-
Using Convolutional Neural Networks to detect Edge Localized Modes in DIII-D from Doppler Backscattering measurements
Authors:
N. Q. X. Teo,
V. H. Hall-Chen,
K. Barada,
R. J. H. Ng,
L. Gu,
A. K. Yeoh,
Q. T. Pratt,
X. Garbet,
T. L. Rhodes
Abstract:
In H-mode tokamak plasmas, the plasma is sometimes ejected beyond the edge transport barrier. These events are known as edge localized modes (ELMs). ELMs cause a loss of energy and damage the vessel walls. Understanding the physics of ELMs and by extension, how to detect and mitigate them, is an important challenge. In this paper, we focus on two diagnostic methods $\unicode{x2013}$ D-alpha spectr…
▽ More
In H-mode tokamak plasmas, the plasma is sometimes ejected beyond the edge transport barrier. These events are known as edge localized modes (ELMs). ELMs cause a loss of energy and damage the vessel walls. Understanding the physics of ELMs and by extension, how to detect and mitigate them, is an important challenge. In this paper, we focus on two diagnostic methods $\unicode{x2013}$ D-alpha spectroscopy and Doppler backscattering (DBS). The former detects ELMs by measuring Balmer alpha emission while the latter uses microwave radiation to probe the plasma. DBS has the advantage of having higher temporal resolution and robustness to damage. These advantages of DBS diagnostics may be beneficial for future operational tokamaks and thus data processing techniques for DBS should be developed in preparation. In sight of this, we explore the training of neural networks to detect ELMs from DBS data, using D-alpha data as the ground truth. With shots found in the DIII-D database, the model is trained to classify each time step based on the occurrence of an ELM event. The results are promising. When tested on shots similar to those used for training, the model is capable of consistently achieving a high f1-score of 0.93. This score is a performance metric for imbalanced datasets that ranges between 0 and 1. We evaluate the performance of our neural network on a variety of ELMs $\unicode{x2013}$ grasssy, suppressed, and wide pedestal $\unicode{x2013}$ finding broad applicability. Beyond ELMs, our work demonstrates the wider feasibility of applying neural networks to data from DBS diagnostics.
△ Less
Submitted 3 July, 2024; v1 submitted 3 June, 2024;
originally announced June 2024.
-
ELM-free Enhanced Dα H-mode with Near Zero NBI Torque Injection in DIII-D Tokamak
Authors:
T. Macwan,
K. Barada,
J. F. Parisi,
R. Groebner,
T. L. Rhodes,
S. Banerjee,
C. Chrystal,
Q. Pratt,
Z. Yan,
H. Wang,
L. Zeng,
M. E. Austin,
N. A. Crocker,
W. A. Peebles
Abstract:
Enhanced $D_α$ H-mode (EDA H-mode), an ELM-free H-mode regime, is explored in neutral beam heated, lower single null plasmas with near zero torque injection. This regime exhibits a good energy confinement ($\mathrm{H}_{\mathrm{98y2}}$ $\sim 1$) with $β_N \sim 2$, high density, regime access at low input power, and no ELMs. This paper further presents the time-resolved measurements of electron and…
▽ More
Enhanced $D_α$ H-mode (EDA H-mode), an ELM-free H-mode regime, is explored in neutral beam heated, lower single null plasmas with near zero torque injection. This regime exhibits a good energy confinement ($\mathrm{H}_{\mathrm{98y2}}$ $\sim 1$) with $β_N \sim 2$, high density, regime access at low input power, and no ELMs. This paper further presents the time-resolved measurements of electron and ion density, temperature, plasma rotation, and radial electric field during the EDA H-mode phase and examines the dynamics of the edge quasi-coherent mode (QCM). Measurements using multiple fluctuation diagnostics reveal the QCM to be a separatrix spanning mode, peaking just inside the separatrix, existing in a wide range of $k_{\perp}ρ_s \sim 0.1-1.2$ with multiple harmonics, and propagating with a very small phase velocity in the plasma frame, where $k_{\perp}$ is the binormal wavenumber and $ρ_s$ is the ion sound radius. Linear gyrokinetic simulations of an EDA H-mode discharge with CGYRO indicates that the trapped electron mode (TEM) and electron temperature gradient (ETG) are dominant instabilities in the region where QCM is unstable. Qualitative analysis indicates that the properties of TEM are consistent with the experimental observed characteristics of the QCM. These similarities suggest that the QCM might be a TEM instability existing in the edge region of the EDA H-mode plasmas.
