A mean deviation-based approach to monitor process variability

Abstract

The study proposes a Shewhart-type control chart, namely an MD chart, based on average absolute deviations taken from the median, for monitoring changes (especially moderate and large changes – a major concern of Shewhart control charts) in process dispersion assuming normality of the quality characteristic to be monitored. The design structure of the proposed MD chart is developed and its comparison is made with those of two well-known dispersion control charts, namely the R and S charts. Using power curves as a performance measure, it has been observed that the design structure of the proposed MD chart is more powerful than that of the R chart and is very close competitor to that of the S chart, in terms of discriminatory power for detecting shifts in the process dispersion. The non-normality effect is also examined on design structures of the three charts, and it has been observed that the design structure of the proposed MD chart is least affected by departure from normality.

Keywords:

• control charts
• normality
• non-normality
• power curves
• MD chart
• R chart
• S chart
• simulations

Acknowledgements

The authors are grateful to the editor, the associate editor and the referees for their recommendations and proposals which helped in improving substantially the earlier version of this article. Muhammad Riaz would also like to pay his gratitude to Prof. Dr. Ronald J.M.M Does, General Manager, Institute for Business and Industrial Statistics, University of Amsterdam, The Netherlands, for his invaluable contributions, in the advancement of his research work in the area of Statistical Quality Control.