Aplicação de estratégias de controle em coluna de destilação

• Ano de defesa: 2010
• Autor(a) principal: Franchi, Claiton Moro
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual de Maringá; Brasil; Programa de Pós-Graduação em Engenharia Química; UEM; Maringá, PR; Departamento de Engenharia Química
• Palavras-chave em Português: Etanol; Química; Controle de processos; Coluna de destilação; Instrumentação; Identificação; Brasil.; Engenharias; Engenharia Química
• Link de acesso: http://repositorio.uem.br:8080/jspui/handle/1/3654

Distillation is the most widely used separation process in industry. It is a process that consumes a large amount of energy and it is essential to understand its operation properly to apply efficient control techniques, thus obtaining the maximum process performance, as well as an end high purity product, with lower energy consumption. This work aims to develop and implement a data acquisition and control card, along with a data monitoring system. The system has been implemented in a distillation column used as a module for Chemical Engineering undergraduate course teaching at State University of Maringá. To apply the control strategies in the distillation column, was first obtained an approximate process model using identification process techniques with artificial neural networks and classical identification methods ARX, ARMAX, OE, BJ and State Space. To implement the identification techniques, algorithms were developed with MATLABTM software, resulting in a model obtained by identification using artificial neural networks with a mean square error of 0,66% on the input-output set. Simulations were made with obtained model in MATLABTM software, applying conventional control strategies (PID), fuzzy and neuro-fuzzy. The simulation software was used to tune controllers, as well to evaluate the performance when applied disturbances. Neuro-fuzzy controller presents the best results, with smaller response times, overshoot and settling time, as well as lower performance criteria (ITAE, IAE, and ISE) by the disturbance application in the load and set point. The control strategies were implemented in the distillation column with control and data acquisition card and supervisory system, integrated with MATLABTM software by one serial interface. The controllers application in the distillation column, tuned and implemented in MATLABTM software play the role to confirm the better neuro-fuzzy controller performance comparing with PID and fuzzy controllers. The experimental results confirm the process model effectiveness and the controllers tuned via simulation, improving the distillation column operation with an appropriate energy consumption to produce a final product within the required specifications.