Meeting overall reliability requirements for heats of steel: A case study

ABSTRACT

In this case study, we examine the reliability of a device whose material was produced over several heats, where the quality metric of the material is the content of a certain element in the material. The analyses revealed a number of issues, including rounded data values as well as periodicity and tilting in the two dimensions of the production process. Novel analysis features include non-standard estimation of certain variance components, and combining several tolerance intervals based on optimization criteria. The results of the analysis were useful to both the customer and supplier of the material.

KEYWORDS:

• analysis of variance (ANOVA)
• arcsine distribution
• rounding
• statistical process control
• tolerance interval
• variance components

About the author

Joseph G. Voelkel is a Professor in the Graduate Statistics Program in the College of Science's School of Mathematical Sciences at Rochester Institute of Technology, Rochester, New York. His interests include experimental design, quality control and improvement, reliability, and non-standard problems in the physical sciences. Before RIT, he had been a statistical consultant for Allied-Signal (now Honeywell) Corporation, working in chemicals, plastics, fibers, water-treatment polymers, and agricultural products, as well as inter-laboratory, epidemiological, and toxicology studies; and was on the faculty at the University of Wisconsin, where he specialized in cancer research. His consulting at RIT has included a wide range of clients, including those in the optics, electronics, resin, plastics, automotive, laser, LED, and bearing industries. He is a Fellow of ASQ, a former Chair of ASQ's Statistics Division, and currently serves on the editorial board of Quality Engineering.

Additional information

Notes on contributors

Joseph G. Voelkel

About the author

Joseph G. Voelkel is a Professor in the Graduate Statistics Program in the College of Science's School of Mathematical Sciences at Rochester Institute of Technology, Rochester, New York. His interests include experimental design, quality control and improvement, reliability, and non-standard problems in the physical sciences. Before RIT, he had been a statistical consultant for Allied-Signal (now Honeywell) Corporation, working in chemicals, plastics, fibers, water-treatment polymers, and agricultural products, as well as inter-laboratory, epidemiological, and toxicology studies; and was on the faculty at the University of Wisconsin, where he specialized in cancer research. His consulting at RIT has included a wide range of clients, including those in the optics, electronics, resin, plastics, automotive, laser, LED, and bearing industries. He is a Fellow of ASQ, a former Chair of ASQ's Statistics Division, and currently serves on the editorial board of Quality Engineering.