A Coalitional Decision Framework for Post-Disaster Interdependent Infrastructure Systems

Abstract

Natural disasters can severely disrupt critical infrastructure systems, including power, water, and transportation networks, which are essential for the functioning of modern cities.  Dynamic decision-making is a key tool for supporting utility and district managers in planning and prioritizing repair actions during the restoration process. When multiple infrastructure managers are involved, these decisions become strategically interdependent, making game theory a suitable analytical framework for modeling their interactions. Non-cooperative game theory models assume managers to be self-interested agents maximizing only their own sector’s recovery, while coalitional game theory models also recognize that sector interdependencies allow for cooperation. Existing approaches model this cooperation at the interdependent component level. We present a coalitional game-theoretic framework with cooperation at the manager level: cross-jurisdictional coalitions whose conveners prioritize repairing interdependent components to indirectly account for component-level interdependencies. Our two-stage framework consists of a coalition formation game and a dynamic resource allocation problem. The water, power, and transportation sectors of each jurisdiction evaluate and join coalitions based on their perceived self-benefit, resulting in a stable coalition partition. Then, each coalition’s convener allocates pooled resources to optimize the group’s recovery over time. Using a hypothetical proof-of-concept scenario inspired by the jurisdictions of Alexandria City, Arlington County, and Fairfax County, we evaluate coalition formation and compare infrastructure resilience against a non-cooperative game-theoretic baseline.