△ Less
Submitted 9 February, 2024;
originally announced February 2024.
-
Effect of mismatch on Doppler backscattering in MAST and MAST-U plasmas
Authors:
Valerian H. Hall-Chen,
Felix I. Parra,
Jon C. Hillesheim,
Juan Ruiz Ruiz,
Neal A. Crocker,
Peng Shi,
Hong Son Chu,
Simon J. Freethy,
Lucy A. Kogan,
William A. Peebles,
Quinn T. Pratt,
Terry L. Rhodes,
Kevin Ronald,
Rory Scannell,
David C. Speirs,
Stephen Storment,
Jonathan Trisno
Abstract:
The Doppler backscattering (DBS) diagnostic, also referred to as Doppler reflectometry, measures turbulent density fluctuations of intermediate length scales. However, when the beam's wavevector is not properly aligned perpendicular to the magnetic field, the backscattered power is attenuated. In previous work, we used beam tracing and reciprocity to derive this mismatch attenuation quantitatively…
▽ More
The Doppler backscattering (DBS) diagnostic, also referred to as Doppler reflectometry, measures turbulent density fluctuations of intermediate length scales. However, when the beam's wavevector is not properly aligned perpendicular to the magnetic field, the backscattered power is attenuated. In previous work, we used beam tracing and reciprocity to derive this mismatch attenuation quantitatively. In this paper, we applied our model, in the small but finite mismatch limit, to a several new cases. We compared our predictions with multiple O-mode channels for the first time. We then identified a $\sim 3^{\circ}$ error in the MAST Q-band's quasioptics, showing that our model is useful for commissioning DBS diagnostics. For both O- and X-mode, we compared experimental data with our model's predictions at multiple times during the shots, unlike our previous work, where only a single time was analysed. Finally, we analysed other contributions to the backscattered signal, evaluating how much they affect our measurements of mismatch attenuation, giving comparisons with data from both MAST and MAST-U. This paper's detailed study systematically validates and demonstrates the usefulness of our model for quantitatively interpreting DBS data from spherical tokamaks.
△ Less
Submitted 5 June, 2024; v1 submitted 30 November, 2022;
originally announced November 2022.
-
Validating and optimising mismatch tolerance of Doppler backscattering measurements with the beam model
Authors:
Valerian H. Hall-Chen,
Julius Damba,
Felix I. Parra,
Quinn T. Pratt,
Clive A. Michael,
Shi Peng,
Terry L. Rhodes,
Neal A. Crocker,
Jon C. Hillesheim,
Rongjie Hong,
Shikang Ni,
William A. Peebles,
Ching Eng Png,
Juan Ruiz Ruiz
Abstract:
We use the beam model of Doppler backscattering (DBS), which was previously derived from beam tracing and the reciprocity theorem, to shed light on mismatch attenuation. This attenuation of the backscattered signal occurs when the wavevector of the probe beam's electric field is not in the plane perpendicular to the magnetic field. Correcting for this effect is important for determining the amplit…
▽ More
We use the beam model of Doppler backscattering (DBS), which was previously derived from beam tracing and the reciprocity theorem, to shed light on mismatch attenuation. This attenuation of the backscattered signal occurs when the wavevector of the probe beam's electric field is not in the plane perpendicular to the magnetic field. Correcting for this effect is important for determining the amplitude of the actual density fluctuations. Previous preliminary comparisons between the model and Mega-Ampere Spherical Tokamak (MAST) plasmas were promising. In this work, we quantitatively account for this effect on DIII-D, a conventional tokamak. We compare the predicted and measured mismatch attenuation in various DIII-D, MAST, and MAST-U plasmas, showing that the beam model is applicable in a wide variety of situations. Finally, we performed a preliminary parameter sweep and found that the mismatch tolerance can be improved by optimising the probe beam's width and curvature at launch. This is potentially a design consideration for new DBS systems.
△ Less
Submitted 30 September, 2022;
originally announced September 2022.
