Parallel Algorithms for General Geometry Navier-Stokes Calculations

一般几何纳维-斯托克斯计算的并行算法

• 批准号: 9405403
• 负责人: Paul Fischer
• 金额: $ 19.31万
• 依托单位: Brown University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1994
• 资助国家: 美国
• 起止时间: 1994-07-15 至 1998-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9405403&HistoricalAwards=false
• 关键词: Parallel; Algorithms; General; Geometry; Navier

Fisher 9405403 Advanced computational fluid dynamics (CFD) software for distributed memory multicomputers will be developed to provide the resolution capabilities (106-107 gridpoints) and high-order accuracy required for large eddy simulation (LES) DG1 DG2 of turbulent flows in complex domain. The choice of distributed memory parallel architectures is motivated by high performance and economy which derives from their intrinsic ability to exploit data locality. This work will be based on the spectral element method for the incompressible Navier-Stokes equations, with emphasis on improving elliptic solver performance and increasing geometric flexibility. Enhanced geometric and resolution flexibility will be achieved through implementation of three-dimensional nonconforming spectral element discretizations. The software will have a variety of applications. The scope and complexity of the proposed activities will foster further research in large scale parallel computation. {PAGE \# "'Page: '#' '"|Page: 1 } DG1 {PAGE \# "'Page: '#' '"|Page: 1 } DG2

将开发用于分布式存储多计算机的高级计算流体动力学（CFD）软件，以提供复杂域湍流大涡模拟（LES） DG1 DG2所需的分辨率（106-107格点）和高阶精度。选择分布式内存并行架构的动机是高性能和经济性，这源于它们内在的利用数据局部性的能力。这项工作将基于不可压缩Navier-Stokes方程的谱元方法，重点是提高椭圆求解器的性能和增加几何灵活性。通过实施三维非一致性谱元离散化，增强了几何和分辨率的灵活性。该软件将有各种各样的应用。拟议活动的范围和复杂性将促进大规模并行计算的进一步研究。{页面\ #”页面 : '#' '"| 页:1}DG1{页面\ #”页面 : '#' '"| 页:1}DG2