Third Technical Meeting on Plasma Disruptions and their Mitigation

3 Sept 2024, 08:30 → 6 Sept 2024, 16:55 Europe/Vienna

ITER Headquarters

Michael Lehnen (ITER Organization)

KEY DATES AND DEADLINES

9 June 2024 Deadline for submission of abstracts through IAEA-INDICO for regular contributions

17 June 2024 Deadline for submission of Participation Form (Form A),  and Grant Application Form (Form C) (if applicable) through the official channels

21 June 2024 Notification of acceptance of abstracts and of assigned awards

---

The most important initiative on fusion R&D is currently ITER: the international tokamak reactor scale experiment being assembled in France. In tokamaks, instabilities can develop under certain operational conditions or as a consequence of a loss of plasma control. These instabilities eventually lead to the rapid loss of thermal and magnetic energy, a phenomenon known as plasma disruption.

Plasma disruptions cause thermal and mechanical loading to the tokamak components. Due to the high amounts of stored thermal and magnetic energies in ITER, the in-vessel components, such as the first wall panels and the divertor, will receive significant thermal loads. Furthermore, the in-vessel components, the vacuum vessel and the coils in the tokamak must also bear substantial mechanical loads. Disruption mitigation will be essential to reducing thermal and mechanical loading in order to guarantee the lifetime of these components.

Objectives

The event aims to serve as a forum to help coordinate experimental, theoretical and modelling work in the field of plasma disruptions with special emphasis on developing a solid basis for possible mitigation strategies in ITER and next generation fusion devices.

Target Audience

The event aims to bring together junior and senior scientific fusion project leaders, plasma physicists, including theoreticians and experimentalists, and experts (researchers and engineers) in the field of plasma disruptions.

Tuesday 3 September

Registration

Opening

Convener: Stefan Jachmich (ITER Organization)

1 Opening

Speaker: Stefan Jachmich (ITER Organization)

2 Welcome from DDG of the ITER Organization

Speaker: Yutaka Kamada (QST)

3 Welcome from IAEA

Speaker: Mr Matteo Barbarino (International Atomic Energy Agency)

4 In memoriam Michael Lehnen

Speaker: Richard Pitts (ITER Organization)

5 ITER project status

Speaker: Alberto Loarte (ITER Organization)

5-Loarte_ITER Project Status.pdf

6 ITER DMS design and plans

Speaker: Uron Kruezi

6-Kruezi_ITER_DMS_design_and_plans_v1_0.pdf

10:15 Coffee

Prediction & Avoidance

Convener: Gabriella Pautasso

7 Control of elongated plasmas in superconducting tokamaks in the absence of in-vessel coils

The roadmap for the commissioning and first operations of superconductive tokamaks envisages the possibility of running discharges with fairly elongated plasmas before the complete installation of the in-vessel components, including vertical stabilization coils, or any other specific sets of coils to be used for the magnetic control of fast transients.

In the absence of dedicated actuators, the magnetic control system shall perform the essential fast control actions by using the out-vessel superconductive coils, if needed. These are typically less efficient in reacting to fast transients, due to the shielding effect of the vessel and imply a coupling with other control tasks relying on the same actuators, such as plasma current, position, and shape control. Hence, effective actuator-sharing strategies must be put in place.

This talk will present a possible control architecture and a related control strategy that is able to cope with vertically unstable elongated plasmas subject to fast varying disturbances, in the absence of dedicated actuator. The architecture exploits a control-oriented plasma model-based for the offline identification of practically decoupled control directions to be assigned to each control task, i.e. plasma current, shape and vertical stabilization. Such model-based actuator-sharing approach allows to accomplish the main magnetic control objectives while minimizing the cross-couplings among the various tasks.

As case study, the setup of JT-60SA during the integrated commissioning and OP1 will be considered. Indeed, in this first phase, this machine has operated with vertically elongated plasmas in absence of dedicated in-vessel control coils.

Speaker: Gianmaria De Tommasi (Università degli Studi di Napoli Federico II)

DeTommasi_IAEA_2024_handout.pdf

DeTommasi_IAEA_2024.pdf

8 Tokamak plasmas in MST with density up to ten times the Greenwald limit

Steady, non-disruptive tokamak plasmas have been produced in the Madison Symmetric Torus (MST) with an electron density up to an order of magnitude above the Greenwald limit [Hurst et al., PRL, accepted for publication]. This result is made possible in part by a high-voltage, feedback-controlled power supply driving the toroidal plasma current. Also important may be the thick, stabilizing, aluminum vacum vessel that closely surrounds the plasma. The achievable density appears to be limited only by hardware and not instability. These MST plasmas were produced with Bt ~ 0.13 T, Ip ~ 50 kA, and q(a) ~ 2.2. Fueled with deuterium via standard puff valves, the circular plasmas are ohmically heated with R = 1.5 m and a = 0.50 m. The minor radius of the plasma-facing aluminum wall is 0.52 m. With a vessel-wall thickness of 5 cm, the wall penetration time is 800 ms, much longer than the typical 50 ms discharge duration. This passively allows stable operation at low q(a). Power and particle handling in MST is relatively primitive. The plasmas are limited, with graphite tiles covering about 10% of the plasma-facing wall, and pumping is achieved thru numerous small holes coupling the vacuum vessel to a pumping manifold.

Increasing the density shot by shot, the Greenwald fraction, $n_e/n_G$, was scanned from 0.5 to 10, with a maximum line-averaged density of about 6 x $10^{19}$ $m^{-3}$. Over this range, the loop voltage increases from 3 V to 65 V, provided automatically by the power supply. The only apparent limit on the achievable density is the flux swing in the iron-core transformer in the poloidal field system. At the highest density and loop voltage, the 2 V-s swing is exhausted before the programmed ramp-down in Ip, such that the discharge can no longer be sustained.

Although the Greenwald limit can easily be surpassed, these MST plasmas exhibit interesting features as the density crosses both $n_e/n_G$ ~ 1 and $n_e/n_G$ ~ 2. Transitioning from below to just above $n_e/n_G$ ~ 1, the sawtooth crashes commonly observed in these low-q plasmas transition to quasi-continuous, rotating tearing modes, and the scalings with density of the ohmic input power and impurity radiation increase sharply. As the density crosses $n_e/n_G$ ~ 2, these scalings become much weaker, and the toroidal current profile abruptly flattens, remaining finite up to the limiter, in contrast with the edge collapse (detachment) observed in other devices at lower $n_e/n_G$. In contrast to the current profile, a broad gradient is maintained in the electron density profile up to very high density.

Work is underway to understand why MST is able to operate with such a high Greenwald fraction. In addition to the obvious role played by the power supply in the global sustainment of these plasmas, it may also be key to sustaining the edge current. The conducting vessel may be important, e.g., in its recently predicted role in slowing resistive-wall-tearing-mode growth and the disruption thermal quench [Strauss et al., PPCF 2023].

