Application of Soft Sensor for Enhancing the Performance of Reactive Distillation Column

Abstract

ABSTRACT newline newline newline newlineReactive distillation is a combination of reaction and separation in a single column. It is beneficial for the reversible reactions in which equilibrium limits the conversion. This process is economical as two processes occurring simultaneously. But this combination makes it highly nonlinear and interacting in nature. Composition measurement using online composition analyzers is not economical as well as it requires frequent calibrations. Thus measurement and control of product composition of reactive distillation is a challenging task. Performance of conventional controllers for such a complex and highly nonlinear process is sluggish and produces large error. The use of soft computing techniques can efficiently overcome these difficulties. In this research work, neural network based soft sensor as intelligent controllers are designed and implemented to enhance the performance of reactive distillation column. The desired product composition is measured and controlled using these soft sensors. Modified neural network using steepest descent method and fuzzy competitive learning based counter propagation neural network are developed for dual reactive distillation column and methyl acetate reactive distillation column. Dynamical mathematical models are developed and considered for simulation study. Simulation is performed in MATLAB environment using MATLAB (2013b). newlineFeed Forward Back Propagation Neural Network (FBPNN) which is modified using steepest descent method is designed and implemented for estimating the product composition of dual reactive distillation column. Transesterification and etherification reaction is performed in dual reactive distillation column to produce Biodiesel and Tert-Amyl-Ethyl-Ether (TAEE) for biodiesel blending process. TAEE is used as novel fuel oxygenate for blending biodiesel in place of commercial diesel. Instead of direct measurement of composition, we have utilized information of secondary process variable for measurement of product composition of reactive distillation. T