Performance of the MEWMA‐CoDa control chart in the presence of measurement errors

Abstract This paper can be considered as an extension of the work of Tran et al (for monitoring compositional data using a multivariate exponentially weighted moving average MEWMA‐compositional data [CoDa] chart) by taking into account potential measurement errors that are known to highly affect production processes. A linearly covariate error model with a constant error variance is used to study the impact of measurement errors on the MEWMA‐CoDa control chart. In particular, the influence of the device parameters (σM,b), the number of independent observationsm, and the the number of variablespare investigated in terms of the MEWMA optimal couples (r,H) as well as in terms of their corresponding ARLs. A comparison between the Hotelling‐CoDaT2and the proposed chart is made in order to show that the MEWMA‐CoDa chart is more efficient in detecting shifts in the presence of measurement errors. A real‐life example of muesli production, using multiple measurements for each composition, is used to estimate the parameters and also to demonstrate how the MEWMA‐CoDa can handle measurement errors to detect shifts in the process.