Speaker: Brett E. Chapman (UW-Madison)

95-Chapman-2-MST_density_limit_2024.pdf

9 Correlation of the tokamak density limit with edge collisionality

A novel database study of the L-mode Density Limit (LDL) in metal- and carbon-wall devices (Alcator C-Mod, AUG, DIII-D, and TCV) identifies a two-variable, dimensionless stability boundary that predicts the LDL with significantly higher accuracy than the widely-utilized Greenwald limit. Historically, there has been broad interest in understanding the operational boundary imposed by the disruptive LDL because density is a critical lever for fusion performance. In this study, we create a multi-machine database of over 150 LDL events with 3000+ non-LDL discharges for evaluating the True and False Positive Rate. We find that data-driven models involving edge density and temperature measurements achieve significantly higher LDL prediction performance than the Greenwald fraction. Additionally, we utilize a Support Vector Machine to identify an analytic, dimensionless, stability boundary that retains the accuracy of the more sophisticated models, such as a Neural Network and Random Forest. The boundary is dominated by the effective collisionality in the plasma edge, $\nu_{*\rm, edge}$. This finding suggests that burning plasmas, with naturally low edge collisionality due to self-heating, may be able to achieve super-Greenwald densities. Additionally, in current and “next step” devices such as ITER, this collisionality boundary can also be deployed for active density limit avoidance.

Work supported by US DOE under DE-FC02-04ER54698, DE-SC0014264. This work has been carried out within the frame-work of the EUROfusion Consortium, via the Euratom Research and Training Programme (Grant Agreement No 101052200 — EUROfusion) and funded by the Swiss State Secretariat for Education, Research, and Innovation (SERI). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union, the European Commission, or SERI. Neither the European Union nor the European Commission nor SERI can be held responsible for them.

Speaker: Andrew Maris (Massachusetts Institute of Technology)

[2024 IAEA TM, Final] Correlation of the L-mode density limit with edge collisionality.pdf

Correlation of the L-mode density limit with edge collisionality

10 RESISTIVE WALL TEARING MODE MAJOR DISRUPTIONS WITH FEEDBACK

Resistive wall tearing modes (RWTM) are closely related to resistive wall modes (RWMs). RWTMs are tearing modes whose linear and nonlinear growth rate depend on the resistive wall penetration time.
The consequence for ITER, with wall penetration time of $250 ms,$ compared to $ \sim 5 ms$ in JET and DIII-D, is that the thermal quench
timescale could be much longer than previously conjectured.
Active feedback is another possible way to mitigate or prevent RWTM disruptions.
Simulations indicate that feedback can make the resistive wall behave effectively as an
ideal wall, preventing major disruptions.

Linear MHD simulations and theory [1,2] show that the RWTM growth time
is asymptotically proportional to
the wall penetration time, like a RWM. The $q = 2$ mode rational surface must be sufficiently close to the
wall for a RWTM disruption to occur. This agrees quantitatively with a DIII-D locked mode disruption database [3],
in which disruptions require the $q = 2$ rational surface radius to exceed $0.75$ of the plasma minor radius.
A nonlinear MHD simulation of a DIII-D locked mode equilibrium reconstruction shows a complete thermal
quench in a time which agrees with experiment.
The Madison Symmetric Tokamak (MST) has a longer resistive wall time $(800 ms)$ than ITER, and disruptions are not observed experimentally
when MST is operated as a standard tokamak. Simulations indicate that
the RWTM disruption time scale is longer than the experimental shot time.

A sequence of low edge current model equilibria [4,5] was produced from MST equilibrium reconstructions,
with higher edge q and with wall distance $1.2$ times the wall radius, similar to DIII-D. Nonlinear simulations showed that only minor disruptions occur with an ideal wall.
Major disruptions occur only for a resistive wall, and with
edge $q \le 3.4.$ This requires that the $q = 2$ minor radius is greater than $0.77$ of the plasma radius,
as in the DIII-D database [3].
Simulations with resistive wall and feedback [5] are similar to an ideal wall. An ideal wall or resistive wall with feedback,
restricts the modes to moderate amplitude, producing only a minor disruption. With the same initial equilibrium and no feedback
the mode amplitude is large, causing a major disruption.
The feedback simulations are consistent with the findings of an
experiment in RFX - mod [6], in which feedback was applied to stabilize
equilibria with edge $q > 2.$

[1] H. Strauss, B. C. Lyons, M. Knolker,
Phys. Plasmas 29 112508 (2022).

[2] H. R. Strauss, B. E. Chapman, N. C. Hurst,
Plasma Phys. Control. Fusion 65 084002 (2023)

[3] R. Sweeney, W. Choi, R. J. La Haye et al.
Nucl. Fusion 57 0160192 (2017).

[4] H. R. Strauss,
Phys. Plasmas 30, 112507 (2023)

[5] H. R.Strauss, B. E. Chapman, B. C. Lyons,
arXiv.2401.07133, submitted to Nucl. Fusion (2024)

[6] P. Zanca, R. Paccagnella, C. Finotti, et al.
Nucl. Fusion 55 043020 (2015).

Speaker: Henry Strauss (HRS Fusion)

48-Strauss-iaea24e.pdf

11 Non-equilibrium perturbations of the vertically unstable mode in tokamaks

The maximum allowable vertical displacement which can be recovered by the magnetic control system is a fundamental quantity for tokamak magnetic control (Gribov 2015 Nucl. Fusion 55 073021). This figure of merit is usually defined relying on a mass-less assumption, i.e. the reaction currents in wall structures are considered to vary in order to guarantee MHD equilibrium during the plasma perturbation. Recently we examined the consequences of considering also non-equilibrium perturbations of the vertical position via a simplified rigid filament model (Isernia 2023 Plasma Phys. Controlled Fusion 65 105007). In this case the reaction currents in wall structures are not determined anymore by the MHD equilibrium constraint during the perturbation. We illustrate the mapping of initial conditions between the full dynamic and the reduced mass-less model, emphasizing the role of the mass-less assumption on the definition of maximum allowable displacement. Moreover we show that the ratio between the perturbation time of the vertical position and the electromagnetic time of the unstable mode in the quasi-static limit $\Delta t / \tau_u$ governs the intensity of reaction currents at the end of the perturbation, hence the overall perturbation of the unstable mode.

Speaker: Nicola Isernia (Università degli Studi di Napoli Federico II)

Non-equilibrium perturbations of the vertically unstable mode in Tokamaks v3.pptx

12:55 Lunch

Posters

All posters will be displayed in both sessions on Tuesday and Wednesday.

12 Advances in High Accuracy Physics-Based Tokamak Disruption Event Characterization and Forecasting Including Real-time Deployment

