An efficient grain remapping algorithm for phase-field modeling of dynamic recrystallization

Abstract

A phase-field (PF) model of dynamic recrystallization was built to predict the microstructure evolution and mechanical responses of maraging stainless steel during hot deformation. Given that the traditional techniques for solving the phase-field model are usually computationally demanding, especially when a large number of grains are present for DRX simulating, an efficient grain remapping algorithm was incorporated into the PF model. Taking advantage of the similarity between order parameters and images, some well-established image processing methods were employed in the algorithm. The algorithm adopted a reduced set of order parameters instead of a one-to-one assignment of order parameters for each grain, but still tracked individual grains for the involvement of unique grain properties and remapped coming-to-contact ones to different order parameters. First, in terms of the PF model, a set of isothermal compression tests were performed to derive necessary model parameters and provide experimental data regarding microstructural and mechanical characteristics for verification purposes. The simulation results, including grain structure and flow stress responses, showed good agreement with experimental observations under various deformation conditions, which indicated the reliability of the established PF model. The influences of deformation conditions on the microstructure evolution and mechanical responses were investigated. Next, a comparative analysis of fidelity and efficiency between the current model and the unmodified model was carried out. Results suggested that the algorithm yielded faithful results of the original model, and significantly reduced the computational memory and time.