Potts-model grain growth simulations: Parallel algorithms and applications

摘要: Microstructural morphology and grain boundary properties often control the service of engineered materials. This report uses Potts-model to simulate development microstructures in realistic Three areas microstructural simulations were studied. They include massively parallel algorithms for grow simulations, modeling mass transport via diffusion these simulated microstructures, a gradient-dependent Hamiltonian columnar growth. Potts growth models supercomputers developed conventional both two three dimensions. Simulations using codes showed self similar no finite size effects previously unapproachable large scale problems. In addition, new enhancements Metropolis algorithm used accelerate calculations. These techniques enable sequential run faster use essentially an infinite number orientation values avoid non-physical coalescence events. Mass phenomena polycrystalline materials studied dimensions numerical on generated Potts-model. The results excellent quantitative agreement with one dimensional problems, however also suggest that transient multi-dimension cannot be parameterized as product coefficient width. Instead, are required. Gradient-dependent mechanisms included by adding extra term Hamiltonian. Under normal growth, primary driving is curvature boundary, which standard