We are always looking for ways to improve our processes and understand our network better.
Working with industry specialists Stormharvester, we’ve harnessed machine learning and more than 1,300 sensors across our network to keep track of rainfall, make predictions and allow us to intervene to prevent spills.
Using more than 1,300 sensors across our assets - everything from storm tanks to pumping stations to manholes - we’ve been able to collect data and use hyperlocal rainfall forecasting to enable us to increase our blockage detection and asset performance, while reducing spills, pollutions, and flood events.
The system collects data form each sensor and machine learns the normal operating thresholds for that site, it can then recognise and highlight anomalies in near real time.
In the first six months of operation, the project with Stormharvester alerted us to possible incidents which resulted in 110 teams being dispatched to investigate further, and 39 pollutions were avoided.