The Brazilian energy matrix is changing following the global trend towards cleaner sources of energy. Decrease of domestic fuel oil market combined with the seasonality of asphalt demand implies a reduction of performance indicators. Imports of oil products rose sharply in recent years to meet the growing demand which contributes to the imbalance of Brazil s trade balance. Given this scenario, Petrobras traced challenging targets to mitigate this problem, which among others involves maximizing the use of assets installed. The model developed will support decisions for corrective and preventive actions allowing a good operating performance of a Petrobras refinery, allowing greater production of light oil products (diesel, jet fuel and gasoline) with high added value. A MINLP (mixed integer nonlinear programming) model was developed and applied in the analysis subsystem (upstream, midstream, downstream) without success. Linearization techniques were applied in order to obtain a MILP model that has been successfully used. The solutions obtained in the case study indicate that the search for the minimization of imports of lights derivatives provide a high operating factor for refinery analysis. Among the decision variables evaluated investment in a new Delayed Coking plant (conversion unit) is has the highest marginal value. The work had a broad approach within the supply chain permeating the strategic and tactical decisions in the organization is characterized as an excellent tool to support decisions.