Modelagem e simulação da fermentação alcoólica em batelada e contínua com múltiplos reatores em série

• Ano de defesa: 2016
• Autor(a) principal: Zanardi, Mayara dos Santos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Química; UFES; Programa de Pós-Graduação em Engenharia Química
• Palavras-chave em Português: Fermentation; Parameter estimation; Nonlinear optimization; Estimação de parâmetros; Otimização não linear; Fermentação; Álcool; Engenharia Química; 66.0
• Link de acesso: http://repositorio.ufes.br/handle/10/7826

Ethanol stands out, compared to fossil fuels, because of its economic and environmental benefits. In Brazil, the second largest world producer, practically all ethanol production is obtained from the fermentation of saccharose (present in sugarcane). Improvements in the fermentation process for ethanol production could make it more competitive. In this context, this study has for objective develop a model for batch fermentation process of literature and a real industrial process that operates in a continuous mode with multiple reactors in series, using kinetic models from literature. The model for each component of interest was developed through transient mass balance on the reactor. The system of ordinary differential equations, resulting from the dynamic model, was integrated by the Runge-Kutta method of 4th order in Matlab® software. The estimation of kinetics parameters was performed by nonlinear optimization of an objective function by Nelder and Mead Simplex method. Two optimization strategies were adopted for the batch process: in the first strategy the µmax parameter is kept fixed and in the second this parameter is estimated. Six different kinetic models were tested. The estimated parameters were sensitive to small changes in experimental data set, moreover, the estimation of the parameter µmax had a strong positive influence on some models. The models Ghose and Tyagi (1979), Jin et al. (1981) and Tosetto (2002) showed the best performance in predicting the behavior of the concentration of the components in the reactor. The parameters of five kinetic models were estimated in the continuous process. The models that consider inhibition by product were better than those without this type of inhibition.