Guided optimization: a fast model-based nested cost optimization technique for existing product family designs

Abstract

Cutting production costs is a crucial instrument for companies to increase profitability and remain competitive. However, companies aim to cut costs without compromising performance requirements of the products. Commonality between components reduces the costs, while diversity between them differentiates the key attributes of the products in a product family. This commonality-diversity trade-off is the essence of the product family design optimization. Current model-based methodologies consider optimizing the design for commonality rather than cost. This article proves that higher commonality does not always amount to minimal cost and, therefore, a simple commonality index cannot replace a cost model. A tunable cost model that includes simplified formulas for standardization benefits is introduced to be used by cost optimization techniques. Current product family optimization methods are often combinatorial and perform inefficiently due to searching large design space. These methods also optimize the product family design from scratch. This limits the applicability of the current methods in industrial settings that are typically complex and brownfield. This article proposes a model-based cost optimization methodology that accelerates the cost optimization by starting from an existing design. The proposed methodology is a two-stage nested optimization algorithm, in which the commonality matrix is optimized with sensitivity analysis on component types in the outer loop and the associated design variables are optimized in the inner loop. The methodology is numerically demonstrated on an industrial example and benchmarked against state-of-the-art cost optimization methods. The proposed methodology ensures a gradual improvement in cost reduction with a significant acceleration in performance.