A new robust optimization model for relief logistics planning under uncertainty: a real-case study

As natural disasters often cause catastrophic damages, disaster management must be considered. Relief logistics planning in disaster management is a dynamic process that includes deliberate processes taken before and after a disaster. This paper presents a robust optimization model for relief logistics planning covering before and after disaster actions to minimize the total cost. In this model, demand and traveling time between nodes are non-deterministic parameters, and scenario-based robust optimization is applied to handle uncertainty. The model is an integrated multi-echelon, multi-period, and multi-commodity vehicle routing problem considering the fair distribution of goods and services in a way that social costs are minimized. Vehicles can enter nodes and exit them several times. In this paper, a genetic algorithm (GA) is considered the solution approach. To estimate the GA parameters, an experimental design based on the Taguchi method is considered. A roulette wheel method is applied to select a child from parents for the next generation. Single-point crossover and swap mutation operators are used to generate the new child. Test problems in different scales are generated and solved using GAMS and GA. The results indicate that GA solutions have 3.75% gaps in average with optimal solutions. Finally, the model is applied with the data of a real case (Tehran). The results also indicate that it is better to consider the appropriate capacity for supply centers and distributors and select vehicles.