Household water leak detection using smart metering generated water consumption data

Residential water leaks pose a significant challenge, leading to water waste and financial burdens for both consumers and suppliers. Traditional leak detection methods, such as visual inspections, acoustic detections, and the use of district metered areas (DMAs), often fall short in terms of efficiency and proactivity. Smart metering infrastructure offers a promising solution to this per- sistent problem. This research explores the advantages of smart meters that provide real-time, high-resolution water usage data, enabling the detection of consumption patterns and leak indi- cations. The case of Italy exemplifies the urgency for improved water management, with high per capita consumption and considerable water loss before reaching consumers. Italian house- holds consume nearly double the European average, and in 2022, 42% of drinking water was lost due to leaks. The adoption of smart meters in Italy could revolutionize water management by reducing waste, saving resources, and alleviating financial impacts on distributors and con- sumers alike. The objective of this thesis is to build upon prior literature on predictive modeling of household water demand in order to develop a scalable model capable of accurately forecast- ing residential consumption patterns and detecting leaks in the water distribution network, solely from existing smart meter data. The model is based on an analysis of real smart meter readings collected across a broad region in northern Italy, thereby eliminating the need for costly addi- tional sensor equipment or manual on-site inspection. By leveraging the extensive smart meter infrastructure already in place, the approach aims to provide a widely implementable and eco- nomically viable solution for predictive demand modeling and leak identification in residential areas.