Info!
UPDATED 1 Sept: The EI library in London is temporarily closed to the public, as a precautionary measure in light of the ongoing COVID-19 situation. The Knowledge Service will still be answering email queries via email , or via live chats during working hours (09:15-17:00 GMT). Our e-library is always open for members here: eLibrary , for full-text access to over 200 e-books and millions of articles. Thank you for your patience.
New Energy World magazine anniversary logo

Celebrating two years of reporting on the industry’s progress toward net zero

ISSN 2753-7757 (Online)

The nuclear fusion race – one small step nearer?

10/8/2022

8 min read

Artist's impressio, with cutaway, of STEP (the Spherical Tokamak for Energy Production) – the UK’s first prototype fusion energy plant Photo: UKAEA
 
The latest research on fusion technology in collaboration with the Hartree Centre will use digital twins and AI on STEP (the Spherical Tokamak for Energy Production) – the UK’s first prototype fusion energy plant

Photo: UKAEA
 

Scientists at the STFC Hartree Centre and the UKAEA are using supercomputing and artificial intelligence to design technologies such as digital twins to help make fusion energy a commercial reality, within decades hopefully, reports Brian Davis.

There’s a long-running joke that commercial-scale nuclear fusion is always 30 years in the future, regardless of when the question is posed. Hopefully, the tide is turning, with a major nuclear fusion breakthrough earlier this year. In February 2022, a team from the Joint European Torus (JET) project in Oxford managed to generate 59 MJ of energy through nuclear fusion. This was nearly double the previous record set in 1997, according to the UK Atomic Energy Authority (UKAEA).

 

The results were good news for advocates of nuclear energy as a low carbon alternative to fossil fuels. But there’s still a very long way to go before commercial nuclear fusion is a readily available, efficient, low carbon means of tackling climate change.

 

Nevertheless, Ian Chapman, CEO of UKAEA, said the results were a ‘landmark’ and will bring us ‘a huge step closer’ to virtually emissions-free energy in the race to decarbonise energy production.

 

Significant steps are underway in the UK to build knowledge and develop the new technology required to deliver nuclear fusion energy. New Energy World was invited to speak to experts at the Science and Technology Facilities Council’s (STFC) Hartree Centre about how it and the UKAEA are using supercomputers and artificial intelligence (AI) to design technologies to make fusion energy a commercial reality.

 

A long-held dream    
Fusion energy is a long-held dream in scientific circles which has the potential to be a revolutionary and limitless energy source that will help reduce dependence on fossil fuels and tackle climate change. Fusion promises some significant advantages, including no greenhouse gases, virtually no long-term radioactive waste (apart from the fusion reactor itself during decommissioning) and an unlimited source of energy which is intrinsically safe.

 

In 2021, the UK government published the UK Fusion Strategy, which set out a path to enable fusion energy. The UKAEA is seen as a global leader in fusion energy science and technology and is responsible for the Spherical Tokamak for Energy Production (STEP), the UK’s first prototype facility to pave the way for commercial development of fusion power stations.

 

Last year, in support of this project, the Hartree Centre and the UKAEA took the first steps towards creating a UK centre of excellence in ‘extreme-scale’ computing for fusion. This is located at STFC’s Daresbury Laboratory, at Sci-Tech Daresbury, Liverpool City region, home to some of the most advanced supercomputing and AI technologies in the UK. ‘The Hartree Centre aims to help scientists develop the necessary technologies quicker and more affordably,’ explains Joseph Weston, Business Development Manager.

 

VIDEO – Fusion is coming    
Source: UKAEA, EUROfusion

 

What is fusion?    
Fusion is a natural process that powers the sun and other stars. When light nuclei fuse to form a heavier nucleus, they release bursts of energy. This is the opposite of nuclear fission, the reaction used in nuclear power stations today, in which energy is released when a nucleus splits apart to form smaller nuclei.

 

To produce energy from fusion here on Earth, a combination of hydrogen gas isotopes found in seawater – deuterium and tritium – have to be heated to extreme temperatures (over 100mn°C) under intense gravitational forces equivalent to those found on the sun. The gas becomes a plasma and the nuclei combine to form a helium nucleus and a neutron, with a tiny fraction of the mass converted into fusion energy. Theoretically, a plasma with millions of these reactions every second can provide a huge amount of energy from very small amounts of fuel.

 

Although record-breaking sustained fusion energy was achieved earlier this year, many complex engineering challenges must be overcome before fusion energy can be produced as part of a large-scale, affordable low carbon energy supply for the future. Key challenges include producing and managing the plasma where the fusion process happens, and developing materials that can tolerate extreme conditions, such as withstanding particles heated to hundreds of millions of degrees.

 

Fortunately, there is a growing armoury of expertise.

 

Current research    
The second phase of the initial five-year project involves over 30 members of staff from the UKAEA and STFC Hartree Centre. Together, they are applying the latest supercomputing, AI and data science expertise to address challenges to the delivery of commercial fusion energy. They are developing technologies for modelling and understanding plasma; modelling digital twins of future fusion power plants; developing prototype tools for advanced data management; and using AI tools to give key insights, such as into machine control and uncertainty quantification.


