Robust Calorimetric Estimation of Semi-Continuous and Batch Emulsion Polymerization Systems with Covariance Estimation

In this paper, the state estimation of semi‐continuous and batch processes for a set of emulsion polymerization reactions is addressed. Specifically, the autocovariance least‐squares (ALS) and the moving horizon estimation (MHE) techniques are implemented to determine the covariance matrices and monitor the reactions in two steps, respectively. A collection of four polymerization reaction data sets that include gravimetric analysis are considered and used for validation of the proposed methods. The major process states estimated are the overall heat transfer coefficient and the conversion. The estimation results are in good agreement with the experimental data, independently of the initial overall heat transfer coefficient employed. They also indicate that a unique set of covariance matrices can represent the four reactions studied. The developed procedure offers a robust alternative for state estimation of industrial and experimental polymerization systems. Moving horizon estimation is an optimization‐based filter that can address non‐linear and constrained systems. A monitoring procedure with statistical parameters estimated by the autocovariance least‐squares technique is proposed and validated with experimental emulsion polymerization reaction data. The estimation results are compared with gravimetric analysis and the effect of the initial estimate is also evaluated, showing that the procedure is robust.