ACCELERATING RESTORATION THROUGH INFORMATION-SHARING: UNDERSTANDING OPERATOR BEHAVIOR FOR IMPROVED MANAGEMENT OF INTERDEPENDENT INFRASTRUCTURE

Abstract

This dissertation examines the roles that organizations and individuals play in restoring interdependent infrastructure following disasters through three studies. In the first study, we focus on how operator heuristics affect the collective restoration speed of three interdependent infrastructure (electric power, chilled water, and IT networks). We do this by developing a novel framework that embeds an interdependent infrastructure network within an agent-based model that mimics the decisions and patterns observed of actual operators. The study sheds light on how coordination and information exchange by separate infrastructure parties affect decisions and thus restoration outcomes. In the second study, we examine recovery times and total unmet demand for the same three interconnected infrastructure systems assuming a variable fraction of node removals. The work is decomposed by the extent to which operators share information and coordinate strategies, enabling us to identify at what fraction of network failure does coordination and information sharing become beneficial. Our study indicates that prioritizing restoration based on node centrality produces the speediest recovery. We also show that communication among organizations may improve collective performance by as much as 50%. Our final research project uses a serious game, Breakdown, focused on restoration of interdependent infrastructure to assess whether engineering graduate students gain a deeper appreciation for the complexity of interdependent infrastructure and socio-technical systems more broadly. This is the first serious game designed to emphasize the value of cooperation, communication, and strategy in times of crisis in the field of interdependent infrastructure. As a result of playing Breakdown, graduate students demonstrated statistically significant improvements in engineering decision-making under uncertainty and sociotechnical systems concepts. As a result of this dissertation, the interdependent infrastructure community gains insight into (1) how individual operators' behavior influences the speed at which interdependent infrastructure systems recover; (2) how policies and procedures, like sharing information and cooperating, can help improve outcomes; and (3) the ways to teach graduate engineering students about socio-technical systems effectively. Using an agent-based model simulation, it quantifies the effects of human behavior, communication, and cooperation on recovery outcomes. By using a serious game, Breakdown, it proposes an innovative way to teach graduate engineering students about socio-technical systems.