Drones for relief logistics under uncertainty after an earthquake

• Published in: European journal of operational research Vol. 310; no. 1; pp. 117 - 132
• Main Authors: Dukkanci, Okan, Koberstein, Achim, Kara, Bahar Y.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2023
• Subjects: Drone delivery; Humanitarian logistics; Relief distribution; Stochastic programming; Uncertainty; Humanitarian logistics; Uncertainty; Relief distribution; Drone delivery; Stochastic programming
• ISSN: 0377-2217; 1872-6860
• DOI: 10.1016/j.ejor.2023.02.038

•We introduce a new relief distribution problem involving trucks and drones.•Our model features an endogenous drone range, a time bound and uncertainty.•We present a novel approach to estimate demand based on earthquake scenarios.•We devise a reformulation and a solution algorithm to solve the model efficiently.•Our model can serve considerably more people compared to a deterministic approach. This study presents a post-disaster delivery problem called the relief distribution problem using drones under uncertainty, in which critical relief items are distributed to disaster victims gathered at assembly points after a disaster, particularly an earthquake. Because roads may be obstructed by debris after an earthquake, drones can be used as the primary transportation mode. As the impact of an earthquake cannot be easily predicted, the demand and road network uncertainties are considered. Additionally, the objective is to minimize the total unsatisfied demand subject to a time-bound constraint on the deliveries, as well as the range and capacity limitations of drones. A two-stage stochastic programming and its deterministic equivalent problem formulations are presented. The scenario decomposition algorithm is implemented as an exact solution approach. To apply this study to real-life applications, a case study is conducted based on the western (European) side of Istanbul, Turkey. The computational results are used to evaluate the performance of the scenario decomposition algorithm and analyze the value of stochasticity and the expected value of perfect information under different parametric settings. We additionally conduct sensitivity analyses by varying the key parameters of the problem, such as the time-bound and capacities of the drones.