High Beta Tokamak-Extended Pulse Research Articles

Low latency optical-based mode tracking with machine learning deployed on FPGAs on a tokamak.

Active feedback control in magnetic confinement fusion devices is desirable to mitigate plasma instabilities and enable robust operation. Optical high-speed cameras provide a powerful, non-invasive diagnostic and can be suitable for these applications. In this study, we process high-speed camera data, at rates exceeding 100 kfps, on in situ field-programmable gate array (FPGA) hardware to track magnetohydrodynamic (MHD) mode evolution and generate control signals in real time. Our system utilizes a convolutional neural network (CNN) model, which predicts the n = 1 MHD mode amplitude and phase using camera images with better accuracy than other tested non-deep-learning-based methods. By implementing this model directly within the standard FPGA readout hardware of the high-speed camera diagnostic, our mode tracking system achieves a total trigger-to-output latency of 17.6 μs and a throughput of up to 120 kfps. This study at the High Beta Tokamak-Extended Pulse (HBT-EP) experiment demonstrates an FPGA-based high-speed camera data acquisition and processing system, enabling application in real-time machine-learning-based tokamak diagnostic and control as well as potential applications in other scientific domains.

Sawtooth suppression by flux pumping on HBT-EP

This study examines the mechanisms underlying sawtooth suppression in the High Beta Tokamak-Extended Pulse (HBT-EP) device. It is observed that strong-intensity sawtooth activities correlate with reduced-amplitude magnetohydrodynamics (MHD) edge modes which are identified as m/n=3/1 external kink modes, while sawtooth suppression correlates with larger and saturated edge mode amplitudes. To further investigate these correlations, the plasma–wall coupling was manipulated by adjusting the positions of the conducting walls in HBT-EP. It was found that strong sawtooth events occur when the normalized wall radius b/a is within a critical value. This implies that the plasma–wall distance must be sufficiently small to ensure effective stabilization of the edge mode. Even slight differences in major radius result in significantly different discharge styles, categorized as ‘sawtoothing discharges’ and ‘sawtooth-suppressed discharges’ respectively. Through a series of mode structure analyses, we confirm the coexistence and coupling of the m/n=1/1 helical core, m/n=2/1 tearing mode, and m/n=3/1 external kink mode during sawtooth-suppression, and that this coupling induces anomalous current broadening. Based on these findings, we conclude that sawtooth suppression in the HBT-EP tokamak is consistent with the process of magnetic flux pumping.

Tangential extreme ultraviolet and soft x-ray diagnostic system for time-resolved temperature measurement on the High Beta Tokamak-Extended Pulse.

The High Beta Tokamak-Extended Pulse has recently incorporated a tangential multi-energy extreme ultraviolet and soft x-ray diagnostic system. This system enables measurements of the electron temperature and the examination of mode dynamics within the tokamak. While other systems have been built for poloidal views over similar temperature ranges, this is the first multi-energy tangential-view system designed to work in a temperature range below 200eV in a tokamak. To facilitate these measurements, a filter wheel comprising five distinct groups of dual-filters has been developed and implemented. By employing a combination of 0.1μm aluminum and 0.2μm titanium filters, the system allows estimation of electron temperature profiles through reconstruction of the emission profile using the standard "double-foil" technique. The influence of impurities and filter oxide layers on measurement outcomes is examined. Results reveal that, while the absolute electron temperature values may exhibit some deviations, key characteristics like the electron temperature profile shape and inversion radius during sawtooth events remain consistent. This consistency confirms the system's suitability for core plasma studies. This system has proven effective in detecting and analyzing internal magnetohydrodynamic phenomena, such as sawteeth.

MHD mode tracking using high-speed cameras and deep learning

We present a new algorithm to track the amplitude and phase of rotating magnetohydrodynamic (MHD) modes in tokamak plasmas using high speed imaging cameras and deep learning. This algorithm uses a convolutional neural network (CNN) to predict the amplitudes of the n = 1 sine and cosine mode components using solely optical measurements from one or more cameras. The model was trained and tested on an experimental dataset consisting of camera frame images and magnetic-based mode measurements from the High Beta Tokamak - Extended Pulse (HBT-EP) device, and it outperformed other, more conventional, algorithms using identical image inputs. The effect of different input data streams on the accuracy of the model’s predictions is also explored, including using a temporal frame stack or images from two cameras viewing different toroidal regions.

