Building on the success of our EDM intelligent sewers challenge, we are launching a further challenge as part of our vision to develop intelligent wastewater networks. For this challenge, we are looking to apply machine learning to data from our rising mains and sewage pumping stations (SPSs). We want to use machine learning to help us identify and respond to issues at these assets faster, thereby reducing the likelihood and impact of pollution incidents.
What are rising mains and sewage pumping stations (SPSs)?
Wessex Water is responsible for around 35,000km of gravity sewers, which transfer wastewater from homes and businesses to water recycling centres (WRCs) using gravity. Here the wastewater is treated before being discharged back into the environment.
Where the topography does not enable our sewers to flow by gravity, we use sewage pumping stations (SPSs) to transfer flows over hills using pressurised pipes called rising mains.
Wastewater arriving at an SPS accumulates in a large tank called a wet well. Once the wastewater reaches a certain depth in the wet well, it is pumped out through the associated rising main. At the other end of the rising main, the wastewater is discharged into a gravity outlet.
We measure various parameters at SPSs (eg, wet well level and pressure) to ensure they are operating appropriately. Measurements are transmitted back to a central control centre in the form of ‘telemetry signals'.
Detecting rising main bursts and SPS issues
Our current approach
If a rising main bursts, or if there is an issue at an SPS, this can result in flooding and pollution. Bursts may not be visible or obvious from the surface so telemetry data is key to detecting these incidents.
We currently use in-house analytics to identify when rising mains have burst and detect anomalies at SPSs or upstream in the network. To some extent this is predictive, i.e. identifying issues as they develop.
We can then direct operational crews to investigate. However, the level of triaging involved in the in-house analysis limits the speed of our response.
Applying a machine learning approach
Machine learning tools are available in this area. As such, we are looking for a machine learning solution that uses real-time wastewater telemetry data to identify when the behaviour of an SPS or rising main deviates from the expected range for the prevailing conditions – which may be an early indication of a problem. The solution should then generate an alert.
In this way, a successful tool would allow us to:
- more rapidly identify bursts and SPS failures that occur, thereby improving response times
- identify developing or imminent issues, allowing us to intervene and reduce the likelihood of a rising main burst, SPS failure or other incident occurring.
As well as a potential reduction in rising mains bursts and SPS failures, we see this approach unlocking numerous other benefits. Our focus would shift from reactive maintenance to proactive maintenance, helping our assets perform better. We would also expect to see fewer alarms in our control room, meaning better response times when incidents do occur.
How to get involved
We are inviting suppliers to use our historical data to develop and demonstrate this predictive machine learning capability. Any solutions must outperform our in-house analytics.
To facilitate this, we have published asset information and historical telemetry and rainfall data for 24 sewage pumping station sites in our region. The two years of data represent a dry year (2022) and a wet year (2023) and include instances of rising main bursts.
We are looking to take up to three solutions forward to proof of concept (POC) trials using a near real-time data feed to test their capabilities. Funding of up to £15,000 per POC is available.
We're running a Teams session on Monday 21 July 2025 for interested parties to find out more, ahead of the deadline for proposals on Wednesday 24 September 2025.
For further details, including registration for the Teams call, requirements for proposals, how we will assess them, and information about the POC trials, please see our additional information document.
We look forward to hearing from you!
Next steps
We will review all proposals to identify those we would like to explore further. We will invite shortlisted suppliers to face-to-face meetings (currently planned for November 2025) to determine who we would like to take forward to the POC stage.
We'll provide feedback to all suppliers. We also plan to give more general feedback on the challenge via our blog page.
If you want to hear about other Marketplace challenges over a range of areas when they are released you can sign up to our mailing list.
Please note that participation in this Marketplace challenge does not guarantee a contract at the end. A tender may be required, and Wessex Water reserves the right to stop the project at any time.