Online model- and data-based leakage localization in district heating networks - Impact of random measurement errors

Pipe bursts and leaks in district heating networks are a problem both for the economic operation and for the supply reliability of the connected consumers. In case of a leakage, pressure and flow rate conditions near the defect change. These changes spread to the entire network within a short period of time and, depending on the size of the leakage, partly lead to drastically changed network conditions. After a leakage is detected, it is necessary to localize the leakage as accurate as possible in order to shut down the affected network segment and maintain the network’s stability. This article discusses and compares three different approaches for leakage localization (pressure wave detection, model-based numeric-analytical and machine learning) that exploit different properties of simulation models and sensor information from the real network.