Dynamic Comparison of Linear and Nonlinear Models for Generic Reactive Distillation System

Abstract

The performance of a linearized dynamic model of a generic reactive distillation system has been compared with that of a rigorous nonlinear model. The effect of steady state sensitivity to different input step changes on the performance of the approximate model has been explored. The accuracy of an approximate linear model is quantified by the use of an average relative error. The approximate linear model nicely averages the process nonlinearities when the magnitude of input change is small and becomes inadequate as the deviation from base steady state increases with increase in the magnitude of input change. The performance and robustness of linear model is improved with the inclusion of internal composition inventory controller in the openloop dynamic model of the system. Simulation studies show that a poor design of an internal composition controller gain could decrease the accuracy of reactive distillation linear process model. Therefore, a well tuned controller gain is a necessity for a good performance of the linear process model.