Disruption prediction and avoidance is critical to maintain steady plasma operation and to avoid damage to device components in ITER and reactor-scale tokamaks. 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 sufficiently early warning to allow disruption avoidance [1]. The DECAF approach has been demonstrated to yield high accuracy disruption prediction and forecasting in both offline analysis and real-time application. A workflow has been established to reach next-steps: high accuracy under all plasma scenarios, and concurrent expansion of real-time deployment. The first real-time application of DECAF was made on the KSTAR superconducting tokamak including initial connection to control actuators. Dedicated plasma experiments focusing on disruptions caused by locking MHD instabilities were forecast with 100% accuracy. An MHD mode locking forecaster, using a torque balance model of the rotating mode, was implemented and utilized in real-time to produce these results. This forecaster was also used several times in these experiments to cue controlled plasma shutdown, trigger disruption mitigation using the KSTAR massive gas injection (MGI) system, and actuate electron cyclotron current drive and n = 1 rotating 3D fields for future disruption avoidance. New hardware and software for real-time (r/t) diagnostic acquisition and DECAF analysis continue to be installed and tested on KSTAR including high bandwidth (500 kHz) electron temperature, Te, profiles from electron cyclotron emission (ECE), and toroidal velocity profiles from a new r/t CES diagnostic with data taken in the 2023-24 run campaign. Several new r/t DECAF modules are being deployed on KSTAR, and first steps are being taken toward DECAF deployment on ITER. Research advances supporting these efforts include high bandwidth Te profile measurements being used to reconstruct ‘crash profiles’ to computationally identify sawteeth, ELMs, and other MHD as NTM triggers and as direct disruption precursors. Different DECAF disruption event chains are observed based on the plasma state at the time of the trigger event. Offline analysis has access to the full databases of many international tokamaks to best understand, validate, and extrapolate models. Fully automated analysis shows for datasets spanning entire run campaigns very high true positive rates, in some cases over 99%. A multi-device study conducted for plasma vertical instability used the DECAF workflow to produce real-time capable modelling with prediction accuracy of 99% -100%. An initial halo current model has been implemented to supplement criteria to determine the need of disruption mitigation.
Supported by US DOE Grants DE-SC0020415, DE-SC0021311, and DE-SC0018623.
U.S. and international patents pending.

References
[1] S.A. Sabbagh, et al., Phys. Plasmas 30, 032506 (2023); https://doi.org/10.1063/5.0133825

Speaker: Steven Sabbagh (Columbia University)

13 Assessment of RE transport induced by the DIII-D REMC with ASCOT5

The inboard-wall runaway electron mitigation coil (REMC) designed for DIII-D was optimized based on vacuum island overlap width (VIOW) [1] and subsequently modeled with the linear MHD code MARS-F [1] and the non-linear MHD code NIMROD [2]. Both the linear and non-linear MHD modeling tracked the confinement of an initial RE test population, with the total loss fraction for the nonlinear model exceeding that of the linear model. Because in reality REs are continuously produced during the CQ by the avalanche mechanism, the single initial test population can not capture the true dynamics of RE generation and loss as a function of space, energy and time. For the SPARC tokamak, the ASCOT5 code [3] was used to evaluate RE transport coefficients based on NIMROD 3D fields [2], and these coefficients were incorporated into DREAM [4] simulations of RE generation and evolution [2]. Here, ASCOT5 modeling is applied for the first time to DIII-D REMC simulations with NIMROD to determine transport coefficients as a function of space and energy. In the simplest model, competition between these transport coefficients and an avalanching population in free-fall can be calculated with an assumed seed to give an estimate of the growth of the RE population. Eventually, the transport results can be incorporated in a more complete RE evolution model like DREAM to predict the expected performance of the DIII-D coil.

[1] Weisberg, D. B., Paz-Soldan, C., Liu, Y. Q., Welander, A., & Dunn, C. (2021). Passive deconfinement of runaway electrons using an in-vessel helical coil. Nuclear Fusion, 61(10), 106033. https://doi.org/10.1088/1741-4326/AC2279
[2] Tinguely, R. A., Izzo, V. A., Garnier, D. T., Sundström, A., Särkimäki, K., Embréus, O., Fülöp, T., Granetz, R. S., Hoppe, M., Pusztai, I., & Sweeney, R. (2021). Modeling the complete prevention of disruption-generated runaway electron beam formation with a passive 3D coil in SPARC. Nuclear Fusion, 61(12), 124003. https://doi.org/10.1088/1741-4326/AC31D7
[3] Varje, J., Särkimäki, K., Kontula, J., Ollus, P., Kurki-Suonio, T., Snicker, A., Hirvijoki, E., & Akäslompolo, S. ¨. (2019). High-performance orbit-following code ASCOT5 for Monte Carlo simulations in fusion plasmas.
[4] Hoppe, M., Embreus, O., & Fülöp, T. (2021). DREAM: A fluid-kinetic framework for tokamak disruption runaway electron simulations ✩. Computer Physics Communications, 268, 108098. https://doi.org/10.1016/j.cpc.2021.108098

Speaker: Valerie Izzo (Fiat Lux)

14 Assessment of vertical controllability using DECAF, and predictive capability of a vertical stability metric for tokamak plasmas

Vertical displacement events (VDEs) in tokamaks involve large displacements of the plasma magnetic axis from the vessel midplane, often leading to disruptions. These events are of particular concern for their potential to cause damage to plasma-facing components, as well as large forces on the vessel due to halo currents generated during the disruption that run through the plasma and vessel [1]. Detection and control of these events and mitigation or avoidance of a potential disruption is crucial. We present the results of an operational space analysis for defining regimes of vertical position controllability, compared across the MAST-Upgrade, KSTAR, and NSTX tokamaks using the DECAF approach [2, 3]. Identification of the vertically-controllable regime based on device data is demonstrated to improve the accuracy of automated vertical displacement event detection to more than 99% for each of the devices studied. These findings can inform the setting of warning levels in real-time plasma control and disruption mitigation systems. Further, we present the results of a vertical instability warning metric that employs a linear model approximation of the current density profile of the plasma. This approach increases VDE early warning times on average by more than a factor of three and achieves 63% accuracy evaluated on full run campaigns of MAST-U data, and 87% accuracy evaluated on all shots from these campaigns without suspected internal reconnection events before the VDE. These results make this a promising method for forecasting vertical displacement events and triggering disruption avoidance procedures in a plasma control system, and motivate further refinement of the current model for robustness to internal reconnection events. This research was supported by the U.S. Department of Energy under grants DE-SC0020415, DE-SC0021311, and DE-SC0018623.
U.S. and international patents pending.
[1] V. Zamkovska, et al., “Implementation of a cross-device model for halo current in the DECAF code as a criterion for the determination of disruption mitigation action”, this conference
[2] S.A. Sabbagh, et al., Phys. Plasmas 30, 032506 (2023). https://doi.org/10.1063/5.0133825
[3] S.A. Sabbagh, et al., “Advances in High-Accuracy Physics-Based Tokamak Disruption Event Characterization and Forecasting Including Real-time Deployment”, this conference

Speaker: Matt Tobin (Columbia University)

15 Cross-tokamak disruption prediction via domain adaptation and generalization

Unmitigated disruptions at high performance discharge are unacceptable for future reactors, while they are not able to provide enough data to train a predictor with acceptable performance along. However, current can bear disruptions, and have accumulated a large amount of data with various disruption patterns. In the meantime, domain adaptation (DA) and domain generalization (DG) are now very promising ways to make full use of knowledge from source domain and adapt/generalize them to target domains with very little data, even 0 shot.
For disruption prediction, a basic idea is to apply DA/DG algorithms to every stage throughout training. While still maintaining the advantages of deep models, multiple DA/DG algorithms are applied and more physical features/constraints are introduced to the model. Several numerical experiments are carried out where J-TEXT and HL-2A serve as the source tokamak and EAST serve as the target. In DA case, which 20 discharges from EAST are added into the training set, AUC reaches .8724 for the best model. In DG case, which uses none data or information from EAST, AUC reaches .8343 for the best. Both DA and DG cases perform acceptable on the target domain.

