Nonlinear Model Predictive Control and Dynamic Real-Time Optimization of Semi-Batch Reactors - A Case Study of Expandable Polystyrene Production
MetadataShow full item record
The aim of this thesis was to develop an advanced two-layer control structure for the semi-batch polymerization of expandable polystyrene (EPS) that minimizes the batch time while still producing a set polymer quality and obeying other constraints. While the use of advanced control structures in the chemical industry has become more common, their use is seen to a lesser extent in the polymer industry. This method of control requires an accurate process model, which can be an arduous task due to the nonlinearities and complex reactions that occur during polymerization. However, research has indicated that the use of such control structures can result in a reduced batch time and improved product purity; therefore, the effort to develop these control structures for polymerization should be considered further. Before diving into the development of the control structure for the production of EPS, some background information on polymerization and semi-batch reactor modeling is provided. Introductory concepts from optimization and control are then presented to highlight ideas necessary to understand before an advance control structure can be designed. Together, these two chapters provided the necessary background information to develop a two-level control structure for the production of EPS.EPS was selected as the case study since it is one of the largest commodity polymers produced. The model equations are outlined with the assumptions that were made in their derivation. These equations were implemented in the programming language C using a template provided by Cybernetica AS; this allowed for the use of their software in the implementation and simulation of the advanced control structure. Offline optimization of the process was performed to identify a starting point for the optimal operating conditions. The two control layers were then constructed and validated. Attention is paid to how the two layers work together to calculate and realize the optimal operating conditions. This work demonstrated that the objective can be achieved using a two-layer control structure where the DRTO level determines the optimal reactor temperature profile and the NMPC level follows the trajectory by minimizing the cooling water flow rate. To further motivate the development and use of advanced control structure, the potential economic advantages of this approach over the current fixed recipe approach are discussed.