A dimensionality reduction algorithm for mapping tokamak operational regimes using a variational autoencoder (VAE) neural network

A variational autoencoder (VAE) is a type of unsupervised neural network which is able to learn meaningful data representations in a reduced dimensional space. We present an application of VAE in identifying the operational stability boundary of tokamak plasma discharges. This model was implemented using a dataset of over 3000 discharges from the high beta tokamak-extended pulse (HBT-EP) device. We found the VAE model to be capable of forming a continuous low-dimensional operational space map and identifying the operational boundaries using a specified warning time window. By projecting the operational parameters onto the same reduced space, this provides an intuitive way for the machine operator or an automated control system to perform disruption avoidance using a relevant control actuator as a discharge approaches a boundary. Pre-programmed GPU control experiments were conducted to demonstrate this control technique using HBT-EP’s saddle control coils as a horizontal position actuator, showing the ability to avoid the oncoming disruptive event and extend the duration of the discharge.

Mode rotation control in a tokamak with a feedback-driven biased electrode.

Rotation of the plasma and MHD modes in tokamaks has been shown to stabilize resistive wall and tearing modes as well as improve confinement through suppression of edge turbulence. In this work, we control mode rotation with a biased electrode inserted into the plasma of the High Beta Tokamak-Extended Pulse's facility in conjunction with its active GPU (Graphical Processing Unit) feedback system. We first characterize a negative linear relationship between the electrode voltage and mode rotation. Using this relationship, we design, simulate, and implement a proof-of-concept, GPU-based active-control system, which shows consistent success in controlling mode rotation in both feedforward and feedback operation. Controllability is limited by operating conditions, the electrode's voltage range, and by the electrode's proximity to the vessel's walls. The final control system has a 15 μs cycle time, but the addition of various signal filters results in a full cycle latency of 200 μs.

Dynamics of MHD instabilities near a ferromagnetic wall

Prospective fusion component testing and DEMO power reactor concepts are expected to employ low-activation ferritic steels because of their ability to withstand the high neutron flux of the reactor environment. However, theory suggests that ferromagnetic material may amplify certain external MHD instabilities. Using its ferromagnetic-resistive wall mode (FRWM) upgrade, the High Beta Tokamak-Extended Pulse (HBT-EP) experiment has observed approximately doubled growth rates when operating with a close-fitting ferromagnetic first wall, compared to operation with a stainless steel first wall. The presence of a ferromagnetic wall correlates with earlier disruptions, and FRWM growth rates increase with decreasing mode rotation, as expected due to the increased skin depth allowing greater mode interaction with the bulk ferromagnetic material. It is also seen that introducing low-n asymmetries into the toroidal distribution of ferromagnetic material, similar to the ITER test blanket module toroidal asymmetry, changes the phase preference of rotating modes; meanwhile, a similar change in purely conducting material does not significantly change the mode’s phase preference.

Measurement of scrape-off-layer current dynamics during MHD activity and disruptions in HBT-EP

We report scrape-off layer (SOL) current measurements during magnetohydrodynamic (MHD) mode activity, resonant magnetic perturbations (RMPs), and disruptions in the High Beta Tokamak—Extended Pulse (HBT-EP) device. Currents are measured via segmented plasma current Rogowski coils, jumpers running toroidally between otherwise-isolated vessel sections, and a grounded electrode in the scrape-off layer. These currents strongly depend on the plasma’s major radius, and amplitude and phase of non-axisymmetric field components. SOL currents connecting through the vessel are seen to reach ∼ of the plasma current during typical kink activity and RMPs. Plasma current asymmetries and scrape-off-layer currents generated during disruptions, which are commonly called halo currents, reach ∼ of Ip. Asymmetric toroidal currents between vessel sections rotate at tens of kHz through most of the current quench, then symmetrize once Ip reaches ∼ of its pre-disruptive value. Toroidal jumper currents oscillate between co- and counter-Ip, with co-Ip being dominant on average during disruptions. A relative increase in local plasma current measured by a segmented Ip Rogowski coil correlates with counter-Ip current in the nearest toroidal jumper. Measurements are interpreted in the context of two models that produce contrary predictions for the toroidal vessel current polarity during disruptions. Plasma current asymmetry measurements are consistent with both models, and SOL currents scale with plasma displacement toward the vessel wall. The design of an upcoming SOL current diagnostic and control upgrade is also briefly presented.

