Control Schemes for Discrete Censored Reliability Data in Two‐Stage Processes

ABSTRACT This study develops three surveillance schemes to monitor a two‐step cascade manufacturing operation with reliability data. The designed surveillance schemes, namely, a lower‐sided cumulative sum (CUSUM), exponentially weighted moving average (EWMA) with and without a reflecting barrier, are based on the proportional odd model to account for the dependent structure of the manufacturing operations. The reliability‐based quality variable corresponds to the second step of the operation and is assumed to follow a discrete Weibull distribution. The efficiency of the proposed schemes is evaluated using zero‐state average run length, steady‐state average run length, standard deviation of the run length, and quartiles of run length distribution. In addition, for a better comparison, the analysis is conducted for both censored and uncensored data. The results show that the EWMA chart with a smaller weighting factor, when applied without a reflecting barrier, demonstrates the best performance as compared to its competing counterparts in both censored and uncensored settings. However, the performance of the EWMA chart declines as the weighting factor increases. The efficiency of the proposed schemes deteriorated as the censoring proportion increased. Lastly, the performance of the designed surveillance schemes is illustrated with the real‐world examples.