Poster_Xue.pdf

16 Electromagnetic and heat loads on tokamak walls as a consequence of disruption

The structural integrity of the Vacuum Vessel (VV) of Pakistan's Metallic Tokamak-I (MT-I), a small spherical tokamak, was tested by simulating a 10 ms input current event on a 180° sector model. During this event, the energy from the plasma is entirely transferred to the VV's first wall. This study is based on the law of conservation of energy, demonstrating that the extent of damage is inversely proportional to the surface area over which the energy is deposited. Simulations are performed using ANSYS to assess the structural response of the MT-1 tokamak VV under different plasma disruption scenarios. Radiation-induced energy deposition lead to uniform heat flux with minimal temperature rise (<3°C) and negligible stress increases. Vertical disruptions induce powerful heat loads, but stresses remain within the elastic limit of stainless steel, though repeated events could lead to fatigue. Horizontal disruptions cause significant localized heating and deformation, posing a risk of irreversible damage. Overall, the VV demonstrates resilience to single-event disruptions, but cumulative effects necessitate further fatigue analysis and enhanced protective measures.

Speaker: Ms Saira Gulfam (Pakistan Tokamak Plasma Research Institute)

17 Heat loading of Pulsed Hydrogen Plasma Stream on Tungsten Substrate

A pulsed hydrogen plasma stream is produced from a pulsed plasma accelerator (PPA) powered by 200 KJ Pulsed Power System (PPS). The PPS, which consists of two modules capacitor banks, is charged up to 15 kV to generates a peak discharge current of 100 kA for a half time period of 500 µs. The high voltage from the capacitor banks thus applied in between two coaxially positioned electrodes to break down the gaseous medium and consequently plasma sheets are formed in between the electrodes. The plasma sheets are driven by the JxB force towards the open end of the electrodes to form a high-density( 1020/m3) plasma stream. A gas injection valve is used to supply the requisite gas during the application of high voltage discharge pulse in between the two electrodes so that the plasma stream can be sustained during the pulse time. To study the heat loading effect on tungsten material, 10 mm diameter tungsten metal samples were placed at 10 cm distance from the electrodes’ end. The measured heat energy density of the Hydrogen plasma stream at this position is 0.205 MJ/m2 while it increases up to 0.224 MJ/m2 under an influence of an external magnetic field with a strength around 0.1 Tesla strength. The effect of heat energy dumped by the pulsed hydrogen plasma stream on the tungsten material creates formation of blister on the surface of Tungsten material for single exposure and it is observed in the micrograph of scanning electron microscope (SEM). However minor and major cracks, displacement of cracked surface, dust formation, re-deposition are more prominent in SEM micrograph for surface exposed for 15 times to the same plasma stream in the same position with and without external magnetic field. Although, the EDX data from the exposed samples as well as emission spectra from the plasma stream shows that carbon impurities, found in the plasma beam, decreases under the influence of external magnetic field. The interaction due to the heat energy density of plasma stream on Tungsten material in this work resembles either a mitigated or lower energy type-I Edge Localized Mode (ELM) [1] and the reported results are highly relevant for fusion reactor.

Reference:
[1] G. Sinclair, et. al., Scientific Reports7, 1, (2017)

Speaker: Dr TRIDIP KUMAR BORTHAKUR (CENTRE OF PLASMA PHYSISCS INSTITUTE FOR PLASMA RESEARCH)

18 Implementation of a cross-device model for halo current in the DECAF code as a criterion for the determination of disruption mitigation action

Disruption of a tokamak plasma is a multi-step process in which the loss of the plasma vertical position control is often among the last events that precede the final plasma deconfinement. A flow of current between the plasma and the vessel components -the halo current [1]- is generated through the contact of vertically displaced plasma with the vacuum vessel, resulting in electromagnetic forces applied on the device as the current crosses the confining toroidal magnetic field. The potential damage of the resulting force scales with the pre-disruptive plasma current level, to an extreme of next-step devices carrying high plasma current in which the peak force amplitude could reduce structural lifetime due to increased metal fatigue. Owing to their potential negative consequences, halo currents were studied across many devices and strategies for their mitigation were developed [2-4]. The present work reports on a cross-device implementation of a halo current model into the DECAF* code [5]. This model is part of DECAF’s capability to evaluate criteria that discriminate between disruptions that require mitigation, and ‘benign’ disruptions, i.e. plasma collapses whose consequences can be routinely handled by the device. This level of disruption classification is of the utmost interest for devices in which the mitigation action is highly perturbative to the subsequent device operation (such as ITER). It is thus meant to be deployed under the elevated risk of device damage from unmitigated disruptions, while avoiding ramifications such as delayed plasma recovery due to unnecessary disruption mitigation.

Speaker: Veronika Zamkovska (Columbia University)

Zamkovska_IAEA_TM_Disruptions_2024_v1.2.pdf

19 Multimodal super-resolution diagnostics for analyzing fast transient events in fusion plasma

We present a groundbreaking multimodal neural network model designed for diagnostics resolution enhancement, which innovatively leverages inter-diagnostic correlations within a system. Traditional approaches have primarily focused on unimodal enhancement strategies, such as pixel-based image enhancement or heuristic signal interpolation. In contrast, our model employs a novel methodology by harnessing the diagnostic relationships within the physics of fusion plasma. Initially, we utilize the correlation among diagnostics within the tokamak to substantially enhance the temporal resolution of the Thomson Scattering (TS) diagnostic. This enhancement goes beyond simple interpolation, offering a "super-resolution" TS (SRTS) that preserves the underlying physics inherent in inter-diagnostic correlation. Increasing the resolution of TS from conventional 230Hz to 500kHz could capture the structural evolution of plasma instabilities and the response to external field perturbations, which is challenging to do with conventional TS.
This physics-preserving super-resolution technique may enable the discovery of new physics that were previously undetectable due to resolution limitations and/or allow for the experimental verification of phenomena that have previously only been predicted through computationally intensive simulations. Furthermore, the proposed approach holds significant potential for disruption prediction and mitigation by enhancing the accuracy of early detection of disruptive events, enabling timely and precise control actions to prevent or mitigate these events.
Figure 1 shows the general diagram of developing the neural network model (Diag2Diag) to generate SRTS data. It also presents an example of generating synthetic SRTS at 500kHz for an ELMy H-mode DIII-D discharge 153764. The synthetic and measured TS match well whenever TS measurements are available. Additionally, synthetic TS captures nearly all the ELM events (indicated by $D_\alpha$ spectroscopy) that are missed by the measured TS even though it is configured in bunch-mode with higher temporal resolution.
This work is supported by US DOE Grant Nos. DE-FC02-04ER54698, DE-SC0022270, DE-SC0022272 and DE-SC0024527, as well as the National Research Foundation of Korea (NRF) Award RS-2024-00346024 funded by the Korea government (MSIT).
1

Speaker: Azarakhsh Jalalvand (Princeton University)

20 Observation of runaway plateaus in EAST disruptions triggered by MGI

Runaway plateaus have been formed in EAST disruptions via massive argon injection into low-density ohmic discharges, which are carried out with circular plasma and limiter configuration in full-metal wall condition. REs are detailly described in the following three phases:
Formation: Formation of runaway plateau is found strongly depends on pre-disruption density and amounts of injected gases. A typical scenario to a reproducible RE plateau is with density of 0.3-0.5(10^19 m-3), injected gases of ~400-500 Pa.L and finally post-disruption density lower than 3. Predisruption REs are also observed and confirmed to
play a key role in plateau formation due to increasing RE plateau with lower pre-disruption ohmic component.
Runaway plateau: At most 90% plasma current can be converted to RE plateau of 270kA and last for almost 400 ms in EAST tokamak. On the phase of RE plateau, they are found to shrink and move outwards and strikes the LFS limiter as observed in SXR array and infrared camera directly.
Termination: A secondary D2 injection during RE plateau phase leads to an earlier termination and increasing density of core, which eventually reduced energy of REs via higher collisionality. At the same time, there appears multiple magnetic fluctuations with higher amplitude accompanied with RE loss events, which is confirmed to convert more magnetic energy to kinetic energy of REs.

