Parallel Computational Fluid Dynamics: Implementations and Results

Computational Fluid Dynamics (CFD) is one of the most important applications areas for high-performance computing, setting the pace for developments in scientific computing. Anyone who wants to design a new parallel computer or develop a new software tool must understand the issues involved in CFD in order to be successful.The demands of CFD, particularly in the aerospace and automotive industries, coupled with the emergence of more powerful generations of parallel supercomputers, have led naturally to work on parallel computational fluid dynamics; and initial results from using parallel machines to study the properties of liquids and gases in motion are promising. Parallel Computational Fluid Dynamics provides the first survey of this rapidly developing field. Drawn. from such different disciplines as mechanical and aeronautical engineering, computer science, and numerical methods, contributions cover implementations of CFD codes on commercially available large-scale parallel systems, studies of parallel numerical algorithms for CFD applications, and discussions of computer science topics with direct application to parallel CFD. Researchers will find that Parallel Computational Fluid Dynamics serves a number of needs. It presents the expertise of multidisciplinary research groups that will make it possible to succeed in solving the "grand challenge" problems stimulated by the new national High Performance Computing and Communication Program. It offers aeronautical or mechanical engineers an excellent introduction to what has been accomplished in the last few years in CFD on parallel machines. And it provides researchers in the areas of hardware, software, and algorithms with a useful survey of CFD's computational requirements as well as a source of applications to test out new ideas.Horst D. Simon is Department Manager for Computer Sciences Corporation and Research Scientist at NASA Ames Research Center.