Mirnov Coils Research Articles

A novel unsupervised machine learning algorithm for automatic Alfvénic activity detection in the TJ-II stellarator

Abstract A novel sparse encoding algorithm is developed to detect and study plasma instabilities automatically. This algorithm, called Elastic Random Mode Decomposition, is applied to the Mirnov coil signals of a dataset of 1291 discharges of the TJ-II stellarator, enabling the identification of the Alfvénic activity. In the presented approach, each signal is encoded as a collection of basic waveforms called atoms, drawn from a signal dictionary. Then the modes are identified using clustering and correlations with other plasma signals. The performance of the proposed algorithm is dramatically increased by using elastic net regularization and taking advantage of GPU architectures. Therefore the signal size and the number of dictionary elements are no longer limiting factors for encoding complex signals. Once the modes are retrieved from the shots, standard clustering and dimensionality reduction techniques are applied to obtain a 2D map featuring of the physical mode characteristics of this subset of TJ-II shots. The clustering features consider the relationship with the plasma current Ip, the diamagnetic energy W, and inverse square root of electronic density 1/√n, profiling different subtypes of Alfvénic activity. The proposed algorithm can potentially create large databases of labeled modes with unprecedented detail.

Synthetic Mirnov diagnostic for the validation of experimental observations.

A synthetic Mirnov diagnostic has been developed to investigate the capabilities and limitations of an arrangement of Mirnov coils in terms of a mode analysis. Eight test cases have been developed, with different coil arrangements and magnetic field configurations. Three of those cases are experimental configurations of the stellarator Wendelstein 7-X. It is observed that, for a high triangularity of the flux surfaces, the arrangement of the coils plays a significant role in the exact determination of the poloidal mode number. For the mode analysis, torus and magnetic coordinates have been used. In most cases, the reconstruction of the poloidal mode number of a prescribed mode was found to be more accurate in magnetic coordinates. As an application, the signal of an Alfvén eigenmode, which has been calculated with a three-dimensional magnetohydrodynamics code, is compared to experimental observations at Wendelstein 7-X. For the chosen example, the calculated and measured mode spectra agree very well and additional information on the toroidal mode number and localization of the mode has been inferred.

Experiments on excitation of Alfvén eigenmodes by alpha-particles with bump-on-tail distribution in JET DTE2 plasmas

Dedicated experiments were performed in JET DTE2 plasmas for obtaining an α-particle bump-on-tail (BOT) distribution aiming at exciting Alfvén eigenmodes (AEs). Neutral beam injection-only heating with modulated power was used so that fusion-born α-particles were the only ions present in the MeV energy range in these DT plasmas. The beam power modulation on a time scale shorter than the α-particle slowing down time was chosen for modulating the α-particle source and thus sustaining a BOT in the α-particle distribution. High-frequency modes in the toroidicity-induced Alfven eigenmode (TAE) frequency range and multiple short-lived modes in a wider frequency range have been detected in these DT discharges with interferometry, soft x-ray cameras, and reflectometry. The modes observed were localised close to the magnetic axis, and were not seen in the Mirnov coils. Analysis with the TRANSP and Fokker-Planck FIDIT codes confirms that α-particle distributions with BOT in energy were achieved during some time intervals in these discharges though no clear correlation was found between the times of the high-frequency mode excitation and the BOT time intervals. The combined magneto-hydrodynamic (MHD) and kinetic modelling studies show that the high-frequency mode in the TAE frequency range is best fitted with a TAE of toroidal mode number n= 9. This mode is driven mostly by the on-axis beam ions while the smaller drive due to the pressure gradient of α-particles allows overcoming the marginal stability and exciting the mode (Oliver et al 2023 Nucl. Fusion submitted). The observed multiple short-lived modes in a wider frequency range are identified as the on-axis kinetic AEs predicted in Rosenbluth and Rutherford (1975 Phys. Rev. Lett. 34 1428).

Magnetic pickup coil systems on the HL-3 tokamak

HL-3 is a new medium-sized copper conducting tokamak, which is dedicated to the plasma shaping control for plasma high performance and flexible operation, especially the advanced divertor configurations. According to the physics needs of the project, three types of magnetic pickup coils are designed and developed on HL-3 for magnetic field measurements. The coils emphasize different geometry sizes and array layouts. The coils are installed carefully by using high-precision positioning equipment to ensure the measurement accuracy of the system. After calibration, the magnetic field measurements are supplied for real-time equilibrium reconstruction and plasma control, e.g., with snowflake-shaped divertor configuration reconstruction. Mirnov coils satisfy the requirements for physical analysis, including determining the mode number from raw signals and detecting the magnetohydrodynamic instabilities.

The advance of magnetic diagnostics system in support of EAST long-pulsed operation

In EAST long-pulsed discharge (hundreds of seconds), electric magnetic diagnosis (EMD) is very important, since EMD not only monitors tokamak security status but also provides accurate measurement accuracy for reconstruction of the plasma boundary. To avoid current measurement drift, a fiber optic current sensor, based on the Faraday effect, is developed and used for poloidal and plasma current feedback control for the first time, relative current measurement accuracy is within 0.5%. To ensure plasma boundary control accuracy, a detailed set of magnetic measurement calibration methods is developed before the plasma discharge. The maximum relative error is less than 1%, the corresponding control accuracy is within 1 cm. To minimize integrator drift error, a long-pulse integrator test is essential, the corresponding drift error needs to be subtracted in plasma control system. Besides, the saddle coil and Mirnov coil not only help to detect MHD issues, but are also utilized for plasma disruption prediction during the long-pulse discharge.

Broadband Alfvénic excitation correlated to turbulence level in the Wendelstein 7-X stellarator plasmas

During the first operational phase (OP1) of the Wendelstein 7-X (W7-X) stellarator, poloidal magnetic field fluctuations, , were measured in several different plasma scenarios with a system of Mirnov coils. In the spectrograms, multiple frequency bands close together in frequency are observed below f = 600 kHz. Furthermore, a dominant feature is the appearance of a frequency band with the highest spectral amplitude centred between kHz. The fluctuations are observed from the beginning of most W7-X plasmas of OP1, which were often operated solely with electron cyclotron resonance heating. The fluctuations show characteristics known from Alfvén waves and possibly Alfvén eigenmodes (AEs). However, the fast particle drive from heating sources, which is generally a driver necessary for the appearance of AEs in magnetic confinement plasmas, is absent in most of the analysed experiments. A characterization of the Alfvénic fluctuations measured during OP1 plasmas is possible using a newly developed tracking algorithm. In this paper, we extensively survey the different spectral properties of the fluctuations in correlation with plasma parameters and discuss possible driving mechanisms. The correlation studies of the dynamics of the possible ellipticity induced AEs indicate that Alfvén activity in the frequency interval between kHz could be excited due to an interaction with turbulence, or profile effects also affecting the turbulence amplitude.

Non-monotonic radial structures of fluctuating temperatures and densities associated with fishbone activities in KSTAR

General characteristics of a fishbone mode in KSTAR are investigated. Fishbone activities are observed with a Mirnov coil, an electron cyclotron emission radiometer (from the core to the edge of plasmas) and an beam emission spectroscopy system (core or edge plasmas) which are measuring fluctuations of poloidal magnetic fields, electron temperatures, and densities, respectively. Temporal evolutions of these fluctuations are similar to the observations from other tokamaks. An interesting and notable feature found in KSTAR fishbone modes is that radial coherence structures of electron temperature and density with respect to magnetic fluctuations are non-monotonic that they have a local minimum at r/a∼0.7 and a maximum at r/a∼0.9 in addition to the usual global peak near the q = 1 surface, where r/a is the normalized minor radius and q is the safety factor. Furthermore, the associated temporal evolution of the electron temperatures in slow-time scale, i.e., less than 1 kHz, with the fishbone activities show that there exist a drop in temperature or increase in temperature depending on inside or outside the q = 1 surface, respectively, from the core to the edge plasmas except that there are almost no temperature changes in the intermediate region which seems to be correlated with the non-monotonic coherence profile. Such a non-monotonic structure and the slow temporal evolution of temperatures are explained with barely trapped resonating fast ions with the banana orbit widths of the order of the minor radius, so that they transit the core and the edge regions simultaneously without trespassing the mid-plane intermediate region.

