An Unstructured Automatic Quadrilateral Mesh Generation On Arbitrary Geometry

AbstractQuadrilateral mesh and triangular mesh are the most widely used mesh recently in two-dimensional mesh generations. At present, the technique for the automatic generation of triangular element mesh has been developed perfectiy. By the contrast, the technique for the automatic generation of quadrilateral element mesh in arbitraiy geometries is not so satisfactory. Therefore, most research on creating fully automatic unstructured mesh generators focus on the automatic generation of quadrilateral element mesh in arbitraiy geometries. Many researchers prefer employing Paving as automatic quadrilateral mesh generators to the other techniques for the three reasons: boundary sensitive, orientation insensitive and few irregular nodes. However; Paving suffered from a lack of low e1~ciency in intersecfion check and bad mesh transition (especially in the region with a large gradation of element sizes). In this paper we present an improved Paving algorithm based on the following aspect1.Firstly by the initial Delaunay mesh obtained on the discrete boundary, we construct a background mesh, which is then used to create Control Space and Neighborhood Space. AlInrwanl on the edge generation of quadrilateral element, by the tools of Control Space, Neighborhood Space and the idea of norm edge, we associate a step size all over the domain easily.2.The tool of Control Space and Neighborhood Space localize the intersection check into a fast local check.3.By the idea of Paving edge condition function and the choice of departure Paving loop and Paving edge in Q-Morph method, we generate a quadrilateral element every time but not a row of elements. And that, we process smooth after the element generation.Then, we can use the improved Paving algorithm to rapidly generate high-quality quadrilateral mesh for an arbitrary planar area. In addition, combined with many smooth and pre-adjustment skills, we can mesh the large graded region effectively. Numerical test shows that the algorithm is efficient, stable and robust...