Session JO03: KSTAR Tokamak

KSTAR Overview

KSTAR focused on exploring the key physics and engineering issues of the high beta steady-state operation for ITER and future fusion reactors utilizing unique capabilities of KSTAR. Most of all, an advanced scenario was developed targeting steady-state operation and significant progress has been made in shape and heating control to address the stationary high ion temperature, hybrid, double barrier and high beta scenario with beta_N ~3 and newly observed I-mode-like regime. KSTAR advanced operation scenario have shown that the electron cyclotron current drive mitigated and suppressed the beam-ion driven toroidal Alfv´en eigenmodes (TAEs) for over several tens of global energy confinement time[1].

Symmetric multiple Shattered Pellet Injections (SPIs) for ITER disruption mitigation operation uniquely demonstrated for its performance in KSTAR. The real time DECAF (Disruption Event Characterization and Forecasting) validated the real time disruption event characterization and forecasting events on KSTAR.

In recent KSTAR 3D magnetic field experiments, ELM optimization and automatic control techniques such as adapted ELM controller [1] and Machine Learning algorithm [2] in order to suppress ELM successfully implemented in KSTAR PCS. Machine learning algorithm successfully suppressed nearly complete ELM-crash with the first ELM. Machine learning, adaptive ELM controller and ERMPs applied to RMP-driven ELM suppression experiments with long pulses for ITER operation.

To enhance understanding of the fundamental processes, the KSTAR research also focused on interaction of island and turbulence in pedestal with complexity-enthropy and the dynamics of turbulence based on broadband ECE diagnostics.

    

Author not Attending

The low q_min≈1, high β_N scenario at a relatively high value of the internal inductance, ??_i is one of the promising candidates for the ITER steady-state scenario. The high ??_i scenario on KSTAR has achieved β_N≈3, G=β_NH₈₉/q₉₅²>0.3, f_NI≈0.9 at q₉₅=5, approaching the normalized performance required for the fully non-inductive, Q=5 goal for ITER. The dRsep>0 (upper single null) operation achieved an “optimized” ??_i≈1 by considering trade-off between the edge pedestal and high ??_i at lower density to maximize the current drive efficiency for the fully non-inductive operation. The no-wall β_N limit calculated by DCON, β_N,limit≈3.9, indicates that higher β_N operation is possible even without wall stabilization. The higher current drive efficiency for the central electron cyclotron current drive (ECCD) compensates for the relatively low bootstrap current fraction for the fully non-inductive operation. No obvious sign of performance degradation has been observed for the pulse length of ~10 sec at β_N≈3. We will report recent progress to achieve higher β_N>3 and longer pulse > 15 sec, guided by the integrated scenario modeling IPS-FASTRAN.    

A plasma regime with long sustained highly peaked ion temperature in the KSTAR tokamak

A new type of plasma operating scenario in KSTAR has been developed by preventing the H-mode transition. The specific method is to maintain a low density and upper single null configuration, which are unfavorable conditions for H-mode transition, when a neutral beam of moderate power is applied. The main feature of the scenario is maintaining a high ion temperature of 10keV or higher for a long time. The overall performance is comparable to a conventional H-mode or above. It also shows a low-loop voltage, which is advantageous for the steady-state operation. Compared to the hybrid plasma, a well-known high-performance plasma regime, we conclude that novelty lies in a high portion of fast ions that stabilize core turbulence during long pulse duration. We coin this new regime as the fast ion regulated enhancement (FIRE) mode and think that the FIRE mode is worth studying for future commercial fusion plasma operations.

