Metodologia de projeto de malha de controle utilizando a técnica SVD modificada aplicada à destilação extrativa.

Abstract

In order for an industrial plant to operate in order to reduce losses, ensure compliance with specifications and safely, it is necessary to use control systems capable of driving process variables to set values, known as set points. One of the most important steps in defining a control system applied to industrial processes is to choose the control structure, where the definition of Process Variable (PV) pairs with Manipulated Variable (MV) is of fundamental importance. The Singular Value Decomposition technique proposed by Moore (1986) aims to determine the PV / MV pairing, in which the control variables respond with greater intensity to the stimuli applied by the manipulated variables of this process, through a linear transformation applied to the gain matrix. static of the process. However, this technique uses the static gain matrix to evaluate the system direction and, therefore, to define the pairing, without considering the information about the process dynamics. Thus, aiming to use the dynamic information as a component part of the definition of the pairing between the variables, this work intends to insert in the classical Singular Value Decomposition technique the information that contemplates the dynamic contribution to the definition of the pairs of variables, resulting in the Dynamic technique. Singular Value Decomposition. The dynamic information parameter considered in this study was the relative time ratio, corresponding to the time required for 95% of process variable modification after the step disorder in MV. The technique proposed in this work was evaluated for the design of controllers applied to the extractive distillation process model for the production of anhydrous ethanol from the azeotrope point. It was found that the control system using the pairings provided by the D-SVD technique had a superior dynamic performance when compared to the control system provided by the classical SVD technique and, therefore, the technique proposed in this work made a significant contribution regarding the controller projects with the insertion of dynamic process information.