Optimal allocation of monitoring stations aiming at an early detection of intentional contamination of water supply systems

A stochastic approach is proposed, capable of specifying the optimal allocation of different sets of monitoring stations aiming at an early detection of the intentional contamination of a water distribution network. The approach is based on the use of the Monte Carlo technique for the generation of different Users’ Water Demand Scenarios (U’WDS), which are variable both within the day and from one day to another. Once the generation of scenarios is performed, a direct analysis of the hydraulic behaviour of the distribution network by a capable hydraulic model allows, for each U’WDS, the evaluation of the hydraulic flow characteristics, such as the mean flow velocities through each link of the network. Given an intentionally contaminated input node, the hydraulic characteristics of the flow calculated for each U’WDS allow the evaluation, by means of a lagrangian transport model, of the arrival time of substance transported at each node of the network. Then, by varying the node chosen as the possible contamination source, it is possible to evaluate, for each node of the network, the arrival times from all potential intrusion nodes. For each operating condition scenario, the locations of monitoring stations are chosen at nodes which maximize the number of upstream nodes characterized by arrival times smaller than a pre-assigned value (Early Warning Time, EWT). Finally, the optimal set of monitoring stations is chosen by a statistical analysis of the results obtained with reference to the generated scenarios.