Sensors Placement in Water Distribution Network Under Dynamic Variations of Common Water Quality Parameters

Detection of pollution in any water distribution network (WDN) is essential to maintain public health. Recent researches have demonstrated common water quality indicators (free Chlorine, pH, Electrical‎ Conductivity, …) response to contaminants, which can facilitate the pollutant detection in WDNs. In this research, EPANET-MSX model is used to simulate the realistic responses of common water quality parameters to a pollutant event in WDN, represented in example 3 of EPANET 2.0. The aim of the study is to apply EPANET-MSX to trace Chlorine changes in response to the potassium cyanide injection order to optimize sensor placements WDN. Optimal sensor placement is derived based on coupled EPANET-MSX _ PSO (particle swarm optimization) framework. PSO is implemented as single and multi-objective algorithm. The objectives of the study are to reduce the pollutant detection time, increase detection likelihood, and reduce the contaminated water consumption as single and/or multi-objective problems. Studies on the sensor numbers, ranging from one to five, indicate 1- the pollutant detection time has decreased from 36.22 to 17.41 hours, 2- the pollutant detection likelihood has increased from 28.8% to 68.6%, and 3- the contaminated water consumption has decreased from 31.87 to 3.78 m3. The results show direct relationship between 1- pollutant detection likelihood and detection time, 2- pollutant detection likelihood and contaminated water consumptions, and 3- pollutant detection time and contaminated water consumption in multi-objective problems. Also, the location of the high-demand nodes in WDN and the contamination events occurrence can affect the location of the sensors.