Re-optimization based control of copolymer quality in semi-continuous emulsion polymerization

The aim of this paper is to re-optimize the semibatch emulsion copolymerization process in order to minimize the batch time while maintaining copolymer composition. This method of model based control combined with minimization of the batch time can be used to reject disturbances and to ensure an optimal future trajectory from the current time step. Orthogonal collocation provides an efficient method for reducing differential equations to algebraic equations and is employed to discretize the dynamic model. The reoptimization strategy starts with a pre-calculated optimal feed trajectory providing initial conditions at the collocation points for the re-optimization of the whole trajectory at the first time step. At every further time step, the process states are estimated using a CEKF. This "measured data" is used to check whether the process deviates from the desired product quality and provides the starting point for the reoptimization. The optimal trajectory from the previous time step provides the initial conditions at the collocation points. This strategy is applied until the end of the batch. The online re-optimization based on the reduced model is applied to a disturbed and to an undisturbed process simulation showing that it rejects the disturbance and achieves the best possible composition under the given conditions.