A novel 3D cellular automata-phase field model for computationally efficient dendrite evolution during bulk solidification

Abstract

Large-domain 3D microstructure prediction that balances both efficiency and accuracy remains a challenging task of materials science and engineering communities. This paper presents a novel 3D Cellular Automata-Phase Field (CA-PF) model that can accurately predict the dendrite formation in a large domain, which combines a 3D CA model with a 1D PF component. In this integrated model, the PF component reformulated in a spherical coordinate is employed to accurately calculate the local growth kinetics including the growth velocity and solute partition at the solidification front while the 3D CA component uses the growth kinetics as inputs to update the dendritic morphology variation and composition redistribution throughout the entire domain. Taking advantage of the high efficiency of the CA model and the high fidelity of the PF model, the 3D CA-PF model saves the computational cost more than five orders of magnitude compared to the 3D PF models without losing much accuracy. By coupling the thermodynamic and kinetic calculations into the PF component, the CA-PF model is capable of handling the microstructure evolution of any complex multi-component alloys. A 2D and a 3D phase-field benchmark problems were first adopted to validate the single dendrites predicted by the 3D CA-PF model. Then, the 3D CA-PF model is applied to predicting the dendrite growth during large-scale solidification processes of directional solidification of Al-30wt.%Cu and laser welding of Al-Cu-Mg and Al-Si-Mg alloys.