Abstract:
Context: Clean drinking water is a very scarce and precious resource, a basic need for life, and requires proper management. Inconspicuous water leaks over a long period of time contribute to the loss of potable water. It also results in inexplicably high water bills, low pressure and damage to property. Therefore, leak detection is a conservation technique that has highly peaked research interest in the recent years. There are quite a number of approaches to leak detection in existence but most are inaccurate or expensive to implement.
Objectives: In order to address unnecessary water losses, the main objective of this work is to implement an efficient method of water leak detection in real-time. As an abundant library of research work on leak detection exits, there are plenty of methods with their own strengths and limitations. Therefore the second objective of this thesis is to review existing literature to better understand their advantages.
Methods:To achieve the objectives stated, a research on the various methods used for water leak detection is conducted, and having considered vibration interpretation as an effective tool, a systematic literature review on the several analysis techniques is done. From which a solution to implement on pipelines is tested and discussed. Results: It was discovered from the literature review that although much effort has been made towards leak detection in distribution systems, there is a limited number of studies on detection in small-diameter pipelines and end-user premises. To address this, an experimental data set is analysed to detect leaks effectively using vibration interpretation. An application of the detection system using a LoRa WAN architecture is discussed.
Conclusion: In Conclusion, from the meta-analysis it was evident that a leak analysis system using vibration interpretation is feasible and that the study of leakage detection in consumer networks is still in its infancy. Therefore, une-leak algorithm that detects leaks by analysing vibration signals was developed and tested. However, the results were not tested in real time.
