Integrating process dynamics within batch process scheduling via mixed-integer dynamic optimization

Abstract

During batch process scheduling, products' batch size, processing conditions as well as operating times are usually established offline and considered out of the scope of the decision making stage. In practice, process dynamics may vary from the ones forecasted, in such a manner that the predicted optimal conditions will not be the best in practice. As a result of this mismatch, the plant usually operates under sub-optimal conditions, but if the process is flexible, its processing conditions can still be adapted to the actual plant needs in order to improve the overall performance. Given this situation, there is a strong motivation for developing models and optimization tools to fully integrate process dynamics into batch scheduling. In this work, the potential of directly including control variables with time varying values and variable batch sizes in the scheduling of batch plants is explored. The optimization of process dynamics, which is time varying, along with scheduling tasks is accomplished using rigorous mixed-integer dynamic optimization techniques. Through several examples, we show that integrating both decision-making levels can lead to significant economic savings.