Abstract:  The failure behavior of metallic materials under complex loading involves a multiscale coupling process ranging from microscopic damage accumulation to macroscopic crack propagation, with the fracture mode being strongly dependent on the stress state. Existing models face challenges in achieving a unified description of microscopic void evolution and macroscopic crack growth, and often fail to adequately account for the influence of the stress state. To address this, this paper develops a stress-state-driven coupled GTN phase-field model. This model achieves the coupling of macroscopic and microscopic damage and the unified characterization of tensile and shear fracture modes by introducing a critical void volume fraction dependent on stress triaxiality and a phase-field degradation function sensitive to the stress state. To address the challenges posed by the large number of model parameters, high nonlinearity, and the requirement for applicability across a wide range of stress triaxialities, a hybrid optimization strategy combining a genetic algorithm with an artificial neural network surrogate model is developed to enable damage parameter identification. Validation based on high-purity oxygen-free copper and 304 stainless steel under spallation, tensile, and shear loading conditions demonstrates that the model can predict the entire process from void evolution to macroscopic fracture, successfully reproduce the mixed tensile-shear failure morphology, and further reveal the regulatory mechanism of the stress state on the damage-fracture coupling process, elucidating the evolution law of mixed tensile-shear fracture. This study provides a unified numerical modeling tool for the damage analysis of metallic materials under complex stress states.
 

Phase-field method; GTN damage model; Tension-shear coupling; Parameter identification method