Product family design optimization is a cost-efficient concept for achieving the best tradeoff between commonalization and diversification of products. When design functions are computationally intensive and thus viewed as black-boxes, the product family design becomes more challenging. In this study a two-stage platform configuration and product family design optimization method with generalized commonality is proposed for scale-based families involving black-box functions. The platform configuration is unknown and multiple sub-platforms are allowed. In this study, the main parameters used towards the family design include a non-conventional sensitivity analysis, the detachability property of each variable, and the variation of individual optimal values for each design variable. Metamodeling techniques are employed to provide both the non-conventional sensitivity and correlation intensities information, which leads to significant savings in the number of function calls. Efficiency of this method is tested through designing a scalable family of universal electric motors. Compared to a number of previously developed methods, the proposed method yields a design solution with acceptable performance loss after commonalization, and better value for the aggregated preference objective function while satisfying all the performance constraints.

This content is only available via PDF.
You do not currently have access to this content.