An Integrated Data-Driven Modeling and Gas Emission Constraints for Large-Scale Refinery Production Planning Framework

The accuracy of secondary unit models including fluidized catalytic cracking, catalytic reforming and delayed coking is essential for the refinery production planning. Through the reaction kinetics, it is clear well that feedstock composition directly determines the product yield of units. Therefore, a hybrid modeling framework was developed in this work, where the feed true boiling point temperature, hydrocarbon group composition, and critical properties were introduced to characterize the lumped kinetic composition of the three secondary units. These indicators were further employed to predict unit product yield, which were embedded in the hybrid models developed using a data-driven approach. This study proposed a production planning framework that coupled data-driven approach and gas emission constraints to improve profitability and reduce emission. Three refinery production planning examples were taken for illustration. The optimal scheme obtained with the proposed hybrid model was compared with that from method based on the fixed yield production planning models. The results demonstrated that the production planning based on the hybrid yield model could achieve an improvement of the refinery profitability by 21.75% and 22.88%. The production planning coupled gas emission constraints optimization scheme achieved reduction in CO2, SO2 and NOx emissions by 16.59%, 16.67% and 16.67% respectively.