There is a lack of studies to assess the economic impact of poor modeling practices in industry. Previous works have focused on interoperability issues [BM02] or presenting the problem of CAD model reusability [JB07]. The CAD-Q project will be the first initiative to scope and size CAD quality-related problems in the industry linked to poor modeling practices.
The starting hypotheses of the project are:
HS1
The cost of low-quality CAD models is a relevant (even dramatic) problem in industry.
HS2
The formal modeling methodologies developed in CAD-Q for solid and surface models result in significantly higher quality CAD models than available methodologies.
The hypotheses are novel as they depart from the well-known but often ignored fact that low-quality CAD models cost time and cause failures with high economic impact. There is little literature on this problem because, as stated above, companies are generally reluctant to openly discuss the prevalence of low-quality models in their organizations, as it would require admitting inefficient business practices. Considering the previous experience of the research group, the problem is approachable based on the next hypotheses:
H1
The effectiveness of a modeling methodology depends on the complexity of the model.
H2
The parent/child dependencies between features are not system-dependent. A CAD model generated by following exactly the same steps in various CAD systems should have the same dependency relationships associated with it.
H3
The six dimensions of parametric CAD quality [*CCO15] can be clustered into three hierarchical levels of quality aimed at ensuring that CAD models are usable, reusable, and semantically rich.
H4
The three levels of CAD quality are useful to define strategies for quality measurement and control, as objective and quantifiable metrics can be defined to measure efficiently and reliably.
H5
The background of the trainees makes a difference since, for instance, a low level of spatial vision negatively affects the ability to generate quality CAD models.