Title: Comparison of control charts for autocorrelated process control

Authors: Canan Bilen, Xianzhe Chen

Addresses: Department of Industrial and Manufacturing Engineering, North Dakota State University, Fargo, ND 58108-6050, USA. ' Upper Great Plain Transportation Institute, North Dakota State University, Fargo, ND 58108-6050, USA

Abstract: Conventional control charts were developed for industrial processes in which successive observations were assumed to be independent and identically distributed; an assumption often violated in practice. Approaches to monitoring autocorrelated data range from using conventional control charts, with adjusted control limits, to using Hotelling T² statistics on a moving window of observations from the univariate autocorrelated process of interest. Although there are many different approaches to monitoring autocorrelated processes, there are no guidelines on choosing the procedure with the best average run length performance for a particular level of autocorrelation and shift magnitude of interest. This paper presents a guideline for choosing proper control charts for various conditions based on average run length performance. A Monte Carlo simulation study is designed to provide a comparative evaluation of the average run length performance of autocorrelated process control charts.

Keywords: autocorrelated process control; control charts; Hotelling T2; batch-means charts; residuals charts; moving EWMA chart; Monte Carlo simulation; SPC; statistical process control.

DOI: 10.1504/IJQET.2009.031127

International Journal of Quality Engineering and Technology, 2009 Vol.1 No.2, pp.136 - 157

Published online: 21 Jan 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article