Speaker: tian tang

IAEA_tt.pptx

21 Progress of mitigating plasma disruption by the shattered pellet injection in EAST

In the past three years, a shattered pellet injection (SPI) system designed for disruption mitigation on the EAST tokamak was successfully developed and integrated into EAST tokamak in 2022. The SPI system is capable of producing Ne pellets with diameters of ~ 5 mm and lengths ranging from 7 to 15 mm. The material gas consumption is approximately 20, 25, and 30 Pa·m3, respectively, and the estimated pellet flight speed ranges from 100 to 400 m/s. A bend tube with an angle of ~ 20° is installed at the end of the pellet flight pipe to ensure pellet fragmentation [1,2]. During bench testing, the condensation process of pellets was simulated using FLUENT through numerical simulation methods. It was found that the best cold head temperature of Ne pellet condensation was 8-10 K with a condensation zone pressure of 60 mbar and heat sinks temperature of 100 K. Experimental results demonstrate a direct relationship between the durations of thermal quench (TQ) and current quench (CQ) during disruption and the parameters of the pellet. Particularly, with the increase in pellet velocity, the durations of tCQ and tTQ will decrease. The timescales for tCQ were approximately 4-6 ms and for tTQ were approximately 0.05-0.2 ms. In comparison to unmitigated disruptions, the total radiation power significantly increased with the implementation of SPI [3]. Subsequently, we replaced the bend tube with a straight tube, and compared the effects of pellets with different degrees of fragmentation on disruption mitigation. The results indicate that relatively fragmented pellets can achieve shorter cooling times, longer tCQ durations, higher particle assimilation rates, and a more uniform poloidal radiation distribution. Subsequently, comparing the injection effects of pellets in L-Mode and H-Mode, it was observed that the Cooling time in H-Mode was shorter than in L-Mode, and most of the plasma's thermal energy would dissipate before the CQ in SPI. These findings from the EAST experiments serve as a valuable reference for establishing SPI technology as the fundamental approach for disruption mitigation in ITER.

[1] J. Yuan, et al. Fusion Engineering and Design, 191 (2023) 113567.
[2] S.B. Zhao, J.S. Yuan, et al. Journal of Fusion Energy, (2023)42:49.
[3] J. Yuan, et al. Nuclear Fusion, 63 (2023) 106008.

Speaker: jingsheng Yuan (Institute of Plasma Physics Chinese Academy of Sciences)

jingsheng Yuam IAEA poster 20240903.pdf

22 Runaway Electrons in JET - Status of RE Data after End of JET Operations in 2023

Plasma major disruptions pose severe threats to the device integrity in future operations of International Thermonuclear Experimental Reactor (ITER). They can cause dangerous excessive electromagnetic forces, heat loads and generation of the intense beams of relativistic runaway electrons (RE). Localized interaction of intense RE beams with surrounding plasma facing components (PFC) inevitably will result in inacceptable PFC damage. To avoid/suppress RE generation and mitigation of other disruption detrimental consequences the Disruption Mitigation System (DMS) is under design for ITER. It will be based on injection of impurities in the form of solid shattered pellets (SPI). Despite significant progress in studies relevant to the ITER DMS design, the set of physical and technology problems remains un-resolved. Development of DMS requires further advances in understanding of the physics of RE and their interaction with plasma, solid pellets and neutral gases (fuel and injected impurities). For this purpose the comprehensive database on RE generation at disruptions in JET has been elaborated recently including the data obtained just before the end of JET operations in 2023.
This report presents the first summary and current status of the JET RE database analysis. The first events of the RE generation have been detected in spontaneous disruptions in JET from the early experiments. A series of dedicated experiments on RE generation have been carried out during whole period of JET operations with divertor (at carbon fiber composite (CFC) tiles till to 2009) and with ITER-like Wall (from 2011)). From the beginning of JET operations there were several attempts to review the data on RE generation events. However, these attempts are still waiting for a compiling into joint database. RE generation in spontaneous disruptions, during those triggered by constant gas puff, at Massive Gas Injection (MGI) and SPI, including latest experiments on benign RE termination, provided the data for JET database. An analysis of this database should stimulate further advances in understanding of the physics of RE and to serve as the basis for possible numerical simulations. The mapping of RE parameters enabled establishing links of pre-disruption and post-disruption parameters. Despite the plasma parameters are poorly known during and after disruptions, this approach enables establishing links between plasma parameters before thermal quench and during current decay (Te and ne, li, and current quench (CQ) rates, data from EFIT, etc.). Obtained data was used to study the trends in RE parameters for a wide range of disrupted JET currents (up to 6.25 MA). Note, that at certain combinations of plasma pre-disruption parameters data analysis yielded the trends, which are in contrary to that obtained early on limited number of points. CQ studies revealed different, accelerating and constraining effects of initial plasma configurations (circular (limiter) or X-point) on RE generation and value of current conversion ratio (Ipl/IRE). One of the important results from the data-base analysis is observation of lower threshold in generation and a decreasing trend in conversion ratio IRE/Ipl depending on CQ rates.

Speaker: Dr Vladislav Plyusnin (Instituto de Plasmas e Fusão Nuclear, Associação EURATOM-IST, Instituto Superior Tecnico)

POSTER_TCM_IAEA_2024version OK.pdf

23 Study of the halo current region resistivity on the DIII-D tokamak

In this work we report measurements of the temperature and density of the halo current region on DIII-D during disruptions using the recently upgraded Thomson scattering diagnostic allowing low-temperature measurements down to a few eV with a sub-ms repetition rate. This is done by employing deliberate downward vertical displacement events (VDEs) and relying on the expansion of the halo current region upward, intercepting the midplane Thomson channels. Both ‘hot’ and ‘cold’ VDEs were studied using ohmic and H-mode target plasmas respectively. ‘Hot’ VDE, having the vertical instability growth rate greater than the current decay rate, characterizes by the electron temperature of the halo expansion region in the range of 1−10 eV increasing towards the core. While ‘cold’ VDE has the opposite time scale dynamics and much lower electron temperature of 1−2 eV with a flat profile. VDE of both types results in the electron density of the halo region comparable with the core plasma density and quickly decreasing to the edge. Poloidal halo currents, measured using the tile current shunts, exhibit values greater by about 20% for the ‘hot’ VDE cases. Modeled poloidal and toroidal halo currents, as well as modeled halo current width are also presented. Implication of the halo region resistivity on the halo current profile and JxB forces is discussed.

