A modified cellular automata （CA） model of dynamic recrystallization （DRX） and a flow stress-based nucleation parameter identification method have been developed. In the method, the modified CA model, which takes the role of deformation degree on nucleation behavior into consideration, is coupled with an adaptive response surface model （ARSM） to search for the optimum nucleation parameter. The DRX behavior of an oxygen free high conductivity （OFHC） copper with different initial grain sizes has been taken as an example to validate the model. Good agreement is found between the simulated and the experimental results, which demonstrates that the new method can effectively improve the simulation accuracy.
Based on hot compression tests by a Gleeble-1500D thermo-mechanical simulator, the flow stress model and microstructure evolution model for SA508-3 steel were established through the classical theories on work hardening and softening. The developed models were integrated into 3D thermal-mechanical coupled rigid plastic finite element software DEFORM3D. The inhomogeneous hot deformation （IHD） experiments of SA508 3 steel were designed and carried out. Meanwhile, numerical simulation was implemented to investigate the effect of temperature, strain and strain rate on microstructure during IHD process through measuring grain sizes at given positions. The simulated grain sizes were basically in agreement with the experimental ones. The results of experiment and simulation demonstrated that temperature is the main factor for the initiation of dynamic recrystallization （DRX）, and higher temperature means lower critical strain so that DRX can be facilitated to obtain uniform fine microstructure.
