LongOps robotics

Above: A UK-Japanese research alliance is looking to develop long-reach robotics

LONGOPS IS A NEW £12 million UK-Japanese research and development programme to explore and extend the functions of digital robotic technologies for use in nuclear decommissioning. The project will focus particularly on the digital tools used to control long-reach robotics in long-term operations — hence it is known as ‘LongOps’.

The key theme is the development of digital twins — a set of technologies including virtual reality (VR) and simulations of robotics that pair the virtual and physical worlds. In nuclear decommissioning, these can enable strategy planning, training, remote operations, storage and analysis of data, including forecasting of maintenance events and potential operational issues.

The aspiration is to create capability and knowledge to enable nuclear end users to become intelligent customers of products and services in this rapidly emerging field. The project is also expected to result in direct benefits such as employment opportunities, advances in ‘fusion-adjacent’ technologies and up-skilling of robotics operations capability in the UK and Japan.

RACE — the UK Atomic Energy Authority’s centre for Remote Applications in Challenging Environments, based in Culham, Oxfordshire — is leading the LongOps project on behalf of a collaboration funded equally by the UK’s Nuclear Decommissioning Authority (NDA), Japan’s Tepco and UK Research & Innovation (UKRI).

Safer, faster, cheaper decommissioning

The delivery of the LongOps programme comes at a crucial time in the world of nuclear decommissioning. In Japan, the Fukushima Daiichi power plant is stable, including the four units severely affected by the earthquake in March 2011, and Tepco has presented an action plan for decommissioning.

In the UK, Sellafield Ltd in Cumbria is engaging with industry to consider how to make best use of robotics and artificial intelligence to conduct decommissioning in a more effective and efficient manner. That includes the Windscale Pile 1 reactor, which was damaged by fire in 1957. In Culham in Oxfordshire, the largest operational fusion experiment in the world, the Joint European Torus (JET), is nearing the end of its life and will face first-of-a- kind decommissioning challenges.

LongOps technologies will open more routes to safer, faster and cheaper decommissioning. It will also promote cost-effective reactor design for fusion and fission.

High-profile challenges

Legacy nuclear and fusion facilities are complex large- scale projects that are time-intensive to decommission safely.

Windscale Pile 1, Fukushima Daiichi and JET are some of the highest-profile use-cases for emerging decommissioning technologies. However, it is hard to imagine a standard approach that would be appropriate for decommissioning all three of them — a 1940s graphite- moderated reactor, 1970s BWR and an experimental fusion reactor.

But while they contain different hazards, the decommissioning strategies overlap. In all three cases, access is via small ports using slender long-reach robotics, and while the internals of the JET device are ‘as-built’, all three environments hold considerable challenges that will evolve over time.

Tepco (through Mitsubishi Heavy Industries) recently received an articulated boom developed by Veolia Nuclear Solutions UK Ltd. The boom, which will reach over 20m at full extension will be used to inspect fuel debris in the primary containment vessels at Fukushima Daiichi.

At Sellafield, Windscale Pile 1 will require robotics to safely and efficiently reach and extract fuel elements housed in the 7.5m-deep reactor core.

These long-term operations involving high-radiation environments that are likely to require long-reach equipment are where LongOps comes in. All three sites will benefit from advanced digital tools throughout the process, in planning, training and potentially automating remote operations, to reduce doses for operators and lower costs.

RACE has experience from decades of operations of the unique JET Remote Handling System. This system, involving two 12m booms and haptic manipulators, has been used to maintain and upgrade JET, including a complete replacement of its inner wall. This means it has been used to disassemble and rebuild the device from the inside.

Next-generation digital tools

At the heart of LongOps is a portfolio of research and development contract opportunities, targeted at industry and academia. Scheduled to start in September 2021, the contracts will explore the potential for a new generation of digital tools to assist with nuclear operations.

These tools, collectively called a digital twin or digital mock-up, make it easier for people performing operations by providing additional information and removing some of the cognitive burden. Outside the control rooms, digital twins can also offer useful information for task planning, operator training, record-keeping and decision-making.

At RACE, workshops are being held with staff from JET, Tepco and Sellafield to ensure that the research and development is informed by user requirements.

Research themes

Physics simulations

One area of research interest is incorporating physics simulations, including control of long-reach flexible booms and haptic manipulators.

Real-world joint positions are hard to accurately measure or predict, and disturbances may cause booms
to deflect, making precise, repeatable motion difficult to achieve. Control algorithms that can account for this are a key LongOps research area, with particular relevance to the tight geometric constraints seen in nuclear environments. RACE’s TARM long-reach boom will be available to test advanced algorithms.

Automation

Automation is another area that can improve operations. Challenging tasks like cutting, sorting, segregating and debris collection within tight operational constraints could be performed faster and with fewer errors if aided by autonomous planning assistance.

Routine inspection tasks could be automated using machine learning, including the ability to detect anomalies and changes in the environment, with this information being stored in the digital twin. Relevant information
can then be provided to operators or used in automated planning and control systems. LongOps will also explore how recent advances in AI and machine learning can be applied to robotic control and perception. However, machine learning relies on the availability of relevant and sufficient datasets, which do not currently exist. LongOps intends to start building these datasets.

Haptics

At JET, remote handling operators rely on haptic feedback when controlling manipulators and tools, which allows them to experience the same forces and torques at the local device that are imposed on the remote device. This allows operators to ‘feel’ items which are being gripped. However, virtual simulations of haptics — which could be used for training — are currently unable to accurately recreate the feeling of manipulators interacting with the environment. Currently, operator training for haptic devices requires the use of real equipment.

LongOps will explore the development of high-frequency, high-fidelity haptic simulators.

Modular software

All this research will be brought together in a modular software package referred to as the prototype next- generation digital mock-up.

Digital tools are often designed to be used independently, or with a limited range of interfacing modules. LongOps will make use of a modular interoperable software architecture, to promote ease of reconfiguration.

This is what makes LongOps such an ambitious and exciting programme — as well as developing new tools to improve the way operations are performed, it requires these tools to work together in a single, modular system.

Different end users may require only some software features or prioritise only some features. By managing a suite of software and basing the requirements on real-world tasks, LongOps aims to create standardised, modular software which can meet the various demands of nuclear decommissioning projects.

Manipulators

Digital twins link the virtual world with the physical world. During 2021, LongOps placed contracts for procurement of two teleoperated manipulators.

Over the next two years, LongOps will undertake performance assessments, benchmarking tests and comparison activities within a test environment. The research programme will seek to enhance their use by using digital twin technologies, building on preliminary work on glovebox automation led by RACE as part of the UKRI-funded Robotics and AI in Nuclear Research (RAIN) academic network.

Meeting end user training needs

For remote operations to be successful, the nuclear sector will need suitably qualified and experienced people. At present, training requires operators to use real hardware, which is expensive and time-consuming. Digital environments that replicate more of the operator experience offer the potential to accelerate training, increase the number of fully qualified operators and maintain and test competence over years. It will also be possible to provide better training for the diverse situations that might be experienced during real operations.

Changing nuclear decommissioning

The next generation of digital tools will provide opportunities to develop decommissioning strategies, and train and conduct operations in high-consequence, constantly changing environments.

RACE, UKRI, TEPCO and the NDA see digital tools and remotely operated machines as a significant route to safer, faster, cheaper decommissioning and cost-effective future reactor design, whether it be in fusion or fission.

More details of the opportunities to contribute to LongOps can be found at: https://ukaeaevents.com/longops-supplier-day