Unitywater taps data and machine learning to manage infrastructure

Queensland’s Unitywater has partnered with Pipe AI to tap into the tech company’s expertise in artificial intelligence and automation for pipe defect detection and failure prediction.

The partnership will facilitate the future development of Pipe AI — a software which uses machine learning to assess the condition of sewerage, water and stormwater pipes.

As part of the Water Source asset management series, we asked Unitywater's Sustainable Infrastructure Solutions Executive Manager Amanda Creevey why this partnership is an important step for the utility's asset management plans leading into the future.

Unitywater provides essential sewerage and water supply services to more than 17% of Queensland’s population and operates infrastructure in Moreton Bay, Sunshine Coast and Noosa, including 6165 km of water mains and 5975 km of sewer mains.

Creevey said the utility’s partnership with Pipe AI is a natural step towards an intelligent network for 24/7 water and sewerage operations, which will ultimately lead to better service for customers.

“By using smarter solutions like Pipe AI, we can manage our assets in efficient and cost-effective ways. Unitywater’s sewer main CCTV review process will transition from manual to automation,” she said.

“Manual effort will be redirected to remediation programs, and our ongoing maintenance and renewals work.

“Smart solutions will also improve our asset risk models by using real-time data that is more accurate and available sooner. Training machine-learning models for our water and sewer pipeline assets will predict the likelihood of asset failure before it happens.

“Combining this with our consequence data we can then prioritise interventions based on risk.

We’re always looking for ways to better serve our customers and we do that best when we save on maintenance and repair bills and minimise disruptions to the customer.”

Developing the future of water asset management

Creevey said Unitywater has been involved with Pipe AI from the beginning of the tech company’s software development, and has helped train the existing software models.

“Learnings from early trials have been embedded in the Pipe AI software to address key challenges faced by utilities when it comes to data quality. We will continue to work collaboratively with Pipe AI to refine the Pipe AI Predict and Pipe AI Review models,” Creevey said.

Pipe AI Predict has been developed to focus on risk classification for pipeline assets, and is designed to be used in conjunction with Pipe AI Review to automate the process of risk attribution to sewerage, water and stormwater pipes.

“For us, asset management is about managing risk and being responsive to change, whether it is increased failures in the network, more complaints or new and emerging trends in our data,” Creevey said.

“Pipe AI will be a single source of truth for asset risk and ultimately the tool that will be used to prioritise investment across our pipeline assets.”

Using predictive analytics and historical pipe maintenance and failures data, Pipe AI Predict creates maintenance schedules for repairs and replacements to ensure the maximum life of the infrastructure can be achieved.

Data transformation

Unitywater’s asset information strategy is driving the utility’s approach to data-driven decision-making, Creevey said.

“One of the key principles we have is to only process data manually where it is not cost effective to automate. [The Pipe AI] partnership opens further opportunities to automate the end-to-end management of condition data which has been a challenge for many years,” she said.

“We continue to digitise our asset base as opportunities arise. This is both digital representations of physical assets and digitising and storing asset information into a dedicated technical library that is searchable and integrated with our asset management information system.

“We will continue to focus on adopting a data-driven approach to our decision-making and building out our digital assets to solve complex problems and continually improve our operations.”

Bolstering this data-driven approach is the utility’s move towards smart metering, with 10,000 smart water meters being installed on the network to produce real-time data about customers’ usage, as well as Unitywater’s pipes, stations, and water quality.