A two-stage stochastic approach to location, production, and inventory planning with demand uncertainty

This study is motivated by the need to restructure Ethiopia’s fertilizer distribution network following the establishment of a fertilizer plant in the country, a partnership between the Ethiopian government and OCP Africa, the state-owned Moroccan phosphate producer. We propose a two-stage stochastic program with recourse to design the new fertilizer distribution network under demand uncertainty. The model considers two types of warehouses, owned and leased, and determines which of these two alternatives to use at which locations. Additionally, it determines how much fertilizer to produce and how best to route fertilizers from the plant to the customers. The goal is to minimize the investment costs and the expected distribution costs subject to operational constraints. Computational results based on real data indicate that a mixed strategy, which combines owned and leased warehouses along with various truck sizes, results in the lowest overall costs. Strategies that rely solely on small trucks or exclusively on spot warehouses can lead to cost increases of up to 14% and 7%, respectively. Sensitivity analysis further confirms that the strategic decisions remain consistent even when demand or scenario probabilities change, demonstrating the model’s robustness. These findings provide practical guidance regarding long-term infrastructure investments and the balance between owned and leased storage.