Algorithms for detection and localization of events in a water distribution network

• Main Author: Zan, Thaw Tar Thein
• Other Authors: Lee Bu Sung, Francis
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/65064
• Institution: Nanyang Technological University

The occurrence of underground pipe bursts in a Water Distribution Systems (WDS) is a pervasive problem affecting water authorities worldwide. These events can be very expensive due to the outage time while the burst pipe is repaired, the cost of repair, and damage to surrounding property and infrastructure. Therefore, the method to detect and quantify them becomes crucial to all water authorities around the world. Moreover, it would be beneficial if there is a method to detect and locate leaks and bursts within the water distribution network in a timely and cost effective manner. The aim of this thesis is to present a novel approach to detect the presence of leaks and bursts and locate them within an urban WDS. This approach has been developed and tested on the real and active WDS. To monitor the WDS, a wireless sensor network is deployed within water distribution network in the downtown supply zone in Singapore. This network is implemented as part of the WaterWiSe@SG project (Wireless Water Sentinel in Singapore), with the aim of developing a generic wireless sensor network to monitor WDS in a continuous manner. This report presents the application of the Joint Time-Frequency Domain Analysis for leak detection in water distribution pipelines. The idea behind this method is based on the detection of fluctuation in pressure transients signal induced by a leak. This new approach to leak detection employs spectrogram, one of the joint time frequency domain approaches. The proposed method is verified using simulated leak-off experiments on real water distribution network in Singapore. Moreover, this method has been validated with real leakage during distribution. The results demonstrate that this method possesses the capability to detect the leaks within WDS. Leak localization is as important as leak detection to facilitate the repair process. Therefore, it is crucial to estimate the location of the detected leak via transient wave. Transient wave is generated due to the sudden changes of flow in the pipe when a leak occurs in a pipeline. The leak-induced transient wave travels in both directions along the pipeline. When it travels through a junction of two or more pipes, part of the wave is reflected back. It is necessary to model the transient wave in order to trace the leak location. Models for pipe failure are included in this thesis. The model is used to assist the leak localization. Two different approaches for leak localization, graph-based localization with time of arrival estimation and energy-based localization with intensity estimation, are proposed. Graph-based localization is based on the difference between the arrival times of the leak-induced pressure wave at each measurement point. Energy-based localization is a ranged based localization method and is based on the intensity of the leak-induced pressure wave at each measurement point. Results from real leak events show that the proposed techniques have potential as a tool for localization in real pipe networks.