‘Together, UKAEA and the STFC Hartree Centre are recreating processes in the sun and bottling them into a power plant with the fusion of elements.’ – Joseph Weston, STFC Hartree Centre

 

‘There are many different types of nuclear fusion, but we are focused on deuterium-tritium fusion, because the temperatures which you require the reactor to reach are relatively lower compared to other types of fusion – but still extremely high at 100mn°C to achieve fusion,’ says Weston. There are some pros and cons with respect to deuterium-tritium fusion, and other isotopes may be used in future. Tritium is not available naturally, but fusion energy power plants will be designed to manufacture tritium from lithium, using breeding blankets.

 

ITER collaboration 

The UKAEA and the Hartree Centre are also partners in the International Thermonuclear Experimental Reactor (ITER), where 35 nations, including the EU, China, India, Japan, Korea, Russia and the US, are collaborating to build the ITER Tokamak in southern France. It will be the world’s largest tokamak, designed to prove the feasibility of fusion on a large-scale as a carbon-free source of energy.

 

On 24 December 2020, a nuclear cooperation agreement was concluded between the UK and Euratom for the UK to remain part of the Fusion for Energy, the European Domestic Agency for ITER, after Brexit.

 

First sub-section of ITER plasma chamber being lifted out of tooling and into the machine wellThe first sub-section of the ITER plasma chamber being lifted out of tooling and into the machine well

Photo: ITER  

 

Hartree and UKAEA are also focused on modelling smaller tokamak designs such as STEP, which will be a lot cheaper to build, and are aiming to have fusion power on the grid by 2040.

 

Other fusion projects in the UK include development of a privately owned prototype reactor by General Fusion at Culham, hosted by the UKAEA, optimistically aiming to build a commercial fusion plant by the early 2030s.

 

Weston explains there are many complex technology issues to address. Particularly, in terms of the design of the powerful magnetic confinement of the plasma within the reactor, using a tokamak or stellarator, for example. In the tokamak, the rotational transformation of a helical magnetic field is formed using external coils together with a poloidal field generated by the plasma current. Whereas in the stellarator (which has been demonstrated by Wedelstein 7-X in Germany), the twisting field is produced by external non-axisymmetric coils.

 

The Hartree Centre has the most powerful supercomputer in the UK dedicated to industry engagement – the Scafell Pike supercomputer, a 4 petaflops* Bell Sequana X1000 supercomputer with Cray ClusterStor storage – and also uses AI to model the behaviour of plasma in fusion reactors, in silico (on a computer chip). ‘This approach can accelerate complex modelling of prototype reactors and ultimately commercial reactors, with faster and more cost-effective R&D timelines,’ says Weston.

 

Digital twin models can be used to test fusion reactor designs with realistic physics for iterative design. ‘This is pushing the fundamental limits of known particle physics. The science is very complex as we are dealing with nonlinear equations to describe plasma behaviour,’ remarks Weston. ‘The key challenge is to ensure that predictions are truly representative of what you would see in the real world – given the high complexity, rather like climate science.’

 

Luke Mason is one of the scientists working on the Farscape Project, looking into the modelling of fusion reactor plasma confinement and computer architecture using exascale computers, which will run at exaflops (ie one quintillion floating-point operations per second). This is an order of magnitude greater computing power for modelling fusion reactors.

 

‘The core focus of our current research is performance portability, which examines our ability to use different computer architectures. Specific components include AI surrogates, training AI to replicate the output of an expensive physics-based simulation. Primarily looking at turbulence in the plasma and around the diverter region, where there is a lot of heat exchange. Then we are looking at particle-in-cell models (used to solve a certain class of partial differential equations in particle physics) to simulate the plasma, and see how these models scale out in exascale machines like the US Department of Energy’s Frontier Exascale Computing Project, in a collaborative programme,’ explains Mason.

 

Looking forward    
There are many competing designs for fusion reactors, with a lot of private investment hoping the promise can be delivered sooner rather than later. ‘But we’re still unlikely to have anything at commercial scale in less than a couple of decades. Tools like exascale computing could be a big advantage in the race to get a working fusion reactor on the grid, but nobody is willing to commit to a date,’ comments Weston.

 

Are we waiting for a big breakthrough?

 

‘It’s not about one big breakthrough, but many years of small incremental stages. It’s taken decades to get to a situation where you can maintain a plasma temperature of 100mn°C. There have to be technology improvements in many areas, including magnets, the physical design and arrangement of magnets, material science and much else besides. Fusion is still a research exercise,’ he admits.

 

Weston emphasises that the energy transition should not be put on hold while we wait for fusion energy to go commercial. ‘The energy transition should be resolved before we have commercial fusion reactors. We can’t wait another 10, 20 or more years until we have commercial fusion reactors but need to invest in renewables now, because the economy needs electrifying to replace fossil energy generation.’

 

*One petaflop is equal to 1mn gigaflops. A 1 gigaflop computer system is capable of performing 1bn (109) floating-point operations per second.