The Detection Of Leakage And Break In Water Supply Distribution Networks Based On The SCADA System

Leakage and pipe bursts of water supply networks are very common in water supply field. Leakage can not only waste water resources, but also increase the cost of water supply. And the most important thing is that pipe bursts may have great influence on people's life and our society. So, with more and more serious of water shortage, it is very important to find an effective and economical method of leakage detection.This study is consist of four parts: supplementary method of pressure measurement points' distribution, improvement of the accuracy of state estimation, bursts' location by fuzzy similar priority comparison and the optimal operation condition of pipe burst detection.(1) Supplementary method of pressure measurement points' distributionBy calculating the influence factors of water supply networks under pipe burst condition, the abnormality detection matrix can be esablished for adding pressure measurement points selectively, which is an improvement of pure fuzzy clustering method. With these additional points, the abnormality of any pipe break can be found to determine the accident location. It has been proved that this method is helpful to find the broken pipes more quickly and accurately.(2) Improvement of the accuracy of state estimationThe accuracy of state estimation mostly depends on the locations and amounts of pressure measurement points and flow measurement points. In this article, the state estimation is on the basis of the supplementary distribution of pressure measurement points and the optimal location of flow measurement points. It is shown that the root mean square of pressure error is reduced from 0.61 to 0.28. That means the estimation is feasible for real-time monitor and pipe burst detection of water supply networks.(3) Bursts' location by fuzzy similar priority comparisonBased on the supplementary distribution of pressure measurement points and the optimal location of flow measurement points, the normal state estimation and the burst state estimation are made by using the information of SCADA system. According to the node pressure changes and node weights, the burst pipe can be found out by fuzzy similar priority comparison. This detection method is effective for most of the time. When it can not find the exact pipe, it can show the accident area as well.(4) The optimal operation condition of pipe burst detectionThrough the calculation of different operation conditions, it is shown that as node flows change synchronously, the pipe detection is more easily under maximum hour condition that that of average hour and minimum hour condition. But when leakage is larger than 80L/s, the accuracy under each operation condition is almost the same. In addition, when node flows don't change synchronously, the detection rate will be influenced by water source flows. That means the detection accuracy can be improved by changing the node flows of larger source flow.All these above provide a useful method for leakage detection and control and it is also the basis of further studies on water supply networks.