SBIR Phase II: Parallel FLUENT/RUG: A Three-Dimensional Finite-Volume Turbulent Flow Simulation Tool

小行星9627661 这个小企业创新研究第二阶段项目是开发一个可扩展的，并行的FLUENT兼容的代码现实的时间依赖性层流和湍流问题。 计算机硬件的快速发展，特别是超级工作站和并行超级计算机，再加上流体动力学的数值和物理建模的进步，为工作的工程师和科学家提供了新的机会，可以在以前只能进行昂贵，耗时的实验室探索的领域中使用流动模拟进行分析和设计。 第一阶段展示了一个新的代码的通用流体动力学计算的并行计算机上的显着潜力，与目前可用的商业代码，即使在温和的问题实现了数量级的加速。 第二阶段的具体研究内容包括：（1）压力求解的代数多重网格加速的实现、并行化和测试，以及对其他非对称求解技术的进一步研究;（2）引入增强的坐标系能力;（3）增强ZFLOW处理湍流传热的能力;（4）并行块非结构化技术的实现、并行化和测试，这些技术可能导致最大可处理问题大小的数量级增加。 商业CFD用户越来越多地需要涉及显著的三维复杂性和/或时间依赖性的问题解决方案。这些解决方案需要使用先进的并行超级计算机。该项目提供了一个独特的机会，可以将先进的CFD和并行计算机技术集成到高度实用的软件产品中，从而快速满足这一需求。 这项新技术将对商业CFD市场产生重大影响。

9627661 Flannery This Small Business Innovation Research Phase II project is to develop a scalable, parallel FLUENT-compatible code for realistic time-dependent laminar and turbulent problems. The rapid development of computer hardware, especially superworkstations and parallel supercomputers, coupled with advances in numerical and physical modeling of fluid dynamics have opened new opportunities for working engineers and scientists to use flow simulations for analysis and design in areas previously only amenable to costly, time-consuming experimental laboratory exploration. Phase I demonstrated the significant potential of a new code for general-purpose fluid dynamics computations on parallel computers, with order-of-magnitude speedups over presently available commercial codes achieved even on moderate problems. The specific research issues to be addressed in Phase II include: (1) implementation, parallelization, and testing of algebraic multigrid acceleration for the pressure solve, as well as further study of other non-symmetric solver technologies; (2) introduction of enhanced coordinate systems capability; (3) enhancement of ZFLOW to handle turbulent heat transfer; (4) implementation, parallelization, and testing of parallel block-unstructured techniques which potentially lead to order-of-magnitude increases in maximum treatable problem sizes. Commercial CFD users increasingly require problem solutions involving significant three-dimensional COIIlplexity and/or time-dependence. These solutions demand the use of advanced parallel supercomputers. This project provides a unique opportunity to integrate advanced CFD and parallel computer technologies into a highly practical software product that will quickly address this need. This new technology should have a significant impact on the commercial CFD market.