Horizontal collaboration in humanitarian logistics can be highly beneficial through pooling and sharing resources, but it can be costly and difficult to create and operate. Moreover, communities may lack funds to establish it and often depend on donors for the initial investment. Integrating the logistical and financial functions within a catastrophic risk management framework addresses these challenges. The role of such a framework is to determine how to integrate these functions together, how to pool risk jointly, and what are the most effective cost and benefit sharing mechanisms.

In this talk, we present how the catastrophe insurance literature, which is rich in sharing mechanisms, can be integrated in a framework for horizontal collaboration for the prepositioning of strategic reserves. The framework consists of a risk-averse umbrella insurer offering multi-year insurance contracts to a portfolio of risk-averse policyholders wanting to collaborate and pool their risks against natural disasters. It encompasses four components, two operational and two financial. The operational functions consist of 1) planning the prepositioning network in preparedness for incoming insurance claims, and 2) insurance contracts design (setting coverage deductibles and limits of policyholders, and providing insurance coverage to the claims in the emergency response phase). The financial functions consist of 3) ensuring the insurer's solvency by efficiently managing its capital, and 4) allocating yearly premiums among policyholders.

We model the framework as a very large-scale non-linear multi-stage stochastic program, and present an exact method to solve it based on a tailor-designed Benders decomposition algorithm. We study the case of Caribbean countries establishing a horizontal collaboration for hurricane preparedness and draw managerial insights from the case study.

Wildfires pose a significant threat to both critical infrastructure and residential areas, especially in regions with a high proportion of wildland-urban interface (WUI) and prolonged fire seasons due to climate change. This study explores the application of Critical Node Detection (CND) to inform strategic landscape management plans aimed at limiting fire spread and intensity while considering the restrictive costs associated with hard-to-access areas.
Our project focuses on a landscape with intensive military training, where ignitions are frequent. Our objective was to identify optimal fuel treatment locations to restrict wildfires from escaping the base and impacting neighboring communities. We integrated CND with fire-growth modeling and structural loss rate modeling to comprehensively assess wildland fire risk. Our preliminary results suggest which particular strategies to adopt according to the desired objective of the mitigation and the particular fire profiles, comparing their costs and relative efficacy.
Our approach encompasses the assessment of fire hazards, impacts, and mitigation strategies, offering valuable insights for proactive wildfire management in comparable settings. This interdisciplinary framework serves as a robust tool for safeguarding communities and bridging the realms of fire science, land management, and military operations, thereby enhancing overall wildfire risk management effectiveness.

Hunger and famine pose great risks to global health and are the second in the UN's seventeen sustainable development goals to be achieved by 2030; however, they prove to be quite challenging to eradicate or alleviate. To mitigate their devastating impact, each year, aid agencies deliver tons of food commodities to populations in need. However, the delivery of food commodities is often expensive, and because of the complex intertwining factors shaping food security, it is very difficult to definitively predict future outcomes and demand for food aid. Without a timely identification of vulnerable populations, food aid often fails to arrive in the right place at the right time. We develop a stochastic optimization framework to assess the value of information: our analytical results quantify the advantages of incorporating food insecurity predictions in decision-making. Such predictions facilitate informed prepositioning decisions. Since acquiring relevant data might require heavy investments, we also analyze the delicate balance between allocating the limited available budget to the prepositioning of food commodities and investing in the acquisition of accurate data.

We present a decision-support tool for flood preparedness developed through a collaboration with the World Bank in Haiti. The shelter network design and evacuation planning problem is formulated as a bi-level optimization model in which the leader is the decision-making authority, and the followers are the evacuees whose reactions to upper-level decisions is modeled. The model considers the inclusion of evacuees' behavior and the temporal evolution of flood disaster over time. The model is tested using socio-demographic, regional characteristics, attributes of hazard, and GIS data.