Masters Theses

Date of Award

12-1994

Degree Type

Thesis

Degree Name

Master of Science

Major

Computer Science

Major Professor

Dinesh Mehta

Committee Members

Whitehead, Pujol

Abstract

Computational Fluid Dynamics (CFD) are an important software algorithmic tool used in the design and analysis of objects that move in a fluid environment. However, due to the large computational times required for CFD problems, parallel CFD grid mapping optimization and simulation are useful tools. These algorithms have been developed to minimize the computational times for these CFD problems and also to provide an approximation of the times required for those problems.

The optimization algorithm, which uses Simulated Annealing to find an mapping of grids to processors which is close to optimum, results in a decrease in execution time ranging from 4-97%. Furthermore, although the problem is NP-Hard, our polynomial-time heuristic found an optimal mapping for the cases tried. The simulation algorithm shows good accuracy at approximating the run times of CFD problems, with under a 7% erгог.

The results of this study warrant the application of these techniques to other parallel CFD flow solvers.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS