Picture: Brian Maloney, Founder and CEO
Storm Harvester's Intelligent Sewer Suite POC has won the Water Industry's Most Innovative New Technology award.
The Judge's said...
"This project tackles a difficult and often neglected problem by using innovative, machine-learning based technology to provide prognostics. The co-operation witha utility company and the trial data is impressive, and the technology will become increasingly important as the sector delivers mitigations for climate change."
The initiative combines StormHarvester's multi-layered machine learning approach with hyperlocal rainfall forecasting to provide each wastewater asset (sewer pipes, pumps etc) the ability to "look six hours into the future". This enables those controlling the assets to amend asset behaviour and optimise future performance, for example early blockage detection.
By combining data from the wastewater network with hyperlocal rainfall predictions, Intelligent Sewer Suite's machine learning technology gives unprecedented accuracy in combined sewer overflow (CSO) and pollution/flooding event predictions, as well as a range of other benefits in network management performance analysis, and reporting.
Wessex Water ran a PoC in Q2/3 2020 across Bath including 89 assets (nine pumping stations) with the aim of using machine learning algorithms and AIto better separate genuine control room alarms from alarms triggered simply because of the high volume of rainfall during wet weather, predict surcharging sewers before they become service failures (pollution or flooding events), and dramatically improve the response time to service failures. StormHarvester became a finalist and deployed its Intelligent Sewer Suite product over Q2 2020. Wessex provided API access for its historic sewer and pump(level and flow sensor) database along with historic rainfall patterns. Bringing AI (or machine learning) to the wastewater industry, to make sewers 'intelligent', is effectively a new category, so competing technologies are very limited. Storm Harvester believes it has established best practice. Bespoke machine learning techniques have been honed over five years and include sewer cluster analysis, level gradient mean analysis and various regression techniques.
Intelligent Sewer Suite has the potential to be applied across any sewer network, pumping station or wastewater treatment works globally, with the only condition being that there are least some level sensors in place. Certainly, widespread adoption across the UK could significantly reduce outages and improve environmental compliance.
The multi-layer machine learning suite learns asset behaviour based on asset history, predicts the future of the asset six hours ahead, so there is no need for hydraulic modelling of any kind. And the smart flow prediction element of the product also enables various process optimisations with pumps and wastewater treatment works.