Abstracts for the Conference on
Grid Adaptivity in Computational PDEs,
Edinburgh, July 96
Next
Previous
Index

Tetrahedral and hexahedral mesh adaptation for rotorcraft CFD problems

R Biswas¹ & R C Strawn

¹RIACS, NASA Ames Research Center,
Moffett Field, CA 94035, USA
(415) 604-4411, { tt biswas@riacs.edu


Abstract

Dynamic mesh adaptation is a powerful tool for computing unsteady flows that require local grid modifications in order to efficiently resolve solution features. For such flows, the coarsening and refinement steps must be executed frequently, so their performance must be comparable to that of the flow solver.

We have developed a dynamic tetrahedral mesh adaptation procedure that uses a data structure based on edges of the mesh. This means that each tetrahedral element is defined by its six edges rather than by its four vertices. This makes the adaptation procedure capable of performing anisotropic refinement and coarsening.

A major drawback of the tetrahedral scheme is that repeated anisotropic subdivision can significantly deteriorate the quality of the mesh. Previous work has demonstrated that isotropic subdivision is required if mesh quality is to be controlled effectively for arbitrary refinement levels. This is a serious limitation when directional flow features are present, leading to an inefficient distribution of grid points. In addition, truly anisotropic subdivision is almost impossible to realize for real problems on tetrahedral meshes.

A remedy for this drawback is to use hexahedral elements which can be subdivided anisotropically without mesh quality problems. Hexahedral meshes also yield more accurate flowfield solutions than tetrahedral meshes for the same number of edges. However, hexahedral adaptation schemes generate ``hanging'' vertices when a hexahedron cannot be split into smaller hexahedra without propagating the mesh refinement into regions where it is not desired. We solve this problem by using pyramids and prisms as buffers between refined and unrefined elements. These buffer elements are never subdivided. If an edge of a buffer is marked for subdivision, it coarsens back to its parent hexahedron and further refinement is performed directly on that hexahedral element.

The talk will describe the tetrahedral mesh adaptation procedure and the modifications that were made for the hexahedral scheme. Sample aerodynamic and acoustic results will be presented for a non-lifting rectangular helicopter rotor blade in hover. These results will be compared to experimental data and solutions from structured-grid calculations. It will be shown that the hexahedral scheme yields superior results with fewer mesh points than its tetrahedral counterpart.


ABSTRACTS
NextPreviousIndex

Last modified Fri Jun 21 19:19:10 GB-Eire 1996 (DBD)