Engineering consultancy Arup London worked with client Network Rail High Speed on an automated solution for tunnel inspections to minimize electrical resources and isolations, reduce track time and collect comprehensive data on the state of the assets.
Article provided by Noemi Roecklinger, iData Analyst, Laser Scanning, Computer Vision and Machine Learning at Arup London
Each year, a total of £8 billion is spent on repairing and maintaining infrastructure in the UK; with inspections alone costing up to £100million. To date, the majority of tunnel inspections are manual and labour-intensive, with inspectors working at night in potentially high health and safety risk conditions. Traditionally, inspectors use a pen and paper approach, manually capturing faults while walking through the tunnels, supplemented with selected photos of the faults.
Understandably, examining over 43km of High Speed 1 (HS1) tunnels requires a lot of resources and planning on the part of Network Rail (High Speed) or NRHS, a client of one of the consultancy firms in engineering. Arup (London). Skilled personnel, track possessions and maintenance vehicles must be reserved up to two years in advance. The NRHS has identified the effectiveness of traditional tunnel examinations as an opportunity for research and development (R&D). Arup helped the NRHS develop approaches to automate inspections. These minimize electrical resources and isolations, reduce tracking time, and collect comprehensive asset condition data.
An automated tunnel inspection and predictive maintenance system
Arup proposed the design, construction and deployment of an automated tunnel inspection and predictive maintenance system using computer vision techniques. Arup’s R&D system consists of two parts: first, a remote-controlled modular construction of a unit mounted on a maintenance vehicle housing an array of sensors. Arup has tested Leica and Faro scanners, FLIR thermal cameras and 360 degree cameras such as the Insta360 Pro. The company has also worked with University College London (UCL) to develop a bespoke camera platform using omnidirectional GoPro cameras. All of the above had to be designed to be compatible with existing NRHS systems. The camera systems provide high quality video images that are equivalent in quality to photos taken during manual inspections. The system captures approximately 1 km of tunnel every ten minutes of operation.
Second, Arup developed a bespoke dashboard and immersive viewer to visualize and analyze the tunnel. The web-based dashboard displays captured data in 3D and enables remote virtual inspections. While working safely from the office, the inspector can mark new defects, search historical data and perform comparative analysis. They are supported by Arup’s Machine Learning (ML) algorithm, which identifies features in imagery, such as line-side equipment and fault lights. The ML algorithm and associated computer vision workflow includes feedback loops to gradually improve the algorithm’s ability to correctly identify features.
The development of Arup’s automated inspection approach was a collaborative effort working closely with the NRHS to better understand their challenges and needs. Through a series of interactive workshops, Arup mapped its current workflow and identified opportunities for improving efficiency. For example, Arup found resource shortages to be a significant bottleneck. Their modular hardware system facilitated a greater possibility of collaborative working between different disciplines, i.e. the sharing of maintenance vehicles, assets and isolation personnel. Collaborating on existing possessions reduces resource demand, isolations, and track time. Additionally, using 360-degree video data capture, Arup’s automated inspections could reduce the number of night shifts from over forty to six. This drastically reduces “starts on ballast” and costs, while collecting more comprehensive asset condition data.
Through the workshops, Arup also identified the need to improve rapid incident response times. Replacing manual capture of defects at separate locations with continuous 360-degree video capture allows engineers to plan their response more efficiently. When every minute of delay counts, examiners benefit from comprehensive and frequent tunnel recordings. The data is more objective, reproducible and of comparable precision. Through benchmarking against historical NRHS datasets, Arup confirmed that they reliably pick up defects like human inspectors would. The image dataset gives a better understanding of asset degradation, enabling better planned interventions and avoiding unplanned closures. The digital dashboard serves as a quick scouting tool to flag where conditions have worsened.
This project ultimately demonstrates how the remaining life of an infrastructure asset can be managed, maintained and understood more effectively. It also moves away from ‘boots on the ballast’, keeping personnel away from potentially hazardous environments, and minimizes disruption to site work. With Arup’s automated tunnel inspections, new, unconventional ideas are embraced to change the industry for the better.