CONTROL OF BATCH DISTILLATION COLUMN USING FUZZY & CONVENTIONAL CONTROL

Abstract

Distillation is most important separation technique usually adopted in the chemical refinery and petro chemical industries. In the distillation process, getting the specified product composition is of utmost importance. Control of a distillation column is a complex issue because of the process disturbances in feed input and reflux flow, modeling error, strong coupling interaction between outputs and nonlinearity of the system. Various control schemes of distillation columns are reported from the literature. Various mathematical models of distillation columns with varied complexities are discussed. For the simulation study and for the application of the proposed control strategies, classical Wood and Berry model and the rigorous model with some assumptions are considered. In the present work, the classic Wood and Berry MIMO transfer function model of binary distillation column is tested with conventional PI controller and Fuzzy controller. Model predictive control is proposed and applied on the Wood and Berry model. There has been a large number of control studies based on unrealistic columns with no flow dynamics, no model error and no measurement delays. A control scheme, which has given a good performance for the unidentical model can not guarantee same performance when applied to real time distillation control. The appreciable results of model predictive control of simple model encouraged for its application on rigorous model. The rigorous dynamic model of binary batch distillation column and multi-component distillation column are developed. The proposed model predictive control strategy is applied on both the models. The performance of all the controllers for the different models developed has been compared. Simulation results confirm that the model predictive control gives superior performance compared to conventional PI control and Fuzzy control even for nonlinear rigorous dynamic model. In this, Integral Square Error (ISE) and Integral Absolute Error (IAE) are used as comparison indices.