1 Copyright © 2003 by ASME DECOMPOSITION-BASED ASSEMBLY SYNTHESIS OF MULTIPLE STRUCTURES FOR MINIMUM PRODUCTION COST Onur L. Cetin and Kazuhiro Saitou ∗ Department of Mechanical Engineering University of Michigan Ann Arbor, MI 48109-2125, USA E-mail: {ocetin, kazu}@engin.umich.edu ∗ Corresponding author ABSTRACT An extension of decomposition-based assembly synthesis for structural modularity is presented where the early identification of shareable components within multiple structures is posed as an outcome of the minimization of estimated production costs. The manufacturing costs of components are estimated under given production volumes considering the economies of scale. Multiple structures are simultaneously decomposed and the types of welded joints at component interfaces are selected from a given library, in order to minimize the overall production cost and the reduction of structural strength due to the introduction of joints. A multi- objective genetic algorithm is utilized to allow effective examination of trade-offs between manufacturing cost and structural strength. A new joint-oriented representation of structures combined with a “direct” crossover is introduced to enhance the efficiency of the search. A case study with two aluminum space frame automotive bodies is presented to demonstrate that not all types of component sharing are economically justifiable under a certain production scenario. Keywords: Assembly synthesis, design for modularity, multi- objective optimization. 1. INTRODUCTION Mechanical products are very rarely monolithic; one of the reasons is that the assembly of components allows simpler forms for the individual components, which are often more inexpensive to manufacture [1]. On the other hand, Design for Assembly (DFA) methodologies [2] often suggests the reduction of the number of components and joints to minimize the assembly cost. Further, the structural products usually favor fewer joints, since very often joints are the weakest points: for instance many fatigue failures are initiated from welded joints. The question is, therefore, “assuming a joint has to be made, what is the best method to do it?” [3]. Recognizing that the decisions on where and how the joints are to be made heavily impact the subsequent design processes of individual components, we have developed decomposition-based assembly synthesis [4,5], a method for the early identification of the joint locations and designs that minimally impact the overall structural strength. Modular product design, which facilitates sharing components across multiple products, is viewed as a convenient way to offer high product variety with low production cost. The basic premise here is that the component sharing would result in less design effort and fewer production varieties with higher volumes, hence reducing overall production cost. However, component sharing has a tendency to result in overdesign of low-end products and more importantly, underdesign of high- end products in a product family [6-13]. This effect, therefore, has to be outweighed by the economical gain of component sharing to justify a decision on component sharing [6]. As an extension of our previous work on decomposition- based assembly synthesis for structural modularity [14-16], this paper presents a method for the early identification of shareable components within multiple structures, posed as an outcome of the minimization of estimated production costs. The manufacturing costs of components are estimated under given production volumes considering the economies of scale. Multiple structures are simultaneously decomposed and the types of welded joints at component interfaces are selected from a given library, in order to minimize the overall production cost and the reduction of structural strength due to the introduction of joints. A multi-objective genetic algorithm is utilized to allow effective examination of trade-offs between manufacturing cost and structural strength. A new joint- Proceedings of IMECE’03 2003 ASME International Mechanical Engineering Congress Washington, D.C., November 15–21, 2003 IMECE2003-43085