Mesh Quality and Control in Boundary Layer Mesh Generation for Viscous Flows

2nd Symposium on Trends in Unstructured Mesh Generation, University of Colorado, Boulder, August 1999

MESHING
RESEARCH
CORNER

Scientific Computation Research Center, Rensselaer Polytechnic Institute, Troy, NY 12180
garimell@scorec.rpi.edu

Abstract
Simulation of viscous flows exhibits strong directionality of gradients. To capture the solution characteristics of these flows with a manageable number of elements, anisotropic meshes are needed in boundary layers and free shear layers. These simulations also require careful control over the sizes, gradation, anisotropy and quality of elements in the mesh.

The Generalized Advancing Layers Method has been developed as a mesh generation technique providing such control for the generation of boundary layer meshes. In addition, the mesh generator is designed to reliably generate boundary layer mesh for large, arbitrarily complex, non-manifold domains. The procedure starts from a surface mesh and constructs highly stretched anisotropic meshes next to surfaces expected to carry boundary layers. The rest of the domain is filled with an isotropic mesh created by a general mesh generator. The Generalized Advancing Layers Method, in conjunction with the surface mesh generator, the volume mesh generator and mesh optimization techniques, is able to create multi-million element meshes with good control of meshes.

The paper focuses on the constructs and techniques for mesh control for boundary layer mesh generation. The issues of validity assurance using concepts of multiple growth curves and topological compatibility checks are discussed. Concepts of blend elements to inherently improve mesh quality are described. Various techniques for a-priori and a-posteriori mesh quality improvement in the boundary layer mesh are also dealt with. The paper discusses issues that arise in combining a highly anisotropic boundary layer mesh with an isotropic mesh. In particular, techniques for shielding the stretched faces of the anisotropic mesh from the isotropic mesh. Finally, the discussion focuses on strategies for controlling the boundary layer mesh to account for changing flow characteristics over surfaces.

Results of mesh generation are presented to demonstrate the viability of meshing complex domains while providing good control over the mesh. Two example simulations are presented to demonstrate the effectiveness of the method in capturing viscous flow phenomena (a) Laminar flow over a flat plate (b) Turbulent flow in a sharply expanding pipe.