Dynamic Inventory Relocation in Disaster Relief^†

This study investigates dynamic inventory relocation to respond proactively the changing relief demand forecasts over time. In particular, we examine how relocate mobile optimally serve nonstationary stochastic at several potential disaster sites. We propose a model using programming (DP) and develop both analytical numerical results regarding optimal policies, minimum cost-to-go function, value of mobility traditional warehouse pre-positioning. Given computational complexity backwards DP algorithm, base state heuristic (BSH) for general problems by exploiting real-world pattern occurrence. For with temporally independent demand, polynomial time exact algorithm based on spatial–temporal graph. spatially design speedup technique implement BSH in The proposed algorithms are further extended consider impact transportation uncertainties. Numerical experiments show that return high-quality decisions only small fraction required an myopic algorithm. applicable any type inventory, facility, or server similar settings.