In this paper, we present a new deep learning disruption prediction algorithm based on important findings from explorative data analysis which effectively allows knowledge transfer from existing devices to new ones, thereby predicting disruptions using very limited disruptive data from the new devices. The explorative data analysis conducted via unsupervised clustering techniques confirms that time-sequence data are much better separators of disruptive and non-disruptive behavior than the instantaneous plasma state data with further advantageous implications for a sequence-based predictor. Based on such important findings, we have designed a new algorithm for multi-machine disruption prediction that achieves high predictive accuracy on the C-Mod (AUC=0.801), DIII-D (AUC=0.947) and EAST (AUC=0.973) tokamaks with limited hyperparameter tuning. Through numerical experiments, we show that boosted accuracy (AUC=0.959) is achieved on EAST predictions by including in the training only 20 disruptive discharges, thousands of non-disruptive discharges from EAST, and combining this with more than a thousand discharges from DIII-D and C-Mod. The improvement of predictive ability obtained by combining disruptive data from other devices is found to be true for all permutations of the three devices. Furthermore, by comparing the predictive performance of each individual numerical experiment, we find that non-disruptive data are machine-specific while disruptive data from multiple devices contain device-independent knowledge that can be used to inform predictions for disruptions occurring on a new device.
Huazhong University of Science and Technology（华中科技大学）；Institute of Plasma Physics, HFIPS, Chinese Academy of Sciences（中国科学院合肥物质科学研究院等离子体物理研究所）；Shenzhen University（深圳大学）
Predicting disruptions across different tokamaks is a great obstacle to overcome. Future tokamaks can hardly tolerate disruptions at high performance discharge. Few disruption discharges at high performance can hardly compose an abundant training set, which makes it difficult for current data-driven methods to obtain an acceptable result. A machine learning method capable of transferring a disruption prediction model trained on one tokamak to another is required to solve the problem. The key is a disruption prediction model containing a feature extractor that is able to extract common disrupti…
Plasma Science and Fusion Center, Massachusetts Institute of Technology（麻省理工学院等离子体科学与聚变中心）；Institute of Plasma Physics, Chinese Academy of Sciences（中国科学院等离子体物理研究所）；General Atomics（通用原子公司）
Next generation high performance (HP) tokamaks risk damage from unmitigated disruptions at high current and power. Achieving reliable disruption prediction for a device's HP operation based on its low performance (LP) data is key to success. In this letter, through explorative data analysis and dedicated numerical experiments on multiple existing tokamaks, we demonstrate how the operational regimes of tokamaks can affect the power of a trained disruption predictor. First, our results suggest data-driven disruption predictors trained on abundant LP discharges work poorly on the HP regime of the…
The purpose of this paper is to further investigate the major disruptions occurring in low-q(a) discharges in Iran Tokamak 1, and to compare the theoretical and experimental results for the rate of island growth. The study of precursor phase of disruption can be predicted and avoided using suitable control systems. In this paper are described the stability analysis and the observed growth rates indicate that the rotating modes are tearing modes.
Understanding generation and mitigation of runaway electrons in disruptions is important for the safe operation of future tokamaks. In this paper we investigate runaway dynamics in reactor-scale spherical tokamaks. We study both the severity of runaway generation during unmitigated disruptions, as well as the effect that typical mitigation schemes based on massive material injection have on runaway production. The study is conducted using the numerical framework DREAM (Disruption Runaway Electron Analysis Model). We find that, in many cases, mitigation strategies are necessary to prevent the r…
Plasma terminating disruptions in tokamaks may result in relativistic runaway electron beams with potentially serious consequences for future devices with large plasma currents. In this paper we investigate the effect of plasma elongation on the coupled dynamics of runaway generation and resistive diffusion of the electric field. We find that elongated plasmas are less likely to produce large runaway currents, partly due to the lower induced electric fields associated with larger plasmas, and partly due to direct shaping effects, which mainly lead to a reduction in the runaway avalanche gain.
