Model predictive control for online optimization of semi-batch reactors

In practical situations the knowledge on discontinuous reaction processes as well as the available measurement information is often very limited. On the other hand, several important variables can be estimated during the batch allowing a short range prediction of the process behavior. This paper discusses online recipe optimization of semi-batch reactors for such an industrial scenario. The derived concept is based on reaction calorimetry for vapor-liquid (VL) systems and formulated in the framework of model predictive control (MPC). In simulation studies of an isothermal semi-batch polymerization reactor the proposed MPC scheme yields the same operational profiles as an off-line optimization using a detailed process model