Otimização do Plano de Compra, Estoque e Consumo de Carvões

Abstract

One of the most important steps for a steel company's competitiveness is the definition of purchasing and consuming its raw materials. Among them are mineral coal and pet coke which correspond to about 30% of the final cost of the slab. At Usiminas, Ipatinga Plant, the annual potential fuel consumption is 2.3 million tons with a purchase budget roughly R$ 1.5 billion. The coal purchasing and consumption planning process at Usiminas is divided into two stages, in which the first defines an aggregate annual demand by a computational tool and the second defines a quarterly purchase of coal and coal consumption by mix independently. The fragmented planning between coal purchase and consumption, without considering all the assumptions in an integrated way, makes this process fragile and does not guarantee a decision making of the best possible scenario. In this work, was developed a mathematical model that integrates the purchase and consumption decisions of coal considering the assumptions of offers and stocks, operational restrictions and coke quality. The model was tested for artificial instances due to demand and coal supply and horizon of analysis sensitivities. The greatest performance was observed for scenarios with a horizon of 30 periods that obtained a mean MIPGAP of 2.37% and mean LINEAR GAP of 4.25% compared to 90 periods that obtained mean MIPGAP of 65.35% and mean LINEAR GAP of 93.68%. A variation of coal demand and offerings had no relevant impact on model performance. The model was applied to a real case study and provide a purchasing and consumption planning of coals that correspond to a potential economy of 5.8 MMR$ when compared to the one done in the analyzed quarter. The indication of the coal mix consumption is feasible and is in line with that already practiced by Usiminas historically. The developed tool shows great applicability as decision support of both purchase and consumption of coals, reducing the time spent on planning with reduction cost potential as it has global vision of the processes.