Water pipeline failure detection using distributed relative pressure and temperature measurements and anomaly detection algorithms

Research output: Contribution to journalArticle

Colleges, School and Institutes


This paper presents the validation of a novel leak detection method for water distribution pipelines, although it could be applied to any buried pressurized fluid flow pipe. The detection method is based on a relative pressure sensor attached non-invasively to the outside of the pipe combined with temperature difference measurements between the pipe wall and the soil. Moreover, this paper proposes an anomaly detection algorithm, originally developed for monitoring website traffic data, which differentiates a ‘leak’ event from ‘normal’ pressure change events. It is compared to two more commonly used methods based on a fixed threshold and a moving average. The validation of the new system in a field trial over a 6-month period showed that all the known leaks were identified with 98.45% accuracy, with the anomaly detection algorithm performing best, making this system a real contender for leak detection in pipes.


Original languageEnglish
JournalUrban Water Journal
Early online date18 Apr 2018
Publication statusE-pub ahead of print - 18 Apr 2018


  • wireless sensor networks, leak detection, leakage, remote sensing, monitoring, pipelines