Description
Spherical tokamaks offer numerous benefits over conventional tokamaks, including enhanced MHD stability, access to a higher ratio of plasma pressure to magnetic pressure (beta), and reduced reactor size. Current and future spherical tokamaks include MAST-U, NSTX-U, and the UK's planned spherical tokamak for energy production (STEP). However, as beta increases, turbulent transport transitions from being primarily electrostatic to electromagnetic, with magnetic instabilities such as kinetic ballooning modes and micro-tearing modes dominating transport.
Non-linear gyrokinetic codes can successfully predict turbulent fluxes in high-beta regimes, but are far too computationally costly to be used in the integrated modelling workflows necessary to identify steady state plasma scenarios for STEP. Reduced models employ various conceptual and numerical simplifications to reduce the computational cost of flux calculations. Quasi-linear codes couple a model of the linear properties of instabilities, usually itself a reduced model, with a saturation rule, which predicts the maximum amplitude of these instabilities. Both reduced linear models and saturation rules commonly include numerical parameters fitted for specific plasma conditions. Refitting these numerical parameters for high-beta conditions is expected to improve agreement between these reduced models and gyrokinetic codes under such conditions.
An analysis of the current leading linear models was conducted by comparing predictions with linear gyrokinetic simulations. By carefully selecting the numerical parameters in the current publicly available leading quasi-linear model (TGLF), the linear performance on high-beta cases was improved. TGLF was then compared to a new Gaussian process regression model trained on a database of linear gyrokinetic simulations from the code GS2. The existing quasi-linear saturation rules will be tested for high-beta ST turbulence, including the effects of flow shear stabilisation.