Face clustering of a large-scale CAD model for surface mesh generation

Computer Aided Design, Elsevier, Vol 33, Num 3, pp.251-261, March 2001

MESHING
RESEARCH
CORNER

The 8th International Meshing Roundtable Special Issue: Advances in Mesh Generation, Computer Aided Design Volume 33, No. 3, March 2001

Keisuke Inoue, Takayuki Itoh, Atsushi Yamada
Tokyo Research Laboratory, IBM Japan Ltd., 1623-14, Shimotsuruma, Yamato-shi, Kanagawa 242-8502, Japan
Tomotake Furuhata
Yamato Software Development Laboratory, IBM Japan Ltd., 1623-14, Shimotsuruma, Yomato-shi, Kanagawa 242-8502, Japan
Kenji Shimada
Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213-3890, USA

Corresponding author. Tel.: +81-46-215-4888; fax: +81-46-273-7428.
E-mail address: inouek@jp.ibm.com (K. Inoue).

Abstract
A detailed CAD model needs manual clean-up, or simplifying operations, before a finite element mesh can be automatically generated because such a model consists of hundreds or thousands of faces many of which may be smaller than a desired mesh element size. We propose an automated face clustering method used as a pre-process of surface mesh generation. By decomposing a model into face clusters so that each region can be projected onto a simple parametric surface such as a plane, we obtain a final mesh as an aggregate of sub-meshes for respective clusters without time-consuming manual preparation work. The projection onto a surface realises re-parameterisation as well as suppression of small details. The main contribution of this work is the integration of (1) a greedy algorithm for combining faces into clusters, and (2) geometric indices that reflect various aspects of a preferable shape for a cluster. The validity of the approach is demonstrated with results of clustering and mesh generation for a realistic-scale CAD model.