2022 – AI for Dam Safety Monitoring

Sam Banzi – WaterNSW, Mohsen Askari – Adasa Systems, Josep Selles – Adasa Systems, Mike Ahmadi – WaterNSW

As the owner of most of New South Wales’s large water supply dams, WaterNSW prioritises the ongoing safety of its dams to protect people, property, and the environment from the harmful effects of failure or misoperation of its dams and reservoirs. As a learning organisation, WaterNSW is constantly scanning for good practices from other dam owners worldwide and other major hazard industries, including aircraft, railway and nuclear. The benchmarking effort has enabled the business to define its desired state in management systems and processes, technology applications, advanced analytics, and data science. As such, WaterNSW is driven to contribute to developing and implementing leading-edge industry practices actively.

WaterNSW views data as the bedrock of a risk-informed, data-driven decision support system for dam safety. Hence WaterNSW built DamGuard, a bespoke mobile and cloud web-based data management solution. This replaced the traditional data capturing the process of using handwritten surveillance sheets for mobile capture and storing data in real or near real-time from both manual and SCADA information. This information is stored in a data lake.

This paper discusses the application of Artificial Intelligence (AI) and Machine Learning (ML) in maximising the value of the historical dam data collected by DamGuard and the previous systems. This approach seeks to develop and incorporate more meaningful types of alarms to enable timely and effective decision-making by dam safety engineers. By analysing decades of accurate historical data, several AI and
ML algorithms are developed to understand the patterns and correlation between different measurements such as seepages, uplift pressures, storage water level, rainfall etc. These help to detect unusual behaviours and generate more reliable and meaningful alarms.

Buy this resource