MULTIVARIATE SPC METHODS FOR PROCESS AND PRODUCT MONITORING

Statistical process control methods for monitoring processes with mult ivariate measurements in both the product quality variable space and t he process variable space are considered. Traditional multivariate con trol charts based on chi(2) and T-2 statistics are shown to be very ef fective for detecting events when the multivariate space is not too la rge or ill-conditioned, Methods for detecting the variable(s) contribu ting to the out-of-control signal of the multivariate chart are sugges ted. Newer approaches based on principal component analysis and partia l least squares are able to handle large ill-conditioned measurement s paces; they also provide diagnostics which can point to possible assig nable causes for the event. The methods are illustrated on a simulated process of a high pressure low density polyethylene reactor, and exam ples of their application to a variety of industrial processes are ref erenced.