Data Reconciliation and Control in Styrene-Butadiene Emulsion Polymerizations

A nonlinear model‐based predictive control (NLMPC) method was developed using a First Principles model of an emulsion copolymerization of carboxylated styrene butadiene rubber (XSBR). Copolymer composition, conversion and average molecular weights of the copolymer were chosen as the controlled variables due to their influence on the final product properties and quality. These properties, however, are rarely measured in‐line due to the operational difficulties associated with their measurement. For this reason a soft‐sensor using reaction calorimetry techniques was developed and used to infer reaction conditions, rates, species concentrations and polymer properties in a industrial scale emulsion polymerization reactor.