Robust design techniques are often applied to the design of manufacturing processes to determine the most robust operating points for a production system. However, such efforts have traditionally been focused on treating the output of each manufacturing operation in isolation. This approach ignores the fact that the sensitivity of each operation to input variation is a function of the operating point, which can only be changed in conjunction with the operating points of all other operations in that system. As such, applying robust design to each operation within a system individually does not guarantee lowest end-of-line variation. This is contrary to commonly held beliefs. What is needed instead is a method for conducting a system-wide parameter design where the operating points of each operation are optimized as a complete set to reduce final product variation. The logistics of such an integrated parameter design scheme become difficult or impossible on processes that may occur in different geographical locations. In this paper we outline the use of mathematical models to conduct system-wide parameter design. We demonstrate this technique on a model of a sheet stretch-forming manufacturing system. Through this example, we show that selecting operating points while considering the entire system results in a greater reduction in variation than Taguchi-style robust design conducted independently on each of the operations within the system.
展开▼
