Unstructured Adaptive Elliptic Smoothing

Steve L. Karman Jr. and Mandar Sahasrabudhe
The University of Tennessee at Chattanooga, Chattanooga, Tennessee, 37403

Forcing functions for controlling grid spacing have been investigated for an unstructured elliptic smoothing scheme. The forcing functions can provide control of grid spacing for feature-based solution adaptation and normal grid spacing for viscous clustering. An alternate approach, using Riemannian metric tensors to define a spacing field, is also described. This approach solves the same Winslow equations, but without forcing functions. Instead, the Riemannian metrics are used to alter the spacing of the computational mesh, which directly affects the spacing of the physical mesh. Two- and three-dimensional results are included to illustrate the use of these two different techniques.