Improving manufacturing quality through planned experiments Statistical methodology

Planned experimentation is a powerful tool to improve quality of industrial manufacturing processes, in an environment placing prime concern on top quality. The goals are to manufacture products that function best possible, to achieve a minimum variability of the production process, and to secure product robustness against noise in the customer's environment. Statistically planned experiments are a tool to achieve these goals. They consist of a few experimental runs to obtain data on the product characteristics. Each run corresponds to a planned choice of levels of production factors and noise factors. Depending on the experimental data, the production factors are classified into signal factors that in the first place affect the process mean, control factors that predominantly determine process variability, and nuisance factors that can be used for cost reduction. Noise factors are an additional category, they mirror the customer's environment. Any such cycle of manufacturing improvement should terminate with a confirmation experiment. (orig.)Available from TIB Hannover: RO 7722(500) / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekSIGLEDEGerman