Efficient linear profile schemes for monitoring bivariate correlated processes with applications in the pharmaceutical industry

Abstract

The complexity of chemical processes warrants that collection of information about the processes will continue as they proceed through the various stages of industrialization. Process monitoring has emerged as an essential tool for confirming that processes stay in control and identify future process improvement opportunities. Linear profile monitoring is an approach that describes the direct relationship between the process or product characteristics and further checks the stability of the relationship by monitoring relevant parameters. In this paper, we propose a new memory-type linear profile control charting scheme that consists of both the homogeneously weighted moving average (HWMA) control charting structure and the Bayesian estimation framework. We utilize the restricted and the pre-test Bayesian framework and propose the HWMAR and HWMAPT control charts, respectively, to monitor the linear profile intercept, slope, and error variance parameters. Comparative analysis revealed the superiority of the proposed charting schemes. Specifically, our simulation results showed that the proposed HWMAR chart outperforms not only the HWMAPT chart but also many other competing charts, already existing in the literature. A real-life example is provided to illustrate the application of the proposed charts in shrinking the variations in the quality of a pharmaceutical product. 2020 Elsevier Ltd