Research On Pressure Management And Leakage Zone Identification For Water Distribution Systems Based On Hydraulic Model

Our country is facing serious water-shortage and water pollution problems.On the other hand,a large amount of leakage exists in water distribution system and has not been effectively controlled,which makes the serious shortage of fresh-water resources worse.It has become an important problem about how to effectively reduce the leakage rate and improve the water supply efficiency in the process of large scale and high density urbanization.The thesis research on pressure management and leakage zone identification for water distribution systribution systems based on hydraulic model,which is used to improve the leakage management level of water distribution system.Firstly,this thesis studies the hydraulic block parallel calculation method and the adaptive adjustment method of parameters for valve / variable speed pump in order to improve the decision-making efficiency of leakage control.The hydraulic block parallel calculation method can greatly improve the computational efficiency of hydraulic model.Compared with traditional method(such as EPANET2.0),the proposed method can increase the computational efficiency.The adaptive adjustment method can adjust the parameters of valve or variable speed pump adaptively according to the actual network pressure,thus reducing the background leakage.The calibration effect of hydraulic model is an important prerequisite for the effective control of leakage.This study proposes a quick calibration method of pipe resistance coefficient and node demand.A real network has been presented here to demonstrate the effectiveness of the method.The comparison results show that the differences between the field-observed and model-simulated of 16 sites are very small(the maximum error was 1.40m).All errors meet the precision requirement of the model(± 2m).The proposed method takes only a few tens of seconds,the computational efficiency has obviously improved compared with the traditional genetic algorithm(usually takes several hours or even days).Then,on the basis of accurate calibration of hydraulic model,this thesis studies the method of smart leakage control from the aspects of pressure partition management and active leakage control.In order to overcome the shortcomings of traditional partitioning methods,to improve the quality of network partition,and to facilitate district meter areas(DMA)measurement and pressure management,a new method is presented in this paper for network partitioning design on the basis of multiscale community detection method and multi-objective optimization.The proposed method has applied to a real network and achieved good application effect.In order to effectively establish the leakage detecting system,this thesis studies the data collection and leakage early-warning approach.Data collection studies the optimal placement of new-added pressure and flow sensors in water distribution systems.Such a method can not only make the collected data fully reflectintg the operating conditions of water supply network,but also provide data support for hydraulic model calibration,leakage early-warning and leakage location.Leakage early-warning studies the screening approach based on data mining theory to realize the automatic screening of three abnormal data types(variable operating conditions,leakage events,singular values).The proposed method has applied to a real network and achieved good application effect.Finally,this thesis presents a leakage zone identification method based on MSVMs to realize rapid leakage localization after abnormal alarm.Three case studies have been presented here to demonstrate the effectiveness of the method.The first case study is used to assess the capacity of the presented method from three aspects: the division of leakage zone,the type of kernel function,and the number and the location of pressure monitor points.The results indicate that using linear or Radial Basis Function(RBF)kernel function and the optimized pressure monitor points can achieve better results than other kernel functions without pressure sensor placement optimization.Case 2 and Case 3 combine the proposed method with pressure dependent Leakage Location(PDLD)method applied to unpartitioned and partitioned network,respectively.The results show that the proposed method can effectively overcome the defect that the search space is too large in the process of PDLD model solution in a large water distribution system.A real network case study shows that the approach of leakage zone identification model together with PDLD model can not only increase the computational efficiency and precision of PDLD model but also shorten the field leakage detection.