Steering tokamak plasmas is commonly viewed as a way to avoid disruptions and runaway electrons. Plasma steering sounds as safe as driving to work but will be shown to more closely resemble driving at high speed through a dense fog on an icy road. The long time required to terminate an ITER discharge compared to time over which dangers can be foreseen is analogous to driving in a dense fog. The difficulty of regaining plasma control if it is lost resembles driving on an icy road. Disruptions and runaways are associated with three issues -- a solution to one tends to complicate the solution to …
Non-rotating (`locked') magnetic islands often lead to complete losses of confinement in tokamak plasmas, called major disruptions. Here locked islands were suppressed for the first time, by a combination of applied three-dimensional magnetic fields and injected millimetre waves. The applied fields were used to control the phase of locking and so align the island O-point with the region where the injected waves generated non-inductive currents. This resulted in stabilization of the locked island, disruption avoidance, recovery of high confinement and high pressure, in accordance with the expec…
Active Disruption Avoidance and Trajectory Design for Tokamak Ramp-downs with Neural Differential Equations and Reinforcement Learning
Plasma Science and Fusion Center, Massachusetts Institute of Technology（麻省理工学院等离子体科学与聚变中心）；Laboratory for Information and Decision Systems, Massachusetts Institute of Technology（麻省理工学院信息与决策系统实验室）；OpenStar Technologies LLC.（OpenStar科技有限责任公司）
Avoidance of the harmful effects of runaway electrons (REs) in plasma-terminating disruptions is pivotal in the design of safety systems for magnetic fusion devices. Here, we describe a computationally efficient numerical tool, that allows for self-consistent simulations of plasma cooling and associated RE dynamics during disruptions. It solves flux-surface averaged transport equations for the plasma density, temperature and poloidal flux, using a bounce-averaged kinetic equation to self-consistently provide the electron current, heat, density and RE evolution, as well as the electron distribu…
Intrinsically explosive growth of a ballooning finger is demonstrated in nonlinear magnetohydrodynamic calculations of high-$β$ disruptions in tokamaks. The explosive finger is formed by an ideally unstable n=1 mode, dominated by an m/n=2/1 component. The quadrupole geometry of the 2/1 perturbed pressure field provides a generic mechanism for the formation of the initial ballooning finger and its subsequent transition from exponential to explosive growth, without relying on secondary processes. The explosive ejection of the hot plasma from the core and stochastization of the magnetic field occ…
In high-current tokamak devices such as ITER, a runaway avalanche can cause a large amplification of a seed electron population. We show that disruption mitigation by impurity injection may significantly increase the runaway avalanche growth rate in such devices. This effect originates from the increased number of target electrons available for the avalanche process in weakly ionized plasmas, which is only partially compensated by the increased friction force on fast electrons. We derive an expression for the avalanche growth rate in partially ionized plasmas and investigate the effects of imp…
Resistive wall mode induced disruptions in an advanced tokamak
State Key Laboratory of Advanced Electromagnetic Technology（先进电磁技术国家重点实验室）；International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics（磁约束聚变与等离子体物理国际联合研究中心）
Disruption prediction and mitigation is of key importance in the development of sustainable tokamakreactors. Machine learning has become a key tool in this endeavour. In this paper multiple machinelearning models will be tested and compared. A particular focus has been placed on their portability.This describes how easily the models can be used with data from new devices. The methods used inthis paper are support vector machine, 2-tiered support vector machine, random forest, gradient boostedtrees and long-short term memory. The results show that the support vector machine performanceis margin…
The potential formation of multi-mega-ampere beams of relativistic "runaway" electrons (REs) during sudden terminations of tokamak plasmas poses a significant challenge to the tokamak's development as a fusion energy source. Here, we use state-of-the-art modeling of disruption magnetohydrodynamics coupled with a self-consistent evolution of RE generation and transport to show that a non-axisymmetric in-vessel coil will passively prevent RE beam formation during disruptions in the SPARC tokamak, a compact, high-field, high-current device capable of achieving a fusion gain Q > 2 in deuterium-tri…
A new physical mechanism of formation of runaway electron (RE) beams during plasma disruptions in tokamaks is proposed. The plasma disruption is caused by a strong stochastic magnetic field formed due to nonlinearly excited low-mode number magnetohydrodynamic (MHD) modes. It is conjectured that the runaway electron beam is formed in the central plasma region confined inside the intact magnetic surface located between $q=1$ and the closest low--order rational magnetic surfaces [$q=5/4$ or $q=4/3$, \dots]. It results in that runaway electron beam current has a helical nature with a predominant $…
