FOUND

Error propagation of uncertainties in pressure-based leak detection

This project is funded by the Federal Ministry of Agriculture, Forestry, Regions and Water Management (BML) (Austria) (Project C300198).

Project partners: Technical University of Berlin, De Montfort University
Author: Robert SITZENFREI, Martin OBERASCHER

Project duration: 04/2024 - 09/2025

Project goals:

The aim of this project is to further improve the accuracy and applicability of pressure-based leakage localisation for subsequent on-site fine detection. The following basic research questions are addressed:

  • How do uncertainties and errors (e.g.: Quality of the measurement data and the numerical pipe network model, sensor placement) affect the error propagation through the process of model- and data-based leakage localisation?
  • How does the draft European Union regulation "Establishing harmonised rules for artificial intelligence" affect the use in critical water infrastructure (explainable artificial intelligence)?
  • Can a hybrid method (in combination with explainable artificial intelligence) support a good transferability of pressure-based leak detection to real WVN in Austria?

Brief project description:

Leaks refer to the unintentional escape of drinking water in a water supply network. Unreported but detectable leaks in particular pose a challenge for incident-free operation, as these leaks can only be found through a proactive search. Due to their long duration, these leaks also cause considerable losses of drinking water and revenue and can lead to damage to the pipe infrastructure or other urban infrastructure facilities due to erosion of the pipe bedding. Rapid localisation and repair are therefore of great interest.
In the literature, system condition-based leakage localisation is increasingly being used, which uses high-resolution measurement data (e.g. pressure data) to enable a spatial pre-limitation of leaks in the water supply network for the subsequent detailed search on site using hardware-based methods. However, the effectiveness of these pressure-based methods is highly dependent on the quality of the measurement data or the numerical pipe network model used. As can be seen from the literature, these uncertainties are hardly taken into account or only in some areas of pressure-based leakage localisation. In addition, the effectiveness of pressure-based localisation algorithms (model- and data-based methods) is rarely compared under realistic reference networks.
To further improve the effectiveness of the pressure-based method for leak detection, this project pursues the following objectives: (1) to combine a variety of factors and investigate the propagation of errors throughout the pressure-based leakage containment process and (2) to compare and evaluate the effectiveness of model-based and data-based methods under a systematic combination of various influencing factors or their uncertainties. A selected reference network from the literature and a real network with leakage simulations are used as case studies. The combination of reference network and real case study enables a detailed determination of the potentials and limitations of pressure-based leakage containment. In addition, this approach also provides insights for practical implementation and recommendations can be derived for the future implementation of pressure-based leakage containment.
In addition, the project will develop a hybrid method that combines model-based and data-based methods in order to optimally reduce the various uncertainties in practical applications. The use of explainable artificial intelligence (or machine learning) is also being investigated in this project in order to fulfil the requirements of the European Union regarding the use of artificial intelligence in critical infrastructure (explainability and transparency). This will ensure a good transfer of the research results to real water supply networks in Austria. This will also further reduce the effort required for hardware-based leak detection on site and support small and very small water supply companies in Austria in particular in water loss management.

Nach oben scrollen