Model calibration for leak localization, a real application

Abstract

The localization of leaks in Water Distribution Networks has a major relevance in terms of environmental and economic efficiency. This localization is generally carried on in situ by human operators using time consuming methods like acoustic loggers. Nevertheless, the automated aid provided to the operators is continuously increasing thanks to the exhaustive use of models. Models that have to be calibrated and updated in order to provide proper help and an improvement in the leak search. This paper presents an experience of leak localization using steady state models combined with a demand calibration algorithm. The calibration produces a notable improvement of the localization accuracy and signals changes in the network configuration. Results presented are based on real data and a real leak provoked for the test.