Statistics in Advanced Manufacturing

Statistical concepts and methods have played a critical role in speeding the pace of industrial development over the last century. In return, industrial applications have provided statisticians with incredible opportunities for methodological research. The richness and variety of these applications have had a major influence on the development of statistics as a discipline; consider, for example, the extensive research in statistical process control (SPC) and changepoint detection, dating back to the pioneering work of Shewhart in the 1920s, and developments in automatic process control, design of experiments, sequential analysis, reliability, and so on. Recent efforts by manufacturers to adopt sustained quality and productivity improvement programs have generated a renewed interest in and appreciation for statistics in industry. In fact, fundamental statistical concepts such as understanding and managing variation form the backbone of popular quality management paradigms and practices. Many of the traditional SPC concepts and techniques grew in response to the manufacturing environments prevalent several decades ago. Current advanced manufacturing and high-technology industries, however, operate under a much more complex and diverse set of conditions. These changes have important implications for research directions in industrial statistics, not only in terms of identifying new problems and developing new methods, but also in reevaluating the paradigms that inspired earlier approaches. In this vignette we use applications from automotive and semiconductor manufacturing to illustrate various issues and to discuss future research needs and directions. The discussion is limited to a few selected topics and is inevitably slanted toward our own experiences.