Operational Models in Humanitarian and Non-Profit Logistics

This dissertation studies operations research models motivated by applications in non-profit and humanitarian logistics; motivating applications include disaster relief aid distribution and non-disaster food aid. The first part of the dissertation is a review of models in vehicle routing for disaster relief, along with insights into future work from discussions with logistics practitioners. The second part presents a stochastic programming model for coordination of disaster relief agencies. The third part presents a location and vehicle routing model for improving recycling operations for non-disaster food aid.;In the survey of disaster relief vehicle routing models, we find potential areas for future work. Practitioners stated that relief activities immediately after a disaster would be challenging to model due to uncertainty in information and rapidly changing conditions. They identified high levels of ambiguity and frequent changes in availability of delivery vehicles, supplies, and location and quantity of need. Practitioners emphasized that every disaster is unique, as are every relief organization's practices. Models providing guidelines and insights for practitioners should be flexible, general, and able to give conclusions with limited data.;Next, we develop a two-stage stochastic programming model motivated by disaster relief aid with uncertainty. This model groups relief organizations, assigning them to deliver goods to aid recipients with uncertainty in whether recipients can be safely reached. The model is a variant of the generalized assignment problem. We discuss potential solution methods, focusing on variants of Benders' Decomposition-type methods.;Finally, we study a problem motivated by a network of hunger relief agencies. The network identified cardboard recycling as a costly expense, given the quantity of donations. Aggregating recycling across the network may be a cost-saving venture. This leads to a variant of the periodic location routing problem with additional dimensions of choice. Choices involve the set of open depots, the capacity of open depots, and the visit frequency. We highlight the challenges introduced by each dimension of operational choice, examining the relative difficulty introduced with each dimension through a heuristic, which can incorporate additional relaxations of model constraints. We present a case study based on data from the network and develop collaboration guidelines.