Dynamic optimization of a cryogenic air separation unit using a derivative-free optimization approach

Optimal dynamic product transition is a challenging and important issue in manufacturing plants. When a reliable dynamic model is available, gradient-based optimization algorithms can be used to achieve this aim. However, in some cases a first principles dynamic model may not be available. In this work, we will assume that only input-output information from a dynamic model embedded in a dynamic process simulator is available for optimal product transitions between products. We present a derivative-free optimization trust region approach to deal with the product dynamic optimization problem of an air separation unit used in several processing plants to obtain pure oxygen. High-purity oxygen is required in intensive energy applications such as steel plants and in combustion processes. A closed-loop model predictive control strategy is used where the system to be optimized is embedded in the ASPEN Dynamics simulation environment. The results demonstrate that black-box dynamic models can be dynamically optimized when model of the dynamic model and/or its gradient information are not available. We have successfully applied the non-linear model predictive control and derivate-free approach to several oxygen composition and productivity transition issues.