Improved feedback control of wall stabilized kink modes with different plasma–wall couplings and mode rotation

A new algorithm for feedback control of rotating, wall-stabilized kink modes in the High Beta Tokamak-Extended Pulse (HBT-EP) device maintains an accurate phase shift between the perturbation and the measured rotating mode through current control, with control power emphasizing fast rotation and phase jumps over fast amplitude changes. In HBT-EP, wall-stabilized kink modes become unstable above the ideal wall stability limit, and feedback suppression is aimed at delaying the onset of discharge disruption through reduction of the kink mode amplitude. Performance of the new feedback algorithm is tested under different experimental conditions, including variation of the plasma–wall coupling, insertion of a ferritic wall, changing mode rotation frequency over the range of 4–8 kHz using an internal biased electrode, and adjusting the feedback phase-angle to accelerate, amplify, or suppress the mode. We find the previously reported excitation of the slowly rotating mode at high feedback gain in HBT-EP is mitigated by the current control scheme. We also find good agreement between the observed and predicted changes to the mode rotation frequency and amplitude. When ferritic material is introduced, or the plasma–wall coupling becomes weaker as the walls are retracted from plasma, the feedback gain needs to be increased to achieve the same level of suppression. When mode rotation is slowed by a biased electrode, the feedback system still achieves mode suppression, and demonstrates wide bandwidth effectiveness.

Design and installation of a ferromagnetic wall in tokamak geometry.

Low-activation ferritic steels are leading material candidates for use in next-generation fusion development experiments such as a prospective component test facility and DEMO power reactor. Understanding the interaction of plasmas with a ferromagnetic wall will provide crucial physics for these facilities. In order to study ferromagnetic effects in toroidal geometry, a ferritic wall upgrade was designed and installed in the High Beta Tokamak-Extended Pulse (HBT-EP). Several material options were investigated based on conductivity, magnetic permeability, vacuum compatibility, and other criteria, and the material of choice (high-cobalt steel) is characterized. Installation was accomplished quickly, with minimal impact on existing diagnostics and overall machine performance, and initial results demonstrate the effects of the ferritic wall on plasma stability.

Active and passive kink mode studies in a tokamak with a movable ferromagnetic walla)

High-resolution active and passive kink mode studies are conducted in a tokamak with an adjustable ferromagnetic wall near the plasma surface. Ferritic tiles made from 5.6 mm thick Hiperco® 50 alloy have been mounted on the plasma-facing side of half of the in-vessel movable wall segments in the High Beta Tokamak-Extended Pulse device [D. A. Maurer et al., Plasma Phys. Controlled Fusion 53, 074016 (2011)] in order to explore ferritic resistive wall mode stability. Low-activation ferritic steels are a candidate for structural components of a fusion reactor, and these experiments examine MHD stability of plasmas with nearby ferromagnetic material. Plasma-wall separation for alternating ferritic and non-ferritic wall segments is adjusted between discharges without opening the vacuum vessel. Amplification of applied resonant magnetic perturbations and plasma disruptivity are observed to increase when the ferromagnetic wall is close to plasma surface instead of the standard stainless steel wall. Rapidly rotating m/n=3/1 external kink modes have higher growth rates with the nearby ferritic wall. Feedback suppression of kinks is still as effective as before the installation of ferritic material in vessel, in spite of increased mode growth rates.

Measurement of 3D plasma response to external magnetic perturbations in the presence of a rotating external kink

The detailed measurements of the 3D plasma response to applied external magnetic perturbations in the presence of a rotating external kink are presented, and compared with the predictions of a single-helicity linear model of kink mode dynamics. The modular control coils of the High Beta Tokamak-Extended Pulse (HBT-EP) device are used to apply resonant m/n = 3/1 magnetic perturbations to wall-stabilized tokamak plasmas with a pre-existing rotating 3/1 kink mode. The plasma response is measured in high-resolution with the extensive magnetic diagnostic set of the HBT-EP device. The spatial structures of both the naturally rotating kink mode and the externally driven response are independently measured and observed to be identical, while the temporal dynamics are consistent with the independent evolution and superposition of the two modes. This leads to the observation of a characteristic change in 3D field dynamics as a function of the applied field amplitude. This amplitude dependence is found to be different for poloidal and radial fields. The measured 3D response is compared to and shown to be consistent with the predictions of the linear single-helicity model in the “high-dissipation” regime, as reported previously [M. E. Mauel et al., Nucl. Fusion 45, 285 (2005)].

