Quantitative prediction of grain size and recrystallized volume fraction is still a real challenge for many alloys, and even for simple materials when subjected to complex thermal/mechanical histories, as in multi-pass (industrial) processing. A first step is therefore taken in the direction of multiscale modelling of recrystallization, by considering digital polycrystalline microstructures. These synthetic mesoscopic microstructures are meshed adaptively and anisotropically, with refinement close to the grain boundaries. Crystal plasticity finite element (CPFEM) simulations are combined with a level set framework to model primary recristallization, following plastic deformation. In the level set method, the kinetic equation describing interface motion uses the calculated stored energy field provided by CPFEM calculations, and works on the same mesh. Discontinuous dynamic recrystallization can be modelled within the same approach, effectively coupling plastic deformation with nucleation and growth processes. Parallel to the finite element approach, a mean field model is developed in the general context of multi-pass processing. The model considers categories of grains based on two state variables : grain size and total dislocation density. As opposed to the finite element approach, there is no crystallographic or topological information. It is computationally much cheaper and therefore suitable for direct coupling at the scale of forming processes, for industrial applications. The parameters of the model can be identified from inverse analysis, using experimental stress-strain curves, recrystallized volume fractions, and grain sizes. Mean field and finite element models are compared, and it is shown that the detailed information provided by finite element simulations can be used to calibrate or optimize the mean field method.
Internal Wave Generation with Level Set Parallel Finite Element Approach
