Abstract Scope |
Computer simulation of dendrite growth with melt flow remains a challenging issue in alloy solidification. In particular, simulations using the phase-field method, which is the most accurate dendrite prediction model, are the most challenging due to their large computational cost. We have been trying to resolve this problem by parallel GPU (graphics processing units) computing using a GPU supercomputer. In this talk, we introduce our previous and state‐of‐the‐art studies regarding phase-field simulations of dendrite growth with melt flow: Dendrite growth in forced [J. Cryst. Growth, 474 (2017) 154-159] and natural [J. Cryst. Growth, 474 (2017) 146-153] convections, permeability prediction [Acta Mater., 164 (2019) 237-249], dendrite growth with motion [Comput. Mater. Sci., 147 (2018) 124–131], introduction of adaptive mesh refinement, and so on. |