Risk-Conscious Operation of Batch Processes

Abstract This contribution suggests an approach to the dynamic optimization of batch process operations that takes into account model uncertainty and inter-run variations. As opposed to the standard nominal optimization, a model with a probabilistic uncertainty description rather than an exact nominal model of the process is used. A probabilistic measure of success, such as the probability of making in-spec product, is optimized. Two simulation examples illustrate the application of the proposed technique and demonstrate the increased robustness to variations of the obtained operating profiles.