Liquefied natural gas inventory routing problem under uncertain weather conditions

We study the liquefied natural gas (LNG) production-inventory control and vessel routing problem under disruptive weather conditions. If extreme weather is expected to strike an LNG plant, all planned LNG loading operations should be rescheduled to prevent expected safety accidents. We propose two mathematical optimization models to cope with the potential disruptions. The first model is formulated as a two-stage stochastic mixed integer program to maximize the overall expected revenue while minimizing the cost caused by the uncertain impact of weather disruptions. The second model is a decision maker's preference model that reflects a decision maker's evaluation of risk. This model enables a decision maker to have a 'what-if' analysis by varying the level of preference for risks. The two proposed mathematical models can be reduced to a vehicle routing problem which is an NP-hard combinatorial optimization problem. Therefore, two computational techniques have developed to improve the optimization performance. First, a probing-based preprocessing technique is developed to reduce the solution space by eliminating obvious infeasible or non-optimal solutions. Second, an optional logical inequality is developed to generate an upper bound for the optimal solution only if an LNG carrier visits one or two customers in a single tour. Computational results indicate our proposed models and computational techniques are well suited to solve the problem within a reasonable time.