Abstract:In a globally proliferating industry to deliver commercialisable fusion power comes the need for a rapid design-cycle from conception through to qualification. The complexity and extreme nature of the fusion environment necessitates a highly-performant digital framework to automatically evaluate and optimise designs of components - and ultimately - entire power-plants. Delivering such a platform will require a coalescing of technologies, including multi-physics modelling, hardware acceleration, parametric geometry, uncertainty quantification, and surrogate modelling.
For magnetic-confinement fusion-reactor concepts, one key component is the breeder blanket, whose multitudinous role is to capture and transport useful neutron heat, to breed tritium fuel that will sustain the fusion reactions, and to shield the rest of the device from damaging irradiation. The physics modelling requirements of a breeder blanket are similar to those of the full tokamak, rendering it an ideal application around which to develop a digital ecosystem.
In this talk, the current status of open-source software tools for high-fidelity multi-physics modelling of breeder blankets developed by the UK Atomic Energy Authority is reviewed. Recent developments for an automated analysis pipeline are described, and a vision for future work is presented.
Biography:Dr Helen Brooks is a HPC Software Specialist at the United Kingdom Atomic Energy Authority (UKAEA), working in the Advanced Engineering Simulation group. She has worked on the UK Spherical Tokamak for Energy Production (STEP) project since 2020, developing performant simulation tools for fusion power plant components, and is a member of the UKAEA Breeder Blanket Team. Her current research interests are multi-physics modelling, developing automated design workflows, and performance portability. She received her MSc in Natural Sciences from the University of Cambridge in 2013, and her PhD in high energy physics from Durham University in 2018.
John Tramm