290

Papers

10/10

Roadblocks Active

6

Connections

Sintered boron pebble rods were mounted in DIII-D's divertor and exposed to heat loads up to 80 MW/m², marking the first test of a granular, self-renewing plasma-facing material in a real tokamak. The pebbles survived but shed significant boron dust — roughly half the eroded mass left as millimeter-scale particles and the rest as fine dust — which then became the dominant source of boron entering the plasma. This matters because a renewable granular divertor could sidestep the erosion lifetime problem of solid tungsten, but only if the dust can be managed to avoid plasma contamination and tritium co-deposition issues.

██████████ 0.9 divertor-thermal Preprint

ReadSaveConnections

Nonlinear gyrokinetic simulations for a reduced-current version of the ARC compact tokamak design show that fusion-born alpha particles substantially reduce ion-scale turbulent heat and particle transport in the inner half of the plasma (r/a ≤ 0.5) compared to a scenario where those alphas are artificially treated as thermal ions. The suppression works through a combination of zonal flow amplification and a nonlinear upshift of the critical ion temperature gradient needed to trigger turbulence. This is important because standard transport models that ignore fast-alpha kinetics likely overestimate how much auxiliary heating is needed to sustain a burning plasma, which would lead to overly conservative (and expensive) power-plant designs.

██████████ 0.9 turbulence-modeling Preprint

ReadSaveConnections

This purely analytical paper extends the classical theory of tearing-mode instability — the mechanism behind magnetic reconnection events that can trigger disruptions — to include a resistivity that switches on only when current density exceeds a threshold, as expected from turbulent or kinetic effects. The key finding is that this threshold-dependent resistivity introduces sharp spatial singularities at the boundary layer and produces a time-dependent growth rate that initially diverges before settling, qualitatively changing disruption onset dynamics compared to classical theory. If confirmed numerically, the result implies that standard MHD disruption models may mischaracterize early-phase growth, with consequences for disruption prediction and avoidance systems in ITER-scale devices.

██████████ 0.9 plasma-disruption Preprint

ReadSaveConnections

Global gyrokinetic simulations using the TRIMEG code reveal that the transport of energetic particles (fast ions from neutral beam injection or alpha heating) is far less stiff than previously modeled when the full nonlinear response of background thermal plasma is included. Specifically, the flux scaling with the energetic-particle pressure gradient drops from a quartic to a quadratic dependence, and the saturation level scales nearly linearly rather than quadratically with mode amplitude. This matters because existing reduced transport models used in reactor design underestimate how well fast ions are confined in the inner core, potentially biasing predictions of heating efficiency and alpha-particle ash accumulation.

██████████ 0.8 turbulence-modeling Preprint

ReadSaveConnections

This theoretical and numerical study shows that high-harmonic fast waves used for plasma heating can decay into shorter-wavelength ion-Bernstein waves in the turbulent scrape-off layer, a process that can cascade into parametric turbulence. The instability exists only within a finite wavelength band of the driving wave and leads to anomalous heating of scrape-off-layer ions. This is relevant because unexplained edge ion heating has been observed in RF-heated tokamaks and poses a risk to plasma-facing components; identifying this parametric decay channel provides a quantitative mechanism that could inform how RF power is coupled and where heat loads actually deposit near the wall.

██████████ 0.8 turbulence-modeling Preprint

ReadSaveConnections

The ECHO algorithm combines a neural-network surrogate of the TORBEAM ray-tracing code with a genetic optimizer to steer electron-cyclotron heating beams to arbitrary target deposition profiles in real time on DIII-D, without needing expensive on-the-fly ray-tracing calculations. The system proved robust when individual gyrotrons failed mid-shot and when plasma parameters changed significantly, redistributing power among remaining beams to maintain the target profile. Real-time control of where ECH power is deposited is critical for suppressing neoclassical tearing modes (a leading cause of disruptions) and for active current profile shaping in steady-state scenarios; demonstrating this robustly in hardware is a meaningful step toward reactor-relevant actuator control.

██████████ 0.7 plasma-disruption Preprint

ReadSaveConnections

This analytical paper argues that toroidal magnetic flux — the flux threading the hole of the torus — has been largely neglected in the tokamak literature despite being essential to a correct application of Faraday's Law and to equilibrium calculations when the plasma boundary is a separatrix rather than a closed flux surface. The author shows that including toroidal flux simplifies and sharpens the equilibrium conditions for diverted plasmas, which are the standard configuration for reactor-relevant devices. While theoretical, the work is relevant to disruption modeling and equilibrium reconstruction codes, where small errors in flux accounting near the separatrix can affect predictions of current evolution and stability margins.

██████████ 0.7 plasma-disruption Preprint

ReadSaveConnections

This paper introduces a two-stage framework that discovers which input variables actually matter for a stochastic system by learning full conditional distributions — not just mean responses — and pruning irrelevant variables using Gaussian-process-based analysis of variance. The key insight is that in noisy or chaotic systems like fusion plasmas, a variable may have zero effect on the average output but large effects on variability or tail behavior, which standard regression misses. For turbulence modeling and digital twins of fusion devices, this offers a principled way to build leaner, more interpretable surrogate models by identifying which plasma parameters (temperature gradient, density, rotation) actually drive transport fluctuations versus which are coincidental correlates.

██████████ 0.6 turbulence-modeling Preprint

ReadSaveConnections

LRC-FNO is a neural operator architecture that compresses particle-in-cell source fields into a latent space, solves for the dominant electromagnetic field response with a coarse Fourier operator, and then applies a residual correction network to recover fine-scale features. On benchmarks including a 2D scrape-off-layer fusion plasma, it achieves relative L2 errors below 5% for electromagnetic potentials and remains stable for simulation times roughly twice the training horizon. This is relevant because full kinetic PIC simulations of edge plasmas are computationally prohibitive for routine use in reactor design; a surrogate that generalizes beyond its training window could enable faster exploration of plasma-wall interaction regimes.

██████████ 0.6 turbulence-modeling Preprint

ReadSaveConnections

This journal paper (identified via the MSF venue) reports on 3C-SiC single crystals subjected to high-temperature annealing between 1700–2100°C, using diffuse X-ray scattering and transmission electron microscopy to track a structural phase transformation as a function of time and temperature. Silicon carbide is a candidate structural and neutron-moderating material for fusion blankets and first-wall components due to its low activation and high-temperature strength. Quantifying the kinetics and activation energy of its phase transformation under extreme thermal conditions is directly relevant to predicting SiC component lifetimes in a fusion neutron environment, where combined radiation damage and operating temperatures could accelerate this transition.

██████████ 0.6 first-wall-materials 🔗 2266 cited Peer-reviewed

Read