Page 26 - Autumn 2024

Industry News

NEW AI TECHNIQUE

ACCELERATES HEAT

PROTECTION CALCULATIONS

FOR FUSION VESSELS

A collaboration between Commonwealth Fusion
Systems, the US Department of Energy’s Princeton
Plasma Physics Laboratory and Oak Ridge National
Laboratory has created a new AI technique
that speeds up calculations that are critical to
protecting fusion vessels from the extreme heat
generated by plasma.
The system, known as HEAT-ML, rapidly identifies
‘magnetic shadows’ within a fusion device, i.e. regions
shielded from direct plasma heat. It is hoped that
this will form the basis for future software that will
accelerate the design of fusion systems and enable
real-time decision making during operations.
In order to harness nuclear fusion, plasma has to be
controlled at temperatures above that of the Sun’s
core, and it is a major challenge to predict how this
heat will interact with the inner walls of the fusion
vessel, known as a tokamak. Accurate and fast
calculations of heat impact zones and protected
regions are essential for the successful design and
operation of such vessels.
Originally the open-source HEAT (Heat
HEAT-ML was developed to simulate a small section Flux Engineering Analysis Toolkit) handled
of SPARC, the tokamak currently being built by these simulations, but the process was very
Commonwealth Fusion Systems in Massachusetts. computationally demanding with each HEAT
This tokamak aims to produce more energy than it simulation requiring half an hour or more for
consumes, and requires a model showing precisely complex geometries. This was because it involved
how plasma heat will affect the reactor’s interior. tracing magnetic field lines and calculating their
intersections with intricate 3D surfaces.
The team focused on a critical area of SPARC’s
exhaust system, namely fifteen tiles close to the HEAT-ML removes this bottleneck by using a deep
bottom which are anticipated to experience the neural network trained on around 1,000 HEAT-
most intensive plasma heat load. In order to generated SPARC simulations. The AI learned
simulate the heat distribution, shadow masks, i.e. to predict shadow masks within milliseconds,
3D maps showing where magnetic field lines shield representing a dramatic improvement that now
components from direct plasma exposure, were enables rapid exploration of design options.
created. These depend on the vessel’s geometry and
Reference: Hookway, P., ‘AI accelerates Heat Protection Calculations
magnetic configuration. for Fusion Vessels’, Electronic Specifier, 20th. October

26
Become a member: https://membermojo.co.uk/ie
Become a subscriber: https://membermojo.co.uk/subscriber
Become a sponsor: https://membermojo.co.uk/sponsor