Improvement Of Dynamic Recrystallization Model For Single-phase Materials Using Cellular Automata Method

Dynamic recrystallization is one of the most important physical and metallurgical processes during the metal thermoplastic processing. Predicting and controlling microstructural evolvement by computer simulation have important significance for determining the best technology parameters and optimizing the microstructure and performance of products.Cellular automata (shortly called CA) model, which is of simple calculating and short running time and can directly investigate the local interaction and its influence on the complex behavior of system, has been widely used in the microstructure evolvement simulation of metal during dynamic recrystallization in recent years. However, because the influence factors and mechanism of dynamic recrystallization are relatively complex and the existing experimental technologies can not completely discover its real behaviors, there still exists great difference between the published simulation results and their corresponding actual ones. Overall, there are two main problems to influence the difference in single-phase metal.①though thermal simulation experiment is the basic method to test the simulation accuracy of dynamic recrystallization, the accuracy of the macro flow stress curve simulated from it remains to be further improved;②several problems need to be solved to improve the simulation accuracy of the models themselves, such as the rationality of constant nucleation rate, the precision of the analyzed dislocation density increment for dynamic recovery stage and preferential nucleation at triple junctions. Considering the first problem belongs to the improvement of experimental techniques, this thesis focuses in solving the second problem.Firstly, based on the theory of dynamic recrystallization simulation as well as the principles of closed loop control and golden section search, a method using golden section search and closed loop control for the difference between the values of simulated and experimental flow stress was proposed to search and identify the transient nucleation rate of dynamic recrystallization. A corresponding dynamic recrystallization CA model was established so that the dynamic recrystallization of HPS485wf steel was simulated. The results show that this method is reasonable and effective; compared to the model with constant nucleation rate, the new model can significantly improve the accuracy and stability of the simulated flow stress so that the maximum value (δmax) and average one (δ) of the flow stress difference decrease by93.269%and87.50%, respectively; at the same time, the new model can enhance the accuracy of the simulated area fraction of dynamic recrystallization as well.Secondly, to improve the accuracy of the simulated flow stress in dynamic recovery, the method to search and identify the transient nucleation rate was again used for identifying the optimal dislocation density increment to replace the one gained by phenomenological model. Then, the dynamic recrystallization of HPS485wf steel was simulated by the dynamic recrystallization CA model based on transient nucleation rate and the improved dislocation density model for dynamic recovery. The results show that the improved dislocation density model for dynamic recovery can not only further improve the accuracy of the simulated flow stress for the whole dynamic recrystallization process, but also ensure the intended simulation accuracy be controlled. Compared to the model with transient nucleation rate, the maximum value (δmax) and average one (δ) of the flow stress difference decrease by26.458%and8.118%, respectively.Finally, based on the dynamic recrystallization CA model including the transient nucleation rate and improved dislocation density model, a method using cellular automata was proposed to identify the triple junctions and nucleate there preferentially. And the conventional dynamic recrystallization nucleation model was improved. The corresponding dynamic recrystallization simulation results of HPS485wf steel shows that R-grains can nucleate preferentially at triple junctions, while the accuracy of the simulated flow stress for the whole dynamic recrystallization process can be controlled properly.