Predição e controle da qualidade no processo de destilação em batelada com uso de sensor virtual

• Ano de defesa: 2017
• Autor(a) principal: Mota, Liliane Oliveira
• Orientador(a): Oliveira Júnior, Antonio Martins de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Programa de Pós-Graduação: Pós-Graduação em Engenharia Química
• Palavras-chave em Português: Engenharia Química; Detectores; Destilação em batelada; Misturas (Química); Sensor virtual; Mistura binária
• Palavras-chave em Inglês: Soft sensor; Batch distillation; Binary mixture
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA QUIMICA
• Link de acesso: http://ri.ufs.br/jspui/handle/riufs/17109

Over the years, scientific advance has enabled the technological development of society as a whole, and in the industrial processes has not been different. Distillation, a widely used unit operation, still has studies focused on monitoring and optimization because there is a need to improve plant efficiency due to the increased competitiveness. Therefore, research related to the simulation area has increased its importance, and the control of product composition in the distillation processes becomes essential for the success of the operation. With this, the soft sensors are presented as a viable alternative, because they allow the reduction of cost and associates the simplicity with the precision of the results. Thus, a soft sensor was developed and validated based on thermodynamic requirements to predict the ethanol composition in a batch distillation process of a binary mixture. For this, experiments were carried out to understand the quasi-stationary distillation process, with water and absolute ethyl alcohol, where an experimental design was done to evaluate the influence of the initial solution concentration (20% and 30%) and the reflux ratio (total and 1: 2) on the alcohol content of the product. These tests were also used for a pre-calibration of the soft sensor. In addition, batch experiments with water and absolute ethyl alcohol in different initial concentrations of alcohol (2% to 30%) and reflux 1: 2 were performed, and later used in the calibration of the sensor. The sensor validation was done with data from jabuticaba wort, and the results showed that the sensor was able to predict the experimental behavior.