Online statistical process control of chemical processes

An online SPC methodology for continuous chemical processes is described. The strategy avoids the model identification problem by focussing attention on statistical input-output relations. The problem then reduces to identifying the main causes of variations in product quality via the application of `principal component analysis´, a multivariate analysis technique. Such an approach is advantageous because guidelines on the required magnitudes of manipulative changes are natural outcomes of the analysis. The effectiveness of this procedure is demonstrated by application to a simulated chemical reaction system. Furthermore, it is shown that the assumption of cost-free adjustments, often stated of continuous processes, may not be always applicable