RörANN – a smart artificial neural network modelfor minimizing leakages in water distributionsystems

In Sweden, 15–20% of drinking water is wasted by leakage. Leakage affects both the environment and the economy, but relates also to social vulnerability and health. With this project we want to support the mind-shift that is going on in Swedish water utilities by further developing a new ANN model, which identifies drinking water pipes at the greatest risk of leakage. The current work on water network maintenance is often reactive, you repair leaks as you become aware of them, sometimes to great costs if the leak is extensive and acute. Renewal planning is often governed by external factors, e.g.that the street office is planning a street renovation. We want to put "glasses" on the water utilities in order for them to clearly see pipes in poor condition. The ANN model we refine and disseminate does not produce any new data, but uses data already collected by the water utilities (material, year of construction, pressure levels, etc.) or has access to through others (soil, traffic loads, demographics, etc.) in order to calculate the probability of failure for each individual pipe. With the model, we give water staff a smart tool for decision support. At the same time, we want to break the old mind-set of many utilities and gather some of the most proactive utilities to spread knowledge about AI technology. It is the end users that have initiated the project. Together with researchers at LTH, they wish to further develop and verify the prototype model. The project will also further develop the output by weighing in the con-sequence if pipe x or y fails. This gives us a full risk assessment tool with AI support. Four large, proactive water utilities participate as partners (SVOA, VA SYD, Vakin & Kretslopp och Vatten)and four others contribute data and experiences. The final goal of the project is two versions (basis and big) of the AI model, which can easily be applied by small or large water utilities. The model will be provided by the industry organization Svenskt Vatten.