Parallel-GPU-accelerated adaptive mesh refinement for three-dimensional phase-field simulation of dendritic growth during solidification of binary alloy

• DOI: 10.1186/s41313-021-00033-5
• Publication Date: 2022-01-06T09:02:24Z
• Authors: Shinji Sakane; Tomohiro Takaki; Takayuki Aoki
• Keywords: Adaptive Mesh Refinement; Dendrite (mathematics); Directional solidification

Abstract In the phase-field simulation of dendrite growth during solidification an alloy, computational cost becomes extremely high when diffusion length is significantly larger than curvature radius a tip. such cases, adaptive mesh refinement (AMR) method effective for improving performance. this study, we perform three-dimensional in which AMR implemented via parallel computing using multiple graphics processing units (GPUs), provide computation GPU computation, apply dynamic load balancing to equalize per GPU. The accuracy condition confirmed through single-GPU computations columnar directional binary alloy. Next, evaluate efficiency by performing multiple-GPU three different simulations moving frame algorithm. Finally, weak scaling tests are performed confirm developed code.