Abstract：In service conditions, metallic materials may undergo various types of failure under mechanical loadings, including yielding, fracture, buckling, wear, fatigue, and so on. Among these, fracture is one of the most significant and destructive form of failure. In pure metals and alloys, ductile fracture, characterised by dimples on fracture surface, is commonly observed. From the microscopic point of view, the ductile fracture of metals and alloys is closely associated with the nucleation, propagation, and coalescence of voids. This microscopic failure process is influenced by numerous factors such as stress state, void size, void volume fraction, void shape, temperature, etc. Micromechanics based models developed for ductile damage considering the void evolution, such as the Gurson model and its extensions, usually assume spherical voids. The development of models considering realistic void shape and their evolution is of great challenge. Furthermore, conducting mechanical analyses of ductile failure at the specimen and component scales requires addressing cross-scale problems. To address these issues, the present study firstly constructed representative volume element models incorporating an isolated void of different initial shapes. Finite element simulations were carried out based on the representative volume elements by adopting a J2 plasticity model for the matrix. A systematic analysis was realised to understand the influence of the initial void shape on the stress-strain response and ductile damage. Triaxial tensile and shear loading conditions were considered. Using the numerical data generated by the simulations, a neural network based surrogate model was trained to approximate the stress-strain responses and damage evolution. The surrogate model was shown to be capable of predicting the influence of the initial void shape on ductile damage. Subsequently, a user-defined material subroutine was developed, and incorporated into a commercial finite element code. The impact of the initial void shape on the ductile failure process of notched specimens was simulated. It was found that reduced aspect ratio of the void decreased the damage rate leading to a delayed softening at the specimen level. The present work shows the potential of surrogate model for predicting ductile damage involving complex microstructural features.
2025, Vol. 46 
