Event-Based Scheduling of Crude Oil Transfer in a Port–Pipeline–Refinery Network Using a MILP Model and a Column Generation-Based Branch-and-Price Algorithm

Efficient scheduling of crude oil transfer from marine tankers to refinery distillation units remains a major logistics challenge due to congestion, demurrage, and feedstock variability. This study develops a fully linear event-based mixed-integer linear programming (MILP) framework for integrated tanker–pipeline–refinery scheduling. All inventory, capacity, and delay relationships are linearized within an event-driven time–space network. A dual-stabilized branch-and-price algorithm, initialized with high-quality columns generated by a genetic algorithm, is proposed to solve the large-scale problem efficiently. Computational experiments on 24 real refinery scenarios show that the proposed GA-seeded branch-and-price reduces average CPU time by 97% relative to direct MILP solving, limits optimality gaps below 1.3%, and decreases vessel demurrage and total logistics costs by up to 55% and 7%, respectively. The results demonstrate that the proposed framework provides a scalable and hardware-light optimization tool for refinery scheduling and decision support.