High performance parallel preconditioning and linear equation solver for CFD applications

适用于 CFD 应用的高性能并行预处理和线性方程求解器

• 批准号: 460872-2013
• 负责人: Groth, Clinton
• 金额: $ 1.82万
• 依托单位: University of Toronto
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2013
• 资助国家: 加拿大
• 起止时间: 2013-01-01 至 2014-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=542342
• 关键词: performance; parallel; preconditioning; linear; equation

Computational fluid dynamics (CFD) has proved to be an important enabling technology in many areas of science and engineering. Its relative maturity and the widespread successes of its application have been driven over the past 10--15 years by significant advances in both numerical methods for partial differential equations and the rapid increase in parallel high-performance computing (HPC) systems. Nevertheless, in order to exploit fully current and future parallel computing systems and enable the more routine solution of fluid flows for practical engineering applications, further advances in numerical methods and parallel algorithm design are generally needed. he goal of the proposed project is to improve on the existing and/or to develop a new high performance parallel preconditioner/linear equation iterative solver for use in the solution of typical linear systems produced by the temporal and spatial discretization schemes of the NIECE CFD solver and software package developed by MAYA Heat Transfer Technologies Ltd. (MAYA). The improved and/or new iterative solver should provide more rapid convergence and be more robust for the flow problems of interest, including solution of problem with meshes having high aspect ratios as frequently encountered in the prediction of high-Reynold-number boundary-layer flows. The improved and more robust parallel iterative solver will be implemented and integrated within NIECE for subsequent release.

计算流体动力学（CFD）已被证明是许多科学和工程领域的重要技术。 在过去的10- 15年里，偏微分方程数值方法的重大进展和并行高性能计算（HPC）系统的快速增长推动了它的相对成熟和应用的广泛成功。 然而，为了充分利用当前和未来的并行计算系统，并使更多的日常解决方案的流体流动的实际工程应用中，进一步的进步，数值方法和并行算法的设计通常是必要的。 该项目的目标是改进现有的和/或开发一种新的高性能并行预处理器/线性方程迭代求解器，用于求解由MAYA传热技术有限公司（MAYA）开发的NIECE CFD求解器和软件包的时间和空间离散方案产生的典型线性系统。 改进的和/或新的迭代求解器应该提供更快的收敛，并且对于感兴趣的流动问题更鲁棒，包括在预测高雷诺数边界层流时经常遇到的具有高纵横比的网格的问题的解决方案。改进的、更强大的并行迭代求解器将在NIECE中实现和集成，以供后续发布。