Quality has been a rallying call in design and manufacturing for the last two decades. One way to improve quality is through variation reduction (VR). VR teams use tools such as Design of Experiments (DoE) and robust design to improve product performance and quality by reducing variation introduced by manufacturing processes. Because VR teams are typically resource constrained, they must carefully select where to focus their efforts. Planning for VR is complex because reduction efforts are executed on individual features and processes but benefits are accrued when the overall product quality improves. The problem is further complicated by the existence of multiple performance criteria and hundreds of processes and dimensions that effect each performance requirement. Consequently, VR teams typically use qualitative assessments to prioritize and schedule their efforts. This paper provides a mathematical model capable of optimally allocating VR resources for a complex product. The VR model has three parts: a model of variation propagation, a model of variation costs, and a model of variation reduction costs. These models are used to directly calculate the optimal resource allocation plan and schedule for a product with multiple product quality requirements. An example from the aerospace industry is used to demonstrate the theory. [S1050-0472(00)00602-4]
A soft-computing framework for automated optimization of multiple product quality criteria with application to micro-fluidic chip production
