Modelagem e otimização para programação de produção em processos químicos em batelada de curta duração

• Ano de defesa: 2021
• Autor(a) principal: Teixeira, Thiago de Oliveira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Engenharia Química
• Palavras-chave em Português: Otimização da produção; Processos em batelada; State-Task Network; GAMS; CNPQ::ENGENHARIAS::ENGENHARIA QUIMICA::PROCESSOS INDUSTRIAIS DE ENGENHARIA QUIMICA; Engenharia Química; Controle de Estoque; Análises de sistemas; Fluxograma
• Link de acesso: https://repositorio.ufu.br/handle/123456789/35258; http://doi.org/10.14393/ufu.di.2022.5314

Reduction in operating costs and reduction of waste are challenges that operational research is able to assist in solving. Batch processes are present in several pharmaceutical and chemical industries. In this present work, an optimization model was developed for short batch processes capable of improving the efficiency of this type of industry. This model uses a common continuous time grid in a State-Task Network representation. The proposed model is suitable for different storage policies, such as unlimited, limited and nonexistent stocks. The main difference of this work for others present in the literature is the non-use of BIG-M restrictions, which brought benefits in relation to similar works. The model was subjected to performance tests in the GAMS software using the CPLEX package and compared with other models already established in the literature, it was also compared with models from the research group PPGEQ / UFU, achieving promising results standing out when the stock limits because the number of variables and restrictions does not differ for when there are unlimited stocks, and when there is a need for utilities as it has the best statistical data, with the least amount of variables and restrictions, in addition to the best relaxing solution of all the tested models, indicating it is very suitable for that situation. The computational and statistical results were presented as tables, and the optimal solutions in a Gantt chart.