Multiperiod Optimization for Production Planning of Petroleum Refineries

Production planning has been the focus of many articles in the mathematical programming literature. In particular, successful operation of oil companies depends on systematic decisions over complex logistics. The present article focuses on the decision-making process involved in petroleum production planning. The proposed model is based on a nonlinear programming formulation that was developed to plan production over a single period. First, the model incorporates multiple planning periods and the selection of different crude oil types. Uncertainty related to petroleum and product prices as well as demand is then included as a set of discrete probabilities. Finally, crude oil handling constraints are added. The resulting models are mixed integer nonlinear programs (MINLP), which were successfully applied to a real-world case at the Petrobras REVAP refinery. Problems of up to 19 time periods and up to five scenarios were solved to optimality.