Multi-objective optimization of a reactive batch distillation process using reduced order model

Abstract

Reactive batch distillation (RBD) processes are highly nonlinear in nature and use of first principles model into the optimization problem demands high computational time. Therefore, a reduced order model for RBD process is proposed which is based on selection of states of prime importance, identification of input state variables, definition of reference trajectories for nominal case, and definition of additional variable transformation to facilitate adaptation of the model to variations in operating conditions. Further, the work focuses on development of a multi-objective optimization formulation for arriving at an optimal operation policy for RBD, and a modified version of Non-dominated sorting Genetic algorithm (NSGA-II) is developed for its solution. The proposed approach is evaluated by application to a simulation case study for production of butyl acetate. The optimization results obtained using the reduced order model are comparable to those obtained with first principles model, while requiring only 1/20th of the time.