Two Phases Leakage Detection Strategy Supported by DMAs

Abstract

The present work proposes a novel two phases model-based strategy for leakage detection. The two phases are: the identification of the district metering area (DMA) and the pipe pre-localization into the identified DMA. The strategy is based on detecting and pre-localizing the punctual leakage as anomaly with respect to the normal working conditions. A further novelty is the fact that the pre-localization phase returns the sequence of pipes to inspect, which makes the strategy attractive for water utilities, whose aim is to identify the anomaly at DMA level and, successively, to localize it with the minimum inspection cost. Furthermore, a random database is useful to test the performance of the strategy with respect to the configuration of DMAs and the pressure metering system. Consequently, a novel strategy to design the location of pressure meters is also proposed. It is demonstrated that the entire strategy limits false positives during the DMA identification phase by using the recently proposed index named Asset Management Support Indicator (AMSI). AMSI is invariant with respect to the deterioration, i.e., it is sensitive to its increase causing punctual leakage. The strategy is studied and discussed using two real Apulian WDNs managed by Acquedotto Pugliese.