Cellular Automaton - Finite Element Modeling of Solidification Grain Structures - Extensions for Welding Processes
Résumé
Cellular Automaton (CA) - Finite Element (FE) modeling is an efficient way to simulate the nucleation and growth of solidification grain structures in the context of casting processes. The methodology is reported to reach predictions for competition between columnar and equiaxed grain structures [1], coupling with microsegregation due to diffusion of species [2], macrosegregation due to fluid flow [3] and solid transport [4]. Recent comparisons with in-situ and real time x-ray radiography observations of solidifying alloys have also reported the current limits of the model and the need for more systematic characterizations of benchmark experiments [5].
One of the limits currently faced is the possibility to predict grain structures for large cast parts. While CAFE simulations can deal with volume sizes that are typical of the investment casting process [6], it requires extensions for other processes. The present keynote will list the efforts recently made to overpass this difficulty. They are based on parallel computations and dynamic allocation strategies in order to focus the computation resources in the region of the mushy zone.
While the above progresses were necessary, they did not permit to handle simulation of the grain structure during welding processes. This is due to the fact that the coupon contains an initial grain structure that must be known at all locations and at any time in order to compute its local melting and the epitaxial growth from partially melted grains, typical of the solidification of the weld pool. A solution is presented in the present paper that permits simulation of multiple passes in welding processes for large computational domains. This achievement allows for grain structure manufacturing in welding processes and remelting surface treatments, as show in the figure below.