Developing Viable Operating Regimes for Fusion Pilot Plants & Predicting their Performance
Abstract: As the magnetic fusion program moves toward pilot plants, we must change the way we operate. Major challenges must be overcome to avoid machine damage, wall erosion & impurity contamination while maintaining high temperatures. High boundary temperatures sputter metallic impurities which are drawn into plasma & degrade performance by radiation. Large edge localized mode (ELM) transients effectively flush these metallic impurities but would damage the divertor & thus intolerable. Scaling arguments suggest that in future machines, turbulent transport is more likely to create naturally ELM-stable scenarios w/ additional benefits.1 However, new challenges arise, such as avoiding metallic impurity accumulation. I'll describe efforts to meet these core-edge integration challenges in future-relevant ELM-stable scenarios such as Quiescent H-Mode in DIII-D. For example, we recently achieved record densities via strong shaping which enabled the first radiative divertor plasma detachment in QH-mode while reducing impurity influxes by a factor of six by tailoring edge pedestal profiles.
Fusion performance is highly sensitive to the edge pressure. For ELM-free regimes, the edge pressure is limited by turbulence rather than MHD stability (which is readily predicted by the EPED model). This exposes a major gap in predictive capabilities. We're addressing this grand challenge in the 11-institution DOE Fusion Innovation Research Engine (FIRE) Collaboratory, APP-FPP: Advanced Profile Prediction for Fusion Pilot Plant Design. Our project is developing accelerated gyrokinetic, multiscale, whole-device predictions of density, temperature & impurity profiles for tokamak & stellarator fusion pilot plants including kinetic plasma-wall interactions & atomic physics. Making these predictions practical requires highly accelerated algorithms & AI/ML techniques. The present focus is on extending, accelerating & coupling edge gyrokinetic codes & widely used fluid & gyrokinetic transport solvers to predict edge profiles. Reduced models also enter for the core & edge, such as our new MIT multiscale gyrofluid code, which can accurately simulate coupled ion & electron scale turbulence on a GPU in 2-3 hours, several orders of magnitude faster than gyrokinetic codes while matching their results.
1D. R. Ernst et al., Broadening of the divertor..., Physical Review Letters 132, 235102 (2024)
Bio: Ernst has been a member of the Theory Group at the MIT Plasma Science & Fusion Center since 2002. Since 2007, he has been the MIT PI for gyrokinetic SciDAC projects, supervising theory postdocs & students in the development of gyrokinetic, gyro-fluid & neoclassical codes & now leads the APP-FPP FIRE Collaboratory. He has been a long-time collaborator on the DIII-D National Fusion Facility, leading an experiment every year. He led the 2022 DOE Joint Research Target & 2022-2023 DIII-D Thrust to Develop High Performance Non-ELMing Regimes & now leads similar international efforts in ITPEA as well as the QH-Mode subgroup of the EU-US Joint Working Group on ELM-Free Regimes. He has presented 32 invited talks at international conferences & is author/co-author of over 100 refereed articles. https://sites.mit.edu/darinernst
Full bio here