Adaptive Local Grid Refinement in Computational Fluid Mechanics.
Abstract
Several promising new techniques for efficient and accurate numerical solution of large-scale fluid flow problems have been developed. These methods include self-adaptive mesh modification techniques for applications requiring front-tracking and local grid refinement as well as new preconditioning ideas for efficient implementation. Properties of systems of hyperbolic conservation laws have been obtained which will aid in development of accurate front-tracking algorithms. Adaptive grid refinement techniques developed include moving grid methods, local, fixed refinement involving linked-list data structures, and certain local patch refinement ideas which have great potential for ease of implementation in existing large scale codes. Domain decomposition concepts for obtaining efficient preconditioners for iterative techniques have proved quite useful both for local patch refinement and for the solution of problems with rapidly varying coefficients. The fast adaptive algorithms being developed have high potential for both parallelization and vectorization. The algorithms have been designed to take advantage of the emerging parallel and vector-oriented computer architectures. Finally, the incorporation of these fast algorithms in accurate finite element, collocation, and finite difference methods is underway.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1987
- Accession Number
- ADA192215
Entities
People
- Eli L. Isaacson
- John H. George
- M. J. Djomehri
- Myron B. Allen
- Richard E. Ewing
Organizations
- University of Wyoming