Losses Identification in Uncalibrated Water Distribution Networks: A Case Study

In a real water distribution network, district metering for the identification of losses cannot always be successfully applied due to the limited number of discharge measurements or poor district identification. In such cases network models should be a valuable aid, but uncertainties on the demand distribution and on the pipe roughness, as well as the unknown location of losses, thwart their calibration. This study suggests and applies in a real case a procedure based on data assimilation techniques (Kalman-filter based) to identify the spatial distribution of leaks and to calibrate, at the same time, the pipe roughness.