How Does the Modeling Strategy Influence Design Optimization and the Automatic Generation of Parametric Geometry Variations?

Abstract

The robustness and flexibility of a feature-based parametric CAD model determines the extent to which the geometry can be modified and reused in other design scenarios. The ability of a model to successfully adapt to changes depends on the type and sequence of the modeling operations selected to build the geometry, the parent–child dependencies defined during the modeling process, and the type and scope of the desired geometric change. Several formal modeling methodologies have been proposed to maximize model reusability, which have been shown to outperform unstructured approaches when designers need to manually modify the geometry. However, the effect of these parametric model strategies on the generation of valid solutions in heavily automated tasks has not yet been investigated. In this paper, we compare and analyze the performance of three well-established parametric modeling methodologies in various design optimization scenarios that involve the automatic generation of a large number of geometric variations. We discuss the results of a study with four parametric models of varying complexity and identify the limitations of each strategy in relation to the internal structure of the model. Our results show that explicit references and resilient modeling strategies are relatively robust for simple parts, but their effectiveness decreases significantly as the complexity of the model increases. In addition, we introduce the concept of intrinsic variability, which impacts the effectiveness of the methodology, and thus the quality of the parametric model, based on how the methodology is interpreted and executed. • The effect of formal parametric model strategies on the automatic generation of geometric variations is investigated. • A study with three formal methodologies and four parametric feature-based models of varying complexity is presented. • The horizontal modeling methodology is not an efficient strategy for automated environments. • The resilient methodology is most effective for simple parts and scenarios that involve manual changes. • Explicit reference modeling is effective for complex models and yields faster model regeneration times.