    

Validation of plasma response and turbulence simulation across KSTAR core magnetic islands

One of the biggest challenges for long-pulse ELM suppression is the control of the heat and particle fluxes' interaction with plasma-facing surfaces while simultaneously minimizing the impurities influx into the core plasma. Achieving these goals relies on our understanding of the transport in the presence of the 3D perturbation fields changing the magnetic topology of the plasma in the edge and core regions. In this work, we present the results from the recent KSTAR L-mode experiments showing increased turbulence near the X-point of the 2/1 core magnetic island resulting from the locked mode. We performed linear MHD simulations of the 2/1 LM island and validated the plasma response models in M3D-C1 and NIMROD against the available KSTAR ECE-I measurements. The results of these MHD simulations were used to initiate the GTC simulations to validate the transport models. Both, the experimental and linear plasma response modeling results with M3D-C1 and NIMROD codes have shown that the physics mechanisms involved are complex and not well understood and require more detailed analysis in order to validate both the plasma response and transport modeling tools for use in ITER and future burning plasma devices.

    

Dynamics and statistics of a self-organized staircase-like electron temperature corrugation in KSTAR plasmas

Turbulent transport near the marginal stability of tokamak plasmas is dominated by non-diffusive avalanche transport events in gyrokinetic simulations. The avalanche transport events are found to interact with self-organized shear flow layers, or the E x B staircase generating a staircase-like pressure corrugation. Various models have been suggested for the relation between the avalanche transport events and the shear flow layers, or the pressure corrugation. While intensive simulation studies have been conducted to advance the understanding of their relation, the experimental researches have been mostly limited to the demonstration of their existence. For example, in KSTAR tokamak plasmas, the co-existence of the avalanche-like electron heat transport events and the staircase-like electron temperature corrugation was confirmed [1]. In this work, further analyses of dynamics and statistics of the self-organized temperature corrugation are provided to understand its evolution and reveal the relation with the avalanche-like events. The avalanche-like events are found to have strong influences on the evolution of the self-organized temperature corrugation.  

[1] M. J. Choi et al., Nuclear Fusion 59, 086027 (2019)   

Safe and efficient access to ELM suppression in KSTAR using low-n edge localized RMP

An edge localization of 3D fields proposes a unique path of a 3D field to safely access the ELM-suppressed High confinement mode (H-mode) from low confinement mode (L-mode) phases with improved plasma performance. Such an edge-localized resonant magnetic perturbation (ERMP) [1] is essential to minimize disruptive core components of low-n RMP while maintaining its efficiency in ELM suppression for the entire period of H-mode. The core reduction in ERMP also expands the ELM suppression window, increasing the safety margin for feedback-based ELM control by 100%. In addition, the ERMP reduces rotation damping and density degradation while maintaining ELM suppression. The confinement improvement from ERMP can be explained by a reduction in neoclassical transport and fast ion orbit losses, as demonstrated in GPEC and NubDEC simulations. The successful validation and applications of ERMP show promising aspects of low-n 3D field optimization to avoid the unnecessary component of the 3D field for ELM control.

 

    

ELM suppressed high performance fusion scenarios achieved with Feedback Adaptive RMP ELM Control

We designed a Feedback Adaptive RMP ELM Controller [1] which, based on the detected ELM activity, changes the RMP such that ELM suppression is achieved. When applied to KSTAR, the controller was successful at obtaining, sustaining and optimizing ELM suppression where up to 60% [2] of the confinement degradation was recovered in feedback.

To eliminate all ELMs, additional capabilities are added, such as ELM suppression-loss precursor detection using Dα-emission. When the detector detects a precursor, an RMP pulse is applied. This mechanism demonstrated its ability to prevent imminent ELMs. As observed, the optimized RMP will change during long pulses because of plasma evolution. Since detailed models are too computationally expensive in real-time, an ML algorithm has been developed that determines the optimized spectrum in real-time. This could allow the controller to adjust significantly during long pulses, in a way where it sustains ELM suppression throughout.

[1] R. Shousha et al., Phys. Plasmas, 29, 032514 (2022).

[2] S. K. Kim et al., Nucl. Fusion, 62, 026043 (2022).

