No abstract

A 2 n tree (capable of supporting binary, Quadtree and Octree) based viscous Cartesian grid generation method has been successfully developed for complex geometries. Compared with an Octree data structure, the 2 n tree data structure supports anisotropic grid adaptations in any of the coordinate directions in an arbitrary manner. This capability enables high resolution of flow features such as shocks, shear layers and wakes with high aspect ratio cells. In order to properly resolve viscous boundary layers, a viscous layer grid is "inserted" between the Cartesian grid and the body surface through a projection technique. The thickness of the viscous layer grid can be determined based on the expected boundary layer thickness. Algorithms to automatically detect and resolve narrow gaps, and geometrically critical features have been developed. The method completely avoids cell-cutting, and produces overall good quality computational grids. Several demonstration cases are included to showcase the capability of the method.