Interplay between beam-driven chirping modes and plasma confinement transitions in spherical tokamak ST40

Experiments on the high field spherical tokamak ST40 have led to the recent observation of interplay between beam-driven modes of sweeping frequency (chirping modes) and transitions to the enhanced global confinement regime (H-mode) and back to the low confinement regime (L-mode). The H-modes of plasma confinement are identified from decreased intensity of Dα signal and from clear distinctions in the edge gradients of the visible plasma boundary (observed as a sharp plasma edge in camera images). The beam-driven chirping modes, identified as ideal magnetohydrodynamics beta-induced Alfvén acoustic eigenmodes modes, are observed in Mirnov coil signals, interferometry, and soft x-ray diagnostics. A moderate amplitude ‘primer’ chirping mode usually precedes an H–L transition. This is followed by a ‘dominant’ chirping mode with higher amplitude during the L-mode. The L–H transition back to the improved confinement occurs on a longer time scale of tens of ms, consistent with the slowing down time scale of fast beam ions. A dramatic decrease in toroidal plasma rotation is systematically observed associated with chirping modes sweeping down to zero frequency. Resonance maps built for the beam-driven chirping modes with the ASCOT (accelerated simulation of charged particle orbits in torodoial devices) code show that the resonant beam ions have orbits near the trapped-passing boundary. The ASCOT modelling assesses how losses of the resonant fast ions caused by the chirping modes with high enough amplitude modify the torque, potentially affecting the plasma rotation.

Current quench and vessel currents characterisation at the COMPASS tokamak

The characterisation of plasma current quench and understanding of its underlying physical processes play a crucial role when designing large fusion devices such as ITER. For the first time, an extensive analysis of the COMPASS tokamak disruption database is presented. A unique set of magnetic diagnostics allows the investigation of local toroidal and poloidal vessel currents, including currents flowing along the open magnetic field lines from the plasma to the vacuum vessel (VV) (i.e., halo currents). Area-normalised current quench times are in agreement with the ITPA 1.67 ms m−2 lower limit. Extremely fast I p quench rates of up to 0.6 MA ms−1 are observed during runaway electron campaigns at COMPASS, which are under the ITPA lower limit. Vertical movement of the plasma column is accelerated by the I p quench during major disruptions. Toroidal vessel currents of around 2% – 4% of the predisruptive plasma current are observed during I p quench. Net poloidal eddy currents are obtained by Mirnov coils and diamagnetic loop, reaching 3% of . Using the Mirnov coils it is shown that the halo current magnitude grows and its poloidal profile broadens with increasing plasma current I p . Geometric features of the VV structure and in-vessel component positions on the poloidal vessel current measurements are discussed.

Gyrokinetic analysis of inter-edge localized mode transport mechanisms in a DIII-D pedestal

In this study, gyrokinetic simulations are used to study pedestal fluctuations for DIII-D discharge 174082 using the GENE code. Nonlinear local simulations indicate that electron heat flux has contributions from electron temperature gradient-driven transport but at levels insufficient to satisfy power balance. We show that microtearing modes (MTM) and neoclassical transport are likely to account for the remaining observed energy losses in the electron and ion channels, respectively. The MTM instabilities found in the simulations are consistent with the high-frequency fluctuations identified in the magnetic fluctuation data from Mirnov coils. The fluctuation data in this discharge also exhibit a low-frequency band of fluctuations. By modifying the equilibrium profiles and plasma β, simulations produce MHD modes, which may be responsible for these observed low-frequency fluctuations. We compare several metrics involving ratios of fluctuation amplitudes and transport quantities for both MTMs and MHD modes. This analysis suggests that the available data are consistent with the simultaneous activity of both MHD modes and MTMs provided that the former is limited largely to the particle transport channel.

