Numerical simulation of dendritic growth during solidification process using multiphase-field model aided with machine learning method

Abstract

Multiphase-field (MPF) method has been widely used for predicting the microstructure formation of technical alloys. Nevertheless, the main reason limiting development of MPF is the quasi-equilibrium constraint which leads to the low efficiency. To solve this challenge, a MPF model aided with machine learning method is proposed. The machine learning method is used to solve the quasi-equilibrium constraint and then invoked in the MPF. Through the validation of the machine learning model, the MPF model and model capabilities, the proposed method are found to be feasible. Besides, time spent of using machine learning method is 1/210 of the traditional Newton iterative method. The accuracy of the model is verified from the LGK model comparison, grid independence and concentration verification. The application of multi-dendritic growth simulation to larger computational domain further verify the accuracy, stability and capability of this model. The model proposed in this paper will further promote the development of the MPF model in the direction of more components and phases, ternary eutectic simulation and 3D simulation.