Efficient Optimization of Product Lines with Design Prohibitions

Customers are no longer satisfied with a one-size-fits-all approach to product design. To be competitive in this new market landscape, companies must better differentiate their offerings from the rest of the field. Product line optimization provides a simple method to design for this challenge. While a great deal of research has already been performed to consider many of the interactions between the engineering and marketing disciplines during product line optimization, the consideration of design prohibitions is one avenue that has not received enough attention. Including design dependencies in a product line optimization is expected to have significant computational impact as more infeasible designs are evaluated. Further, large-scale applications of product line optimization, such as those faced in the automotive industry, are likely to render an optimization intractable without technological breakthroughs. This research investigates two important research questions related to these challenges. The first question looks to identify the best approach for considering design prohibitions in a product line optimization problem. The results suggest that design prohibitions should be included as constraints in the optimization and that there is a potential benefit to considering design prohibitions when estimating customer preference models. The second research question investigates how the added computational cost of either including design prohibitions or looking at large-scale optimization problems can be reduced. The results show that strategically creating the initial population helps to considerably improve the rate of convergence for both cases.