Control charts for flexible and multi-variety production systems

This study presents two new proposals of self‐start statistical process control procedures for implementing quality control charts in mass customized production environments. Mass customization is characterized by high variety and/or low production volumes, or by current production strategies that increase the necessity of flexibility. The proposed approaches consider that the quality characteristic under monitoring is the same among the items, allowing the possibility of different averages and standard deviations for different products into the series. Analysis of the proposed approaches include assumption violations and the comparison of the new approaches to a Phase I–Phase II residual control chart, as benchmark. Simulation results show that one of the proposed approaches ties in performance with the benchmark, and the other can produce faster detection of deviations from the process mean, both without the need of a Phase I – retrospective analysis, thus increasing suitability for application in real environments. Also, the proposed approaches are made to handle with unitary production lots as well. Analysis performed in an extensive simulation procedure presents that normality assumption violation (with higher asymmetry) has the most degenerative effect on all tested control charts. The study also shown decreasing in performance on scenarios that violated the independence of the observations assumption.