Maxwell Prize: Efficient, Accurate Plasma Turbulence Simulations (Remembrances of Prof. Dorland will be shared during this time.)

Remembrances of Prof. Dorland will be shared during this time. His original abstract was:

Turbulent fluctuations of electromagnetic fields, temperature, density, and so on, typically degrade the quality of the insulation provided by the magnetic field in magnetic confinement fusion (MCF) devices, because fluctuations in the “core” plasma lead to loss rates that worsen as the plasma temperature is increased to the required thermonuclear values. This is true even when the fluctuations are small fractions of their background values (e.g., ) and radial correlation lengths are much smaller than the plasma itself. Over the last thirty-five years, simulations of turbulent plasmas have become efficient enough and accurate enough to enable a deeper theoretical understanding of the fluctuations, and to enable detailed tests of predictions of turbulence-induced losses in hundreds of specific experiments. The job is not completely finished, but already the best fusion start-up companies rely heavily upon simulations of turbulence and transport (along with many other in silico tools) to optimize concepts and to develop detailed fusion power plant designs, and theoretical descriptions of plasma turbulence are very often tested against ab initio simulations more fruitfully than when they are tested against experimental observations directly. The challenges are particularly acute for stellarators, which have the largest design space. In this short talk, I will review a few key developments that got us to this state, such as how the tension between efficiency and accuracy has evolved; cross-pollination of MCF simulations with space and astrophysical turbulence simulations; and (one example of) the growing role of machine learning in this space.