Multivariate SPC for Sulfite Batch Control, 2000 Pulping / Process & Product Quality Conference Proceedings

R. Monette, M. Champagne--The applications of multivariate data analysis and process control are growing, especially for continuous processes. These multivariate techniques can also be applied to batch processes. This paper describes the use of PCA/PLS on a batch sulfite pulp digester. Historical data from the digester process containing acceptable batches was used to build reference models. Models are built on two levels, the observation level and the batch level. The observation level detects process deviations with control charts. The control charts have limits of ± 3 standard deviation and can be used to monitor the evolution of new batches. The batch level model can be used to predict final batch characteristics and classify batches as good or bad. The advantages of multivariate batch process control are that deviations are detected easily and quickly, and all variables affecting the batch process are monitored equally as opposed to observing only the critical.