Large Eddy Simulations of Separated Flows on Massively Parallel Architectures (Postdoctoral Research Associateship in Computational Science and Engineering)

大规模并行架构上分离流的大涡模拟（计算科学与工程博士后研究助理）

• 批准号: 9404934
• 负责人: Danesh Tafti
• 金额: $ 4.62万
• 依托单位: University of Illinois at Urbana-Champaign
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1994
• 资助国家: 美国
• 起止时间: 1994-06-01 至 2000-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9404934&HistoricalAwards=false
• 关键词: Large; Eddy; Simulations; Separated; Flows

9404934 Tafti This proposal plans to construct a scalable computational fluid dynamics computer program for use on massively parallel processing platforms. The parallel algorithm will be designed in a structured modular manner with emphasis on easy portability to various MPP machines. This will allow the computer program to be used to evaluate various MPP platforms in solving large-scale fluid flow computations. The modular and general code structure will allow one to attack a wide variety of fluid dynamics problems as well as accommodate supplementary equations. It will focus on understanding and predicting separated flows in the moderate to high Reynolds number range. Accurate prediction techniques based on Pade or compact difference schemes will be developed and used to evaluate subgrid-scale turbulence models in a simple yet fundamental separated flow configuration. The extraction of embedded two and three-dimensional structures represents a major part of proposed research.

该提案计划构建一个可扩展的计算流体动力学计算机程序，用于大规模并行处理平台。 并行算法将被设计在一个结构化的模块化的方式，强调易于移植到各种MPP机。 这将允许计算机程序被用来评估各种MPP平台在解决大规模流体流动计算。 模块化和通用的代码结构将允许一个攻击各种各样的流体动力学问题，以及容纳补充方程。 它将侧重于了解和预测分离流在中等到高雷诺数范围。 将开发基于Pade或紧凑差分格式的精确预测技术，并用于评估简单但基本的分离流配置中的亚网格尺度湍流模型。 嵌入式的二维和三维结构的提取代表了拟议的研究的一个主要部分。