Disaster Digital Twins: AI World Models for Predicting Cascading Infrastructure Failures

Abstract: This study investigates how disaster digital twins can be operationalized as AI world models to anticipate cascading infrastructure failures under compound hazards. The central premise is that critical services—power, water, telecommunications, healthcare, and transportation—form a tightly coupled, multi-layer network in which disruptions propagate nonlinearly through physical dependencies, cyber-physical controls, and human operational responses. A unified digital-twin architecture is