Impact of stochastic travel times on the military port selection problem

This research analyzes the impact of stochastic inland and oceanic travel times on port selection for large-scale deployment mobilization. A stochastic mixed-integer programming model seeks to minimize the average arrival time of equipment into theaters. In the first stage, it selects ports to open and assigns installations and ships to open ports. In the second stage, it considers realized travel times and routes equipment to ports and onto sealift ships for overseas transit. Testing applies Sample Average Approximation to a realistic scenario for deployments from the United States to two overseas theaters. Over instances manifesting a range of parametric value combinations, results indicate a critical point of diminishing marginal returns for opening additional ports, as well as relative solution quality for alternative assumptions about the correlation of travel times’ stochasticity. Uncorrelated stochasticity reduces port congestion and enables a 2.3% faster average arrival time. For different port-opening limitations, subsequent analysis identifies up to a 90% degree of solution similarity, indicating robustness for selected port-opening decisions. Testing also identifies that correct assumptions about parametric stochasticity reduce average arrival times by up to 1.2%, indicating the value of validating such information for mobilization planning.