Multimode observations and 3D magnetic control of the boundary of a tokamak plasma

We present high-resolution detection and control of the 3D magnetic boundary in the High Beta Tokamak-Extended Pulse (HBT-EP) device. Measurements of non-axisymmetric radial and poloidal fields are made using 216 magnetic sensors positioned near the plasma surface. Control of 3D fields is accomplished using 40 independent saddle coils attached to the passive stabilizing wall. The control coils are energized with high-power solid-state amplifiers, and massively parallel, high-throughput feedback control experiments are performed using low-latency connections between PCI Express analogue input and output modules and a graphics processing unit. The time evolution of unstable and saturated wall-stabilized external kink modes are studied with and without applying magnetic perturbations using the control coils. The 3D dynamic structure of the magnetic field surrounding the plasma is determined through biorthogonal decomposition using the full set of magnetic sensors without the need to fit either a Fourier or a model-based basis. Naturally occurring external kinks are composed of multiple independent helical modes. Smooth transitions between dominant poloidal mode numbers are observed for simultaneous n = 1 and n = 2 modes as the edge safety factor changes. Relative amplitudes of coexistent m/n = 3/1 and 6/2 modes depend on the plasma's major radius and edge safety factor. When stationary 3/1 magnetic perturbations are applied, the resonant response can be linear, saturated, or disruptive, depending upon the perturbation amplitude and the edge safety factor; increased plasma–wall interactions from the perturbed plasma are proposed as a saturation mechanism. Initial feedback experiments have used 40 sensors and 40 control coils, producing mode amplification or suppression, and acceleration or deceleration depending on the feedback phase angle.

In situ “artificial plasma” calibration of tokamak magnetic sensors

A unique in situ calibration technique has been used to spatially calibrate and characterize the extensive new magnetic diagnostic set and close-fitting conducting wall of the High Beta Tokamak-Extended Pulse (HBT-EP) experiment. A new set of 216 Mirnov coils has recently been installed inside the vacuum chamber of the device for high-resolution measurements of magnetohydrodynamic phenomena including the effects of eddy currents in the nearby conducting wall. The spatial positions of these sensors are calibrated by energizing several large in situ calibration coils in turn, and using measurements of the magnetic fields produced by the various coils to solve for each sensor's position. Since the calibration coils are built near the nominal location of the plasma current centroid, the technique is referred to as an "artificial plasma" calibration. The fitting procedure for the sensor positions is described, and results of the spatial calibration are compared with those based on metrology. The time response of the sensors is compared with the evolution of the artificial plasma current to deduce the eddy current contribution to each signal. This is compared with simulations using the VALEN electromagnetic code, and the modeled copper thickness profiles of the HBT-EP conducting wall are adjusted to better match experimental measurements of the eddy current decay. Finally, the multiple coils of the artificial plasma system are also used to directly calibrate a non-uniformly wound Fourier Rogowski coil on HBT-EP.

High resolution detection and excitation of resonant magnetic perturbations in a wall-stabilized tokamak

We report high-resolution detection of the 3D plasma magnetic response of wall-stabilized tokamak discharges in the High Beta Tokamak-Extended Pulse [T. H. Ivers et al., Phys. Plasmas 3, 1926 (1996)] device. A new adjustable conducting wall has been installed on HBT-EP made up of 20 independent, movable, wall segments instrumented with three distinct sets of 40 modular coils that can be independently driven to generate a wide variety of magnetic perturbations. High-resolution detection of the plasma response is made with 216 poloidal and radial magnetic sensors that have been located and calibrated with high-accuracy. Static and dynamic plasma responses to resonant and non-resonant magnetic perturbations are observed through measurement of the step-response following a rapid change in the toroidal phase of the applied perturbations. Biorthogonal decomposition of the full set of magnetic sensors clearly defines the structures of naturally occurring external kinks as being composed of independent m/n = 3/1 and 6/2 modes. Resonant magnetic perturbations were applied to discharges with pre-existing, saturated m/n = 3/1 external kink mode activity. This m/n = 3/1 kink mode was observed to lock to the applied perturbation field. During this kink mode locked period, the plasma resonant response is characterized by a linear, a saturated, and a disruptive plasma regime dependent on the magnitude of the applied field and value of the edge safety factor and plasma rotation.

