Physically Informed Fracture Criterion for Ultralow-Cycle-Fatigue of Structural Steel

A triaxiality-dependent fracture prediction criterion is presented for structural steel subjected to ultralow-cycle-fatigue (ULCF) loading. The model is grounded in direct observation and quantitative assessment of the depth of ductile dimples that occur on the fracture surface. An automated statistical analysis is employed to extract key geometric parameters of the dimples from scanning electron microscope (SEM) fractographs. The results indicate that dimple depth intensity is a function of the compressive excursions experienced during ULCF loading, as well as the level of triaxiality, thus enabling development of the physically informed fracture criterion (PIFC). The proposed model demonstrates superior calibrability, yielding the lowest optimized mean squared error compared to existing methods and achieving high accuracy in predicting ULCF fractures.