Data-driven approaches and model-based methods for detecting and locating leaks in water distribution systems: a literature review

Water distribution systems are made up of interconnected components that should allow water systems to meet demand, but leaks can waste enough water to limit supply. To limit financial losses, water utilities must quickly determine that a leak is occurring and where it is referred to as the localization of the leak. Over the years, there have been various methods proposed to detect and locate leaks. This literature review summarizes many of the methodologies introduced, categorizes them into data-driven approaches and model-based methods, and reviews their performance. Data-driven approaches demand efficient exploitation and use of available data from pressure and flow devices, and model-based methods require finely calibrated hydraulic models to reach a verdict. Data-driven approaches can manage uncertainty better than model-based methods.