Adjoint-Based Probabilistic Source Characterization in Water-Distribution Systems with Transient Flows and Imperfect Sensors

Abstract

If a contamination event occurs in a water distribution system, sensors in the network may observe water quality changes. The data from these sensors can be used to identify the source of the contamination. The sensors can be binary sensors that record the presence or absence of contamination, fuzzy sensors that measure concentration within a set range, or perfect sensors that measure the exact concentration within the bounds of measurement uncertainty. This work presents an adjoint-based probabilistic approach for identifying the source node, source release time, and source strength for an instantaneous release of contamination based on sensor observations and known system hydraulics. In the adjoint approach, information is propagated upgradient from the sensors to the possible source nodes.