Robust Design and Control of Extractive Distillation Processes under Feed Disturbances

Abstract

A mixed-integer nonlinear programming optimization framework based on the steady-state model of the extractive distillation process for benzene/acetonitrile azeotropic mixture separation with dimethyl sulfoxide solvent is developed. Two single-end control structures are evaluated for the designed plant for the nominal feed composition. Because the adoption of inferential control strategies cannot guarantee the specification of the products, benzene feed composition is assumed to be a Gaussian variable. Then, the optimization problem is reformulated taking the set-points of the controllers together with the design parameters as decision variables which leads to a closed-loop stochastic optimization problem in which unscented transform is used as the uncertainty propagation tool. Investigation of the steady-state properties, as well as the dynamic responses of the optimal solution under feed composition disturbances, reveals that the use of the control scheme with fixed reflux-to-ratio for the extractive an...