Analysis and Control of Grid Quality in Computational Simulation

The objective of this project was to determine and formulate true measures of grid quality that can be used to evaluate numerically generated grids. Empirical studies were conducted on the relationships between grid metric quantities and actual numerical solution error for physical field problems for which experimental data was available. Correlations of the actual solution error were made for numerical solutions of partial differential equations for which analytical solutions were available. Emphasis was placed on equations that model some aspect of some physical problems. In these studies, detailed localized comparisons were made throughout the field. This is particularly important since certain truncation error relations may be important only in regions of high gradient. Correlations of actual solution error were made for physical field problems. The EAGLE grid and flow codes were used as test beds in this task. General graphical tools for grid analysis and procedures for automatic adjustment of the grid to improve the quality were developed. These tools and procedures were incorporated in the EAGLE code.