Contamination Event Detection in Water Distribution Systems using a Model-Based Approach

This work describes a model-based approach for contamination event detection in water distribution systems using chlorine measurements. The proposed method considers the known chlorine input injection signals, and uses multiple Monte-Carlo simulations which run in parallel to the real system, in order to compute at each time step, bounds of the expected chlorine concentration at the different chlorine sensing locations. The sensor measurements are then compared with the estimated bounds and according to a certain event logic, an event alarm flag is raised when these bounds are exceeded. The methodology is applied on a realistic benchmark network, taking into account uncertainties in the hydraulic dynamics. (C) 2014 Published by Elsevier Ltd.