An effective disruption mitigation system in a tokamak reactor should limit the exposure of the wall to localized heat losses and to the impact of high current runaway electron beams, and avoid excessive forces on the structure. We evaluate with respect to these aspects a two-stage deuterium-neon shattered pellet injection in an ITER-like plasma, using simulations with the DREAM framework [M. Hoppe et al (2021) Comp. Phys. Commun. 268, 108098]. To minimize the obtained runaway currents an optimal range of injected deuterium quantities is found. This range is sensitive to the opacity of the pla…
For ITER-relevant runaway electron studies, such as suppression, mitigation, termination and/or control of runaway beam, obtaining the runaway electrons after the disruption is important. In this paper we report on the first achieved discharges with post-disruptive runaway electron beam, entitled "runaway plateau", in the COMPASS tokamak. The runaway plateau is produced by massive gas injection of argon. Almost all of the disruptions with runaway electron plateaus occurred during the plasma current ramp-up phase. Comparison between the Ar injection discharges with and without plateau has been …
Collisional-radiative data for tokamak disruption mitigation modeling
Los Alamos National Laboratory（洛斯阿拉莫斯国家实验室）；Curtin University（科廷大学）；Korea Institute of Fusion Energy（韩国聚变能源研究所）；Queensland Quantum and Advanced Technologies Research Institute, Griffith University（格里菲大学昆士兰量子与先进技术研究学院）；School of Environment and Science, Griffith University（格里菲大学环境与科学学院）
We present experimental results of magnetic feedback control on the m=2, n=1 tearing mode in RFX-mod operated as a circular ohmically heated tokamak. The feedback suppression of the non-resonant m=2, n=1 Resistive Wall Mode (RWM) in q(a)<2 plasmas is a well-established result of RFX-mod. The control of the tearing counterpart, which develops in q(a)>2 equilibrium, is instead a more difficult issue. In fact, the disruption induced by a growing amplitude m=2, n=1 tearing mode can be prevented by feedback only when the resonant surface q=2 is close to the plasma edge, namely 2<q(a)<2.5, and the e…
The tearing mode instability and the associated magnetic islands can lead to a degradation of tokamak plasma performance and eventually to a disruption. A crucial issue for the extension of advanced tokamak scenarios to long pulse operation is to avoid these MHD instabilities. Prevention of disruptions and limitations of adverse effect on energy confinement depend on the avoidance of formation of magnetic islands, and it depends also on the successful control of the growth modes excited by intrinsic plasma effects such as the neo-classical drive of unfavourable bootstrap current gradient at so…
Results from the last FTU campaigns on the deuterium large (wrt FTU volume) pellet REs suppression capability, mainly due to the induced burst MHD activity expelling REs seed are presented for discharges with 0.5 MA and 5.3T. Clear indications of avalanche multiplication of REs following single pellet injection on 0.36 MA flat-top discharges is shown together with quantitative indications of dissipative effects in terms of critical electrical field increase due to fan-like instabilities. Analysis of large fan-like instabilities on post-disruption RE beams, that seem to be correlated with low e…
Force-free plasmas are a good approximation where the plasma pressure is tiny compared with the magnetic pressure, which is the case during the cold vertical displacement event (VDE) of a major disruption in a tokamak. On time scales long compared with the transit time of Alfven waves, the evolution of a force-free plasma is most efficiently described by the quasi-static magnetohydrodynamic (MHD) model, which ignores the plasma inertia. Here we consider a regularized quasi-static MHD model for force-free plasmas in tokamak disruptions and propose a mimetic finite difference (MFD) algorithm. Th…
Non-thermal electron cyclotron emission during runaway plateau in tokamak disruptions from a highly anisotropic dielectric tensor
Seoul National University（首尔大学）；Nuclear Research Institute for Future Technology and Policy, Seoul National University（首尔大学未来技术与政策核研究所）；POSTECH（浦项工科大学）
The surface current on the plasma-vacuum interface during a disruption event involving kink instability can play an important role in driving current into the vacuum vessel. However, there have been disagreements over the nature or even the sign of the surface current in recent theoretical calculations based on idealized step-function background plasma profiles. We revisit such calculations by replacing step-function profiles with more realistic profiles characterized by strong but finite gradient along the radial direction. It is shown that the resulting surface current is no longer a delta-f…
University of Science and Technology of China（中国科学技术大学）
