Modelagem e controle de unidades de craqueamento catalitico - FCC / Modeling and control of catalytic cracking units - FCCU

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

Catalytic cracking is one of the most studied conversion processes due its importance in the petrochemical industry. In this process high weight molecules are broken into lighter ones, yielding more economically valuable products. The study about the process, its mathematical modeling, and its control systems may guarantee the necessary knowledge needed to reach optimal operational conditions which lead to maximum load conversion. In this context, the present work thesis has as main objectives the research and use of a mathematical modeling for a FCC (Fluid Catalytic Cracking) conversion unit, the study of the system under PID control loop widely used in refineries, structuring of an artificial neural network (ANN) with its use as an internal model of a predictive control system developed for catalytic cracking. Parallel to this, it was developed a computational tool which in a friendly and practical way made possible to carry out simulations needed to accomplish this work. FCCGUI (Fluid Catalytic Cracking Graphical User Interface), name given to this simulator, allows to choose simulation parameters, different control systems (PID, DMC, and MPC-ANN Based), graphical visualization of many different variables, such as manipulated and controlled variables plus valve signals, also it allows for and easy means for disturbing the system while the simulation continues. Through the simulations it was possible to follow the behavior of many process variables under different tested operational conditions, to structure the neural network, and to obtain controlled variables responses for both types of control systems studied. The obtained results were quite satisfactory since the chosen mathematical modeling represents well the system and the developed MPC-ANN controller showed to be efficient to maintain the process stable and close to the set points even after the introduction of disturbances in the process. Taking all these into account, compared to PID controller, MPC-ANN showed to be less oscillatory and to be more effective

ASSUNTO(S)

craqueamento catalitico predictive control controle preditivo redes neurais (computação) catalytic cracking neural networks

ACESSO AO ARTIGO

Documentos Relacionados