“CityLeak Zero” is an advanced urban water system digitization project aimed at significantly reducing physical losses of drinking water in cities with aging networks. It combines distributed sensors, hydraulic modeling, artificial intelligence, and remote monitoring. The proposed technical architecture integrates field detection technologies with predictive analytics platforms, transforming opaque and reactive infrastructures into intelligent, transparent, and climate-resilient systems.
The intervention includes the installation of acoustic correlator sensors to detect underground vibrations associated with invisible leaks, differential pressure loggers to record instantaneous drops linked to early-stage breaks, and clamp-on ultrasonic flowmeters that measure flows without service interruption. All collected data is transmitted via NB-IoT or LoRaWAN networks to a cloud analytics platform, integrated into a dynamic digital twin of the network.
This digital twin, built with topological, historical, and real-time data, simulates expected hydraulic behavior and detects deviations within seconds due to loss events, abnormal pressures, or localized overconsumption. Based on these deviations, machine learning algorithms are triggered to issue early alerts, classify events by criticality, and prioritize interventions based on estimated water return. This predictive maintenance logic reduces leak detection times from weeks to hours and optimizes repair crew allocation, materials, and shutdown scheduling.
The integrated urban management platform allows operators to visualize network status in real time, consult historical data, project loss scenarios, and generate periodic reports on hydraulic efficiency, associated energy, and environmental metrics. The project’s technical goal is to achieve a cumulative reduction of 40% to 55% in physical losses within prioritized zones, with full traceability of recovered water volume, avoided climate impact, and operational benefits for public services.
Cities operating with drinking water networks built between the 1950s and 1980s generally exhibit structures highly vulnerable to non-visible losses. These leaks, caused by internal fissures, household connection failures, or material degradation, do not surface or generate public complaints but account for substantial volumes of extracted, transported, and treated water ultimately lost before reaching consumers. This phenomenon represents not only a significant water waste but also an economic burden on service operators, who bear the energy and chemical costs without consumption recovery.
Technical studies in Spanish cities indicate that Non-Revenue Water (NRW) levels routinely exceed 25%, reaching 35% or more in some urban sectors, particularly during demand peaks or thermal overpressure periods. This situation worsens in extended summers, where material fatigue and continuous pumping pressures lead to progressive microfailures.
The most frequent structural causes include internal corrosion of cast iron pipes, cracking due to structural loads or road traffic, progressive degradation of rubber joints, obsolescence of isolation valves, and unauthorized or substandard connections. Additionally, geotechnical phenomena like urban subsidence, differential settlement, or saline intrusion can accelerate hydraulic degradation and loss processes.
Compounding these issues is a low level of hydraulic sectorization (DMA), lack of continuous per-section monitoring, and reactive maintenance planning in many municipalities. In contexts of water stress and energy transition, such inefficiencies not only result in direct resource and financial losses but also intensify pressure on overexploited aquifers, increase the urban water cycle’s energy footprint, and heighten reputational risks amid stricter regulatory and financial environments.
“CityLeak Zero” proposes a high-resolution technical solution based on progressive digitization, intelligent sensing, and advanced data analytics applied to urban potable water distribution networks. This architecture is grounded in interoperability between field devices, hydraulic modeling platforms, and cloud-based operational intelligence systems.
The technical strategy includes installing geolocated acoustic correlator sensors in master valves, critical break points, and vulnerable branches. These sensors detect anomalous acoustic signatures generated by turbulence from non-visible microleaks. They are complemented by differential pressure loggers that monitor real-time pressure drops, unusual nighttime declines, and transient events with sub-second resolution. Additionally, non-intrusive clamp-on ultrasonic flowmeters are deployed to establish hydraulic balances in sectorized areas without service disruption or network modifications.
All data is continuously transmitted via NB-IoT or LoRaWAN gateways to a central cloud processing platform. There, it is integrated with historical, topological, and climatic variables to feed a hydraulic digital twin, a dynamic digital replica of the network simulating expected physical behavior in each section. This simulation incorporates nominal hydraulic conditions, seasonal variability, hourly demand profiles, and structural material age characteristics.
Through machine learning techniques and multivariable time-series analysis, the system identifies persistent deviations between the expected model and real-world data: leak acoustic patterns, sustained overpressures, inconsistent flows, and unexplained pressure fluctuations. These anomalies are classified, scored by hydraulic criticality, and converted into operational alerts.
Alerts are displayed on an interactive geospatial dashboard where operators visualize network health status, critical zones, temporal evolution of anomalies, and receive automated recommendations on intervention prioritization. The system allows real-time calculation of estimated water savings per intervention (m³ recovered), equivalent energy savings (kWh avoided), and reductions in indirect CO₂ emissions. This technical-operational integration reduces leak detection times from weeks to under 24 hours, improves corrective maintenance efficiency, and provides verifiable evidence for sustainability and water governance reporting frameworks
The deployment of the “CityLeak Zero” project will follow four sequential and integrated phases over 24 months, combining technical planning, sensor installation, digital modeling, and operational intelligence. Each stage is designed to maximize traceability, enable independent validation of the generated water benefits, and facilitate replication in other urban environments.
Phase 1 – Diagnosis and Design (Months 0–6): This stage involves a comprehensive technical assessment of the existing water distribution network. It includes mapping pipe materials, diameters, age by section, historical break data, operational zoning, topography, and nighttime minimum consumption patterns. Based on this assessment, between 5 and 8 District Metered Areas (DMAs) will be selected, prioritizing those with older infrastructure, high population density, and critical urban functions (e.g., heritage sites, hospitals, schools). For each DMA, a detailed sensorization plan will be developed, specifying the placement of pressure loggers, acoustic sensors, and flowmeters according to strategic nodes and risk history.
