatiguestrength model was formulated using the true ultimate tensile strength, true yield strength, and postyield hardening reserve. The strength–Vhardening coupling imposes β=–α, leaving three independently calibrated coefficients whose values are material-family dependent. Predictive performance was assessed by leave-one-state-out cross-validation for material families containing at least four reported states. Compared with conventional strength-only correlations, the formulation incorporates descriptors of the true deformation response while retaining a compact form for engineering assessment.