A model for tokamak discharge through deep learning has been done on a superconducting long-pulse tokamak (EAST). This model can use the control signals (i.e. Neutral Beam Injection (NBI), Ion Cyclotron Resonance Heating (ICRH), etc) to model normal discharge without the need for doing real experiments. By using the data-driven methodology, we exploit the temporal sequence of control signals for a large set of EAST discharges to develop a deep learning model for modeling discharge diagnostic signals, such as electron density $n_{e}$, store energy $W_{mhd}$ and loop voltage $V_{loop}$. Comparin…
Based on the analysis of data from the numerous dedicated experiments on plasma disruptions in the TEXTOR tokamak the mechanisms of the formation of runaway electron beams and their losses are proposed. The plasma disruption is caused by strong stochastic magnetic field formed due to nonlinearly excited low-mode number magnetohydrodynamic (MHD) modes. It is hypothesized that the runaway electron beam is formed in the central plasma region confined by an intact magnetic surface due to the acceleration of electrons by the inductive toroidal electric field. In the case of plasmas with the safety …
Runaway electron populations seeded from the hot-tail generated by the rapid cooling in plasma-terminating disruptions are a serious concern for next-step tokamak devices such as ITER. Here, we present a comprehensive treatment of the thermal quench, including the superthermal electron dynamics, heat and particle transport, atomic physics, and radial losses due to magnetic perturbations: processes that are strongly linked and essential for the evaluation of the runaway seed in disruptions mitigated by material injection. We identify limits on the injected impurity density and magnetic perturba…
The key basis for tokamak plasma disruption modeling is to understand how currents flow to the plasma facing surfaces during plasma disruption events. In ITER tokamak, the occurrence of a limited number of major disruptions will definitively damage the chamber with no possibility to restore the device. In the current exchange plasma-wall-plasma, according to the Helmholtz decomposition theorem, our surface current density in the conducting shell - the unknown of our problem - being a vector field twice continuously differentiable in 3D, has been splited into two components: an irrotational (cu…
Implementation and verification of the avalanche source in a 3D full-f particle-in-cell model of relativistic electrons for studies of tokamak disruptions
As tokamaks are designed and built with increasing levels of stored energy in the plasma, disruptions become increasingly dangerous. It has been reported that 95% of the disruptions in the Joint European Torus (JET) tokamak with the ITER-like wall are preceded by the growth of large locked islands, and these large islands are mostly produced by off-normal events other than neoclassical tearing modes. This paper discusses the use of RF current drive to stabilize large islands, focusing on nonlinear effects that appear when relatively high powers are used to stabilize large islands. An RF curren…
During tokamak disruptions the profile of the net parallel current is observed to flatten on a time scale that is so fast that it must be due to a fast magnetic reconnection. After a fast magnetic reconnection has broken magnetic surfaces, a single magnetic field line covers an entire volume and not just a magnetic surface. The current profile, given by $K\equivμ_0j_{||}/B$, relaxes to a constant within that volume by Alfvén waves propagating along the chaotic magnetic field lines. The time scale for this relaxation determines the commonly observed disruption phenomena of a current spike and a…
Injection of relatively high density spheromaks with significant helicity-content into a tokamak has been proposed as a means for fueling and current drive. The CHI (Co-axial Helicity Injection) device was devised to inject current to the STOR-M tokamak. Various circuit modifications were made to the CHI controls, enabling testing of various injection configurations. The charge/discharge circuits for CT formation/acceleration and stuffing field were modified, and the power supplies and power converters were replaced. Various modifications were implemented to solve the original slow bank trigge…
For the simulation of disruptions in tokamak fusion plasmas, a fluid model describing the evolution of relativistic runaway electrons and their interaction with the background plasma is presented. The overall aim of the model is to self-consistently describe the nonlinear coupled evolution of runaway electrons (REs) and plasma instabilities during disruptions. In this model, the runaway electrons are considered as a separate fluid species in which the initial seed is generated through the Dreicer source, which eventually grows by the avalanche mechanism (further relevant source mechanisms can …
Parametric coupling of lower hybrid pump wave with low frequency collisionless/weakly collisional trapped electron drift wave, with frequency lower than the electron bounce frequency is studied. The coupling produces two lower hybrid sidebands. The sidebands beat with the pump to exert a low frequency ponderomotive force on electrons that causes a frequency shift in the drift wave, leading to the growth of the latter. The short wavelength modes are destabilized and they enhance the anomalous diffusion coefficient.
