Agent-Based Modeling for Planning Emergency Response to Contamination Emergencies in Water Utilities

Texas A&M Engineering Experiment Station, College Station TX

Investigators

Abstract

This research project bridges the disciplines of hydraulic engineering and social science to provide answers to the questions a water utility manager faces in a contamination event, "What do I need to know and what tools can I use to effectively minimize sickness and deaths from this event?" Approximately 90 percent of the U.S. population receives water from one of 170,000 public water utilities. Despite the ubiquity of this infrastructure and its importance for public health, many aspects of emergency management for water systems remain at an undeveloped stage. Accidental contaminants in the water distribution system erode public trust and result in sickness and death, and malicious contamination may cause even more destructive results. As a contamination event unfolds, water utility managers need to make decisions in an extraordinarily difficult environment: information that they receive is incomplete and subject to great error; the water distribution system is dynamic and extremely complex; and consumer reactions are uncertain and affect the operation of the water system, making the dynamics of the system even more complex and uncertain. Throughout an event, consumers may reduce their water consumption based on official notices, such as boil water notices, or informal (peer-initiated) warnings, and the actions of consumers in response to these alerts will change hydraulic conditions in the network. Thus, any further decisions by water utility managers should take into account the fluctuations of the contaminant plume. The project addresses several unknown facets of the water distribution threat management problem. This research will collect data needed to empirically model consumers' decision-making process and compliance rates to protective action recommendations. Empirical models of consumer behavior and hydraulic simulation of water distribution networks will be coupled through an agent-based modeling (ABM) framework. The simulation framework will be coupled with optimization algorithms to evaluate large sets of response options and develop effective emergency response plans that account for uncertainty and the dynamic nature of population-infrastructure interactions. Project results will be used to develop a protocol for responding to contamination events for municipalities and water utilities. While intentional contamination is a realistic threat to U.S. water utility infrastructure, little research literature exists on the complex interactions of emergency warnings, public response, and management effectiveness for water utilities. The Environmental Protection Agency has published documents to provide some guidance for responding to a water distribution event, but can recommend only very generic actions, as the best response for a utility depends on the hydraulic characteristics of the system, the characteristics of the contaminant release, and the interactions of the public, media, and decision makers. This framework will allow a specific water utility to identify response options for a range of events based on its particular characteristics, thus preparing and equipping public officials and water utility operators to better protect public health. When a utility uses the framework, a set of documents and flowcharts will be produced that can later be used to guide management actions in real-time as a contaminant event unfolds and the characteristics of the event are understood as information is gathered.