Dynamic Optimization of Processing Systems with Mixed Degrees of Freedom

Procedures like start-up, product change-over, or shutdown of processing systems usually involve the manipulation of continuous and discrete controls. To optimize such procedures, the use of hybrid models is often appropriate to account for the change of dynamics during the transition. Some recently developed approaches for hybrid system optimization are based on mixed-integer linear programming. Applications reveal that the complexity inherent to these approaches considerably limits the applicability to industrial-size problems. This paper suggests as an alternative a tailor-made algorithm that optimizes the discrete and continuous degrees of freedom in a two-stage procedure. While the discrete controls are selected by a graph search algorithm, the continuous controls are obtained by embedded nonlinear programming. The method is illustrated for the start-up of a CSTR.