Quadrilateral Meshing for Hexahedral Mesh Generation Based on Facet Normal Matching

Abstract

Because performance testing using actual products is costly, manufacturers use lower-cost Computer-Aided Design (CAD) simulations. In this paper, we focus on hexahedral meshes, which are more accurate than tetrahedral meshes, for finite element analysis. Our final objective is automatic hexahedral mesh generation with sharp features to precisely represent the corresponding features of a target shape. Our hexahedral mesh is generated using a voxel-based algorithm. In our previous works, we fitted the surface of the voxels to the target surface using Laplacian energy minimization and used normal vectors in the fitting to preserve sharp features. However, we were unable to precisely represent sharp concave features using the method. In this proposal, we improve the previously used Laplacian energy minimization by adding a term that depends on facet normal matching for multi-normal vectors, instead of using normal vector matching.