Agent-based modeling and evolutionary computation for disseminating public advisories about hazardous material emergencies

In the event of a large-scale disaster, an important aspect of humanitarian logistics is the distribution of information or warnings to the affected population. This research develops the problem formulation and solution approach for a specific routing for relief problem, in which warnings should be disseminated to an affected community, using public announcement systems mounted on emergency vehicles. The problem statement is formulated to maximize the number of individuals of a community who are protected. An evolutionary algorithm framework is developed by coupling an agent-based model with a variable-length genetic algorithm to route emergency vehicles. The dynamics of interactions among consumers, emergency vehicles, and the spatiotemporal trajectory of the hazard are simulated using an agent-based modeling approach, and a variable-length genetic algorithm approach selects routes to warn a maximum number of consumers before they are affected by the emergency. The example that is explored in this research is contamination of a water distribution network. A fleet of emergency vehicles is equipped with public address systems and is deployed to warn consumers to stop using contaminated water. The framework is demonstrated for an illustrative virtual city, Mesopolis. The results of the evolutionary algorithm framework are compared with two conventional routing optimization approaches, including a covering tour problem approach and a manual routing approach, for four contamination scenarios. The evolutionary algorithm can be applied to route emergency service vehicles to broadcast information for other emergencies, such as flash flooding, hazardous materials incidents, and severe weather.