    

Nonlinear MHD modeling on RMP-induced pump-out under the KSTAR geometry

This work introduces nonlinear 3D MHD simulations and validations with KSTAR geometry that reveal a hybrid particle-MHD transport as a key process for driving the bifurcating dynamics in particle transport under resonant magnetic field perturbation (RMPs). Here, JOREK [1,2] and PENTRC [3] codes are coupled for the modeling. It turns out that the threshold characteristics of pump-out originate from resonant field penetration and island opening [4], showing a good agreement with the previous modeling [5]. However, these toroidal simulations also show that the pump-outs are attributable to both polarization and neoclassical toroidal viscosity (NTV) [6] effects. In particular, the first pump-out observed in KSTAR experiments is reproduced only when both transport mechanisms are integrated into the simulations. [1] G.T.A. Huysmans et al., PPCF 51, 124012 [2] M. Hoelzl et al., NF 61, 065001 [3] N. C. Logan et al., POP 20, 122507 [4] R. Nazikian et al., PRL 114, 105002 [5] Q. Hu et al., POP 26, 120702 [6] Y. Liu NF 60, 036018

*Supported by the US DOE Contract DE-SC0020372.   

Disruption Event Characterization and Forecasting Research and First Real-time Application on KSTAR

Disruption prediction and avoidance is critical for ITER and reactor-scale tokamaks to maintain steady plasma operation and to avoid damage to device components. Physics-based disruption event characterization and forecasting (DECAF) research determines the relation of events leading to disruption and aims to provide event onset forecasts with high accuracy and early warning for disruption avoidance. Real-time application of DECAF was recently made on the KSTAR superconducting tokamak. Experiments focused on locking MHD instabilities produced 40 plasmas with nearly equal disrupted / non-disrupted cases that are forecast with 100% accuracy. These real-time forecasts triggered controlled plasma shutdown and disruption mitigation. The warnings were issued well before the expected plasma disruption time and early warning guidance given for ITER disruption mitigation. Offline analysis has access to data from several tokamaks (e.g. KSTAR, MAST, NSTX) to best understand, validate, and extrapolate models. Recent code improvements allow fully automated analysis of up to entire device databases. Such initial analysis shows very high true positive success rates over 99%. *This research is supported by the U.S. DOE under grants DE-SC0020415, DE-SC0018623, and DE-SC0021311.   

Real-time locking dynamics analysis of rotating MHD for disruption prediction and avoidance in KSTAR

Tokamak reactors require low disruptivity to support commercial viability. An important precursor to disruptions is the locking dynamic of rotating MHD events that are often neoclassical tearing modes (NTM). The drag of electromagnetic and fluid viscosity torques can cause the slowing down of NTM's and lock them to device conducting structure. A balance of the driving torque from the NBI, and drag from perpendicular viscous diffusion drag and electromagnetic forces on the mode, as well as its inertia, are used to model the mode rotation dynamics. Rotation frequencies below which the mode rotation is expected to lead to locking serve as a disruption forecaster. Mode identification is computed most accurately by Fourier analysis of a toroidal array of magnetic probes, or using simpler approaches more amenable to real-time calculation. From the rotation, the torque components are then calculated based on conditions for the expected drag torque ratios at the mode onset, changes in frequency, and amplitude. The technique is employed for database and real-time analysis of KSTAR plasmas. Database analysis is carried out to validate the model. In real-time, the forecaster was used to trigger controlled shut-down or disruption mitigation systems.   

Enhancement of detachment control with simplified real-time modelling on the KSTAR tokamak

Detachment control based on ion saturation current I_sat measurements from Langmuir probes (LPs) is implemented in the KSTAR tokamak and shown to be capable of following dynamic and constant target trajectories with good accuracy, in H-mode, by moderating the flow rate of nitrogen or deuterium. I_sat controllers normalize I_sat in order to form attachment fraction (A_frac) as their control parameter. The KSTAR implementation of A_frac control differs from previous work in that it continuously calculates a model for attached I_sat and uses that as the denominator in A_frac, whereas prior implementations either record peak I_sat at rollover as they pass it or take estimated I_sat,rollover as a manual input prior to the shot. The KSTAR controller therefore does not need to keep track of rollover status and keep separate targets for pre- and post-rollover states, and it can automatically adapt to changes in scenario at any time. It is also less vulnerable to noise as it will not lock in an outlier as a rollover point.

