Performance of the Double Sampling Coefficient of Variation Chart under Measurement Errors Using a Linear Covariate Error Model for Monitoring NPK Fertilizer Manufacturing Process

Abstract

The double sampling (DS) coefficient of variation (CV) chart has been investigated in the literature under the assumption that the quality characteristic of interest is measured with no measurement error (ME). However, in practical applications, measurement errors caused by gauge equipment are often unavoidable and can lead to inconsistent process monitoring. In this manuscript, the performance of DS CV chart with measurement errors, referred to as the DS CV - ME chart is studied. The linear covariate error model, based on the constant error variance and that based on the error variance which is a linearly increasing function of the mean, is utilized. The DS CV - ME chart is evaluated using the average number of observations to signal (ANOS) and expected ANOS (EANOS) performance metric. The findings demonstrate that measurement errors diminish the DS CV - ME chart’s detection ability. Using a larger value of the coefficient B of the linear covariate error model can ameliorate the DS CV - ME chart’s performance. Additionally, the findings show that taking multiple measurements per item does not notably reduce the negative effect of measurement errors. A real-world application in manufacturing is provided to showcase the DS CV - ME chart’s working.