Design for Optimal End-of-Life via Product-Embedded Disassembly Shingo Takeuchi and Kazuhiro Saitou Department of Mechanical Engineering, University of Michigan, Ann Arbor, MI, USA Abstract This paper presents a method for designing products that disassemble via a domino-like “self-disassembly” process triggered by the removal of a few fasteners. Given component geometries, the method optimizes the spatial configurations of components, locators and fasteners, and the end-of-life (EOL) treatments of components and subassemblies for maximum profit and minimum environmental impact. Extending our previous work, it incorporates the end-of-life (EOL) treatments of components as additional decision variables, and Life Cycle Assessments (LCA) for evaluating environmental impact. A multi-objective genetic algorithm is utilized to search for Pareto optimal design alternatives. An example on Power Mac G4 cube® is discussed. Keywords Design for disassembly, configuration design, disassembly sequence planning, life cycle assessments 1 INTRODUCTION Economic feasibility of an EOL scenario of a product is determined by the interaction among disassembly cost, revenue from the EOL treatments of the disassembled components, and the regulatory requirements on products, components and materials. While meeting regulatory requirements is obligatory regardless of economic feasibility, EOL decision making is often governed by economical considerations [1]. Even if a component has high recycling/reuse value or high environmental impact, for instance, it may not be economically justifiable to retrieve it if doing so requires excessive disassembly cost. Since the cost of manual disassembly depends largely on the number of fasteners to be removed and of components to be reached, grabbed, and handled during disassembly, it is highly desirable to locate such high-valued or high-impact components within a product enclosure, such that they can be retrieved by removing less number of fasteners and components. As a solution to this problem, we have previously introduced a concept of product-embedded disassembly [2, 3], where components are spatially arranged within a product enclosure such that they can be reached and removed in the sequence for the optimal EOL scenario. In order to minimize the number of fasteners, the relative motions of components are constrained, wherever possible, by locators (eg, catches, lugs, tracks and bosses) integrated to components. As an illustration, consider two assemblies in Figures 1 (a) and 1 (b). In Figure 1 (a), three components A, B and C are fixed to a case with three fasteners, whereas in Figure 1 (b), the motions of components B and C are constrained by the locators integral to the components and the case. With a high labour cost for removing fasteners, only component A in Figure 1 (a) can be disassembled for reuse and recycle, with the reminder sent to landfill. For the assembly in Figure 1 (b), on the other hand, the removal of the fastener fixing component A activates the domino-like self-disassembly process of A, B, and then C. Since no additional fasteners need to be removed, components B and C can also be disassembled, allowing the recycle/reuse of all components as well as the case. This paper presents an extension of our previous work [2, 3] on the computational method for designing the products with such embedded disassembly means, where the problem was posed as optimization of the arrangements of components, locators and fasteners, to maximize the profit of disassembly. In [2, 3], the profits of components via EOL treatments are considered as constant and given as inputs to the problem. Although one should always assume the most profitable EOL treatments (or non- treatments) for maximizing overall profit, it is well known that this would not be always optimal for minimizing environmental impact. In order to examine a trade-off between the overall profit and the environmental impact of disassembly, the present work newly incorporates the end-of-life treatments of disassembled components and subassemblies as additional decision variables, and the Life Cycle Assessments (LCA) as a means to evaluate environmental impact. A multi-objective genetic algorithm [4] is utilized to search for Pareto optimal designs in terms of 1) satisfaction of the distance specification among components, 2) efficient use of locators on components, 3) profit of the EOL scenario of the product, and 4) environmental impact obtained by LCA. The method is applied to a simplified model of Power Mac G4 cube® for demonstration. Figure 1: (a) conventional assembly (b) assembly with embedded disassembly. 2 RELATED WORK 2.1 Design for disassembly, configuration design, and disassembly sequence planning The present work belongs to the area of design for assembly, and relates to the areas of configuration design, disassembly sequence planning. The previous works on design for disassembly [5] address local deign modification for the ease of disassembly, but not the global rearrangement and redesign of components for an optimal disassembly sequence addressed in this paper. landfill reuse reuse reuse recycle recycle (a) (b) A B C A B C B C A B C C B C 423