Economic-statistical design of EWMA-semicircle charts under the Taguchi loss function

Abstract

A single exponentially weighted moving average (EWMA) chart is effectively used to monitor the process mean and/or variance simultaneously. An EWMA-semicircle (EWMA-SC) chart designed from the economic-statistical perspective is proposed, which incorporates Taguchi's quadratic loss function into Lorenzen and Vance's cost model. Moreover, economic-statistical performance and the effect on process capability index are compared to those with sum of square EWMA (SS-EWMA) and maximum EWMA (MaxEWMA) charts. The optimal decision variables - namely, sample size n, sampling interval time h, control limit width L and smoothing constant λ - are obtained by minimising the expected cost function. Via simulations, the EWMA-SC chart is found to incur the smallest expected cost when a process mean and variance simultaneously shift. However, the MaxEWMA chart incurs the lowest cost of defective products when a process means shifts on its own. [Received: 1 May 2017; Revised: 22 August 2018; Accepted: 3 January 2019]