Multivariate SPC for startups and grade transitions

TLDR: In this article, a multivariate statistical method based on PCA and PLS is proposed to improve process transition performance using historical records of transition data, which is aimed at reducing the amount of off-specification materials and reducing transition time.

Abstract: Process transitions (grade changeovers, startups, and restarts) are very frequent in industry, and usually lead to the loss of production time, the production of off-grade materials, and to inconsistent reproducibility of product grades. Two aspects of using multivariate statistical methods based on PCA and PLS to improve process transition performance using historical records of transition data are discussed. First, multivariate SPC approaches are proposed to determine if the process conditions for the commencement of a transition (“startup readiness”) are correct and to assess the successful completion of a transition (“production readiness for the new grade”). The latter is illustrated using a simulated fluidized-bed process for the production of different grades of linear low-density polyethylene. Second, analysis tools are suggested for diagnosing the reasons for past transition problems and for monitoring new transitions to ensure repeatable high quality transitions. The latter methods are aimed at reducing the amount of off-specification materials and reducing transition time, as illustrated on industrial data from restarts of a polymerization process.