00005 ™c83d1e05ebea51ca527b5a4071b83a

4

Keats & Montgomery

integration of these two generał classes of techniąues would potentially be effective.

The first paper in the section, co-authored by W. S. Messina, D. C. Montgomery, J. B. Keats and G. C. Runger discusses strategies for monitoring a process where the adjustments are driven by a form of integral control. They show that monitoring either the output deviation from target or the seÄ…uence of adjustments madÄ™ to the manipulatable variable by the integral controller is an effective strategy. In some situations, combinations of these two approaches are shown to be effective.

The second paper, "Applications of the EWMA for Algorithmic Statistical Process Control," by R. V. Baxley, Jr., shows how the EWMA lends itself to a simple but effective monitoring and adjustment procedurÄ™. He presents two implementations at Monsanto in their textile fibers manufacturing operations. The first is a manuaÅ‚ chart for control of an important variable in the manufacture of spun carpet staple fiber, and the second is a computerized application involving a combined EWMA/Shewhart procedurÄ™ for improving uniformity of a parameter on textured continuous-filament carpet yam.

The last paper in the section is "Statistical Process Monitoring with Integrated Moving Average Noise," by S. A. Vander Wiel. The author studies monitoring forecast errors from a time series to detect a level shift in the process. Four monitoring schemes are eyaluated; CUSUM, EWMA, Shewhart indiyiduals charts, and a scheme based on likelihood ratio statistics. Average run length results show that the CUSUM can be designed to be at least as good as, and frequently superior to, the other methods. Graphical aids are provided for designing the CUSUM and EWMA procedures.

Section IV is devoted to design of experiments. The first paper, by D. C. Montgomery, provides a brief discussion of guidelines for successful design of an experiment in a industrial setting, where the primary application environment is process development or process improvement. A seven-step approach is presented, and the importance of pre-experimental planning is emphasized. The effective integration of engineering or process knowledge with sound statistical thinking at this stage is crucial to success. Computer-aided design of experiments and the sequential approach to experimentation are also briefly discussed.

The second paper, co-authored by D. C. Montgomery and G. C. Runger, is an introduction to experimental design problems with random factors. This is a vast subject, and the authors have elected to concentrate on a special case, the study of measurement system capability with factorial designs. Many of these studies can be treated as a two-factor factorial. Several methods of yariance component estimation are briefly suryeyed, including the usual