SA Water Transients Audio 1
Fast and repeated pressure events can reduce the lifespan of water mains and may result in breaks in weakened sections. However, the presence and source of these pressure events is difficult to identify and prioritise for targeted field investigation. SpiralData and SA Water have co-designed a beta solution for transient detection and characterisation using machine learning that is fast, scalable, and reliable, with minimal human input on an artificial intelligent platform.
OUR APPROACH
SA Water engaged experts in algorithms for water networks, SpiralData, commencing with a series of workshops with subject matter experts (SMEs) to:
clearly define key pain points
co-design a solution using an Agile framework.
The key outcome from the workshops was a shared understanding of the significant value in finding transient events in a water network to enable identification of root-causes.The POC was run as a series of milestone sprints starting with data discovery, followed by the development of a big data snipping algorithm and a machine learning algorithm for clustering shapes. Sprints de-risked the investment and ensured continual course correction by SMEs to a workable transient detection solution.
OUTCOMES
Big data challenge - fast, reliable method to identify transients
An efficient and reliable method of extracting both transients and significant edge cases from the high-frequency (128 Hz), big dataset (5 TB of raw data)
Machine learning delivers automation and productivity gain
A method to rank transient events by severity including magnitude and frequency, reducing the effort by SMEs to interpret transient events by a factor of 15-20
Reproducible results
When tested by SMEs for two sensors over one month, the large magnitude transient extraction was 97.5 percent accurate
WANT TO HEAR MORE?
Visit us online spiraldata.com.au
Rapid troubleshooting tool
Fast and repeated pressure transients can reduce the lifespan of water mains and may result in breaks in weakened sections. The reliable and accurate identification of repeatable transient events, and the ability to prioritise the events, has given insight previously not available.
FUTURE APPLICATIONS
The monitoring of sites used in the POC is continuing and expanding to new geographical locations to refine the tool and increase applicability, thereby reducing pipe fatigue, leaks and breaks informing strategic planning investment. This aligns to SA Water’s Strategy of ‘Driving Customer outcomes,’ and the related goal of reducing unplanned customer interruptions and third-party disruption.
The next step is to turn the artificial intelligence platform into an operational tool for SA Water and the water industry more widely, so that triage can enable field investigation and mitigation of root causes, resulting in a calmer network, improving asset life and potentially reshaping the design process for asset planning.