Quantifying, predicting, and controlling turbulent transport in fusion devices

Determine effective methodologies to quantify, predict, and control turbulent transport in fusion devices so as to enable the design of optimized fusion power plants. This field-level problem concerns establishing reliable approaches for assessing and managing turbulence-driven energy transport that sets confinement performance in reactors.

Background

The paper emphasizes that turbulent transport sets the energy confinement time in fusion devices, directly impacting reactor performance and design. Despite advances in first-principles simulations, routine tasks like optimization or uncertainty quantification require computationally efficient yet reliable models.

The authors address part of this broader challenge by proposing an advanced surrogate scaling law for ETG-driven electron heat flux in tokamak pedestals, including new dependencies on safety factor q and electron beta βe and a quantification of prediction uncertainty via bootstrapping. Nonetheless, the overarching task of quantifying, predicting, and controlling turbulent transport remains a central open problem motivating such model development.

References

Quantifying, predicting, and controlling this turbulent transport is a prerequisite for designing optimized fusion power plants and is therefore considered a key open problem in fusion research.

— Advanced surrogate model for electron-scale turbulence in tokamak pedestals  (2405.09474 - Farcas et al., 2024) in Section 1 (Introduction)

However, the level of predictive accuracy required in practice will ultimately depend on the intended transport-model application and remains to be established in future studies.

— Machine learning methods for modelling local, linear gyrokinetic simulations of MAST-U pedestal turbulence  (2608.25945 - Niemelä et al., 26 Aug 2026) in Discussion and conclusions

While some works report favorable comparisons when ensuring sufficient length flux tubes, it remains future work to determine if and how more detailed gyrokinetic calculations can be included in stellarator optimization problems.

— AI-Accelerated Gyrokinetic Predictions of Turbulent Transport for Stellarator Design Optimization and Experimental Planning  (2609.28730 - Churchill et al., 23 Sep 2026) in Section 2, subsection “GX AI Surrogate Model,” subsubsection “Simulation code”

Nevertheless, robust scalings for $\kappa_\perp$ and, consequently, for $\lambda_q$ in the SOL of stellarator devices remain open problems.

— Physics-Informed Neural Networks to Infer the Perpendicular Energy Conductivity in the Scrape-Off Layer of Stellarator Devices  (2609.11628 - Gallego et al., 10 Sep 2026) in Section 1, Introduction

However, it is not clear that this behaviour can be extrapolated to the SOL, where open field lines, lower temperature and higher collisionality can substantially modify the transport regime.

— Physics-Informed Neural Networks to Infer the Perpendicular Energy Conductivity in the Scrape-Off Layer of Stellarator Devices  (2609.11628 - Gallego et al., 10 Sep 2026) in Section 2.1.2, Existing $\kappa_\perp$ scalings