Construção de modelos matemáticos para processos fermentativos – avaliação de efeitos das condições de cultivo

• Ano de defesa: 2019
• Autor(a) principal: Piazzi, Ana Carolina Ferreira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Engenharia Química; UFSM; Programa de Pós-Graduação em Engenharia Química; Centro de Tecnologia
• Palavras-chave em Português: Modelagem; Microalga; Levedura; Carotenoides; Condições de cultivo; Estimação de parâmetros; Modeling; Microalgae; Yeast; Carotenoids; Culture conditions; Parameter estimation; CNPQ::ENGENHARIAS::ENGENHARIA QUIMICA
• Link de acesso: http://repositorio.ufsm.br/handle/1/19696

The mathematical modeling of fermentative processes is presented as a tool for improving and understanding of such processes, which has the main objective to predict predicting the dynamics of the process. Besides, the models allow organizing the information about the process in a simultaneously and organized way. However, there is a lack of specific models to describe and predict fermentations through the various modifications made to enhance the process performance. In the present work, the modeling and simulation of 4 case studies (3 of them from previous works) were developed and analyzed to predict the growth of microorganisms when culture conditions influence part of the process. For this purpose, the influence of the application of the magnetic field and the illuminance on the growth of the microalgae Spirulina sp. was evaluated. For the Saccharomyces cerevisae yeast, the influence of the application of magnetic field, agitation and aeration on its growth, substrate consumption and, glutathione yields were evaluated and, for yeast Phaffia rhodozyma cultivated without the influence from external factors, its growth, substrate consumption, and carotenoid yield was evaluated. The estimation of the parameters of the model equations was performed by a hybrid combination of the optimization particle swarm and the nonlinear least-squares algorithm. Experimental data were used to validate the models and demonstrate their accuracy and reliability. The growth model of Spirulina sp., which takes into account the influence of the magnetic field and illuminance, presented a higher predictive power when compared to the Verhulst model, presenting the best fit for the growth curve of this microorganism. The use of artificial neural networks to predict Saccharomyces cerevisae growth by application of the magnetic field, the agitation and the aeration showed a high capacity of predicting the experimental data, being able to entirely evaluate the influence of these variables in the process. The proposed model to evaluate the interaction between the microorganism and the substrate proved to be effective to predict and describe the substrate consumption for submerged fermentation, when compared to the Pirt equation, demonstrating that the microorganism/substrate interaction needs investigation. Phaffia rhodozyma yeast growth as well as substrate consumption were better described and predicted using the Contois equation and the proposed model for substrate consumption, respectively. Models for bioproducts such as carotenoids produced by Phaffia rhodozyma need further study to improve their predictive capacity. Further, the results also demonstrated that the development of models that consider the effects of culture conditions is of utmost importance to describe more accurately the real processes.