Adjoint-based Probabilistic Characterization of Contaminant Sources in Water Distribution Systems under Realistic Flow and Sampling Conditions

摘要

If a chemical or biological agent is released into a water distribution system, sensors that are installed in the pipe network may detect the contamination as it travels through the system. We developed an adjoint-based backward modeling approach that uses the collected sensor data to characterize the source of contamination in a probabilistic manner. In prior work, we used a travel time probability density function (PDF) conditioned on the sensor measurements to identify the time of release and node of release for an instantaneous source of contamination in a system with steady demands. In this work, we generalize the approach to characterize sources of contamination that occur over a finite duration under transient flow conditions, with imperfect sensor measurements. We demonstrate the method using a hypothetical water distribution system and a hypothetical release scenario.