Advanced control of a bio-reactor and a solid-state polymerization system

Two industrially significant control problems are dealt with in this thesis. The control objectives for each problem are described. The control technology used to solve these problems and the results obtained are clearly described.; The first problem deals with the control of a solid-state polymerization process. This part of the work is a scouting study on the applicability of Constrained Model Predictive Control (CMPC) technology to a solid-state polymerization process. Optimal operation of the solid-state polymerization process is demonstrated in the presence of process constraints and disturbances. A fundamental model of this process is developed and described. This portion of the work was performed at the DuPont Experimental Station, Wilmington, Delaware. Most of the information presented in this part of the thesis has limited detail due to the proprietary nature of the information involved.; The second problem deals with the control of a bio-reactor. Various control strategies were studied on this complex nonlinear problem. Both linear strategies (PID, CMPC) and nonlinear strategies (IMC, RNCL, Neural IMC and Neural RNCL) were tested on this process. Each control strategy is evaluated with respect to the operational objectives for this process and the advantages and disadvantages are described. The constraint handling and optimization capabilities of CMPC are also demonstrated.