    

Investigation of the effects of krypton seeding under various experimental conditions in KSTAR H-mode plasmas

Controlled krypton gas seeding experiments under various experimental conditions were performed in KSTAR H-mode discharges. ELM mitigation/suppression, divertor heat flux reduction, and ion internal transport barrier (ITB) formation were observed with increasing krypton levels. ELM was mitigated and suppressed in an intermediate level of krypton seeding (c_{Kr, local maximum} ~0.1%). When the krypton amount was increased to a high level (c_{Kr, local maximum} ~0.25%), ELM was suppressed and H-L back transition occurred. In an even higher level of krypton seeding (c_{Kr, local maximum} > ~0.5%), an ion ITB discharge with T_i,core > 10 keV and a significant reduction in particle flux onto both inner and outer divertor targets were achieved. To investigate the effects of krypton on the core plasma, linear gyrokinetic simulations for the ion ITB case have been conducted using the GKW code. The maximum growth rate (γ_lin,max) of instability was significantly reduced in the region around the peak of krypton concentration n_kr/n_e. To further explore the effect of krypton seeding on heat flux reduction, SOLPS-ITER modeling of divertor plasmas was conducted. As the concentration of krypton increased, radiated power fraction of the core region increased while that of the divertor region decreased.   

A Study of Shattered Pellet Injection Scheme Using Multiple Injection System In KSTAR

ITER plans to systematically inject multiple shattered pellets to reliably dissipate thermal and electromagnetic energies which are rapidly released in a plasma disruption. It is expected that the effectiveness of disruption mitigation in ITER will heavily depend on the material composition of the pellets and also their injection timing since, depending on the order of injection, subsequent pellets will encounter plasmas of completely different properties. KSTAR with two shattered pellet injectors (SPIs) which form a symmetry in the toroidal direction can independently inject three different pellets from each SPI. KSTAR has conducted experiments to validate the disruption mitigation scheme of ITER using multiple shattered pellets by varying their composition and timing. For instance, in order to prevent runaway electron generation, when pure deuterium pellets are first injected and then Ne-doped pellets follow them, there is a tendency to show more localized radiation compared to the case in which only Ne-doped pellets are injected. To measure the characteristics of the plasma during disruption mitigation that happens in a short period of time within a few milliseconds, we also operated dedicated diagnostics that can measure related physical process such as fast bolometry, fast visible camera, and short wavelength interferometry. This result will contribute to establishing efficient injection scheme of ITER.   

Development of Virtual KSTAR

Digital twin technologies are expanding their applications in many scientific and engineering researches. As an attempt to apply the digital twin technologies to fusion, the development of Virtual KSTAR (V-KSTAR) is reported. The graphic engine Unity® is employed for fast and interactive visualization of KSTAR main device. Based on HDF5 format, a new data framework is also developed to systematically process and integrate fusion data from experiment and simulation into V-KSTAR. It is demonstrated that V-KSTAR can visualize key experimental data such as plasma equilibrium, temperatures of plasma facing components in real time, which can be utilized for the real time monitoring of KSTAR operation and experiment. V-KSTAR can incorporate fusion simulation. To demonstrate this, three simulation codes are selected: 1) Monte-Carlo simulation code for neutral beam injection, 2) ray tracing simulation for electron cyclotron wave heating, and 3) 3D magnetic field simulation for resonant magnetic field perturbation. It is demonstrated that V-KSTAR can provide a fitting platform to perform and analyze these fusion simulations in virtualized environment.