As part of their Smart Water Grid (SWG) program to monitor more than 6,000 kilometers of Singapore’s pipeline, PUB initiated a grid anomaly detection and localization project for near real-time analysis of SWG monitoring data and detection and localization of anomaly events. While permanent sensors offer 24/7 monitoring, anomaly and leak detection practices still required biannual costly, labor-intensive network surveys. To make their anomaly and leak detection practice more cost effective, PUB sought a machine learning-based, data-driven approach.

Leveraging Bentley’s iTwin and OpenFlows applications with AI and machine learning, PUB developed an integrated anomaly leak finder technology solution, using hydraulic models recalibrated daily with monitoring data to generate a high-fidelity digital twin. Since January 2024, it has led to the detection of two significant underground leaks, alerting PUB’s leak detection crew and localizing the leaks to less than one kilometer. The intelligent digital twin solution shifted PUB’s previous time-driven network surveillance schedule to data-drive-driven, reducing total workforce costs, supporting proactive processes, and redefining the concept of high-fidelity digital twins for smart water grids.