Avaliação do desempenho e sintonia de controladores de nível

Abstract

Industrial processes are of a dynamic nature, varying continuously in time due to external disturbances or changes in the process and, therefore, the need for automatic control of variables is a reality, mainly due to the complexity increase of the systems installed in last decades. Due to the importance of controllers, this research brings the study of techniques for performance assessment and tuning of level controllers, as well as guides the choice of indexes to be used for the performance assessment of level controllers, according to the control objective, presenting the magnitude of the error obtained when used an inadequate performance assessment index. It addresses level control systems that aim to maintain the level in the reference accurately or minimize the outlet flow variance. It is presented a proposal of methodology for performance assessment and tuning of level controllers through process modeling and solving an optimization problem. The proposed methodology is applied to a self-regulatory process. The effectiveness of the performance assessment index for level controllers is explored, considering different types of perturbation in the inlet flow. Traditional techniques and optimization techniques for controller tuning are compared, being that optimization techniques present better results than traditional techniques. It is presented the uncertainties impact in the model used to obtain the controller parameters through optimization techniques in which, depending on the control objective and the type of input disturbance, there are discrepancies to be considered in relation to the need for controller retuning. The consequences of using inadequate performance assessment index are analyzed. Through the obtained results it was verified that to determine the adequate index for the performance assessment level controllers one must identify correctly the control objective, which does not always occur, since the systems of support to the decision in general provide performance assessment indexes that consider the process variable, but some control meshes may consider the manipulated variable.