Water losses in drinking water distribution systems are a globally recognised problem. Drinking water utilities have a number of reasons to address this challenge. In the Netherlands image considerations are of particular importance while, elsewhere in the world, financial, ecological and regulatory reasons also apply.

In this project we are developing a tool for the assessment and operational application of several leakage-detection and localisation methods in combination. The tool’s validation will involve applying it to three areas with different characteristics (including size), using historical and experimental data.

Once available, the Dutch water utilities can apply the project’s knowledge and tool w to reduce their water losses. Moreover, the project partners will be able to use the knowledge and tool to provide Dutch expertise in the field to assist other countries reduce water losses in their distribution systems. The international market in particular offers a huge potential, given that a large part of the world is struggling with water losses of tens of percentage.


The tool was tested with synthetic data (in the computer-simulated leaks) and with leaks simulated in the field (flushing tests) in two different DMAs in the Netherlands. For the synthetic leaks, the results show that the tool functions correctly and is capable of identifying the right area of the leakage (usually within 100-200 meters of the actual leakage point). This also applies for significant noise levels in the time series (up to 10%), which can be considered representative both for uncertainties in the measurements and the effects of the stochastic water demand. The tool’s application to data from flushing tests in the field was, in the first instance, less successful, but the insights obtained offer good prospects for successful field application.

The tool can be successfully applied to practical data under the following conditions: 1) For the leakage detection, a thorough analysis needs to be conducted to determine both the timing and the extent of the leaks. The former is identified by all methods implemented in the tool, and the latter by means of CFPD (Comparison of Flow Pattern Distributions) and spectral analysis, in particular. 2) For the leakage localisation, one needs to ensure that the hydraulic model used is sufficiently representative of the real hydraulic conditions in the field, both in terms of topology/connectivity, but also especially in terms of the water demand. Both aspects are achievable. In this way, effective leakage detection and localisation, based on various methods from the scientific literature, is brought within reach of the drinking water utilities.


Typical sources of leaks in water pipes.


Methods for detecting leaks in a water distribution system. CALLISTO focuses model-based methods and data-driven data-driven methods.