Phase 2 – Installation and Connectivity (Months 6–12): Physical deployment of monitoring and communication devices will be carried out. Acoustic correlator sensors will be placed in valve boxes and access chambers to capture subsurface vibrations typical of microleaks. Clamp-on ultrasonic flowmeters will be installed at DMA inflows and outflows for non-intrusive flow monitoring. High-resolution differential pressure loggers will detect transient pressure drops. All equipment will connect to NB-IoT or LoRaWAN gateways, transmitting real-time data to Azure IoT Hub with full encryption. The network will maintain a sensor density of at least one unit per 300 meters, enabling effective triangulation and coverage.
Phase 3 – Modeling and Machine Learning (Months 12–18): A dynamic digital twin of the urban network will be developed using hydraulic modeling platforms (e.g., EPANET) integrated with Microsoft Azure Digital Twins. Real-time and historical data will calibrate the model to simulate pressure, flow, and consumption behaviors. Machine learning algorithms will be trained to detect patterns such as abnormal pressure drops, repeated acoustic anomalies, or flow inconsistencies. Alerts will be classified by severity and visualized on dashboards connected to Microsoft Sustainability Manager. Copilot AI will provide automated operational recommendations (e.g., “Investigate segment C21-B18: pressure drop >7%, acoustic peak, estimated leak 0.8 L/s”).
Phase 4 – Validation, Reporting, and Scaling (Months 18–24): Detected events will be verified in the field using portable correlators or geophones. Confirmed leaks will be documented with location, type, estimated volume lost, and time active. This empirical feedback will enhance model precision. Microsoft Sustainability Manager will issue monthly reports detailing water saved, leaks resolved, response times, energy savings, and CO₂ reductions. These outputs will align with CDP Water Disclosure, ESRS E3, and VWBA 2.0 reporting standards. Based on results, the model will be scaled across additional districts or adapted to other Mediterranean cities.
“CityLeak Zero” is a digital transformation initiative aimed at solving one of the main structural challenges in Mediterranean cities: invisible potable water losses in aging urban networks. The project is based on a combination of distributed sensing technologies, hydraulic modeling, artificial intelligence, and predictive analytics, deployed on Microsoft’s digital infrastructure (Azure, Sustainability Manager, Copilot). Its objective is to detect, quantify, and continuously eliminate real water losses in urban systems, enabling more efficient, sustainable, and resilient water resource management.
The proposed pilot city is Seville, located in the Guadalquivir River Basin, one of Spain’s most water-stressed regions, and a capital with a complex, aging water infrastructure that is highly visible to the public. Seville’s drinking water distribution network exceeds 1,500 km in length, with an average age over 40 years in many districts. Non-Revenue Water (NRW) studies report physical loss levels above 25%, equivalent to tens of millions of m³ annually that are extracted, treated, and pumped without reaching end users.
These losses are often invisible: microleaks in valves, internal pipe ruptures, household connection failures, overflows, undetected overpressures, or slow breaks under pavement that never surface. The lack of operational visibility and the system’s late reaction increase structural damage, economic losses, and pressure on surface and groundwater sources in a climate context that demands greater efficiency.
CityLeak Zero’s proposed solution is to fully digitize the network through an intelligent mesh of acoustic correlator sensors, pressure loggers, and non-invasive ultrasonic flowmeters. These devices will capture real-time hydraulic data (pressure, vibration, flow, temperature) and transmit it to the cloud via LoRaWAN or NB-IoT networks. A dynamic digital twin of the hydraulic network will be built on Microsoft Azure Digital Twins, simulating expected hydraulic behavior and contrasting it with real-world data to identify deviations associated with losses.
Using machine learning algorithms trained on historical and meteorological data, the system will identify leak patterns, transient events, and structurally vulnerable zones, and generate predictive intervention alerts integrated into an intelligent control panel. This panel will be deployed in Microsoft Sustainability Manager and operated in collaboration with EMASESA (Seville’s public water utility), enabling real-time visualization of active leaks, estimated lost volume, priority zones, and environmental and economic impact reports.
The project will unfold in four phases:
The water benefit will be quantified using the VWBA A-2 methodology, considering the difference between baseline loss flows and post-intervention registered values, multiplied by the duration of the effect. This volume will be reported monthly, with traceability of site, detection time, leak type, and associated energy benefit. Additional benefits will include savings in pumping energy, reduction of indirect emissions, extended network component life, and enhanced urban resilience to drought events.
The project will also include a citizen engagement component via a mobile app where users can report visible leaks, view savings achieved by zone, and promote responsible water behavior, reinforcing participatory governance of urban water.
The system can be integrated with EMASESA’s SCADA systems and later replicated in other Andalusian or Mediterranean cities, positioning Seville as a European benchmark for urban water digitization. For Microsoft, it represents a strategic opportunity to showcase the power of its technology stack applied to the SDGs, Spain’s PERTE Water initiative, and corporate ESG standards.
In terms of traceability, the system will operate under robust environmental and social certification frameworks. Each recorded benefit can be audited and converted into quantifiable indicators aligned with Act4Water’s Positive Water Credit (PWC) standard, enabling Microsoft’s participation in voluntary water offset markets or certifiable corporate sustainability programs.
Ultimately, “CityLeak Zero” is not just an efficiency project, it is an operational platform for achieving leak-free cities, with accountable water systems, predictive infrastructure, lower emissions, and greater transparency in managing the 21st century’s most critical resource. A public-private-technological alliance in service of water resilience.
© 2025 Aquapositive. All rights reserved. Further distribution is not permitted without authorization from Aquapositive.
