DESIGNING CONTROL CHARTS BASED ON RESIDUALS FOR MONITORING SIMPLE LINEAR PROFILES IN SHORT RUN PROCESSES

In some applications, performance of a process or quality of a product is characterized by a relationship between a response variable and one or more explanatory variables, referred to as profile in the literature. Certain methods have been developed to monitor various profiles. On the other side, nowadays due to diversity of customer demand and short time for presenting products in market, manufacturing strategy is focused on short run processes characterized by high diversity and low volume. Therefore, statistical process control for such processes, due to inspection restrictions in a short period is a special practice. In such circumstances, control charts in Phase I cannot be performed and correct estimations are not available for estimating process mean and standard deviation. To overcome the situation, self-starting methods are developed to update the parameter estimations along with new observations and simultaneous checks of the out-of-control conditions. Hence, implementing traditional control charts for monitoring short run processes is not practical, and new methods and control charts should be developed to monitor such processes. In this paper with aggregating two above-mentioned subjects, quality characteristics which are pertained to short run processes and which are modeled by simple linear profiles, have been monitored. Suitable methods and new control charts are developed to monitor process effectively. In this paper, we focus on monitoring residuals and propose new control charts to monitor mean and dispersion of residuals simultaneously. In order to monitor residuals in short run processes whose quality characteristics are modeled by simple linear profiles, we propose two control charts for monitoring mean and one control chart for monitoring standard deviation. Then, with combination of these control charts, we develop two distinct control charts named QMCC and TMCC to monitor mean and variance of residuals concurrently. Performance of the proposed control charts have been compared with competitor control chart using simulation studies and average run length (ARL) criterion. The results of simulation studies show that our proposed control charts in some parameters have better performance compared to competitive control chart under moderate and large shifts in terms of out-of-control ARLs.