Optimal Design of Product Families Using Selection-Integrated Optimization (SIO) Methodology

A Single-Stage Gradient-Based Approach for Solving the Joint Product Family Platform Selection and Design Problem Using Decomposition

“…Simpson et al [1] classify approaches for solving the joint a posteriori platform selection and design problem based on the number of stages used for finding the optimal solution: Singlestage approaches optimize both platform variable selection and the design of the family of products simultaneously (Akundi et al [24]; Cetin and Saitou [25], Fujita et al [38], Fujita and Yoshida [29], Gonzales-Zugasti and Otto [30], Hassan et al [36], Simpson and D'souza [37], Khire and Messac [39], Khajavirad et al [40]), whereas two or multi-stage algorithms select the platform within the first stage and fix the selection while optimizing the product family design in the second stage (de Weck et al [26], Hernandez et al [31], [32], Messac et al, [33], Nayak et al [34], Fellini et al [27], [28], Rai and Allada [35]). There is some tradeoff between single and twostage approaches: Optimizing the platform and corresponding design variables in two separate stages may lead to sub-optimal solutions.…”

“…According to this classification, some methods limit scope in order to reduce complexity by assuming that design variables defining product platforms are known a priori and are not treated as variables in the optimization process (Allada and Jiang [2]; Blackenfelt [3], D'souza and Simpson [4], Dai and Scott [5], Farrell and Simpson [6], Fellini et al [7],;Gonzales-Zugasti et al [10], [11], Hernandez et al [12], Kokkolaras et al [13], Kumar et al [14], Li and Azarm [15], Messac et al [16], Nelson et al [17], Ortega et al [18], Seepersad et al [19], [20], Simpson et al [21], [22], Willcox and Wakayama [23]). However, other approaches optimize for the platform selection and product family design simultaneously; that is, platforms are specified a posteriori (Akundi et al [24], Cetin and Saitou [25], de Weck et al [26], Fellini et al, [27], [28], Fujita and Yoshida [29], Gonzales-Zugasti and Otto [30], Hernandez et al [31], [32], Messac et al [33], Nayak et al [34], Rai and Allada [35], Hassan et al [36], Simpson and D'souza [37], Fujita et al [38], Khire and Messac [39], Khajavirad et al [40]). Fujita [41] provides a related classification by defining three classes of product family optimization problems: In class-I problems, product attributes are optimized under a fixed platform assumption (i.e.…”

Industrial application

“…However, the two-stage approaches have been shown to lead to suboptimal solutions (Messac et al 2002); therefore, single-stage approaches are preferred for optimality. Single-stage Class III problems typically employ commonality restrictions to reduce computational cost (box 5) (Simpson and D'Souza 2004;Hassan et al 2004;Khire et al 2006), and therefore, suffer from suboptimality. Fujita and Yoshida (2001) addressed generalized commonality for the joint problem (box 6) by hybridizing GA, branch and bound, and sequential quadratic programming (SQP) to determine platform configuration, direction of similarities and variant design respectively.…”

“…Designing a family of products is a difficult task that embodies all of the challenges of product design while adding the complexity of coordinating the design of multiple products in an effort to increase commonality across the variants without drastically compromising their individual performance (Simpson et al 2001). This challenge manifests early in the design process wherein designers must not only specify the platform configuration-also referred to as platform variable selection or platform selection (Khire et al 2006)-but also optimize the design of the platform as well as the individual variants derived from the platform.…”

A Decomposed Genetic Algorithm for Solving the Joint Product Family Optimization Problem

“…Designing a product platform and corresponding family of products is a difficult task that embodies all of the challenges of product design while adding the complexity of coordinating the design of multiple products in an effort to increase commonality across the set of products without compromising their individual performance. This challenge manifests early in the design process wherein designers must not only specify the platform configuration (i.e., selecting which design variables are shared across the products in the family -also referred to as platform variable selection or platform selection 2 ), but also optimize the design of the platform and the individual variants by choosing design variable values while maintaining commonality defined in the platform configuration. Resolving the inherent tradeoff between platform commonality and product distinctiveness is paramount: Increasing the degree of commonality among variants in a product family generally reduces total cost, but it can also compromise the ability of each variant to fully achieve the desired characteristics that make it distinct and attractive to different market segments.…”