Controle plantwide do processo de tratamento de óleo de um FPSO do pré-sal brasileiro empregando lógica neuro-fuzzy com otimização por algoritmos genéticos e controle override

• Ano de defesa: 2023
• Autor(a) principal: Chaves, José Andersands Flauzino
• Orientador(a): Silva, Gabriel Francisco da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Programa de Pós-Graduação: Pós-Graduação em Engenharia Química
• Palavras-chave em Português: Inteligência artificial; Aprendizado do computador; Machine learning; Processamento em FPSO
• Palavras-chave em Inglês: Artificial intelligence; Machine learning; FPSO processing
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA QUIMICA
• Link de acesso: https://ri.ufs.br/jspui/handle/riufs/17818

With the growing demand for fuels and with the prospect of exhaustion of conventional oil deposits around the world, it is necessary to make oil production viable in the most hostile and complex existing environments. One of these environments is the Brazilian pre-salt, which has several technological challenges for its production, such as the high water depth that covers it and the composition of the produced fluids that contain high levels of N2 and CO2. In parallel with this scenario, we observe the emergence of the 4th industrial revolution, also called Industry 4.0, which is accompanied by the application of artificial intelligence and machine learning as ways to enable industrial production at its maximum efficiency with safety and control. This work proposes the use of advanced automation and process control techniques through: Plantwide Control, Fuzzy Logic, Expert Systems, Neural Networks, Genetic Algorithms, Gain Scheduling Control and Override Control. First, a non-linear and concentrated parameter modeling of the oil treatment process of an FPSO (Floating, Production, Storage and Offloading) that currently operates in the Brazilian pre-salt by Petrobras was elaborated, using the Simulink® of Matlab®, validated from the comparison with the real data of the process plant and its design. Then, the project was made and the control loops of the modeled and simulated equipment were implemented using the Plantwide control technique to evaluate the most productive way for the process to be controlled based on the ways of controlling the plant's power supply, its recycles and restrictions. imposed by the process output from neuro-fuzzy logic, expert system and override control. In addition, it was proposed the automation and control of the opening of the chokes of the wells via fuzzy logic and neural networks, and the automation of the setpoint of the separators parameterized by a study carried out in genetic algorithms with an expert system. With this work, it was possible to find a way in which the plant operates with greater stability and profitability, with a gain in production and a significant increase in revenue for the oil production process of the FPSO under analysis, thus obtaining a much better result. than originally planned. As a future perspective is the application of this technology also for the treatment of the gas and water of the platform and the application of computational fluid dynamics for the evolution of the created systems.