Rethinking Public Transit as a Means of Distributing Social Aid: A Learning-Based Decomposition Approach

Problem definition: Social aid programs require regular distribution of aid in non-disaster conditions to beneficiaries living in urban areas where poverty and hunger concentrate. These programs are often led by local governments or non-profit organizations that suffer from high transportation costs and operational difficulties while delivering donations to people in need. In this study, we investigate the use of excess capacity of the public transportation network (PTN) to support aid distribution in large urban areas. In particular, we propose forming a two-echelon delivery network where PTN serves as a backbone network for efficiently transporting donation boxes between the strategically located loading and distribution points (LDPs), and in the second echelon, voluntary couriers handle the short-distance transfers from the LDPs to beneficiaries' doors. The strict dependence of the operations in the two echelons requires jointly deciding the transfers in both echelons, presenting a highly complex operational problem for system management.

Methodology/results: To address this challenge, we use an innovative learning-based decomposition method and investigate the theoretical properties of the solutions it produces.

Our numerical experiments, based on real-world data, demonstrate the merits of the proposed distribution system and the efficiency of the solution methodology, which can be used in addressing similar two-echelon distribution problems arising in different applications.

Managerial implications: The efficacy of the proposed distribution method is evident in its ability to generate significant utility with a low supply-demand ratio when all donations are available at the start of the operation time window. Additionally, the decomposition-based approach demonstrates efficiency, achieving notable utility with a small number of public transit trips per day. Finally, the critical role of volunteer courier availability emerges as a key factor influencing the success of the distribution system.