Design of Robust High-Purity Distillation Columns under Feed Quality Disturbances

The main contribution of this work is the development of a control system with improved robustness against the feed composition disturbances in a high-purity distillation column separating the mixture of water and 40 mol% methanol. Despite the superior performance of an inferential control system with one temperature measurement along with a constant reflux-to-feed ratio in comparison to a fixed reflux ratio, however, it fails to provide acceptable performance. Methanol feed composition is taken as a Gaussian variable to obtain a more robust single-end control system. The set of decision variables is constructed considering the design parameters and the set points of the PI controllers. The first and second moments of the constraints and objective function are provided using the unscented transform for the solution of the stochastic optimization problem in the chance-constrained sense. The dynamic responses of the modified single-end control system against unmeasured disturbances in feed composition confirm the effectiveness of the proposed algorithm since product quality can be guaranteed to lie above the desired purity.