Blog

Rail electrification specialist Furrer+Frey has developed an AI tool that can analyse railway line data and present thousands of options for electrification of the route, presenting key figures such as cost and carbon for each. Now, with the backing of the UK Space Agency (UKSA), it is working with Airbus Defence and Space to adapt the tool onto earth observation imagery to make it even more effective for identifying and optimising rail electrification opportunities.

“Essentially, the goal of this project is to speed things up dramatically and make it more cost effective and therefore make rail electrification more attractive to invest in and therefore more likely to happen,” Furrer+Frey GB head of UK projects Noel Dolphin told NCE.

The UK lags behind other European countries on electrifying its railways and over the decades the rail electrification sector has gone through periods of boom and bust dependent on whether it is a government priority or not. Recent developments such as the pause of the Midland Main Line electrification suggest it has fallen down the political agenda again due to cost and other factors – something that has been confirmed by comments from former Network Rail CEO Andrew Haines and rail minister Lord Hendy.

“Rail electrification is great for increasing capacity or making journeys more reliable, but I think Treasury sees it as a huge up-front cost and very slow to deliver on returns,” Dolphin said.

To combat this, Furrer+Frey has for the last couple of years been working on an AI tool to reduce the cost of electrification that it calls Ebot: Electrification Benefits Optimisation Tool. Ebot can design an electrification scheme for any rail route with “every single possible combination of different systems, different distances between structures, different tensions and about 50 other variables”, according to Dolphin.

“It can automate an outline design for any electrification route based on real construction costs from Network Rail,” he continued. “It can not only design a route 1,000 or 10,000 times, it can also rank each of those with the construction costs and embodied carbon so at the optioneering stage you can see what system would be best for the route.”

Network Rail has already tasked Furrer+Frey with using Ebot on an already electrified line to benchmark it and see if it was designed cost effectively. Dolphin isn’t allowed to reveal which line it was used on, but he said Ebot found that “it wasn’t optimised”.

“It came up with lots of interesting findings of how you can tweak existing design rules to make them more efficient,” he added. “Running through every combination told us very quickly that making minor tweaks can have large financial savings. For example, if you can save on 100 foundations, there’s obviously saving on the cost of the steel piles but also the bigger cost saving is if you can pack up the project early.”

However, up until now Ebot has been working on manually inserted data and spitting out spreadsheets. That will all change with the addition of Airbus’ satellite data.

“Until now, someone’s gone through all the existing record drawings, route surveys or Lidar and manually input it, metre by metre, noting the structures on the route – that has taken weeks,” Dolphin said. “Albeit you get thousands of options out of it, which would normally take many years, but with Airbus’ data we can automate all of that.”

Additionally, the earth observation imagery will enable engineers to “really see what’s there”, which brings many other benefits, Dolphin explained.

“Airbus provides us really detailed imagery of the railway and it can be filtered by lots of different spectrums,” he said. “You can use the spectrums to highlight things like vegetation growth, which is useful because you can’t have trees touching live wires.

“And because the satellite is collecting new imagery every time it goes over the same spot, you can easily see the movement and changes. So you can spot, say, where the embankments are unstable, which tells us where there are trouble spots to worry about and where we will need to consider the designs of our electrification foundations.”

Dolphin said it can visualise the options by overlaying them onto the satellite imagery, which effectively enables design in the real space.

The task over the next six months, with the funding from UKSA, is to bring the automation system up to speed with the satellite imagery so that it can quickly identify things like level crossings, underbridges or weak embankments, so that it can account for these in the options it presents.

“The project runs until March next year and then we’ve got to demonstrate it as a workable product,” Dolphin said. “In the interim we’re training it on almost every unelectrified line in the UK and we will be demoing it on a couple of real lines where there are proposals to electrify but no design.”

Dolphin is confident that this project will be a success. “On top of the UKSA grant funding we, as an SME, are investing a considerable amount of money into it and we wouldn’t be doing it – and the UKSA wouldn’t doing it – unless we were all confident,” he said. “It’s already built, it’s just taking the existing tool to the next step.”

Furrer+Frey and Dolphin acknowledge people’s fears around using an AI tool to do the optioneering, replacing what has always been a human job.

“Even internally there’s a concern about whether we are just replacing engineers with AI, but I think traditionally we get graduates and apprentices to learn from doing this really basic and boring thing over and over again,” he said. “You might learn from the first three or four or even dozen times, but you’re not going to carry on learning by doing a repetitive task for three or four years.

“We want to free up people from doing the really mundane stuff so they can focus on the complex. You want an engineer to focus on something like a high-speed junction, not a simple straight bit of track where you have identical structures.”

There will also be new roles opening up with the use of Ebot.

“Some of our engineers have moved from what they did before onto this,” Dolphin said. “And there will still need to be engineers to help make decisions. It’s not here to replace people. We still need engineers, especially for the challenging elements like working with architects at historic stations or getting complex foundations into a difficult site. That’s where the real engineering is.

“Ultimately, we want more people to work in electrification because we want more electrification and a big part of this project is encouraging more investment in electrification so hopefully it ends up with more people on site building things.”

This fundamentally comes back to the main point of the Ebot project: to bring down time and costs for electrification projects, especially in the design phase.

“I think we can reduce the build costs, but also massively save on the projects costs,” Dolphin said. “If you can speed up the optioneering and outline design phase by a year and not need to have a huge project team sat in an office optioneering, that can have a huge cost saving.”

Like what you’ve read? To receive New Civil Engineer’s daily and weekly newsletters click here.