University of Exeter logo

Research and Innovation blog

Home About Contact Toggle navigation Open menu

Machine learning to reduce flooding as a result of sewer blockages

20 September 2019

2 minutes to read

Machine learning to reduce flooding as a result of sewer blockages

Flooding and pollution as a result of sewer blockages are a major cause of concern for water companies. Blockages not only cause inconvenience and distress, but can be environmentally damaging and are costly events to rectify.

As part of a Knowledge Transfer Partnership between the University of Exeter and DC Welsh Water, the sixth largest water company in England and Wales, a research project was undertaken to determine if a machine learning tool could be developed that could more accurately predict and identify potential blockages across the sewer network.

Identifying blockages

Previously, DC Welsh Water had been identifying likely blockage locations through a review historical data alone. This system, although accurate to a degree, was not sufficiently precise which in-turn led to unnecessary blockages throughout the network. Through their research and a review of this historical data, the team were able to develop a set of machine learning techniques to more accurately model the propensity of sewers to block based on a variety of factors, including the age, construction and condition of the sewer in addition previous incidents reported on that sewer.

This system was embedded in DC Welsh Waters GIS maintenance planning system and provides a colour-coded layer of blockage prediction. This system is currently being used by the operational team at DCWW to inform the selection of sewers to undergo proactive and planned maintenance.

Machine learning

Since it was embedded into the GIS maintenance planning system, the machine learning tool has contributed directly to the strategic operation of DCWW (serving 3 million customers in England and Wales through 30,000km of sewer main). The machine learning system is being used in the central operations hub in the company’s Linea office in Cardiff, providing the proactive maintenance team with predictive information on the likely blockage hotspots around their network. This information is being used to direct crews to areas of greatest blockage risk, allowing them to systematically prevent blockages from forming and moving from a reactive to a preventative strategy.

This development has led to a 27 per cent reduction in blockages since 2015 for DC Welsh Water, representing a significant reduction in costs and regulator penalties for the water company.  In addition, this innovation has led to fewer incidents suffered by customers and the environment.

 



Researchers

Collaborators

DC Welsh Water
Back home
TOP