This paper presents a method for eliminating unnecessary parts, features, and dimensions from feature based models in computer aided tolerance analysis systems.
A typical tolerance analysis involves a dozen or so parts, or subassemblies. Each part may have dozens of features, and hundreds of dimensions. Many of these subassemblies, parts, features, and dimensions do not effect the tolerance analysis, and therefore are not needed. Computational effort can be reduced by eliminating the unnecessary items from the model.
Currently, tolerance analysis models are implicitly simplified by the user. The user examines the problem, and determines which parts, features, and dimensions can be ignored. The user then constructs his model, leaving out the unneeded items. This is true whether the analysis is accomplished with a computer aided tool, or with paper and pencil. Simplification of the model is essential. Practical tolerance analysis problems become overwhelmingly complex if every detail is included.
The method discussed in this paper is intended for use in computer aided tolerance analysis systems that use feature based, and dimension driven, solid modeling. It uses a combined tree and graph data structure. The tree structure represents the hierarchy of assemblies, parts, and features in the model. The graphs represent the dependence between features in a part, or parts in an assembly. An algorithm has been developed that searches this tree/graph model, locating the parts and features needed to accomplish the tolerance analysis.