Work supported by the US DOE under DE-FC02-04ER54698

Speaker: Dr Andrey Lvovskiy (General Atomics)

Lvovskiy_IAEA TM DisMit_2024.pdf

24 Summary of dispersive shell pellet injection experiments on DIII-D

Dispersive shell pellet (DSP) injection is currently being developed as an alternative disruption mitigation technique to massive gas injection and shattered pellet injection. The main advantage of DSP injection is the core deposition of the payload which is expected to result in higher assimilation fractions and an inside-out thermal quench (TQ). DSPs have been successfully launched into DIII-D Super H-Mode plasmas, resulting in the rapid shutdown of the plasma current. Core payload deposition was most readily achieved using high density carbon (HDC) shells with 3.6 mm outer diameter and 40 μm wall thickness. Both low-Z (boron dust) and high-Z (tungsten grains) payloads have been investigated. Initial experiments have shown that the pellet penetration increased with pellet velocity, however the assimilation fraction remained approximately constant (0.5 – 0.9). The assimilation fraction decreased when using larger 5 mm diameter pellets with larger payloads, but this may be due to added perturbation from the increased shell mass and surface area. Both small and large HDC shells have been found to be too perturbative to produce a true inside-out TQ. To remedy this, materials with lower atomic numbers than carbon have been proposed as an alternative coating/shell material. The development of multi-micrometer thick lithium coatings for future DSP experiments on DIII-D is underway.

Work supported by US DOE under DE-FC02-04ER54698, DE-AC05-00OR22725, and DE-FG02-07ER54917.

Speaker: Dr Grant Bodner (General Atomics)

Bodner_IAEA_TM_Disruption_2024_A0.pdf

25 Thermal Quench and its diagnostics at JET

Disruptions are an inherent property of tokamak plasmas, which cannot be completely eliminated. The consequences of disruptions are especially dangerous for large machines like ITER and even more so for DEMO. Thermal Quench (TQ) is the initial phase of disruption followed by plasma Current Quench (CQ). Essential diagnostics for the TQ are magnetics (dB/dt), Electron Cyclotron Emission (ECE) and soft X-rays. Of course, the signals must be recorded with high time resolution of more than 100 kHz. Some magnetics signal (e.g. Ip, Locked and rotating MHD amplitude, etc.) can be recorded at moderate time resolution in the range of (5-10) kHz. TQ is characterized by a sharp drop in plasma thermal energy (and hence electron temperature) and MHD bursts. The MHD disruption phase continues from the beginning of the TQ to the end of the distinctive plasma current spike. The TQ duration can vary from several hundred microseconds to milliseconds, and the MHD phase can vary within a few milliseconds. The dynamics of the TQ and the entire MHD phase can shed light on the nature of the MHD instabilities underlying these events. The widespread explanation of a TQ is associated with a Neoclassical Tearing Mode (NTM) or Resistive Wall Tearing Mode (RWTM), which drives islands on separate resonant surfaces nq = m and ultimately leads to global internal reconnections. However, the remarkably fast TQ of order of ~ 100 µs suggests that the Wall Touching Kink Mode (WTKM) can be another (or additional) underlying the initial phase of disruption. The WTKM mode is a free boundary kink mode which is the strongest and fastest kind of MHD instability in tokamaks.

Speaker: Dr Sergei Gerasimov (CCFE)

3TMPDM-Poster_v2.pdf

26 Transfer learning with a parsimonious disruption predictor: from JET C-wall to the metallic wall

The Learning Using Privileged Information (LUPI) paradigm allows training classifiers with data not available at execution time. Recently, an application of the LUPI paradigm to the prediction of disruptions with extreme data scarcity was demonstrated [J. Vega et al. Nuclear Fusion 64 (2024) 046010 (12 pp)]. The objective of the previous reference was to test the development of an adaptive disruption predictor based on LUPI and its potential application to JT-60SA. To this end, a line integral density (LID) signal used for both training and real-time prediction is complemented with the mode lock signal as privileged information (i.e. the mode lock signal is only used for the training). The predictor was developed before the start of the JT-60SA operation and, therefore, it was trained with a JET specific database.
The present contribution investigates the potential of LUPI for transfer learning. The goal is to develop a parsimonious disruption predictor with JET C-wall discharges and to perform transfer learning to ITER-like Wall (ILW) shots. The term ‘parsimonious’ indicates two important properties of the predictor: that it is a purely data-driven approach without any a priori assumptions or conjectures, and that it performs adaptive predictions from scratch. Again, only a LID signal is used for real-time predictions and the predictor is trained with the LID signal and the mode lock as privileged information. The database consists of 439 C-wall discharges (409 non-disruptive shots and 30 unintentional disruptions) together with 471 ILW discharges (392 non-disruptive and 79 unintentional disruptive shots). An adaptive predictor is trained with the C-wall data. This means that a first predictor is generated with one disruptive and one non-disruptive discharge and re-trainings are carried out after missed or tardy alarms. After processing the 439 C-wall discharges, the last predictor is applied to ILW shots. Again, the predictor is re-trained adding ILW data after missed or tardy alarms. It is important to note that in both cases, C-wall and ILW, the predictor that results from a re-training only replaces the previous one when the outcomes (in terms of success, tardy and false alarm rates) with all the preceding discharges are better.
For the whole C-wall database, only two predictors have been generated (the initial model plus one re-training). The overall performances are: two tardy predictions, one missed alarm and three false alarms. This means a success rate of 96.7% (90% with positive warning time and 6.7% tardy detections) and 0.7% false alarms. After the transfer learning to the ILW, two re-trainings have been performed and the performances are: 5 missed alarms, 8 tardy detections and 32 false alarms. Therefore, the overall statistics of the transfer learning, taking into account both C-wall and ILW discharges, are: success rate 94.5% (85.3% with positive warning time and 9.2% of tardy alarms) and 4% of false alarms. The average warning time is 269 ms and, therefore, this approach to transfer learning based on a) prediction from scratch, b) two unique signals and c) the LUPI paradigm can be considered adequate for mitigation purposes.

Speaker: Prof. Jesús Vega (CIEMAT)

JVega.pdf

27 Ultrafast actuator development for shattered pellet injectors - fast, high pressure valve and ITER DMS Shutter

The ITER Disruption Mitigation System (DMS) is based on Shattered Pellet Injectors (SPI), which accelerates a large protium, neon or mixture pellet with high pressure gas and shatters it prior to the entrance into the plasma, creating a plume of smaller pellet fragments. The ITER DMS Support Laboratory is part of the ITER DMS Task Force programme to establish the physics and technology basis for the ITER DMS. The laboratory is located at the HUN-REN Centre for Energy Research (HUN-REN CER), Budapest Hungary.
A key component of an SPI system is a fast, high-pressure gas valve which can release propellant gas with about 10 m3/s volumetric flow rate, within a few milliseconds after the arrival of the control signal. This necessitates the opening of a gas reservoir of approximately 1 litre volume, filled with gas up to 15 MPa pressure, which requires an ultrafast actuator capable of releasing ~20 kN force within a millisecond after the trigger. A fast gas valve, using eddy current principle has been developed fulfilling these requirements.
The propellant gas released by the fast valve can overtake the pellet, triggering an early disruption and deteriorating the disruption mitigation efficiency. A suppressor chamber is foreseen to be utilised in which most of the propellant gas can be retarded for some time, however, a certain fraction of the gas will still arrive at the shatter device exit ahead of the fragments.
An optional fast shutter is considered after the suppressor to block the path of the propellent gas when the pellet has passed. The piston of the valve must be accelerated to tens of meters per second to close the 40 mm orifice in a few milliseconds and decelerated after the closure to avoid high velocity impact at the end position. The device must survive several thousands of cycles because maintenance access will be limited. The setup needs to be compact due to space restrictions, to operate in 400 mT external field, to withstand a high neutron dose rate and be tritium compatible as it is a part of the ITER main vacuum system.
This contribution describes the development and the laboratory testing of the ITER DMS Support Laboratory fast gas valve and the ITER DMS Fast Shutter from the physics design, through the electromechanical prototyping and the model validation, including the final mechanical design and the laboratory test results.

