Development and validation of reduced models for transport in spherical tokamaks.

Creating a complete and accurate digital model of a Fusion Tokamak is an alluring goal, allowing design decisions to be evaluated and iterated upon.

This is particularly important for the UK next-generation fusion reactor STEP (Spherical Tokamak for Energy Production), where the design is still being finalised.

Creating a digital model is the aim of integrated modelling, a method of combining computational models of different physics within the tokamak.

Modelling heat flow from the centre of the reactor to the edge is a key part of this physics, as it influences and is influenced by a wide range of other phenomena.

Nonlinear gyrokinetics is the current gold standard for calculating heat flow, however it is too computationally expensive for integrated modelling.

Quasilinear modelling uses results from Nonlinear Gyrokinetics and experimental data to inform a simplified description of plasma instabilities, allowing faster evaluation whilst preserving accuracy.

However, descriptions of the instabilities are based on traditional tokamak geometry and miss physics associated with spherical tokamaks. To have accurate integrated modelling for STEP, we need to modify our quasilinear models for spherical tokamaks. This project will use machine learning to analyse data from MAST-U, the world's largest spherical reactor.

This analysis will inform a new quasilinear code which will be validated against fusion experiments. This code will then be implemented into a fully integrated modelling code and used to make predictions about STEP.