Inroads in sewer system efficiency with first-of-its-kind machine learning
University of Exeter KTP makes inroads in sewer system efficiency with first-of-its-kind machine learning
A recent Knowledge Transfer Partnership between the University of Exeter and South West Water has developed a landmark AI tool to improve the efficiency of sewage systems. They are now considering a delivery model to continue the project to develop this technology for the entire water industry.
Company: South West Water Ltd.
Industry Sector: Water services.
Size: Large enterprise (1,650 Employees)
The UK’s vast network of over 525,000km of sewers are notoriously expensive to maintain. Currently, sewer pipes’ conditions are assessed by time-consuming CCTV surveys, where trained technicians work to identify each pipe’s faults and features. However, these manual surveys can sometimes be unreliable, due to a number of factors.
To improve survey efficiency, South West Water – a company which provides drinking water and wastewater services throughout Devon and Cornwall – wanted to develop an AI tool capable of accurately detecting faults in sewers. This will assist in the effective maintenance of the network, supporting positive outcomes for SWW customers and the environment.
The KTP emerged as the perfect vehicle to transfer existing research into a ‘real world’ environment, by using South West Water’s wealth of archived CCTV footage to improve the efficiency of automated annotation software.
For the KTP, Associate Josh Myrans was recruited to help develop and implement the new AI tool, building on his PhD research into detecting sewer faults using computer vision and AI technologies, and with ongoing academic supervision from Prof Zoran Kapelan (KTP project lead) and Prof Richard Everson.
The Key Results:
The two-year KTP project developed a hugely promising AI-base software solution, with the ability to detect over a third of the UK standard classified faults with an impressive accuracy of 90% . Thanks to these successes and the algorithm’s future potential, the project has since received prestigious Future Leaders Fellowship funding and support from a network of key industrial partners to develop this technology for future commercialisation.
— The fledging software has been developed to a prototype tool with a dedicated user interface, which the company now aims to develop to full UK industry standards. Currently, the AI tool is reliably achieving over 90% detection rates  and 70% type classification rates  on the most common 18 sewer faults.
— This technology will help South West Water to manage faults more efficiently and increase the reliability of sewer network condition assessments. This will assist in the prevention of sewer collapses, pollution and flooding incidents, and negative customer experiences. As the KTP’s associate Josh Myrans put it, “with the technology, it’s a case of recognising the situations where these surveys are likely to be impacted – by poor weather conditions, for example – and allowing the technology to intervene. In turn, this frees up staff capacity and can improve the accuracy of sewer surveys.”
— The company expect to see benefits and positive impact through the reduction in resource required to annotate surveys and possible re-inspections, as well as improved health, safety and wellbeing by reducing the time spent in hazardous conditions, as they fully integrate the software solution within their working practices.
— The associate has been awarded a prestigious UKRI Future Leaders Fellowship to continue this line of research from within South West Water (SWW), alongside other partner organisations. They will use this UKRI and SWW funding (£587k) to drive forward the AI software solutions capabilities to meet full UK standards and begin the process of integration with hardware to explore the possibilities of autonomous sewer inspection devices.
Benefits for the Company:
“South West Water continue to make significant strides in this field of innovation and the company is increasingly regarded as the leading UK water company for the development of AI automation of sewer surveying”
South West Water now has a prototype AI tool, ready for initial trials across the company. This will assist in identification of faults, and (when fully developed to UK standards) improve the operational efficiency of internal CCTV surveying processes.
Working so closely with the technology’s development throughout the KTP has helped SWW to develop and enhance their CCTV survey management and archiving practices. Furthermore, the company’s current practices in this area have undergone a detailed data quality review, identifying areas for improvement with the implementation of this technology.
Finally, through the industry engagement seen in the project, South West Water continue to make significant strides in this field of innovation and is increasingly regarded as the leading UK water company for the development of AI automation of sewer surveying.
Benefits for the Associate:
“The KTP has been a career defining process and a massive boost to my professional prospects.”
“The KTP associate role has enabled me to make genuine progress with my professional development and working in an academic-industrial partnership, I have built a rapidly expanding network of contacts for future collaborations and continued evolution of practical deployment of my research.
“In January 2020, for example, I presented what we’d achieved at a conference for the wastewater industry. A day’s session there was devoted entirely to this topic, which gave me the chance to interact with the whole industry and [between us] figure out what we wanted from this kind of technology. This level of industry exposure has not only increased my understanding of the world in which my research fits, but the confidence to develop and disseminate research in a whole new environment.
“The KTP’s success is best shown by the prestigious Future Leader’s Fellowship award. This personal funding will allow me to continue leading the way in developing automated technologies and defining the future of surveying practices.
“My favourite aspect of the work is that, essentially, I get the best of both worlds. I’m able to be involved with discussions about cutting-edge technology and the latest academic developments, while I equally take some of those technologies into the world and see how their practical application works. Thanks to this process, I’ve been able to see the fruits of my labour, which is something I’ve really enjoyed.”
Benefits for the Academic Team:
“The KTP is a great example of the possible achievements of close collaboration between industry and academia”
Over the course of the KTP project, the academic team (Professors Kapelan and Everson) has seen its research input into the software solution flourish, embarking on the development of a real-world tool with the potential to deliver significant impact in the water industry. This success has led to further industry engagement, acting as a great example of the possible achievements of close collaboration between industry and academia.