Abstract
This work is concerned with the application of optimal control methods to a pilot scale distillation column. A dynamic model of the column is derived from physical equations and the model results are verified by comparison with those obtained in practice. A reduced order model of the column was also obtained by calculating model frequency responses. Optimal control systems to regulate the column are designed using Kalman's formulation of Bellmans dynamic programming principle. A quadratic cost function and a linearised plant model are used to obtain a state feedback and feedforward control system. Methods of reducing the number of measurements required for optimal control are investigated, including the use of an observer system and the elimination of control coefficients using a penalty function. The optimal control was tested using computer simulation and online to the distillation column under direct digital computer control. The results were compared with those obtained using discrete-time two-term control systems. An approximate method of achieving time optimal control was also investigated.