Development of SX chart for quality control

Abstract

This project develops the algorithm and presents the methodology for designing the optimal xbar , SX, and CUSUM charts. The optimal design parameters are obtained by minimizing the loss function. The out-of-control detection effectiveness, as well as

performance of the charts, would also be explored based on two key measures – the Average Time to Signal (ATS), which is the time taken to signal an out-of-control condition upon its onset, and the loss function. The performance of the charts is compared over a range of process mean shifts, and for three different specified values of minimum allowable ATS0 (average time a false

alarm is signalled). The findings from the comparative study revealed that the SX chart largely outperforms the Shewhart xbar chart. Furthermore, while it is nearly as effective as, or slightly inferior to the CUSUM chart in detecting process mean shifts, the SX chart is much simpler in its operation.