Otimização do fluxo de produtos de uma empresa mineradora

• Ano de defesa: 2009
• Autor(a) principal: Tulio Angelo Machado Toffolo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais
• Palavras-chave em Português: Programação inteira; Pesquisa operacional; Otimização combinatória; Computação
• Link de acesso: https://hdl.handle.net/1843/SLSS-7WMGW5

It is well-known that the mineral extraction industry is very important to Brazil. In this context, the development of technologies that can help these industries is of great relevance. Periodically, the mining companies make decisions related to the production and the transportation of the minerals, considering their logistic and productivity capacities as well as the demands of the market. Such decisions generate a plan of products flow, which consists in determining the flow of the minerals produced in the different mines, from production to distribution, having the goal to minimize the quality gap between the demanded and the final product while optimizing the logistics chain. In this process, the quality of the minerals to be used in the composition of the final product and a complex transportation system that includes mine pipes, long-distance belts, roads, railroads terminals and harbors must be considered. This dissertation proposes algorithms to deal with the Products Flow Problem, which includes some classic problems in the literature, such as the Ore Blending Problem, Transport Planning Problem and Sequence Production Planning Problem. The problem was considered in different planning horizons: annual, quarterly, monthly and daily. A multiobjective model based on goal programming was proposed for the problem, being able to solve only annual and quarterly term instances in acceptable time. To deal with the monthly and daily instances, heuristics algorithms based on relax-and-fix, GRASP and ILS techniques were developed. The different methodologies were validated through tests on instances based on the reality of a major Brazilian mining company.