Finding Big Leaks with Big Data: Case Studies from an Internet-of-Things Leak Detection Platform

摘要

This paper presents results from an "Internet of Things" leak detection technology that remotely and non-invasively monitors pipelines at the scale of a distribution network. Findings are presented from the deployment of 10,000+ acoustic leak detection nodes with dozens of utility partners. The design and installation of the nodes, the type of data collected, and the approach to data analysis is outlined. A series of case studies are described where the technology has found leaks that failed to surface: large or significant leaks, leaks that were difficult to pinpoint, and unexpected signals found in the process. Both the data analysis and utility perspectives are shown, comparing remotely collected acoustic data to observations from field technicians. The successes and limitations of the technology are both discussed.