Refinery-wide planning operations under uncertainty via robust optimization approach coupled with global optimization

Abstract The paper focuses on the refinery-wide planning operations under uncertainties in product demand and price via robust optimization framework coupled with global optimization. Industrial/simulation data sets are used to build explicit nonlinear surrogate models for correlating the yield/property of products and the operational conditions of secondary processing units. A large-scale nonconvex mixed integer nonlinear programming (MINLP) model is formulated. The product demand and sale price uncertainties are incorporated into the proposed model via the “interval” and “interval + ellipsoidal” uncertainty set, respectively, to obtain the robust counterpart optimization model. A global optimization framework which combines the enhanced normalized multiparametric disaggregation technique (ENMDT) and the optimality-based bound tightening is developed. The proposed model and the global optimization algorithm are applied to an industrial refinery site with three illustrative cases. The results show the value of formulating nonlinear models for secondary processing units and the benefits from applying ENMDT-based algorithm.