🔬 physics

Modeling and Experimental Comparison of Tungsten Impurity Profiles for rotating plasmas in KSTAR

This study validates an ad-hoc 2-D model that couples short-time parallel dynamics with long-time radial transport to accurately predict tungsten impurity profiles and radiation distributions in high-rotation KSTAR plasmas, demonstrating reasonable agreement with experimental infrared video bolometer measurements.

Original authors: Hyojong Lee, E. A. Belli, J. Candy, D Fajardo, Jekil Lee, Jungpyo Lee

Published 2026-07-15

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Hyojong Lee, E. A. Belli, J. Candy, D Fajardo, Jekil Lee, Jungpyo Lee

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Cosmic Dance of Heavy Particles

Imagine trying to keep a pot of soup from burning while stirring it with a giant spoon. In the world of fusion energy, scientists are trying to do something even trickier: they are heating a cloud of gas (plasma) to temperatures hotter than the center of the sun to make clean energy. But there's a catch. If even a tiny bit of heavy, dirty stuff—like a speck of tungsten metal from the reactor walls—gets into the soup, it can act like a heat sink, sucking away all the energy and cooling the plasma down before it can make power. This is a major headache for building future fusion reactors.

To solve this, scientists need to understand how these heavy "impurities" move inside the swirling, super-hot plasma. It's not just a simple drift; the plasma spins incredibly fast, creating forces that push heavy particles to one side or the other, making the distribution uneven. Think of it like a spinning merry-go-round: if you stand on the edge, you feel pushed outward, but if you're holding onto a pole, the forces get complicated. The challenge is that these heavy particles react to the spin and friction in a split second, but they move across the plasma much more slowly. Scientists have long wanted a way to predict exactly where these heavy particles will gather and how much heat they will steal, but doing the math for the whole 3D mess at once is like trying to solve a Rubik's cube while juggling.

The Paper's Story: A Two-Step Dance

This paper, titled "Modeling and Experimental Comparison of Tungsten Impurity Profiles for rotating plasmas in KSTAR," introduces a clever shortcut to solve that Rubik's cube. The researchers, working with data from the KSTAR fusion device in South Korea, developed a new way to model how tungsten impurities behave. Instead of trying to calculate every single movement of the heavy particles in all directions at the same time (which is computationally exhausting), they used a "time-scale separation" trick.

Imagine the heavy tungsten particles are dancers in a crowded room. The paper suggests we can split their dance into two parts. First, there's the "fast dance": the particles quickly adjust to the spinning of the room, creating a lopsided crowd where more people gather on one side due to centrifugal force (the push you feel when a car turns a corner) and friction. This happens almost instantly. Second, there's the "slow dance": over a longer period, the entire group slowly drifts across the room. The authors' model calculates the fast, lopsided arrangement first, then uses that result to figure out how the group slowly drifts over time. They combined two existing tools—one for the "friction" part (called FACIT) and one for the "turbulence" part (called TGLF)—to create this hybrid 2-D model.

When they tested this model against real data from a KSTAR experiment (specifically discharge #40085), they found some very encouraging results. The model successfully predicted where the tungsten would gather. For instance, at 5 seconds into the experiment, the model predicted the radiation peak would be at a radius of about $r/a \approx 0.61$ , which was very close to the experimentally observed peak at $r/a \approx 0.67$ . By 9.05 seconds, the model predicted a peak at $r/a \approx 0.51$ , matching the observed $r/a \approx 0.55$ .

However, the paper is careful not to claim this is a perfect, finished solution. While the location of the heat loss matched well, the amount of heat was off by a huge margin. The simulation calculated a tiny amount of cooling power (0.001 MW), while the actual measurement was much higher (0.14 MW)—a difference of about 160 times. To make the comparison fair, the researchers had to artificially scale up their simulation numbers by a factor of 160 just to see if the shapes matched. They found that even with this scaling, the shapes of the radiation profiles looked similar.

The authors explicitly note that their method has limits. Because they split the fast and slow movements into separate steps, the model misses some subtle interactions where the slow drift might change the fast spin in real-time. Also, they didn't include a specific source for where the tungsten came from in their simulation; they just assumed it was there. Despite these gaps, the study suggests that this "two-step" modeling approach is a useful tool. It shows that even with its simplifications, the model can capture the complex, lopsided behavior of heavy impurities in a spinning plasma, offering a promising path for future research into keeping fusion reactors clean and hot.