Magnetohydrodynamic Mode Identification for Golem Mirnov Coil Signals Using Singular Value Decomposition and Multichannel Variational Mode Decomposition Method for Analyzing Time–Frequency

Magnetohydrodynamic Mode Identification for Golem Mirnov Coil Signals Using Singular Value Decomposition and Multichannel Variational Mode Decomposition Method for Analyzing Time–Frequency

Current density limitation during disruptions due to plasma-sheaths

The presented experimental study realized in the COMPASS tokamak demonstrates, for the first time, that the current density that flows from the plasma into the vacuum vessel during disruptions is limited by the ion particle flux. Such a limitation shows that, at least in COMPASS, the sheath that forms between the plasma and the first wall dominates the halo current flow. This observation is achieved by measuring simultaneously the ion saturation current with negatively biased Langmuir probes and the halo current with grounded probes to the vacuum vessel. These comparative measurements, which were never performed during disruptions in other machines, directly confirm that the halo current density remains below the ion particle flux in COMPASS. The study also shows, using Mirnov coils measurement, that the total electric current entering the wall grows with the plasma current while the current density obtained by Langmuir probes remains unaffected. This, together with the current density limitation, leads to a novel finding that the halo current width increases with the pre-disruptive plasma current, which limits the local forces. The new findings reported here could also provide potential constraints on the modeling of disruption-induced loads on future reactor scale tokamaks and motivation for further experiments on existing devices.

Behavior of multiple modes before and during minor disruption with the external resonant magnetic perturbations on J-TEXT tokamak

The behavior of multiple modes before and during minor disruption with the external resonant magnetic perturbations (RMPs) has been studied on a J-TEXT tokamak. The main component of RMPs is m/n = 2/1, where m and n are the numbers of the poloidal and toroidal modes, respectively. During the mode-locking caused by RMPs, it is found that before a minor or a major disruption (if there is no minor disruption), strong oscillations in both electron temperature and density occur if the edge safety factor q a > 3. The analysis shows that the oscillations are caused by the m/n = 3/1 mode. In addition, using the ECE, Mirnov coils and 2D electron cyclotron emission imaging diagnostic systems, it is found that a thermal collapse occurs on the inner side of the 2/1 magnetic island during the minor disruption, and before the collapse, a 3/2 island increases, after the collapse, the 3/2 island may disappear. Moreover, the study also shows that these 3/1, 2/1 and 3/2 modes play roles in the thermal collapse of disruptions.

Identifying the microtearing modes in the pedestal of DIII-D H-modes using gyrokinetic simulations

Recent evidence points toward the microtearing mode (MTM) as an important fluctuation in the H-mode pedestal for anomalous electron heat transport. A study of the instabilities in the pedestal region carried out using gyrokinetic simulations to model an ELMy H-mode DIII-D discharge (USN configuration, 1.4 MA plasma current, and 3 MW heating power) is presented. The simulations produce MTMs, identified by predominantly electromagnetic heat flux, small particle flux, and a substantial degree of tearing parity. The magnetic spectrogram from Mirnov coils exhibits three distinct frequency bands—two narrow bands at lower frequency (∼35–55 kHz and ∼70–105 kHz) and a broader band at higher frequency (∼300–500 kHz). Global linear GENE simulations produce MTMs that are centered at the peak of the ω * profile and correspond closely with the bands in the spectrogram. The three distinctive frequency bands can be understood from the basic physical mechanisms underlying the instabilities. For example (i) instability of certain toroidal mode numbers (n) is controlled by the alignment of their rational surfaces with the peak in the ω* profile, and (ii) MTM instabilities in the lower n bands are the conventional collisional slab MTM, whereas the higher n band depends on curvature drive. While many features of the modes can be captured with the local approximation, a global treatment is necessary to quantitatively reproduce the detailed band gaps of the low-n fluctuations. Notably, the transport signatures of the MTM are consistent with careful edge modeling by SOLPS.

