Research On Online Monitoring And Leak Location Model In Water Distribution System

Leak of urban drinking water distribution network is a worldwide problem thathas been bothering both domestic and overseas drinking water industry. Distributionnetwork leak not only leads to a waste of resource and energy, but also threatens thedrinking water safety. Therefore, with the boom of economy water industry drawincreasing attention to the detection and control of leakage in nowadays. Because ofthe large and complex topology of supply network, it always takes water companiesa considerable time to identify the leakage point when it occurs. As a result, thisdissertation focuses on the rapid positioning of urban drinking water networkleakage. Fuzzy Clustering, Genetic Algorithm and Bayesian Decision Theory wereapplied in the model of leakage location.For ensuring the accuracy of leakage location, the model was further refinedbased on conventional hydraulic model. Specifically, global microcosmic hydraulicmodel was built and DMA partition was conducted in this network. In conventionalhydraulic model, the demand of other junction was assumed to be constant beforeand after leakage, and this is inconsistent with the facts. Therefore, the PressureDrive Demand hydraulic model was the first time used to simulation leak in thispaper. The demand of leak location was divided into actual demand, backgroundleakage and leakage. And through analyzing a water distribution example, it showedthat the leak model based on PDD can be more accurate simulate the distributioncondition when leak accident occurred than conventional hydraulic model.Pressure monitoring points in the pipe network can effectively reflect therunning conditions, and can locate the leaking point faster when accident occurred.Due to the demand of cost control, the number of pressure monitoring points isrestricted. Combined with the theory of Fuzzy Clustering, sensitivity analysis forpressure monitoring points was carried out and optimization scheme of monitoringpoints was obtained. The process of optimization was described combining withcalculation example.Leakage accident would, to varying degrees, result in pressure fluctuation inthe whole network. Based on this theory, leak location model was established,respectively, using Model Calibration and Bayesian Decision Theory afteroptimization scheme of monitoring points was built. This study put forward animproved leakage positioning model which had certain immunity to hydraulic modelerror and measurement error and could response timely to the leakage of pipelinenetwork.Combined with M-City’s pipeline network, Microscopic model of the entire network was established and Leakage positioning test in HY district was also carriedout. The study compared the application effect of the three kinds of leakagepositioning model in HY district. The results showed that the improved leakagepositioning model, which could determine the missing node or region by calculatingleakage probability under different model parameters, had high detection accuracy.