An Efficient MILP Approach for the Short-Term Planning of Refined Products Blending and Distribution Operations

AGUSTÍN MONTAGNA, DIEGO C. CAFARO Y JAIME CERDÁ

Computer Aided Chemical Engineering

Elsevier

Lugar: Amsterdam; Año: 2013; p. 649 - 654

Among the refined petroleum products obtained in an oil refinery, gasoline contributes with more than half of the total profits. This work is focused on the last stage of the gasoline processing in which different ?cuts? are blended together with additives and oil fractions in order to fulfill customers? demands with the right product at the right time. The short-term planning of these operations becomes a complex task due to the need of managing a large number of orders, delivery time-windows, blending equipments, intermediate and final product tanks, quality specifications, and variable product recipes. Early works have adopted mixed-integer linear programming (MILP) models, based on unit-specific time slots, requiring extremely large CPU times to obtain the optimal solution. This work introduces an alternative formulation using the General Precedence sequencing concept for the integrated short-term planning of gasoline blending and distribution operations in order to minimize production costs, inventory levels and quality giveaways; improve customer satisfaction and avoid ship demurrage. Six industry-scale examples have been solved. Compared to previous works, results show similar solutions with reductions in CPU times of up to three orders of magnitude.