Speaker: Daniel Imre Refy (HUN-REN Centre for Energy Research)

28 Validation of Extended-Magnetohydrodynamic Modeling of KSTAR Disruption Mitigation with Collisional-Radiative Impurities

Future tokamaks will require disruption mitigation systems (DMS) to prevent machine damage during the uncontrolled loss of plasma confinement. Massive impurity injection, particularly shattered pellet injection (SPI), is the leading candidate for a DMS. Validated, predictive models are needed to project these systems to future devices, which require models for the macroscopic plasma evolution as well as the injected impurity dynamics. To simulate these conditions, the M3D-C1 extended-magnetohydrodynamics code has traditionally been coupled to a coronal non-equilibrium model for impurity ionization and radiation, which is formally valid only in the low-collisionality limit. We present an overview of the M3D-C1 code as well as an upgrade to the impurity model by coupling M3D-C1 to the full ADAS model. This provides a density-dependent, collisional-radiative model for ionization and radiation and greatly expands the number of impurities that can be considered by M3D-C1. We show that for neon, the collisional-radiative model predicts significantly less line radiation than the coronal model, resulting in longer thermal quench times and increased density & material assimilation in SPI simulations. We also present M3D-C1 modeling of several realistic SPI scenarios on the KSTAR tokamak. The results are compared to data for single and dual-symmetric injection of neon-doped pellets as well as dual, time-staggered injection with a pure-deuterium pellet followed by a neon-doped pellet. We find that the simulations generally underpredict the density after neon-doped injection but overpredict the density after pure-deuterium injection. While the radiation in the staggered injection modeling agrees very well with experiment, simulations of neon-doped injection without the initial pure-deuterium injection generally see a radiation spike that is too rapid and too toroidally distributed. We explore the effects of the localization of the plasma source, equilibrium plasma rotation, and coronal versus collisional-radiative impurities on these results. Future work will seek to determine the sources of these discrepancies.

This work is supported by the U.S. Department of Energy, Office of Science, Office of Fusion Energy Sciences under Awards DE-FC02-04ER54698, DE-FG02-95ER54309, and DE-SC0020299. This work has contributed to the ITER DMS Task Force.

Speaker: Brendan Lyons (General Atomics)

29 Validation of Miyamoto’s formula (for the total force on the vessel following disruption) with ASDEX Upgrade measurements

Abstract
The magnitude of the electromagnetic forces expected during and after a vertical displacement event (VDE) of a tokamak plasma is a necessary design parameter for the structural design of the many components of the tokamak device. During and after a VDE, the plasma current moves, changes in time and induces currents in the conductors around the plasma. The total vertical force
on the device is then the integral over all the conductors of the vertical component of the acting
force. The total vertical force (and its spatial and temporal evolution) can be found by simulating
the VDE with a 2D or 3D MHD code.
S. Miyamoto [1] found an approximate way of deriving the total maximum vertical force due to a
VDE in a tokamak device without modeling the whole phenomenon. This formula is of interest
because it allows to dimension the vessel and its supports during the conceptual design of a new
device without having to use numerical codes for detailed simulations.
In this contribution, the formula is validated against ASDEX Upgrade measurements. Since the
validation is not successful, possible errors in the derivation of Miyamoto’s formula are discussed
and a correction is suggested.

[1] S Miyamoto 2011 Plasma Phys. Control. Fusion 53 082001

Speaker: Gabriella Pautasso

30 WEIBEL INSTABILITY DUE TO NONLINEAR INVERSE BREMSSTRAHLUNG ABSRPTION IN MAGNETIZED FUSION PLASMA

Weibel instability due to nonlinear inverse bremsstrahlung absorption (WINLIBA) in magnetized plasma has been investigated in the frame of the relativistic kinetic theory (RKT). In this study the magnetized plasma is described by relativistic Fokker-Planck equation with an ameliorated Krook collision term which takes into account the relativistic effect and the Landau microscopic collision theory. The dispersion relation of the Weibel modes is obtained from the perturbed Fokker Planck equation and the growth rate is explicitly calculated as a function of the physical parameters of the plasma, the magnetic field and the electromagnetic wave. The Langdon effect due to the distortion of the isotropic component of distribution function is taken into account in this study. Applications are given for magnetic confinement fusion (MCF) plasma heated by micro-waves and for magneto-inertial fusion (MIF) scheme. The numerical analysis of model equations shows a moderate unstable modes for MCF and highly unstable Weibel modes (~〖10〗^11 s^(-1)) for MIF plasma.
Key words: Weibel instability, Nonlinear inverse bremsstrahlung absorption, magnetized plasma

Speaker: Prof. Abdelaziz Sid (PRIMALAB laboratory, University of Batna 1)

Prediction & Avoidance

Convener: Dr Carlo Sozzi (Istituto per la Scienza e Tecnologia dei Plasmi ISTP-CNR Milano Italy)

31 Cross Tokamak Disruption Prediction with Different Methods and from Different Perspectives

Currently machine learning disruption predictor is the most promising way of solving the disruption mitigation triggering problem. But it does need data from the target machine to be trained. However, the future machine may not be able to provide enough data both in quality and quantity to satisfy the training. In this paper we first explained why just simply mixing limited data from target machine is not the best way to build a cross tokamak disruption prediction model. Then, we attempt to address this issue from 3 different perspectives. First, we try to extract machine independent features with expert knowledge. Combined with domain adaptation method, we can port a disruption predictor to target tokamak with limited data from the target tokamak. Then, we tried to use deep neural networks to learn common representations of disruption instead of using expert knowledge, we fine-tuned the parts of the model with limited data from the target tokamak to transfer the predictor to the target tokamak. This is expected to get better performance if there are large enough data from the existing tokamaks. Finally, anomaly detection method was attempted. This eliminates the need of disruption shots which the future reactor cannot tolerant. However, for future reactors the goal is not just to build a high-performance disruption predictor that works on all the shots the reactor generated. Based of the above 3 different methods we suggested an integrated disruption prediction strategy. The goal of this strategy is to ensure the safety of the reactor though the different stages of the reactor which may have different requirements on the disruption predictors. The preliminary results suggested it’s a promising way of building disruption predictors for future tokamak reactors.

