Abstract
Keywords CAD, FEA, MAT, Interface, Medial
Abstract The design and manufacture of automobile structural parts is now predominantly a digital process: Designers create the parts using computer aided design (CAD) software, and manufacturers use numerical control (NC) software to import the CAD design and instruct the machine tools to manufacture the parts. As an important element of the virtual design cycle, these parts must also be tested for their structural integrity using finite element analysis (FEA) software. Whereas, the interface between CAD and NC products is excellent, the interface between CAD and FEA is imperfect. The process of preparing CAD models for FEA consumes a great deal of the stress analysts time. Existing automatic CAD to FEA translators tend to treat all part features as solid; this leads to longer computation times and less accurate results for features that can be better characterized as thin or long. In addition, many features of CAD parts (e.g. fillets and chamfers) are important for their size and shape in the manufactured product, but have relatively little impact on the strength of the part and needlessly complicate the stress analysisthese features are usually removed by the analyst prior to FEA; they may need to be evaluated with additional analyses to test if it is safe to remove them. The Automatic CAD-FEA Interface Project (ACFI), at Vanderbilt University, is developing algorithms to make the translation from CAD to FEA seamless and automatic; these algorithms are based on mathematical theory and the principles of theoretical mechanics.
This paper presents the latest ACFI advances for (i) automatically evaluating and reworking three-dimensional CAD part geometries to prepare them for finite element meshing, (ii) exporting the revised geometries to a preprocessor, and (iii) identifying element type to be associated with each feature geometry. The algorithms used in this work approximate the medial axis transform (MAT) of the CAD part, a power shape that represents the three-dimensional solid part. This part can then be evaluated for its geometric properties. This approach has been shown to be a robust method for shape interrogation of three-dimensional geometries.
The methods presented in this paper will make virtual prototyping of automobile structural parts simpler and faster. The net result will be shorter design cycles and faster time to market.