Comparison of disruption mitigation from shattered pellet injection with massive gas injection on J-TEXT

The mitigation of disruption damage is essential to the safe operation of a large-scale tokamak. In order to achieve the safe operation of ITER, the shattered pellet injection (SPI) has been considered as a primary measure of disruption mitigation. A dedicated argon SPI system, focusing on disruption mitigation has been designed for the J-TEXT tokamak. In the J-TEXT SPI system, a pure argon pellet can be formed in the freezing tube, then separated from the tube and accelerated by a punch mechanism. The pellet can be injected with a speed of 150–300 m s−1. The performance of disruption mitigation by Ar SPI has been compared with Ar massive gas injection (MGI). The cooling process observed from the ECE indicates that the SPI has deeper deposition, with the cold front that can reach the q = 1 rational surface in case of SPI, but stops at the q = 2 profile in MGI. The increase of core plasma density during a fast shutdown is higher than that with Ar MGI, which proves a deeper penetration of SPI. In disruption, the magnetohydrodynamic (MHD) activities, measured at the Mirnov coils, have similar behavior in both MGI and SPI shots. The m/n = 2/1 mode dominates the MHD activities from the penetration to the end of the thermal quench (TQ). Subsequently, in the current quench (CQ) phase, the m/n = 3/1 mode grows to become the dominant mode. The radiation power is much stronger in the SPI shot. The radiation asymmetry is compared in the two plasma disruption mitigation methods. Changing the pellet velocity can effectively adjust the TQ process and CQ rate, which can achieve a higher impurity assimilation rate when the pellet velocity increases in a certain range.

Plasma column position measurements using magnetic diagnostics in ADITYA-U tokamak

Due to several forces acting on the plasma column in a tokamak, the plasma column tends to move horizontally and/or vertically leading to many adverse events including termination of plasma. Precise measurement of plasma column position throughout the discharge with good temporal resolution in the real time is hence necessary in order to restrict the column movement using the stabilizing equilibrium magnetic fields. The plasma column position is measured by several magnetic diagnostics such as magnetic pick-up probe, Mirnov coil, Sine-Cosine coil along with flux loops in ADITYA Upgrade tokamak (ADITYA-U). In spite of the simplicity of the measurement principle of these probes as well as of their construction, obtaining the position of the plasma column is not very straightforward. The diversity of all these magnetic diagnostics in terms of their structures, installation location, mounting scenario etc leads to improper position estimations as these probes are being affected differently by the unwanted magnetic pick-ups. These unwanted pickups, especially those arising from vessel eddies, are successfully removed by introducing a scaling method for all the above-mentioned position probes in ADITYA-U based on the in situ calibration of these probes. The correction factors are deduced by estimating the known positions of a central current carrying conductor correctly using the probe measurements during the in situ calibration. These correction factors are used for estimating the plasma column position in the tokamak experiment along with proper nullification of magnetic fields arising from other magnets during tokamak operation. The plasma column position estimation from different magnetic probes are compared with each other and also with those estimated from diagnostics other than the magnetic ones.

Automatic identification of MHD modes in magnetic fluctuation spectrograms using deep learning techniques

The control and mitigation of magnetohydrodynamic (MHD) oscillation modes is an issue in fusion science because these modes can contribute to the outward particle/energy flux and can drive the device away from ignition conditions. It is of general interest to extract the mode information from large experimental databases in a fast and reliable way. We present a software tool based on deep learning that can identify these oscillation modes taking Mirnov coil spectrograms as input data. It uses convolutional neural networks that we trained with manually-annotated spectrograms from the TJ-II stellarator database. We have tested several detector architectures, resulting in a detector area under the curve score of 0.6 on the test set. Finally, it is applied to find MHD modes in our spectrograms to show how this new software tool can be used to mine large databases.

