Developing and validating a fully coupled model of non-local crystal plasticity and probabilistic cellular automata for dynamic recrystallization simulation

This study presents a fully integrated model combing non-local crystal plasticity finite element method (CPFEM) and probabilistic cellular automata (CA) to capture the coupled effect of heterogeneous deformation, morphological evolution and mechanical responses during dynamic recrystallization (DRX). The developed model incorporates a non-local methodology that accounts for geometrically necessary dislocations (GND) and a probabilistic CA model that describes DRX microstructural evolution, both of which are integrated into CPFEM formulations that handle multiscale heterogeneous deformation. Based on the periodic polycrystalline grids serving as both finite elements and CA cells, the non-uniform distribution of mechanical responses at grain-level, including two types of dislocation densities is calculated through CPFEM. The microstructural evolution of DRX, synchronized with deformation, is predicted through CA model with probabilistic switching rules. The DRX induced changes in dislocation densities and crystallographic orientation are then fed back into CPFEM to determine the subsequent mechanical response and plastic deformation. The proposed model is validated against experimental data for SA508–3 steel during hot compression bonding (HCB) process. It’s demonstrated that the proposed model effectively integrates predictions of macroscale mechanical response, mesoscale dislocation density distribution, and microscale microstructural evolution during DRX. Furthermore, the model can be extended to other problems by adapting corresponding CA switching rules.