doi:10.1093/imaman/dps005 Operational planning of oil refineries under uncertainty Special issue: Applied Stochastic Optimization GABRIELA P. RIBAS ∗ Center of Excellence in Optimization Solutions (NExO), Industrial Engineering Department, Pontifical Catholic University of Rio de Janeiro, Brazil ∗ Corresponding author: gabriela.ribas@labnexo.com [Received on 29 October 2010; accepted on 11 February 2012] In this paper, we describe a non-linear programming model for operational planning of oil refineries, considering exogenous (external) and endogenous (internal) uncertainties. Three mathematical models based on stochastic programming (two-stage stochastic model) and robust programming (min–max regret model and max–min model) are developed to address these uncertainties. The main purpose of this paper is to address the impact of uncertainty on the operational planning of oil refineries by using different risk profiles. The stochastic approach corresponds to a risk-neutral attitude and optimizes the expected value of the objective function. The robust approach, on the other hand, corresponds to a risk-averse attitude and hedges the decision-maker against the worst values of all possible scenarios, although it does not require the estimation of scenario probabilities. A study based on real data from a Brazilian refinery demonstrates the performance of various approaches. After analysing the oil purchase decisions, we identify a clear relationship between the adopted risk attitude and the quantity and quality of the purchased oil. We also show the strong influence of the product specification constraints on the model decisions. Keywords: optimization under uncertainty; refinery planning; scenario analysis. 1. Introduction Oil refineries are increasingly interested in improving the planning of their operations. One of the major driving factors is the dynamic nature of the refining business. Companies want to assess the potential impact of various refinery shifts on the overall performance, such as the final product specifications, the crude oil composition as well as other operational variations including the available capacity of the refinery. It has also been shown that the integration of new technologies for process operations is an essential profitability factor (Joly et al., 2002). The use of mathematical programming in the planning activities was shown to lead to potential gains of US $10 per ton of refined product, which corresponds to savings of more than 1 million dollars per year in a large refinery (Moro, 2003). However, such a gain is extremely difficult to achieve because of the complexity of oil refining activities. In the literature, many operational planning models have been tested in real refineries around the world. Gao et al. (2008) developed a mixed-integer linear programming (MILP) model to address the production-planning problem of a large-scale fuel oil-lubricant plant in China. The authors considered the choice of operational modes at each processing unit as the main optimization decision of the model. The MILP model proposed by Micheletto et al. (2007) optimizes the operation of a refinery plant in Brazil by considering mass and energy balances, operational mode of each unit and demand satisfaction over multiple periods of time. Moro et al. (1998) also employed their model for studying a refinery in Brazil. They developed a non-linear planning model, which was applied to the particular case of diesel production to maximize the profit of the refinery. c  The authors 2012. Published by Oxford University Press on behalf of the Institute of Mathematics and its Applications. All rights reserved. IMA Journal of Management Mathematics (2012) 23, 397–412 Advance Access publication on 26 April 2012 ,ADRIANA LEIRAS SILVIO HAMACHER AND at PontifÃ-cia Universidade Católica do Rio de Janeiro on November 26, 2012 http://imaman.oxfordjournals.org/ Downloaded from