Dynamics and dependencies of the configuration-dependent 1–2 kHz fluctuation in W7-X

• A 1–2 kHz fluctuation observed in W7-X is studied using cameras, probes, and magnetic coils. • The fluctuation appears to rotate poloidally in the E × B direction between 1 and 10 km/s. • The strength of the fluctuation depends on several plasma and machine parameters. A 1–2 kHz electromagnetic mode is present in a large fraction of W7-X discharges in the standard magnetic field configurations with 5/5 edge island structure. The power in the mode is significantly reduced in other magnetic configurations. The mode can appear spontaneously and persist for the full duration of a discharge. Its full width at half maximum is typically 100–300 Hz. The 1–2 kHz mode was first observed in visible light emissions from the divertor [1] . Matching spectral peaks were found in measurements of both global and edge plasma parameters, including line-integrated density, diamagnetic energy, and edge floating potential and ion saturation current [2] . An infrared camera recording shows a temperature modulation of ~ 2.5 K due to this mode on a divertor tile where the average temperature was 600 °C. Separate correlation analyses of the high-speed camera recordings, Langmuir probe signals, and Mirnov coil signals all indicate a poloidal velocity of 0.5–10 km/s and m = 2 as a likely poloidal mode number. Langmuir probes mounted on the Multi-Purpose Manipulator show that the fluctuation is present in one of the magnetic islands over a radial distance of at least 5 mm. A database of various diagnostic measurements and dB/dt signals from a Mirnov coil was created using a majority of the discharges from the 2018 campaign on W7-X, in order to shed light on the plasma parameters that are correlated with the existence of the 1–2 kHz mode. The power in the mode is positively correlated to the toroidal current, total external heating power, and core electron temperature. It is negatively correlated with plasma density.

A machine learning approach to identify the universality of solitary perturbations accompanying boundary bursts in magnetized toroidal plasmas

Experimental observations assisted by 2-D imaging diagnostics on the KSTAR tokamak show that a solitary perturbation (SP) emerges prior to a boundary burst of magnetized toroidal plasmas, which puts forward SP as a potential candidate for the burst trigger. We have constructed a machine learning (ML) model based on a convolutional deep neural network architecture for a statistical study to identify the SP as a boundary burst trigger. The ML model takes sequential signals detected from 19 toroidal Mirnov coils as input and predicts whether each temporal frame corresponds to an SP. We trained the network in a supervised manner on a training set consisting of real signals with manually annotated SP locations and synthetic burst signals. The trained model achieves high performances in various metrics on a test data set. We also demonstrated the reliability of the model by visualizing the discriminative parts of the input signals that the model recognizes. Finally, we applied the trained model to new data from KSTAR experiments, which were never seen during training, and confirmed that the large burst at the plasma boundary that can fatally damage the fusion device always involves the emergence of SP. This result suggests that the SP is a key to understanding and controlling of the boundary burst in magnetized toroidal plasmas.

Обнаружение высокочастотных альфвеновских колебаний в омических разрядах сферического токамака Глобус-М2

In Globus-M2 ohmic discharges with low density, by means of Mirnov coils array, magnetic field oscillations with frequencies in 1 MHz range were detected. Frequency range of these oscillations significantly exceed the range of TAE and RSAE frequencies, which were previously observed on Globus-M and Globus-M2 tokamaks, and their amplitude, contrary, turned out to be up to an order of magnitude lower. It was found that high frequency oscillations are interrelated with suprathermal electron fraction. At the same time the observed instability seems to have Alfvenic nature, since its frequency correlates well with Alfven frequency scaling. It was also found that magnetic perturbation always forms standing wave with predominantly low toroidal wavenumbers, including n = 0 structure, which makes gap (e.g. TAE) mode excitation impossible. Frequency chirping during single bursts with δω ~ √t is consistent with hole-clump model predictions.