Advanced Computational Techniques for Boundary Element Analysis

边界元分析的先进计算技术

• 批准号: 9019852
• 负责人: James Kane
• 金额: $ 11.95万
• 依托单位: Clarkson University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1992
• 资助国家: 美国
• 起止时间: 1992-07-15 至 1995-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9019852&HistoricalAwards=false
• 关键词: Advanced; Computational; Techniques; Boundary; Element

This is a joint Clarkson-Yale University project for the investigation of novel computational techniques to exploit intrinsic characteristics of boundary element analysis (BEA), for truly large scale problems on contemporary/emerging computers, removing bottlenecks to high performance computing by examining steps currently consuming up to 95% of overall computational resources. Special regard is given to the effectiveness of these techniques in the design process where solution of sequences of evolving problems and design sensitivity analysis are required. Iterative equation solving, vector/parallel numerical integration, and iterative reanalysis and design sensitivity analysis, in a high performance computing environment, is being investigated. The knowledge gained, will facilitate the use of significantly more refined BEA simulations, predicting system response with increased fidelity, earlier in the engineering process, add to the general efficiency of BEA vis-a-vis finite element and finite difference analysis, and help make tractable the coupling of accurate three-dimensional simulation with programmed intelligence in overall analysis and optimization systems of practical utility in the mainstream of engineering and design.

这是一个克拉克森-耶鲁大学的联合项目，旨在研究新的计算技术，利用边界元素分析（BEA）的内在特征，在当代/新兴计算机上解决真正的大规模问题，通过检查当前消耗高达95%的总体计算资源的步骤，消除高性能计算的瓶颈。特别考虑到这些技术的有效性，在设计过程中，解决序列的演变问题和设计敏感性分析是必需的。在高性能计算环境下，对迭代方程求解、矢量/并行数值积分、迭代再分析和设计灵敏度分析进行了研究。所获得的知识将有助于使用更精细的BEA模拟，在工程过程的早期以更高的保真度预测系统响应，增加BEA相对于有限元和有限差分分析的总体效率。并有助于在工程与设计的主流实用的整体分析与优化系统中实现精确的三维仿真与可编程智能的耦合。