Research On Real-Time Hydraulic Modeling And Applications In Urban Water Supply Network DMAs

Network hydraulic model is widely used in the planning,design and dispatching decision of water supply system,however,limited by off-line modeling condition and manual verification and maintenance way,the off-line hydraulic model has met more and more difficult to deal with the emergent challenges like pipe burst,water pump mechanical accident and water quality pollution.Therefore,the traditional off-line model capacity is questioned and underestimated in recent years.It is the hot research spot of water supply network hydraulic modeling research to do the real-time modeling with the connection of on-line data in SCADA system,which can fully exerting the hydraulic model performances to meet the actual requirements.Due to the restraints on the previous informatization basic conditions and relavant theoretical research,there is no precedent of real-time modeling research and application in domestic.With the gradual improvement on SCADA,GIS and other information system,the condition of real-time modeling research are increasingly mature.So,the real-time hydraulic modeling and applicating of water supply network are studied,the main contents are as follows:1.Real-time hydraulic model improvement research against dynamic uncertainty of the water supply network: the improvement of real time hydraulic model to overcome dynamic uncertainty factors,which the traditional off-line models are difficult to overcomeis expounded;the sensitive factors that influence the accuracy of real time hydraulic model are analyzed.2.EPANET-RTX real time-modeling technology is introducted and improved:(1)For EPANET-RTX online data preprocessing method can not effectively eliminate network abnormal noise,the resampling and moving smoother methods is improved.(2)For EPANET-RTX water demand estimation algorithms would easily make the node estimated demandis too smooth to lead to big deflection,a dynamic base value with on-line demand method is put forward.(3)For EPANET-RTX not yet provide the related methods for updating status of values and updating head of negative demand water supply dividing points,the corresponding methods are proposed respectively.3.DMA real-time hydraulic model is founded.An DMA independent real-time modeling method is put forward,and an real-time hydraulic modeling test for city S DMA is carried out.Test results show that the designed method for DMA real-time hydraulic model improve the average simulation performance,the dynamic simulation performance and the real-time tracking performance relatively,compared with EPANET-RTX and previous off-line hydraulic model;4.DMA real-time hydraulic model application in water supply network pipe explosion locating problem: Due to the inherent flaws of the traditional off-line hydraulic model in detecting and locating the pipe explosion,a method based on the real-time hydraulic model is proposed.The genetic algorithm is adopted to optimize and verify DMA real-time hydraulic model and to calculate the location of the bursting point and leakage.To make the solution converge and a precise locating,a coding method based on natural space search is put forward.A method based on the analysis of the trends of pressure of monitoring points is proposed for initial locating and the real-time hydraulic model provides a precise locating with the initial parameters from the initial locating.5 pipe explosion experiment results show that the 2 location bias controled within 250 m.While the other 3 pipe explosion has little effect on pressure of the measuring point,with larger locating difficult,the location bias still controls within 1km,proved that method based on real-time model maybe more practical than traditional method,the locating accuracy can further improved.5.DMA monitoring points placement optimization improvement applications: To improve the correction accuracy and the exception detection capabilities of the real-time hydraulic model,the monitoring points optimization placement method based on cluster analysis and based on sensitivity optimization are used respectively.The Kmeans algorithm combinated with fuzzy similar matrix is used to quickly cluster the sensitivity of the nodes because of the slow efficiency of the traditional net method.a standard deviation factor of space distance of monitoring points is introduced to make the monitoring points distribution uniform because of the traditional sensitivity scatter factor cannot ensure the relative dispersed distribution of monitoring points,Results show that the two types of optimization design are more reasonable than the existing monitoring point distribution,achieve the expected goals and lay the foundation for further real time modeling research and application.