Fast and accurate simulations of turbulence for fusion energy reactors – FASTER

Fusion energy is an ideal solution for sustainable non-intermittent electricity production. Accurate simulation of fusion plasma turbulence is required for reactor operation and control, but is either too slow or lacks accuracy with present techniques.
This project aims to circumvent these conflicting constraints of accuracy and tractability and provide real-time capable turbulent transport models with increased physics fidelity for tokamak temperature, density, and rotation velocity prediction.
Advanced theoretical models of the non-linear saturation of plasma turbulence will be combined with surrogate models of the linear plasma response to micro-instabilities obtained from neural network regressions of high physics fidelity simulations to yield the turbulent fluxes at both unprecedented accuracy and speed. The application of neural network technology for emulation of transport models is a novel idea in fusion. It bridges between theory and the pragmatic constraints of control room tools, combining tokamak plasma physics and machine learning to improve the speed of transport calculation.