Controle plantwide do processo Tennessee Eastman empregando lógica fuzzy

Abstract

When the term Plantwide Control is used, it is to identify how variables can be manipulated, what associations can be made between them and how to control the plant as a whole in a way that is close to the operational optimum. These questions are important and appear in the daily life of a process engineer, since for most control theories it is assumed that the effective structure of sensors and actuators is provided and the operations involved in an industrial plant can be analyzed individually, without considering the interaction between them, which is not always true in a chemical industrial plant. The study of Plantwide control is of great importance mainly in the chemical transformation industry, where the presence of interdependent phenomena can cause problems in both production and industrial safety. These phenomena must be taken into account when aiming to create a safe control loop that avoids instabilities, as is the case of the snowball effect caused by the increase in temperature impacting the rate of chemical reaction of exothermic reactions. The present work seeks to replace classic PID control strategies used in the literature for controllers based on fuzzy logic. This modification aims to improve parameters of the industrial process, since the Tennessee Eastman Process (TEP) have different nonlinearities and has the potential for numerous control strategies. After that, from the plantwide control methodology, the process performance index will be evaluated. The designed controllers were build using the fuzzy Mamdani method and controlled the process more effectively, decreasing the pressure variability by 42.4% and reducing the amount of product loss in the purge by 8.2% when compared to classical PID controllers commonly used in the literature. For this comparison, several performance indicators will be applied, such as the IAE, the control effort and the quantification of product lost in the purge.