Development of Soft Concentration Sensor for Esterification Reaction Using Electrochemical Features

ABSTRACT Esters are industrially important products with a broad range of applications from flavoring agents to jet fuels. In this study, butyl butyrate ester is produced in a batch reactor at different operating parameters and electrochemical signatures (Cyclic Voltammetry) were recorded for samples taken at certain intervals. Predictive models that is, Multi linear regression (MLR) and Partial least square (PLS) regression for prediction of unconverted butyric acid/conversion were developed. Molar ratio, temperature, and current peak heights were used as input variables for MLR model while reduction current signal was used as input variables for PLS model. The developed soft sensor is advocating as a replacement for physical concentration sensor. As proof of concept, closed loop simulation of the process using PID controller was undertaken to assess the performance of the soft sensor in regulator mode at different ensemble of operating conditions. In closed loop sensing, disturbance rejection is slightly improved when PLS model was used as an estimator.