ROBUST PRODUCT DESIGN OPTIMIZATION METHOD USING HIERARCHICAL REPRESENTATIONS OF CHARACTERISTICS

ABSTRACT Product design optimizations usually require the optimization of not only all performance characteristics, but also the robustness of certain performance characteristics. Obtaining optimum design solutions is far from easy, since this requires evaluations of numerous related characteristics that usually have complicated and conflicting interrelationships. Some of these characteristics can include variations of one type or another, such as manufacturing process variations, variations pertaining to the environments where the product is used, variations in how long-term use affects certain product characteristics, and so on. The difficulty of obtaining optimum design solutions is thus compounded by the need to carry out specific optimizations that provide sufficient robustness to safely accommodate anticipated ranges of variations. This paper expands the hierarchical multiobjective optimization method based on simplification and decomposition of characteristics so that optimizations can be concurrently conducted for both performance characteristics and maximization of robustness against characteristic variances. A principal cause of variations in performance characteristics is variations in the contact conditions of joints, and the utility of the proposed robust product design optimization method is demonstrated by applying it to machine-tool models that include joints. INTRODUCTION In today's product manufacturing environment, a wide range of factors such as product performances (such as the speed, accuracy and efficiency in accomplishing certain tasks), product qualities (such as performance variance and robustness) and operational and manufacturing costs must be considered when designing and producing machine products. To create successful products, numerous product characteristics need to be concurrently evaluated so that the specific requirements need to satisfy the product performances and qualities can all be satisfied to the highest possible degree. To accomplish these tasks, appropriate system optimization methods must be constructed and then used effectively. There are two main obstacles to achieving sophisticated system optimization results for product designs. One pertains to the problem of obtaining suitable optimum design solutions, and the other to whether or not the obtained solutions have sufficient robustness and can successfully accommodate the variations in design variables and parameters that will occur during product manufacturing and later usage of the product. These two problems are of equal importance in product design optimizations, since insufficient robustness can ultimately spoil what might be considered a superior design in situations where the need for adequate robustness is not adequately recognized