Nonlinear model predictive control of integrated process systems

The paper considers the NMPC of process systems that feature a dynamic behavior with multiple time scales, as a consequence of process integration. The process dynamics are modeled via a singularly perturbed system of differential equations in nonstandard form. A framework for simultaneous model reduction and composite controller synthesis is introduced, relying on linear feedback control for the fast dynamics and NMPC for the slow dynamics (using the corresponding reduced order, non-stiff model). Precise dimensions for the state-space realization of the fast and slow components of the system dynamics are provided and the stability and optimality properties of the proposed control framework are characterized. Furthermore, it is argued that this framework reduces online computation times and a simulation case study is presented for illustration.