Controle de temperatura em reatores de polimerização baseado em modelos simplificados do balanço de energia com estimação de parâmetros em tempo real

Abstract

This work aimed to develop a control strategy through the estimating parameters based on the real-time signal filtering technique with the Kalman Filter, which has potential for the development of more effective control strategies for batch polymerization reactors. This study sought to develop, using the extended Kalman filter, the simultaneous estimation of values of the heat of reaction (Qr) and the overall coefficient of heat transfer and heat exchange area (UA) through the simplified models of the energy balance in reactor and in jacket using only measurements of process temperatures. The stochastic process defined in extended Kalman filter, used the strategy of random walk and was developed to design a feedforward controller, through which the estimates of Qr and UA determines the new setpoint for the slave controller (PI) in a loop cascade. In order to test the control strategy proposed in this dissertation and due to the unavailability of a real reactor polymerization, we used the technique of computer simulation to describe the dynamic behavior of a real reactor. Therefore, we opted to use the benchmark polymerization reactor proposed by Chylla e Haase (1993) to test control strategies, whose mathematical model was validated by the authors in a real polymerization plant. After completion of this study, it was concluded that the control strategy based on the estimation of parameters in real-time is capable of maintaining the reaction temperature properly controlled against various disturbances, such as varying environmental conditions, physical and chemical and parameters barely tuned. The proposed control strategy proved to be robust and general since it is able to estimate process parameters based on simplified models of energy balance, with the only variables measured the temperatures of the reactor and jacket, showing that the controller is suitable for multipurpose applications.