Non-invasive and Efficient Leak Detection System for Long Pipelines

Our society relies on an extensive pipeline networks to transfer and deliver water, oil, etc. In many cases the pipeline extends over hundreds of miles and runs through inhospitable environments. The pipes are often subject to erosion over time due to the parametric conditions. Moreover, pipes carrying valuable commodity may be subject to theft, sabotage, etc. Leakage not only would waste resources, but also can be harm and hazardous. Therefore detecting leakage and containing its negative effect is very important. Current pipeline monitoring systems are inefficient and costly. They lack responsiveness and often report the problem after significant fluid is spilled. Furthermore, current systems involve mobile equipment and significant manpower.
An articulation of the network architecture for the leak detection system. The dots represent sensor nodes that are externally mounted on the pipe wall.
In this project we develop a solution that is based on a wireless sensor network that continuously monitor a pipeline and provide an early warning when leakage starts. Our proposed leak detection system employs non-intrusive methods, which makes it a very attractive solution for existing pipeline system and facilitates maintenance. The system is composed of a set of ultrasonic transducers that are mounted around pipes in multiple locations. The approach is based on tracking the flow speed using the ultrasonic transducers and interprets the drop of flow volume as indication of liquid loss through a crack in a pipe. The analysis combines the transit-time and Doppler based flow speed measurements in order to deal with various liquid characteristics. The data collected by the individual components are sent over radio links via multi-hop routes to a base-station for further analysis to confirm the occurrence of leakage and determine where it happens. We are working on the development of circuit boards for data acquisition and on-node processing. In addition, algorithms are being devised for fusion of the collected data by both methodologies in order achieve high fidelity detection of leakage and minimize the occurrence of false alarms, and for data routing and in-network aggregation as well as other network management functionality. A prototype system is being built to validate the effectiveness and practicality.

Recent Publication
  1. A. Santos and M. Younis, "A Sensor Network for Non-Intrusive and Efficient Leak Detection in Long Pipelines," in the Proceedings of the 4th IFIP Wireless Days Conference, Niagara Falls, Ontario, Canada, October 2011