MA12 - Humanitarian Logistics

As disasters have become more frequent and severe, their impacts on households and homes have intensified accordingly. Addressing post-disaster housing remains a persistent challenge facing emergency management community. This research addresses the critical question of how governments can optimally plan housing resources to accommodate displaced populations effectively throughout the post-disaster recovery period. The study develops an optimization framework to maximize the expected well-being of displaced households throughout the designated temporary housing period. Central to this research is the incorporation of households' utility function that quantify preferences based on displacement distance, housing type attributes, and temporal considerations across recovery phases, recognizing that housing preferences evolve over time. Through a case study of Louisiana's hurricane response, we assess the tool's performance across multiple analytical dimensions.

This study proposes a stochastic selective routing problem for planning field visits for post-disaster needs assessment surveys, motivated by a project deployed in Türkiye after the 2023 earthquakes. Two challenges are addressed: (1) selecting a representative sample of communities, and (2) designing efficient routes under uncertainty in key informant availability. Representativeness is ensured using k-means clustering based on socio-demographic, geographic, and economic attributes, with proportional cluster coverage. Uncertainty is handled through a novel main–backup assignment logic. We propose a two-phase matheuristic combining route enumeration strategies with a set-partitioning model to jointly determine sampling and a priori routing decisions. Each route is represented as a decision tree, enabling practitioners to follow a predefined yet adaptive plan. The approach is evaluated on a dataset covering 11 provinces in Türkiye and supports field decision-making. Results show that performance depends on cluster structure and informant availability.

Wildfires pose significant risks to human life, property, and ecosystems, and their increasing frequency and severity are placing growing pressure on emergency response systems. Effective response relies on the prepositioning of resources to enable rapid intervention; however, this task is challenging due to uncertainty in the timing, location, and intensity of fire events. We collaborate with SOPFEU (Société de protection des forêts contre le feu), the agency responsible for wildfire management in Quebec, Canada, to provide a decision-support tool for designing a resource prepositioning network to facilitate timely response. We develop a two-stage stochastic optimization model to determine prepositioned resource quantities and locations. The proposed approach is evaluated using realistic wildfire scenarios.

Many patients in northern Ontario rely on air ambulance services for transport to hospitals; however, reaching these patients is challenging due to the uneven distribution of air ambulance bases across the province. To address this challenge, we study the Air Ambulance Location Problem, which consists of patients, aircraft, and base locations, with patients characterized by their pickup locations. Our goal is to find the best possible base locations for aircraft, assign aircraft to bases, and decide which aircraft and base provide service to each patient. We propose several problem definitions at varying levels of abstraction and detail, each including different variations. Given Ontario's geographic and demographic diversity, we examine several scenarios to reduce regional disparities in healthcare access. We develop several optimization models that are computationally efficient and easily adaptable to future changes. Using five years of call data from Ornge, Ontario's air ambulance service, we evaluate the performance of these models.