OpenStar is a New Zealand company, founded here in Wellington, whose goal is to bring nuclear fusion to the grid. Our approach differs significantly from the currently accepted standard: Where most designs for magnetically confined fusion surround the superheated fusion plasma with magnets, we replicate what we find in nature and surround the magnet with the plasma. This kind of configuration powers the Earth's aurora and the powerful radiation belts of Jupiter, plasma systems that have been stable for an incredible amount of time. However, designing this system in a compact way on earth creates some interesting challenges. Smaller devices require far more extreme conditions that that found in the sun in order to produce useful fusion power. At these temperatures the plasma would destroy any material unfortunate enough to come into contact with it. One common approach to avoid this is to use strong magnets to direct the plasma away from the walls and back onto itself. The strength of magnet needed to achieve this is immense and requires the use of exotic materials such as superconductors to handle the incredibly hight currents. Designing such a magnet is also not trivial, the strong fields result in forces that act to rip the magnet apart and the 100 million degree fusing plasma in close proximity can create some thermal management challenges.

Closing this design problem is hard, and is one of the reasons why fusion has always been 30 years away. At OpenStar, we have used Rust to create a full system optimisation tool. This tool connects together a large range of physical models, some in Rust, some in other languages, and uses them to optimise the size and shape of the core superconducting magnet to achieve the best possible performance. This talk will go into the design of a Levitated Dipole Reactor, how this influences the optimisation and design tooling, and how Rust is a key enabler in the rapid iteration and development of our in house tooling.