Application of multivariate statistical techniques for monitoring emulsion batch processes

Abstract

The power of multivariate statistical methodologies, such as multi-way principal component analysis and projection to latent structures (PLS), for batch process analysis, monitoring, fault diagnosis, product quality prediction and improving process insight is illustrated in this work. The techniques were applied successfully to several emulsion polymerization batch processes; one of them is illustrated in this article. A key feature of this work is that reaction extent was used as the common reference scale to compare batches with varying time duration. Results indicate that variations in one ingredient trajectory and heat removal related variables contribute primarily to viscosity variability. A PLS model, relating product viscosity with process variables was developed. This model shows great promise as a predictive tool for new batches.