End-to-End AI Application for Water Supply Chain Optimization: NRW, Leak Detection, and Resilience Using SIOTH

Context

Advanced Solutions department — PFE 2026–2027

This project aims to develop an end-to-end AI application covering the full water supply chain, from production and transmission to storage and distribution, in order to reduce Non-Revenue Water (NRW), detect leaks and strengthen supply-chain resilience and continuity.

The trainee will integrate and analyze operational data from production plants, pumping stations, reservoirs and distribution networks—including flows, pressures, levels, demand and metering data—together with GIS data such as network topology, pipe attributes, asset locations, elevations and DMA boundaries. SIOTH and Python with relevant libraries will be used. The study will apply water-balance methodologies, anomaly detection, demand forecasting, spatial analytics and scenario analysis.

Main topics and deliverables

• Collect, integrate and preprocess supply-chain data in SIOTH across production, transmission, storage and distribution assets.
• Integrate GIS data and link spatial assets to operational tags.
• Develop water-balance and NRW-quantification models distinguishing real and apparent losses by zone or DMA.
• Develop leak-detection and localization models using flow/pressure anomaly detection, Minimum Night Flow analysis and Machine Learning techniques.
• Apply GIS-based spatial analytics to localize leaks and losses, prioritize pipes by risk based on age, material, pressure and failure history, and assess disruption impacts by zone.
• Develop demand-forecasting models for production planning and storage management.
• Assess supply-chain resilience and continuity through disruption scenarios including plant outages, pipeline failures and demand surges, as well as storage autonomy and redundancy analysis.
• Generate AI recommendations for loss reduction, leak-response prioritization and continuity actions.
• Build interactive SIOTH UHMI dashboards with GIS map views, forecasts, alerts, KPIs and recommendations. KPIs include NRW percentage, ILI, MNF, supply-demand balance, storage days and resilience index.
• Produce technical documentation, a final report and a presentation covering methodology, models, results and potential business impact.

Keywords

AI, Python, water supply chain, leak detection, Non-Revenue Water (NRW), spatial analytics, water balance, demand forecasting, Machine Learning, supply-chain resilience, business continuity, anomaly detection, GIS, SIOTH, SIOTH UHMI.