چكيده به لاتين
Detection of intentional or accidental pollutant injections in any water distribution networks (WDN) is essential to maintain public health. Recent technological advancements in wireless and wired sensor networks make it possible to use fix and/or mobile sensors, which are capable to continuously collect and transmit water quality measures at the fine temporal resolutions. The studies indicate, common water quality parameters (e.g. chlorine, pH, electrical conductivity, total organic carbon, etc.) respond to network contaminations. This dynamic water quality changes in WDN facilitate contaminant detections. In this regard, EPANET-MSX is used to simulate pollutant injections to water distribution network, realistically. In this research, the optimal layout of stationary or mobile sensors is aimed to 1- reducing contaminant detection time, 2- increasing contaminant detection likelihood, and 3- reducing contaminated water consumption considering as single-purpose and/or multi-purpose problems. Significant changes and reductions of Chlorine concentration in responses to Potassium Cyanide (KCN) injection have been defined as criteria in network contamination. Particle Swarm Optimization (PSO) algorithm has been coupled to EPANET-MSX to determine optimal sensor placement in WDN. Due to uncertainties in network contamination event scenarios, 1000 random scenarios have been generated based on uniform distribution function. The uncertainties in injection time, mass injection rate, injection location, and injection continuity have been considered in random scenario generations. Evaluation the number of stationary or mobile sensors on the accuracy of contaminant detection likelihood, expected time of contaminant detection, and contaminated water consumption indicate the direct relationship with WDN security objectives. The results show mobile sensors have lower capabilities in contaminant detection than the fixed ones, due to their spatial variability and limited presence in the WDN. The profound studies indicate direct relationship between contaminant detection likrlihood and contaminant detection time, direct relationship between contaminant detection likelihood and contaminated water consumptions, and inverse relationship between contaminant detection time and contaminated water consumptions. Furthermore, in the combined layout of stationary and mobile sensors, increasing the number of mobile sensors lead to decrease of contaminated water consumption and contaminant detection time. Increasing the coverage ratio of sink nodes in the WDN results in contaminant detection time reduction. However, coverage rate changing from 80 to 100 indicate constant contaminant detection time. Battery lifetime reduction in WDN with one to three mobile sensors have resulted in adverse effects on water quality security. Based on the results, the battery lifetime is far more effective than coverage ratio of sink nodes in protecting water quality security in WDN.
