Pilot study on Distillation Column

This book deals with model development and control of a distillation column at Chennai Petroleum Corporation Ltd., Manali, Chennai. Real time data are collected from this 40-tray distillation column. Developing a conventional model for this process is difficult; and even if it were developed, the model would be of very high order and hence complicated. We have developed a Statistical Model, which incorporates techniques like Canonical Variate Analysis (CVA), Singular Value Decomposition (SVD), Akaike Information Criteria (AIC) and Auto Regressive exogenous (ARX). There are hundreds of variables associated with a distillation column. Developing a model involving all these variables would be very complicated. We have employed statistical technique to reduce the number of variables, by choosing only those variables that affects the system significantly. The variables so found to significantly affect the system agrees with the results obtained by computation of Relative Gain Array method. We have also found the order of the process using CVA. State space models have been found for the distillation column. The model so determined is validated conventionally and statistically.