Affiliations: [a] Department of Quantum Engineering and Systems Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113, Japan | [b] Research into Artifacts, Center for Engineering, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113, Japan | [c] Tokyo Gas, Co. Ltd., 2-3 Nakase, Chiba, Chiba 260, Japan
Correspondence:
[*]
To whom all correspondence should be addressed.
Abstract: This article describes a novel automatic finite-element mesh generation method for complex geometries based on the fuzzy knowledge processing and computational geometry. This mesh generation process consists of three subprocesses: (a) definition of an analysis domain, i.e., a geometric model, (b) generation of nodes, and (c) generation of elements. Two commercial three-dimensional geometric modelers for solid and shell structures, respectively, are employed for the subprocess (a). In the subprocess (b), several local nodal patterns are first chosen from the nodal pattern database constructed in advance, and their node density distributions are superposed on one another over the analysis domain by using the fuzzy knowledge processing. Thanks to this procedure, a node density distribution in a whole analysis domain can be controlled quite well. Then, nodes are generated by the bucketing method whose processing speed is almost proportional to the total number of nodes. In the subprocess (c), triangular elements are generated over a two-dimensional domain, and tetrahedral solid elements over a three-dimensional domain using the Delaunay triangulation, whose speed is also proportional to the total number of nodes. As for two-dimensional plane and three-dimensional shell structures, triangles can be completely converted into quadrilaterals. Practical performances of the present method are demonstrated through the mesh generation of a nozzle corner of pressure vessel, and that of a portion of car body.
