Surface Gridding from Discrete Data

An advancing front surface gridding technique that operates on discretely deened surfaces is presented. Diierent aspects that are required to make the procedure reliable for complex geometries are discussed. Notable among these are a) the recovery of surface features and discrete surface patches from the discrete data, b) ltering based on point and side-normals to remove undersirable data close to cusps and corners, c) the proper choice of host faces for ridges, and d) fast interpolation procedures suitable for complex geometries. Post-generation surface recovery or repositioning techniques are discussed. Several examples ranging from academic to industrial demonstrate the utility of the proposed procedure for ab initio surface meshing from discrete data, such as those encountered when the surface description is already given as discrete, the improvement of existing surface triangulations, as well as remeshing applications during runs exhibiting signiicant change of domain. 1. INTRODUCTION The rst and by far the most tedious step of any mesh generation procedure is the deenition of the boundaries of the domain to be gridded. This may be accomplished in two ways: a) analytically, i.e. via functions, or b) using a tesselation or triangulation. From a practical point of view, it would seem that an analytic deenition of the surface is the method of choice, given that nowadays most engineering data originates from some CAD-CAM package. However, in many instances, the boundaries of the domains to be gridded are not deened in terms of analytical functions, such as splines, B-splines,