The high beta tokamak-extended pulse magnetohydrodynamic mode control research program

The high beta tokamak-extended pulse (HBT-EP) magnetohydrodynamic (MHD) mode control research program is studying ITER relevant internal modular feedback control coil configurations and their impact on kink mode rigidity, advanced digital control algorithms and the effects of plasma rotation and three-dimensional magnetic fields on MHD mode stability. A new segmented adjustable conducting wall has been installed on the HBT-EP and is made up of 20 independent, movable, wall shell segments instrumented with three distinct sets of 40 saddle coils, totaling 120 in-vessel modular feedback control coils. Each internal coil set has been designed with varying toroidal angular coil coverage of 5, 10 and 15°, spanning the toroidal angle range of an ITER port plug based internal coil to test resistive wall mode (RWM) interaction and multimode MHD plasma response to such highly localized control fields. In addition, we have implemented 336 new poloidal and radial magnetic sensors to quantify the applied three-dimensional fields of our control coils along with the observed plasma response. This paper describes the design and implementation of the new control shell incorporating these control and sensor coils on the HBT-EP, and the research program plan on the upgraded HBT-EP to understand how best to optimize the use of modular feedback coils to control instability growth near the ideal wall stabilization limit, answer critical questions about the role of plasma rotation in active control of the RWM and the ferritic resistive wall mode, and to improve the performance of MHD control systems used in fusion experiments and future burning plasma systems.

A high-power spatial filter for Thomson scattering stray light reduction

The Thomson scattering diagnostic on the High Beta Tokamak-Extended Pulse (HBT-EP) is routinely used to measure electron temperature and density during plasma discharges. Avalanche photodiodes in a five-channel interference filter polychromator measure scattered light from a 6 ns, 800 mJ, 1064 nm Nd:YAG laser pulse. A low cost, high-power spatial filter was designed, tested, and added to the laser beamline in order to reduce stray laser light to levels which are acceptable for accurate Rayleigh calibration. A detailed analysis of the spatial filter design and performance is given. The spatial filter can be easily implemented in an existing Thomson scattering system without the need to disturb the vacuum chamber or significantly change the beamline. Although apertures in the spatial filter suffer substantial damage from the focused beam, with proper design they can last long enough to permit absolute calibration.

A Kalman filter for feedback control of rotating external kink instabilities in the presence of noise

The simulation and experimental optimization of a Kalman filter feedback control algorithm for n=1 tokamak external kink modes are reported. In order to achieve the highest plasma pressure limits in ITER, resistive wall mode stabilization is required [T. C. Hender et al., Nucl. Fusion 47, S128 (2007)] and feedback algorithms will need to distinguish the mode from noise due to other magnetohydrodynamic activity. The Kalman filter contains an internal model that captures the dynamics of a rotating, growing n=1 mode. This model is actively compared with real-time measurements to produce an optimal estimate for the mode’s amplitude and phase. On the High Beta Tokamak-Extended Pulse experiment [T. H. Ivers et al., Phys. Plasmas 3, 1926 (1996)], the Kalman filter algorithm is implemented using a set of digital, field-programmable gate array controllers with 10 μs latencies. Signals from an array of 20 poloidal sensor coils are used to measure the n=1 mode, and the feedback control is applied using 40 poloidally and toroidally localized control coils. The feedback system with the Kalman filter is able to suppress the external kink mode over a broad range of phase angles between the sensed mode and applied control field. Scans of filter parameters show good agreement between simulation and experiment, and feedback suppression and excitation of the kink mode are enhanced in experiments when a filter made using optimal parameters from the scans is used.

Active control of 2/1 magnetic islands in a tokamak

Closed and open loop control techniques were applied to growing m/n=2/1 rotating islands in wall-stabilized plasmas in the High Beta Tokamak-Extended Pulse (HBT-EP) [J. Fusion Energy 12, 303 (1993)]. HBT-EP combines an adjustable, segmented conducting wall (which slows the growth or stabilizes ideal external kinks) with a number of small (6° wide toroidally) driven saddle coils located between the gaps of the conducting wall. Two-phase driven magnetic island rotation control from 5 to 15 kHz has been demonstrated powered by two 10 MW linear amplifiers. The phase instability has been observed and is well modeled by the single-helicity predictions of nonlinear Rutherford island dynamics for 2/1 tearing modes including important effects of ion inertia and finite Larmor radius, which appear as a damping term in the model equations. The closed loop response of active feedback control of the 2/1 mode at moderate gain was observed to be in good agreement with the theory. Suppression of the 2/1 island growth has been demonstrated using an asynchronous frequency modulation drive which maintains the inertial flow damping of the island by application of rotating control fields with frequencies alternating above and below the natural mode frequency. This frequency modulation control technique was also able to prevent disruptions normally observed to follow giant sawtooth crashes in the plasma core.

A photodiode for the measurement of soft x-ray radiation from plasma

Photodiodes with three layer zirconium, titanium, and aluminum have been used as detectors on the high beta tokamak-extended pulse tokamak for soft x-ray radiation measurement. Signals from the photodiodes show sawteeth instabilities and magnetohydrodynamic oscillations.