Algorithm for optimal loss reduction in serial manufacturing

Quality improvement efforts and process planning for quality both focus on the manufacturing process and the effect of each processing step on the quality of the manufactured part. A probabilistic model that describes the serial effects of subprocessing rates and times on the quality of a part characteristic is proposed. The Taguchi loss function is used as a guide to action on the process, because expected loss per part operationally defines the consequences of off-target process centring and excessive process variation. Continual reduction of loss is conceived as a sequence of specific actions on subprocessing rates or times, each action providing an increment of loss reduction via an adjustment to a mean, a reduction in a variance, or both. An algorithm is reported that automatically generates the optimal sequence of loss reduction actions on a serial manufacturing process. Each action prescribed to the quality engineer is the one that, within the technology limits of the subprocesses, provides the most loss reduction per dollar expended on the change. Several examples are solved via the algorithm, and extensions to more complex cost of improvement functions are suggested.