Optimizing alert occurrence in the Cincinnati contamination warning system

Abstract

A system designed for rapid detection of contamination in drinking water distribution systems is a novel concept, and there are no benchmarks against which to evaluate the performance of a contamination warning system (CWS). In Cincinnati, Ohio, during evaluation of the first full‐scale CWS, a primary design objective was to minimize invalid alerts without compromising the system's ability to detect true deviations from acceptable water quality conditions. This objective was important to user acceptance of the system because excessive invalid alerts can undermine user confidence and burden staff members with inconsequential investigations. During the final six months of the two‐year evaluation period, a steady‐state rate of approximately one alert per day requiring investigation was achieved. This alert rate was acceptable to the utility, and nearly all alerts were investigated by appropriate personnel. The primary conclusion drawn from the data presented is that a CWS can be optimized to achieve a reasonable and manageable rate of invalid alerts. Furthermore, the investigation of these invalid alerts required only a modest level of effort (approximately seven hours of staff time per month).