Mesoscale Modeling of Dynamic Recrystallization: Multilevel Cellular Automaton Simulation Framework

The main attraction of cellular automaton (CA) method used in computational material science lies on not only the simulation of recrystallization without the complicated differential equations calculation, but also the visualization of nucleation and grain growth during discontinuous recrystallization. In this work, by incorporating the idea of multilevel cellular space into the classical CA simulation framework and formulating cellular state transformation rules and data transfer rules between different levels of cellular space, the multilevel cellular automaton (MCA) model for dynamic recrystallization (DRX) is constructed for the first time. The developed MCA model includes a multilevel recrystallized nucleation (MRN) module and a full-field multilevel grain topological deformation (FMGTD) module. The thermal compression experiments of 316LN stainless steel are carried out, and the developed MCA model is applied to the numerical simulation of DRX for 316LN steel. The accuracy and reliability of this model are verified by comparing simulation results with experimental results. The influences of simulation parameters such as the number of levels N in the FMGTD module and the discrete strain increment on simulation results are discussed. The discrete cellular space area (i.e., grain topology mapping accuracy) in the MCA model increases with N but decreases with the discrete strain increment. The results show that the developed MCA model can not only describe the grain topological deformation in the DRX process more accurately but also more compatible with the physical mechanism of recrystallized nucleation. The calculation accuracy of the MCA model is higher than the existing CA model. Besides, the MCA model can be closer to the real deformation process while ensuring the high grain topology mapping accuracy and solve the problem of the loss of grain boundary area in the existing CA model.