290

Papers

10/10

Roadblocks Active

5

Connections

Injecting a nitrogen-deuterium gas mixture into the EAST tokamak completely suppressed large periodic plasma eruptions called ELMs, while simultaneously boosting energy confinement by roughly 33% (H98 factor rising from 0.9 to 1.2). The suppression appears linked to a new edge oscillation mode (20–50 kHz) that regulates the plasma boundary more gently, preventing the violent energy dumps that would otherwise damage the tungsten metal wall. This matters because ELMs are one of the main threats to the lifetime of reactor-scale divertors, and demonstrating benign suppression without sacrificing confinement is a key step toward viable reactor operation.

██████████ 0.9 elm-control Preprint

ReadSaveConnections

Researchers integrated a neural network running on a dedicated FPGA chip directly into DIII-D's plasma control system, enabling real-time prediction of ELMs from 160-channel beam emission spectroscopy data sampled at one million points per second. The key engineering advance is that the network's weights can be swapped on the fly — allowing the same hardware to switch between different classification tasks without time-consuming chip reprogramming. This matters because fast, hardware-accelerated ML inference is a prerequisite for closed-loop control that can suppress plasma instabilities before they cause damage, rather than reacting after the fact.

██████████ 0.9 elm-control Preprint

ReadSaveConnections

This is an archived diagnostic dataset from Japan's Large Helical Device (LHD), containing FFT-processed phase contrast imaging (PCI) signals that track plasma density fluctuations over time across the 50–500 kHz frequency band. PCI measurements in this frequency range can reveal turbulent structures that drive energy transport losses — a central challenge for sustaining fusion conditions. The dataset itself does not present new findings but contributes to the growing archive that future statistical studies of turbulence behavior will draw upon.

██████████ 0.6 turbulence-modeling Peer-reviewed

Read

A second archived PCI FFT dataset from LHD shot #176721, covering the same 50–500 kHz spectral bands at high sampling frequency, enabling time-resolved tracking of plasma density oscillations. Such datasets form the raw material for training machine-learning turbulence classifiers or validating gyrokinetic simulation codes against real experimental fluctuation spectra. Scientific value depends entirely on how these records are used in aggregate analysis.

██████████ 0.6 turbulence-modeling Peer-reviewed

Read

An archived LHD diagnostic record using lower-sampling-rate FFT analysis of PCI signals for shot #190069, covering frequency content relevant to lower-frequency MHD and turbulent modes. The slower sampling rate complements high-frequency datasets by capturing dynamics in the 3–150 kHz range where large-scale pressure-driven instabilities can appear. As with other LHD archive entries, the scientific value lies in future comparative or statistical studies.

██████████ 0.5 turbulence-modeling Peer-reviewed

Read

Archived high-sampling-rate PCI FFT dataset from LHD shot #167109, with 156,249 time points spanning 50–500 kHz, capturing amplitude modulation patterns in plasma density fluctuations. These records are building blocks for empirical databases of turbulence behavior across different LHD plasma conditions. Without accompanying plasma parameters or analysis in this record, direct scientific interpretation is deferred.

██████████ 0.5 turbulence-modeling Peer-reviewed

Read

Archived LHD PCI FFT dataset (shot #192945) providing high-frequency spectral amplitude time series across three frequency bands. The dataset adds to the LHD turbulence diagnostics archive and is relevant to studies of plasma confinement degradation driven by high-frequency density fluctuations. Standalone scientific content is minimal without cross-referencing plasma conditions.

██████████ 0.5 turbulence-modeling Peer-reviewed

Read

High-sampling-rate PCI FFT archive record for LHD shot #189794, documenting spectral amplitude evolution in the 50–500 kHz range. As part of the LHD NIFS data repository, this record supports future multi-shot turbulence studies but contains no independent scientific findings. Its relevance to plasma disruption and ELM research would require correlation with plasma operational data.

██████████ 0.5 plasma-disruption Peer-reviewed

Read

Low-sampling-rate FFT PCI diagnostic archive from LHD shot #184929, covering four frequency bands from 3 kHz to 500 kHz with 1,220 time points — a coarser temporal resolution suited to tracking slower MHD phenomena. This complementary sampling regime alongside fast datasets could help distinguish turbulent from coherent oscillation contributions to confinement loss. Scientific utility depends on integration into broader shot-database analyses.

██████████ 0.5 plasma-disruption Peer-reviewed

Read

Archived high-rate PCI FFT dataset from LHD shot #175997 (acquired December 2021), providing 156,249-point time series of integrated spectral amplitudes across the 50–500 kHz band. This record extends the LHD turbulence diagnostics archive but, like the other LHD entries today, contains no standalone scientific analysis. Its value will emerge when researchers combine it with plasma parameter logs to study correlations between density fluctuation spectra and confinement quality.

██████████ 0.5 plasma-disruption Peer-reviewed

Read