292

Papers

11/11

Roadblocks Active

6

Connections

Using the SIMSOPT optimization code on a university-scale device geometry, researchers show that a single array of high-temperature superconducting dipole coils can produce tokamak, quasi-axisymmetric stellarator, and hybrid equilibria by adjusting coil currents alone, without any mechanical changes. The rotational transform can be tuned from near-zero up to values typical of stellarators, while plasma shaping (elongation and triangularity) matches standard tokamak targets. If built, such a device would let experimenters compare multiple confinement concepts on a single machine, dramatically reducing the cost and time of systematic optimization studies.

██████████ 0.9 hts-magnets Preprint

ReadSaveConnections

Global gyrokinetic simulations with the TRIMEG code show that when the thermal ions and electrons are allowed to respond nonlinearly — not just the energetic particles — the amplitude of unstable Alfvén waves grows nearly linearly with the energetic particle drive rather than quadratically, and the resulting particle flux scales quadratically rather than quartically with the pressure gradient. This means transport is substantially less stiff than models that only treat energetic particles nonlinearly predict. For burning plasmas where alpha particles provide self-heating, this changes the predicted losses and could alter ignition margin estimates.

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

ReadSaveConnections

This study compares multiple three-dimensional numerical approaches for simulating how thin-wall high-temperature superconducting bulk materials magnetize under pulsed fields, assessing both accuracy and computational cost. Getting magnetization simulations right is critical for fusion magnets that use HTS stacks or bulk inserts, because pulsed magnetization determines the trapped field profile and thus the effective field seen by the plasma. Identifying which 3D model balances fidelity and speed informs which codes should be used at the magnet design stage.

██████████ 0.9 hts-magnets Peer-reviewed

Read

Using a GdBCO bulk disc at 77 K and a 2D finite-element model validated against experiment, the study shows that how uniformly or non-uniformly the applied magnetic field is distributed across the surface of an HTS screen significantly changes its shielding effectiveness. In fusion magnets, field non-uniformity arises from coil geometry, joints, and off-normal events, so mispredicting the screen response leads to errors in field quality estimates inside the plasma. The open-source Life-HTS toolkit used here makes these simulations accessible to magnet design teams.

██████████ 0.8 hts-magnets Peer-reviewed

Read

The authors cast the relativistic Fokker-Planck equation for runaway electrons into an adjoint form and train physics-informed neural networks to predict the runaway current, mean energy, and full energy distribution for arbitrary initial conditions — achieving speedups of several orders of magnitude over traditional kinetic solvers. Runaway electrons are a serious disruption hazard because even a small runaway beam can punch through the first wall; fast prediction of their evolution is essential for real-time mitigation systems. The code is to be released publicly, which would lower the barrier for integrating runaway modeling into tokamak control frameworks.

██████████ 0.8 plasma-disruption Preprint

ReadSaveConnections

This analytical paper argues that toroidal magnetic flux has been underused in tokamak equilibrium theory and that adopting it as the primary variable clarifies how the loop voltage drives poloidal-to-toroidal flux slippage — the mechanism behind plasma current evolution. Near the separatrix (the boundary between confined and open field lines), this formalism simplifies the mathematical description of equilibrium constraints and makes the conditions for disruption more tractable. Cleaner analytic foundations for separatrix behavior support better disruption-prediction models.

██████████ 0.8 plasma-disruption Preprint

ReadSaveConnections

Using a kinetic Vlasov-Poisson model, this study shows that high-harmonic fast waves used for plasma heating can decay into Bernstein-mode waves near the scrape-off layer, launching a cascade of parametric instabilities that drive anomalous turbulent heating of ions in that region. The scrape-off layer connects the confined plasma to the divertor, and anomalous heating there increases the heat load on plasma-facing surfaces in ways not captured by standard transport models. This mechanism could explain unexpectedly high ion temperatures observed in the near-wall region during radio-frequency heating on several machines.

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

ReadSaveConnections

The ECHO algorithm combines a neural network trained to emulate a full ray-tracing code (TORBEAM) with a genetic optimizer to control where electron cyclotron heating power is deposited in real time on the DIII-D tokamak, and demonstrates robustness when individual gyrotrons fail mid-shot. Precise deposition control matters for suppressing neoclassical tearing modes that degrade confinement, and the system must respond in real time to changing plasma conditions. Demonstrating this on an operating tokamak — including fault handling — brings this class of AI-assisted control a step closer to reactor relevance.

██████████ 0.7 long-confinement Preprint

ReadSaveConnections

A physics-informed neural operator trained on roughly 10,000 TCV discharges reconstructs the full plasma shape in under 100 microseconds — fast enough to close control loops at 10 kHz — while enforcing the Grad-Shafranov equilibrium equation through automatic differentiation. Accurate, real-time knowledge of plasma shape is a prerequisite for controlling the boundary conditions that determine ELM stability and heat-load distribution on the divertor. Deployment and validation directly on TCV's plasma control system distinguishes this from laboratory demonstrations.

██████████ 0.7 elm-control Preprint

ReadSaveConnections

Gyrokinetic simulations with the ORB5 code show that geodesic acoustic modes driven by energetic particles do not saturate at a fixed amplitude but instead oscillate nonlinearly, with the oscillation frequency scaling with mode amplitude in a way that resembles beam-plasma instability physics. EGAMs can cause bursts of energetic particle transport that degrade fast-ion confinement and reduce plasma self-heating efficiency. Identifying this scaling law provides a testable prediction and a path toward incorporating EGAM saturation into transport models without running expensive full simulations.

██████████ 0.7 turbulence-modeling Preprint

ReadSaveConnections