Predicting Water Quality: A Vital Need
Melbourne Water, a Victorian-owned corporation, is embracing cutting-edge technology to forecast recycled water quality two days in advance with an impressive 75 percent accuracy rate. This innovative approach aims to provide timely alerts to the various users, including farmers, businesses, and households, who rely on Melbourne Water’s Class A recycled water for non-drinking purposes. Anticipating water quality is crucial as it enables proactive measures when quality deviates from the standard.
IoT Sensors and Data Insights
Before venturing into predictive analytics, Melbourne Water had already made significant investments in IoT technology and developed an in-house virtual model for real-time insights into treatment plant discharges. With over 700 sensors and meters in water treatment plants, the organization was well-equipped to collect real-time data about recycled water quality.
The Path to Predictive Analytics
To transition from merely gathering real-time data to predicting water quality in advance, Melbourne Water embarked on an ‘Intelligent Network Enablement’ program. In collaboration with AWS partner Arq Group, they took a comprehensive approach to unify data sources and leverage cloud-managed services. The program aimed to create a single platform to view all data sources, utilizing services like Snowflake and AWS to build a data lake and analytics platform known as the ‘unified data store.’
Digital Twins and Machine Learning
Arq Group leveraged the unified data platform to construct a virtual model, referred to as a ‘digital twin,’ representing Melbourne Water’s recycled water production. This digital twin relies on AWS IoT SiteWise to gather and analyze real-time IoT sensor data, laboratory data, and weather data. These insights are then fed into Amazon SageMaker, which employs machine learning models to predict the impact of various factors on water quality.
Proactive Water Quality Management
Thanks to these innovative efforts, Melbourne Water has transitioned from a position of having real-time data to being able to predict water quality two days in advance with a remarkable 75 percent accuracy rate. This transformation enables the utility to detect water quality issues swiftly, allowing for proactive measures and the timely resumption of water production.
A Broader Digital Transformation
Melbourne Water’s journey into predictive analytics aligns with a broader expansion of its IT, OT, and IoT capabilities. With plans to delve into real-time video analytics for drones, the organization continues to advance its intelligent network enablement program, building on the successes of previous digital transformation initiatives.