Hierarchical decomposition heuristic for crude oil scheduling: a SINOPEC case

This work addresses the large-scale crude oil scheduling problem of China’s SINOPEC Maoming refinery importing various types of crude oil from two terminals via long-distance pipelines. A hierarchical decomposition approach is employed to construct a two-stage MILP model of the practical refinery operations. In the upper level model, storage and charging tanks are aggregated to determine the inflows and outflows among the two terminals and the refinery. Detailed loading and unloading operations located on storage and charging tanks are solved in the lower level sub-models. In order to further improve the computational efficiency, we develop a practical rule-based tank selection strategy to efficiently obtain feasible detailed schedules. While state-of-the-art commercial solvers cannot obtain a feasible solution of the relaxed monolithic MILP model within reasonable time limit, the decomposition heuristic can obtain near-optimal and flexible scheduling results with much less computational effort compared to the one obtained manually.