Jones Seminar: Enhancing the Effectiveness and Responsiveness of Humanitarian Food Aid Delivery Services Through Data-Driven Optimization

Humanitarian organizations must plan and operate two distinct aid delivery services on a shared network for responding to both ongoing (OG) demand in a reliable and cost-efficient way and sudden-onset (SO) emergencies rapidly. We study the joint planning problem and present scenario-informed, data-driven delivery service recommendations that preserve managerial interpretability. We develop an integrated two-stage model to jointly plan OG and SO services over a shared network under common operational constraints. In the first stage, the model optimizes over which warehouses to open and SO-dedicated safety stock levels; in the second stage, operations over the network are decided for serving the known OG demand and the SO requests revealed over time. To support planning under limited historical data, a deep-learning pipeline labels historical demand as OG or SO and generates realistic, diverse demand scenarios that preserve temporal and spatial structure while introducing controlled variation. Rather than solving a single large stochastic program, we solve the model per scenario to build a reusable decision library and reveal how first-stage designs vary with scenario conditions.

From this library, we derive (i) a baseline recommendation using only historical scenarios and (ii) a pattern-based recommendation based on the frequency of first-stage decision patterns over all scenarios generated. Using a pre-specified demand-feature set, we train a feature-to-decision recommender that maps scenario features to representative first-stage designs, enabling fast, interpretable, scenario-specific guidance without re-solving the optimization model. Computational experiments show that, relative to the historical baseline, pattern-based recommendation improves per unit cost efficiency and increases early fulfillment and service coverage. Feature importance analysis from feature-to-decision recommender indicates that total sudden-onset demand and the budget constraints are the primary decision drivers, with commodity compositions, seasonality, and geography refining the recommended design.

Hosted by Professor Wesley Marrero.

About the Speaker(s)

Özlem Ergun
Professor of Mechanical and Industrial Engineering, Northeastern U

Özlem Ergun is a College of Engineering Distinguished Professor in Mechanical and Industrial Engineering at Northeastern University. Ergun has applied her work on network design, management, and resilience to problems arising in many critical systems including transportation, pharmaceuticals, and healthcare. She has worked with organizations that respond to emergencies and humanitarian crises around the world, including USAID, UN WFP, UNHCR, IFRC, OXFAM America, CARE USA, FEMA, USACE, CDC, AFCEMA, and MedShare International. Ergun also served as a member of the National Academies Committee on Building Adaptable and Resilient Supply Chains after Hurricanes Harvey, Irma, and Maria and the National Academies Committee on Security of America's Medical Supply Chain. She received a BS in operations research and industrial engineering from Cornell and a PhD in operations research from MIT.

Contact

For more information, contact Joyce Xiao at joyce.xiao@dartmouth.edu .