Optimal planning of liquefied natural gas deliveries

• We investigate the problem of designing an annual delivery plan for an LNG producer.
• We describe a novel exact arc-flow formulation.
• The proposed formulation accommodates several practical constraints.
• Computational experiments were carried out on a set of real data.
• Optimal solutions for large-scale instances can be provided in a reasonable CPU time.

We investigate the problem of designing an optimal annual delivery plan for Liquefied Natural Gas (LNG). This problem requires determining the long-term cargo delivery dates and the assignment of vessels to the cargoes while accommodating several constraints, including berth availability, liquefaction terminal inventory, planned maintenance, and bunkering requirements. We describe a novel mixed-integer programming formulation that captures important industry requirements and constraints with the objective of minimizing the vessel fleet size. A peculiar property of the proposed formulation is that it includes a polynomial number of variables and constraints and is, in our experience, computationally tractable for large problem instances using a commercial solver. Extensive computational runs demonstrate the efficacy of the proposed model for real instances provided by a major energy company that involve up to 118 cargoes and a 373-day planning horizon.