GeoTran-HC: Geometric transformation of highly coupled variable topology multi-body problems

Manipulating geometry to prepare a CAD design model for mesh generation is an important step in the finite element analysis (FEA) process. However, complex problems such as electronic chip packages often consist of hundreds or even thousands of sub-features with varying materials, complex geometrical shapes and changeable connectivity configurations. We define a subclass of such problems, termed highly coupled variable topology multi-body (HCVTMB) problems, where configuration and relative geometric sizes cause meshing changes in one body to propagate throughout much of the model. Today creating FEA models for HCVTMB problems typically requires manual geometrical manipulation which is labour-intensive and error-prone. In this paper we introduce an automated geometric transformation method-GeoTran-HC-within the context of a knowledge-based modeling method for CAD-FEA integration. Through five steps-geometry extraction, cell decomposition, associativity maintenance, feature recognition and associativity recovery-this transformation method automatically decomposes an HCVTMB analytical model into a meshable model. It also manages mesh quality and attribution of non-geometrical information such as material properties and boundary conditions. A plastic ball grid array chip package thermo-mechanical case study demonstrates the efficacy of GeoTran-HC within the context of our knowledge-based FEA modelling method. This and similar studies show reductions in total FEA modeling time from days/hours (using traditional methods) to hours/minutes (using the presented method), as well as enhanced modularity and reusability. © 2007 Elsevier Ltd. All rights reserved.