Multiphase macroscale models for macrosegregation and columnar to equiaxed transition during alloy solidification

In the field of metal casting, solute composition inhomogeneities at the macroscale are called macrosegregation, and the transition from the elongated grains in the outer portions of a casting to the more rounded grains in the center is termed Columnar to Equiaxed Transition (CET). Simultaneous prediction of macrosegregation and CET is still an important challenge in the field. One of the open questions is the role of melt convection on the CET and the effect of the CET on macrosegregation. A three-phase macroscale model for macrosegregation and CET was developed. The model accounts for numerous phenomena such as columnar dendrite tip undercooling, undercooling behind the columnar tips, and nucleation of equiaxed grains. This three-phase model was used to develop a less complex model that consists of two phases only and disregards undercooling behind the columnar tips and nucleation of equiaxed grains. An in-house parallel computing code on the OpenFOAM platform was developed to solve the equations of these models. The models were used to perform columnar solidification simulations of a numerical benchmark problem. It was found that the predictions of these models are nearly identical. It was also found that the dendrite tip selection parameter, which appears in the constitutive relation for the dendrite tip velocity, plays a key role in these models. With a realistic value for this parameter these models account for columnar dendrite tip undercooling, but as its value is increased in the simulations, predictions of these models converge to predictions of a model that neglects undercooling. Next, the three-phase model was used to perform CET simulations in the numerical solidification benchmark problem in the presence of melt convection. It was found that accounting for stationary equiaxed grains does not change the overall macrosegregation pattern nor the form of channel segregates. Finally, for the first time in the field of solidification, we developed accurate constitutive relations for macroscale solidification models that are based on a formal mesoscale analysis on the scale of a representative elementary volume that is used in developing volume-averaged macroscale models. This upscaling enabled us to present relations that incorporate changes in the shape of grains and solute diffusion conditions around them during growth. The models and constitutive relations we developed can now be used to predict critical phenomena such as macrosegregation, channel segregates, and CET in castings.