Water infrastructure is the lifeline of every community. Adequate water service relies on resilient water systems. This session will report on our ongoing research and development project, in the context of a water system digital twin, for Smart Water Grid Anomaly Detection and Localization. It is funded by Singapore National Research Foundation (NRF) under its Competitive Research Programme (CRP) (Water) and administered by Public Utility Board (PUB), Singapore’s national water agency.

1. Digital Twin Intelligence and Smart Water Grid, Zheng Yi Wu, Bentley Fellow, Bentley Systems Incorporated, Watertown, CT, USA

With the adoption of smarter sensors and temporary and/or permanent data loggers increasing, it is imperative to leverage a digital twin paradigm for improving the quality and efficiency of water service. This talk will present an overview of digital twin intelligence for advancing infrastructure in general. The presentation will also exemplify the water system digital twin developed for the NRF-funded PUB project and how its outcomes can help water utilities improve smart water grid operation by quickly detecting, localizing, and evaluating anomaly events. Water utilities can then take corresponding actions to prevent and mitigate the impact of possible water service disruption.

2. Hydraulic Time Series Data Analysis for Anomaly Detection, Jocelyn Pok, Software Engineer, Bentley Systems Singapore.

To embark on a smart water grid initiative, it is essential to monitor flows and pressures throughout the distribution network.  The analysis of the monitored time series data will enable water utilities to understand the system and detect anomalies.  This talk will elaborate on the research and development of hydraulic data analysis as the integral component of a water digital twin.

3. Acoustic Data Analysis for Anomaly Detection, Rony Kalfarisi, Software Engineer, Bentley Systems Singapore.

Detecting anomaly events, such as hidden pipe bursts or leaks, is a challenging task. In addition to monitoring pressures and flows, acoustic sensors, such as hydrophones, are used for recording the acoustic signals generated by the leaks and propagated through water flows and pipelines.  A comprehensive analysis solution has been developed for prescreening acoustic data quality, visualizing acoustic data, and detecting anomaly events. 

4. Integrated Data Analytics Framework for Anomaly Detection, Jianping Cai, Senior Software Engineer, Bentley Systems Singapore.

An effective data-driven model is the integral component of digital twins for smart water grid operation.  Integrated software has been designed and implemented for near real-time anomaly detection using hydraulic time series data (flows and pressures) and high-sampling rate acoustic data.

5. Data-driven Model Training and Validation for Near Real-time Anomaly Detection, Meng Xue Software Quality Analysts, Bentley Systems Singapore.

To validate the effectiveness of developed methods and software for anomaly detection, it is imperative to test them on large datasets. This presentation will demonstrate how to apply the developed software to train the data-driven models for SWG anomaly detection in near real time.

6. Daily Model Calibration with Water Loss Estimation and Localization, Alvin Chew, Research Engineer, Bentley Systems Singapore.

Hydraulic simulation models have been conventionally used for the design and maintenance of water distribution networks. With continuous monitoring data, the hydraulic model can be calibrated in near real time for smart water operation, especially when estimating water losses and localizing anomaly events, such as hidden pipe leaks, before they evolve into disruptive water main breaks.