Economic model-predictive control for chemical processes

This thesis addresses important challenges that have to be overcome to facilitate a wider application of economic nonlinear model-predictive control in industry. One of the reasons justifying its application in place of other control methods is that the economic control performance may be significantly improved. Hence, the main focus of this thesis lies on achieving the best possible economic control performance at any time. To achieve this, a comprehensive formulation of the economic optimal control problem integrating the control task, the optimization task and the scheduling task as well as a specifically tailored numerical solution strategy are presented. Furthermore, a suitable model-predictive control class is chosen. After addressing feasibility and stability issues, existing fast nonlinear model-predictive control schemes are reviewed. These schemes substantially reduce computational delay such that real-time applicability becomes possible and performance losses are diminished. Supported by the results of the review, a fast economic nonlinear model-predictive control scheme is proposed which also accounts for all aspects ensuring the best possible control performance at any time.