Integration of the nonlinear distillation cutpoint modeling into the online refinery scheduling optimization

Abstract

For high-performance production scheduling optimization in the crude-oil refinery processing, a fundamental step is to properly determine the yields of the outputs in the crude-oil distillation units. Better predictions on the amounts of distilled products and their properties is the base ground for better control and optimization of the whole refinery network. Although linear (LP) models for cutpoint optimization of the distillates in refinery planning and scheduling optimization have been extensively used in the industry and literature, their inaccuracies may under- or over-predict the quantities and qualities of the final products. However, by using rigorous distillation process considering nonlinear (NLP) models for cutpoint optimization, complications in the solution of the refinery planning and scheduling problems may be imposed. In this research the scope of the crude-oil refinery scheduling optimization problem (covered commonly from the crude-oil unloading and storage up to the distillation straight-run streams) is extended by comprising the entire process-shops and product liftings downstream to the distillation units. In such complex modeling of the refinery production scheduling, cutpoint optimization methods of distilled streams in cascaded towers are addressed, whereby previous formulations for cutpoint optimization are compared with the proposed hybrid model that combines rigorous and surrogate formulations. The integration of the NLP distillation cutpoint modeling into the online scheduling optimization is necessary to reduce the gap between the model production and the actual plant operation. (AU)