A multiattribute cumulative sum-np chart

Abstract

Multiattribute charts can be adopted in various processes as there are numerous quality characteristics that cannot be represented quantitatively or are not easily expressed on a numerical scale. However, only a limited amount of research has been done on multiattribute charts. The multiattribute np (Mnp) chart has been developed to monitor attribute data with multiple quality characteristics. In this paper, a binomial multiattribute cumulative sum (MCUSUM) chart is developed. For a better detection effectiveness, the multiattribute cumulative sum-np (MCUSUM-np) chart that merges the binomial MCUSUM and Mnp charts to monitor upward shifts in fraction nonconforming is further proposed. A comprehensive comparison of the MCUSUM-np chart with its individual counterparts, in terms of the average number of nonconforming units and average time to signal criteria, shows that the MCUSUM-np chart has a better or at least equal performance compared with its basic charts. The implementation of the proposed charts is illustrated with a real-life application on the cold rolling process. The findings of this paper will improve the detection of shifts in fraction nonconforming so that practitioners can take swift corrective actions to return an out-of-control process to its in-control condition.