Technical Summary: Modeling and Experimental Comparison of Tungsten Impurity Profiles for Rotating Plasmas in KSTAR

Problem Statement
Characterizing high-Z impurity profiles, specifically tungsten (W), is critical for predicting the operational limits of future fusion reactors. High-Z impurities tend to accumulate in the plasma core due to friction with main ions, leading to increased core radiation and degraded plasma performance. In tokamak environments with high toroidal rotation, such as KSTAR and JET, centrifugal forces significantly influence impurity transport. Modeling these dynamics is complex because it requires resolving the interplay between parallel force balance (driving poloidal density asymmetries) and radial transport. While parallel dynamics occur on a much faster time scale than radial transport, comparing experimental 2-D radiation profiles requires a scheme that can simulate the temporal evolution of the full 2-D density distribution.

Methodology
The authors developed an "ad-hoc" 2-D modeling scheme based on the time-scale separation between parallel force balance and radial transport. The approach decouples the physics into two distinct components:

Short-time Parallel Dynamics: This component calculates the poloidal density asymmetry ( $n_{Z0}(\psi, \theta)$ ) driven by centrifugal forces, parallel friction, and pressure gradients. The model utilizes a simplified fluid model (FACIT) to account for friction across the full collisional regime, solving the parallel momentum balance equation. This step determines the poloidal inhomogeneity (both in-out and up-down asymmetries) based on the flux-surface-averaged density.
Long-time Radial Transport: This component evolves the impurity density over the transport time scale using a 1-D radial continuity equation applied to the 2-D density variable. The effective cross-field diffusion ( $\langle D_\perp \rangle$ ) and convection ( $\langle V_\perp \rangle$ ) coefficients are computed via linear superposition of neoclassical contributions (from the FACIT model) and turbulent contributions (from the TGLF-SAT0 code). To address the limitations of standard TGLF in high-rotation regimes, an analytical correction is applied to incorporate centrifugal effects.

The scheme iterates between these two time scales: the radial transport updates the flux-surface-averaged density, which is then fed back into the parallel dynamics calculation to update the poloidal asymmetry and transport coefficients until a steady state is reached.

Key Contributions

Development of a 2-D Transport Scheme: The study introduces a computational framework that combines short-time parallel physics with long-time radial transport to simulate the 2-D evolution of tungsten impurities without solving the full 2-D equations simultaneously.
Integration of Centrifugal Effects: The model explicitly incorporates centrifugal forces into both the parallel force balance (inducing out-in asymmetry) and the turbulent transport coefficients (via an analytical correction to TGLF).
Experimental Validation: The model is benchmarked against existing JET data (discharge #82722) and applied to new experimental data from KSTAR discharge #40085.
Radiation Profile Reconstruction: The study demonstrates a method to calibrate simulated tungsten density against Infrared Video Bolometer (IRVB) measurements by scaling the volume-integrated cooling power rate to match total radiated power.

Results

Benchmarking: The model's prediction of poloidal asymmetry trends (specifically the enhancement of up-down asymmetry with plasma elongation) aligns with theoretical scaling laws and previous JET benchmark studies. The calculated transport coefficient ratios ( $RV_W/D_W$ ) and density gradients show good agreement with JET discharge #82722 data.
KSTAR Comparison: For KSTAR discharge #40085, the model successfully reproduces the spatial distribution of core radiation. While there is a significant discrepancy in absolute magnitude between the raw simulated cooling power and IRVB measurements (a factor of ~160 at $t=9.05$ $t = 9.05$ s), the model accurately captures the radial location of the radiation peaks.
- At $t=5$ s, the simulated peak is at $r/a \approx 0.61$ (experiment: $\approx 0.67$ ).
- At $t=9.05$ s, the simulated peak is at $r/a \approx 0.51$ (experiment: $\approx 0.55$ ).
Steady State: The simulation indicates that the impurity particle flux stabilizes and the tungsten density distribution reaches a steady state within the transport time scale ( $t \approx 0.5$ s).

Significance and Limitations
The paper claims that despite the limitations of the "ad-hoc" approach, the reasonable agreement between the modeled spatial distribution of core radiation and experimental IRVB measurements demonstrates the usefulness of this 2-D modeling scheme for analyzing tungsten behavior in rotating plasmas.

The authors explicitly acknowledge the following limitations:

The time-scale separation scheme cannot resolve pure 2-D dynamics where radial and parallel physics occur simultaneously; it misses perturbed poloidal asymmetries caused by transport within the short-time evaluation.
The simulations conserve the total amount of impurity without an explicit tungsten source term.
The model relies on a calibration factor to match absolute radiated power, rather than predicting absolute density from first principles alone.

Future work is suggested to extend the modeling to include more consistent treatments of the physics and explicit impurity sources.

The Cosmic Dance of Heavy Particles

The Paper's Story: A Two-Step Dance

More like this