Stochastic last mile relief network design with resource reallocation

We study a last mile distribution network design problem for situations where there exist local distribution centers (LDCs) with prepositioned supplies. Given the information on the existing pre-disaster relief network, the problem determines the locations and capacities of LDCs and points of distribution in the relief network, while capturing the uncertain aspects of the post-disaster environment. We introduce a new accessibility metric and develop a two-stage stochastic programming model that would allow more accessible and equitable distribution of relief supplies. Since solving the proposed stochastic optimization model is computationally challenging, we employ a scenario decomposition-based branch-and-cut algorithm. We perform a computational study—based on the real-world data from the 2011 Van earthquake in Turkey—to provide insights about the model and demonstrate the effectiveness of the solution method.