4.1 KNOWLEDGE-BASED EVALUATION FOR DESIGN
Not all manufacturing problems can be anticipated, let alone encoded in a knowledge base. But there is substantial, routine manufacturing knowledge, just as mundane as the knowledge for correct spelling, that can be applied at the design stage and largely isn't.
There is an "80:20" rule-of-thumb within expert system research that purports that 80% of the advise an expert is called upon for is mundane and can easily be encoded into a knowledge-based system. The other 20% is more difficult to encode. Acting as proxy for an expert, 80% of the problems usually dealt with by the expert would be handled by the advisor, leaving the expert more time to address a more interesting 20%.
Three general manufacturing concerns are addressed within specific knowledge domains in the DFM system. These include feature-based evaluation, tooling, and operation scheduling. In the feature-based evaluation domain of PIMES, features and their parameters are abstracted from the Noodles solid geometric model. The evaluation of these parameters, along with material specifications, detects potential for non-ejectability, cooling sinks, and other manufacturing defects.
The mold dimensioning aspect is a tooling concern. Based on part cavity dimensions, cavity orientation, and number of cavities, an injection mold is dimensioned. Mold dimensioning is one step toward machine selection, an operations concern. In addition to mold dimension, tonnage specifications and other information come into play for the selection of machines available to mold a given part.