On Designing a Triple Sampling Coefficient of Variation Chart

Process monitoring is essential for maintaining a stable production process. The triple sampling (TS) charts have been extensively investigated to monitor the process mean. However, the TS charts are unable to reliably detect process mean shifts when the mean and standard deviation are proportional to one another. To address this setback, in this study, a TS coefficient of variation (CV) chart which improves upon the double sampling (DS) CV chart's performance in detecting shifts in the process CV is developed. Optimization procedures in computing the optimal charting parameters by minimizing the out-of-control average number of observations to signal (ANOS), average run length (ARL), expected ANOS (EANOS) and expected ARL (EARL) values are provided. Tables of optimal parameters are given in this study to facilitate the quality engineer in implementing the TS CV chart. The findings reveal that the TS CV chart outperforms the DS CV chart in detecting all sizes of CV shifts. A real dataset involving the NPK fertilizer production process is used to illustrate the implementation of the TS CV chart.