Quality Detection Sensors Location in Water Distribution Networks for Uncertain Water Demand Conditions

Accidental or intentional entry of pollution into water distribution networks as one of the vital life lines of any country, causes extensive damage in society. Early detection and warning of contaminant injection in the water distribution network is of significant importance. In order to optimize the location of quality sensors, this study presents a new algorithm based on uncertainty in pollution entry location and time and water demands. For the first time, this study presents a bi-objective algorithm using the results of single-objective PSO optimization simultaneously for one of the NSGA-Ⅱ constraints based on water requirement uncertainty by the Minimax method and robust approach. The two goals are to minimize the maximum possible damage caused by the contaminated water consumed and to minimize the cost of sensor preparation and installation. In this study, considering three scenarios (S1, S2 and S3), arsenic injection with concentrations of 50, 100 and 150 mg / l from different nodes and in different time steps are investigated. The results of the developed model on a reference network reveal that in each of the three injection rate scenarios S1, S2 and S3, the installation of only one sensor reduces the damage by 75, 70 and 61%, respectively. Also, the entry of pollution from 19:00 and 20:00 leads to a critical scenario. Although, the location of the sensors depends on the concentration of the injected contaminant, the highest repeated location in the generated responses was found in nodes 5, 17, and 19. The proper formation of the Pareto front in all three scenarios of injection rate showed that for all parameters with uncertainty, the robust approach generates justified responses with a minimum deviation of the objective function from its optimal value.