That was the message at COP29 today, November 14, 2024 during an international collaboration panel hosted by fusion mega-project ITER and the Clean Air Task Force.
A new source of carbon-free energy that can meet rapidly growing electricity demand and mitigate climate change is essential for a sustainable future. Recent breakthroughs have shown fusion energy no longer belongs in the realm of science fiction, and could realistically be contributing to national energy grids by the 2030s.
Collaboration by the public and private sectors, ambitious national strategies and bold power plant programmes in the UK, United States and Japan are all accelerating fusion development, boosted by rising investment.
This development can be accelerated further, especially in the field of materials development, by successfully harnessing artificial intelligence (AI) and machine learning (ML), it was announced at the UN Climate Conference in Baku, Azerbaijan.
Success in fusion depends on building the world’s collective understanding of the novel materials that can withstand the extreme conditions inside fusion power plants – where plasma temperatures reach 150 million degrees Celsius, 10 times hotter than the sun’s core.
MORE FROM ENERGY & ENVIRONMENT
Obtaining performance data on fusion-relevant materials is challenging and expensive, so to achieve it at pace requires a group effort. International collaboration would ensure data quality, facilitate licensing, and prevent duplication of effort. Critically, the amassed, centralised data would enable the use of rapidly developing AI and ML approaches to build confidence in materials, reduce material verification times, validate power plant designs, and significantly reduce the costs of material development.
The global fusion materials database, harnessing machine learning, can be used to provide guidance on future global materials testing strategies, moving away from the current, inefficient and piecemeal system.
Current algorithms will help build the database, allowing supercomputers to data mine up to 6,000 published papers per hour and analyse results at high speeds.
The world needs fusion and fusion needs teams around the world working together on solving big challenges. An open-access, single source of truth for high-quality fusion materials data, at the heart of a suite of machine learning tools, is therefore essential for this new phase of technology development.
The approach could revolutionise the way humans conduct scientific research.
Ultimate vision for a single source of truth
Our ultimate vision is for a fusion materials database with quality-controlled data supplied by entities from all countries, accessible to all. It would cover all properties relevant to fusion device design, and act as a single source of truth.
The database would include all supporting data, including the production process of each test material, and span all fusion approaches.
Where are we now?
Tokamak Energy, a private fusion company formed in Oxford, UK, in 2009, is part of a Clean Air Task Force (CATF) International Working Group alongside other leading players like UK Atomic Energy Authority (UKAEA) and EUROfusion actively exploring how to make this happen.
The group is consulting with a number of major AI global players and investigating how the evolving techniques can be applied to fusion challenges. Momentum is building.
Creating a centralised data store is an ambitious goal, but we are taking a pragmatic approach, initially using easy to access data and focusing on the information most useful to the most players. We will build from there, demonstrating the high value being generated to encourage more sharing.
We know powerful supercomputers can simulate what takes place inside a fusion device. A digital twin is invaluable in optimising and validating experimental scenarios and developing plasma control systems at a much faster rate than contemporary machines can handle. Such a digital twin requires a vast amount of material data for it to be valid, and I believe the global fusion materials database is a prerequisite for producing accurate results.
A machine learning-powered materials database would be another giant step in removing barriers on our ultimate path to commercial fusion, in time to make a difference for our warming planet.
Jim Pickles is head of materials at Tokamak Energy
Heat network to set bar for decarbonising heat
The report is somewhat outdated and seems to give emphasis to hydrogen and carbon capture. Thermal energy sources do not include those for electricity...