AWWA WSC61381

This paper describes ongoing research to address concern about deliberate attacks on water distribution systems that involve the injection of toxins. The thesis of the research described here is that a combination of conventional sensors and intelligent software can provide reliable, low cost security monitoring for water distribution systems. The types of sensors envisaged are already widely used to measure water chemistry properties, such as specific conductance, pH, and residual chlorine. The capabilities of the intelligent software, called a Decision Support System (DSS), should include automatically learning and remembering "normal" patterns of water quality behavior; detecting unprecedented patterns (anomalies) that could indicate an attack; generating alarms; and making recommendations to operators about what to do. Additional capabilities could include automated recognition (classification) of toxin types, and predicting concentrations and fate and transport within the distribution system. It should be noted that developing and deploying these capabilities would necessarily be achieved without ever risking toxin injection in an actual distribution system. This research involves experiments in a flow loop to collect data about water chemistry responses using combinations of different field grade sensors, toxin types, and toxin concentrations. These data will then be used to develop an intelligent prototype decision support system (DSS) with the above capabilities. A field trial will be conducted in the distribution system in Charleston, South Carolina to test the performance of the sensors and DSS together. The trial will include periods in which fluoride concentration is modulated to simulate an attack. Includes 39 references, figures.