Automation is now routine in most factories. However, it is harder to implement in more variable environments. The nuclear sector is tackling how to use robots in processes like this.

Wherever we can see a repeatable step in a process, we are able to design robots to automate them. But some parts of the economy aren’t predictable. Fruit pickers, distribution centre workers, and recycling plants are presented with a stream of materials that must each be considered anew. Every movement of a hand towards a raspberry, or of a vacuum gripper onto a package, must be analysed and modified. 

Much of the un-automated part of the economy is like this. The next step in automation is to develop robots that can analyse and optimise tasks in a highly variable process in real time, as well as a human. This will take a combination of physical and digital tools. In some of the world’s oldest nuclear plants, now undergoing decommissioning, this process is driving innovations that will shape the future of automation.

Demonstrating the benefits of robotics

Atkins have, over the last four years, been working with clients in the nuclear sector on a series of ‘active demonstrators’. These have been practical tests of robotic tools’ ability to handle tasks in a highly variable environment, where safety is always at the front of everyone’s minds.

Christian Pilon developed his skills in robotics working in the aerospace and packaging industries. “I always like to start with the three D’s of robotics: dirty, dull and dangerous tasks,” he says. “if it’s one of these tasks, we think a robot should be doing it. And we happen to have many of these tasks in nuclear. So this is what’s enabling the business case, the reason for automating and using robots.” 

The challenge though is that this sector is far less predictable than those where robots now dominate. But that is changing, Pilon says. “There’s an element of technology advancement that enables robotics for the nuclear sector right now. If you look at other industries, we see robots being excellent at repeatable tasks. This is perfect for robotics to enable automation, because you have a low requirement for intelligence in the robot, if you will. You can have a very simple system doing the same thing all day long, in a very standardised environment.”

That’s not the case in nuclear decommissioning. In the UK, Sellafield is one of only two plants in the world currently dedicated to decommissioning. The plant went through many iterations, including, in the late 20th Century, being used to reprocess waste from other reactors. Today, all of the materials at the site must be sorted, packaged safely, and stored. Eventually, they will be placed into long term storage.

One of the first tasks at any nuclear plant is to inspect all of the many buildings used in the nuclear sector. For this, at Sellafield, Atkins have been making use of Boston Dynamics robot dog, Spot, and other mobile robotics. 

These have reduced the risks associated with humans inspecting potentially hazardous spaces. And it has saved time and money, as Spot can enter these areas without needing to don a protective suit. As Spot and other mobile robots explore the site, there are some tasks that are already being automated, as part of the active demonstrator programme underway at the site.

Cut to fit

Some waste at Sellafield is stored in skips. These are metal boxes, containing some radioactive materials, and a lot of empty space. They are currently being stored in pools at the site. But they must be moved first to temporary storage, and eventually into a permanent site. The space needed to store them for centuries will be hugely expensive. By cutting and repacking them, making better use of space in each skip, these long term costs will be reduced.

If these were identically sized and in identical condition, this would be an easy task for robots. But Sellafield has been in use for decades. It is a place where lessons have been learned. And, in its later years, it received radioactive materials from many other sites.

Robert Marwood joined Atkins from Sellafield, where he had worked on the decommissioning process. Seeing the potential of early uses of robotics at the facility, he wanted to work more closely on their development, and potential deployment beyond the nuclear sector. It is, he explains, an ideal environment for testing new approaches to automation.

“There’s a fleet of reactors across the UK from the 70s and 80s called Magnox power stations. They produced nuclear fuel, andnd that fuel either has to be disposed of or, in the case of the UK, we decided to try and recycle that fuel. So Sellafield had a programme of reprocessing waste. The fuel from the Magnox reactors was sent to Sellafield where it had to be stored in ponds under water because that keeps it cool and protects the environment from radiation. So to store the fuel, under water, it was placed in skips.” 

The task now is to remove those skips. They are all classified as ‘intermediate level waste’ or ILW. Not nuclear fuel, but building surfaces, machinery and other materials from the site that have come into contact with radiation. They must be size reduced, packed more efficiently, before storage. While humans could put on diving suits and size reduce them in their storage pools—this is not highly radioactive material—the number of skips and the cumulative dose that these workers would receive is today consider unacceptable.

“They’re essentially a hollow box that isn’t very efficient for storage,” explains Marwood. “There’s a need to try and come up with ways of making that long term storage more efficient. You have to build a building to store them in and then keep that building running for at least 100 years, until we’ve got an underground storage facility developed.” 

This will be an expensive facility, and size reducing the skips will mean it can be smaller, and more efficient. The solution being demonstrated as Sellafeld is largely based on typical industrial robotics you would find in a car manufacturing plant 

“It’s industrial robots, with laser cutting tools,” says Pilon. “It’s something that in a different industry, is pretty standard technology. It’s quite new for nuclear. There’s an extra level of preparation and management when you implement robots on nuclear sites. That’s the challenge for us.”

The skips will be packaged for safe transportation. When they arrive onsite the packaging is removed, they are placed on a rolling conveyor and clamped to a rotating table. 

“A small welding robot, like you’d find on an industrial car factory-floor, cuts the skip into pieces,” says Marwood. ”It cuts a skip into five pieces. Another larger robot holds the piece you’re cutting off the side of the skip with a magnetic gripper. Once you’ve cut it, it holds it in place and then packs it into an export skip. So essentially, you’re cutting a box into five sides and then packing it into another skip for export.”

The skips come in from the pond, two of them are cut up and placed in the third skip and sent on. That reduces the size of the skips by two thirds, making it possible to reduce any storage facility by a similar scale. That has real financial benefits.

“It will cost something like £100,000 per cubic metre of ILW stored, for its lifetime,” says Marwood. “For every skip you manage to pack into another you’re saving tens, if not hundreds, of thousands of pounds of lifetime costs.”

In order to cut the skips, the robots must examine each one individually. In the current demonstrator, a human supervisor checks the planned cut paths against the actual skip. The data captured from each skip, can then help automate the process. And it can provide insights that inform the broader decommissioning project.

“By automating the cutting of the skips with robots, we can also take the opportunity to capture data on the skip,” says Pilon. “We can, for example, use a 3d scanner to verify the dimensions of these skips. And we are already finding that where we expect it to be only one type of skips, one dimension, we discovered that over time, there had been a few variations created. “

Marwood adds, “By laser scanning the skip, that gives you a 3d model of the real skip and we can overlay that with a theoretical model, which the cut paths for the laser are based upon. An operator can then work out if that’s within tolerance.”

“The cut paths are planned based on a theoretical perfect skip with tolerances. And then the skip that’s imported, is scanned to confirm it sits within the envelope of deviations that are allowable. We haven’t designed it based on perfect skips, it’s designed to tolerate deviations and take that into account.”

Despite the challenges, the Atkins and Sellafield teams have made rapid progress. “It’s gone from concept design to operations in about four years,” says Marwood. “For a facility that size-reduces ILW with lasers and robots, that’s pretty much an industry first.”