Cost-Effective Sensors Placement and Leak Localization

This papers extends the research activities performed by the authors in developing and applying an approach to analytically identify leaks within a Water Distribution Network (WDN), by combining hydraulic simulation (EPANET) and Network Science based Data Analysis techniques.

The software model of the WDN is used to run several “leakage scenarios”, by varying leak location (pipe) and severity, and to build a dataset with the corresponding variations in pressure, at nodes, and flow, on pipes, induced by the leak.

All junctions and pipes are considered for potential pressure and flow sensors deployment; a clustering procedure on these potential locations is then performed to identify the most relevant nodes and pipes, while costs for pressure and flow meters are considered to select the combination which guarantees the best trade-off between reliability in localizing leaks and overall cost.

A graph is then generated from the dataset, having scenarios as nodes and edges weighted by the similarity between each pair of nodes (scenarios), in terms of pressure and flow variation due to the leak. Spectral Clustering is adopted to group together similar scenarios in the eigen-space spanned by the most relevant eigenvectors of the Laplacian Matrix of the graph.

This approach proved to be more effective than other traditional techniques which work directly in the space of pressure and flow variations. Finally, Support Vector Machines classification learning is used to learn the relation between variations in pressure and flow at the deployed meters and the most probable set of pipes affected by the leak.

Media

Taxonomy

• Water
• Sensor Systems
• Leakage Detection
• Leakage
• Leakage Reduction
• Monitors
• Sensorial networks