Speaker: Wei Zheng (International Joint Research Laboratory of Magnetic Confinement Fusion and Plasma Physics, Huazhong University of Science and Technology)

Cross Tokamak Disruption Prediction with Different Methods and from Different Perspectives-2.pptx

32 Implementing deep learning-based disruption predictor in a shifting data environment of new tokamak：HL-3

A deep learning-based disruption prediction algorithm has been implemented on a new tokamak, HL-3, for the first time. An Area Under receiver-operator characteristic Curve (AUC) of 0.940 has been realized, despite the limited training data obtained during the first two campaigns. Besides the well-known issue of lacking training data, a new issue is addressed that the data environment of new device is quite unstable. The plasma scans from low to high parameter space, bringing a shifting distribution of disruption causes and diagnostic data. This problem is often overlooked in previous implementations on steadily operating tokamaks and calls for more attention in future tokamaks like ITER. To address these challenges, novel modules including predict-first neural network (PFNN), data augmentation, pseudo data placeholders are developed and implemented, which promotes the accuracy by up to 20%. A series of advantages are also brought by the modules, including the robustness to function with missing input channels, and the interpretability to identify which parameter of plasma is under abnormal condition. The results demonstrate that the deep learning-based algorithm can provide reliable disruption alarms on a new tokamak, with the support of dedicated data collection and algorithm implementation.

Speaker: Zongyu Yang

Zongyu Yang-Implementing deep learning-based disruption predictor in a shifting data environment of new tokamak HL-3.pptx

16:00 Coffee

Prediction & Avoidance

Convener: Dr Carlo Sozzi (Istituto per la Scienza e Tecnologia dei Plasmi ISTP-CNR Milano Italy)

33 Combining Adaptive Learning and Physics-Guided Machine Learning for Prediction and Control of Anomalies and Disruptions in Tokamaks

Disruptions are one of the most critical issues in tokamak operation. In fact, the rapid termination of plasma magnetic confinement leads to significant heat and electromagnetic loads on the plasma-facing components, threatening the integrity of the reactor. Moreover, continuous early terminations of plasma discharge can cause variations in the energy production, limiting the amount of energy produced and the economic and environmental advantages of nuclear fusion. Despite the development and testing of effective strategies for disruption mitigation and prevention in recent years, the ultimate goal clearly remains the avoidance of disruptions. However, this solution requires the recovery and/or avoidance of plasma anomalies. Such an approach necessitates that the control system can act on the plasma when instabilities are in their early stages, small enough not to require termination of the discharge or mitigation actions.
Therefore, an algorithm capable of predicting disruptions in advance and detecting and classifying plasma anomalies is essential to provide the control system with all the information required to perform the correct countermeasures. Historically, disruption predictors and anomaly detectors have been based on two types of approaches: physics-based indicators, which are simple, easy to transfer, and interpret but often inaccurate, and AI data-driven algorithms [1], which are performant but usually obscure and not fully transferable.
This work aims to present a new approach that combines data-driven and physics-guided methodologies into a hybrid physics/data-driven AI algorithm capable of detecting, classifying, and predicting anomalies and disruptions. The algorithm is easy to interpret, highly performant, and easily transferable between tokamaks thanks to the combination of physics-informed machine learning with adaptive learning [2,3]. The algorithm has been tested in various JET campaigns, including the Deuterium-Tritium (DTE2) campaign, without a pretraining from previous campaigns. Very high detection and classification performances have been obtained, suggesting that a hybrid physics/data approach may play a relevant role in future reactors.

References:
[1] J.Vega, A.Murari et al Nat. Phys. 18, 741–750 (2022) https://doi.org/10.1038/s41567-022-01602-2
[2] A. Murari, R.Rossi et al. 2024 Nature Communications 15 https://www.nature.com/articles/s41467-024-46242-7
[3] R. Rossi et al 2024 Nucl. Fusion 64 046017 https://iopscience.iop.org/article/10.1088/1741-4326/ad2723
[4] R. Rossi et al 2023 Nucl. Fusion 63 126059 https://iopscience.iop.org/article/10.1088/1741-4326/ad067c

Speaker: Riccardo Rossi (Department of Industrial Engineering, University of Rome Tor Vergata)

Rossi R. - ITER - Disruptions - Final.pdf

Rossi R. - ITER - DIsruptions - Final.pptx

34 Advancing Interpretability in Artificial Intelligence Disruption Prediction Models: A Cross-Tokamak Perspective

Artificial Intelligence (AI) techniques, such as Machine learning and Deep Learning, have been extensively investigated for the construction of disruptions predictive models in tokamaks. Although the excellent performance has demonstrated the applicability of the paradigm to the experimental machines currently in service, the development of cross-tokamak models is still in its infancy [1]. These models, trained on existing machines, should be applied to newly constructed or next-generation machines. To achieve this ambitious goal, several issues still need to be resolved, including the unavailability of experimental data for newly constructed machines and the scalability of the data available from existing machines to new ones. The choice of the appropriate AI model also plays a fundamental role in the transportability of a predictor in a cross-tokamak approach.
For JET the literature has proposed AI supervised approaches such as Support Vector Machines [2-4], Decision Trees [5,6], Convolutional Neural Networks (CNN) [7-10], and unsupervised Manifold Learning approaches such as Generative Topographic Maps (GTM) [11-13] and Self-Organizing Maps (SOM) [14-16]. However, in both these supervised and unsupervised approaches, it was necessary to provide information on the duration of the pre-disruptive phase in the disrupted discharges of the training set during model training. This information could only be obtained heuristically [2-7] or statistically [13]. Recently, the authors of this contribute proposed, for JET, an unsupervised predictor [17], based on SOMs that does not require this information, relying only on the termination condition of the discharge (disrupted or regularly terminated), with undeniable advantages for the transportability of the model to another machines.
In this work the authors will initially conduct an analysis of their AI disruption predictors developed for JET [8-10,12-15,17] with a view towards developing a cross-tokamak model, highlighting their strengths and limitations.
Another aspect of fundamental importance for the use of AI models in control systems is their interpretability and their ability to identify the chain of events leading to disruption, so that appropriate avoidance actions can be implemented. The predictors presented in the literature so far also differ in this aspect. Manifold learning methods, such as SOMs and GTMs, allow to track the trajectory of a new discharge on the 2D map of machine's operational space, associating each time instant with the risk of disruption and, potentially, with the type of preceding event. On the other hand, despite their undeniable proficiency in predicting disruptions, deep neural network models exhibit a black-box behavior. However, recently, Explainable AI techniques [18] have emerged, aimed at interpreting the model's responses in relation to its input. In [19] a contribution to the explainability of a CNN predictor for JET was proposed.
The current work will thoroughly delve into these aspects and briefly introduce an innovative method for extracting rules from the SOM map of JET's operational space, facilitating a clear interpretation of the model's decisions throughout the discharge evolution.

Speaker: Prof. Giuliana Sias (ENEA -University of Cagliari)

IAEAthecnical meeting_GSias.pdf

35 Time-to-Disruption Estimation Using LSTM Networks

The application of machine learning methods has aided to improve the accuracy of disruption predictors in the last 15 years. However, these models are normally just a trigger and they do not provide a crucial piece information: the remaining time to the disruption. This is detrimental for their practical utility in order to develop efficient control actions.
This study tackles this limitation by employing Long Short-Term Memory (LSTM) recurrent neural networks to estimate the time from the detection of a precursor until disruption occurrence based on the alarms activated by an existing predictor trained with data from the Joint European Torus (JET). To this end, first, we trained the LSTM system on two intervals: “mitigation” (for imminent disruptions) and avoidance/prevention (for alarms triggered with a warning time > 100 ms), achieving classification accuracies above 87%. Second, a more detailed classification into three timeframes (mitigation, avoidance and prevention) further demonstrates the robust detection capabilities of the classification system, as detailed in the confusion matrix of Figure 1.
This methodology can improve the outcomes of existing predictive models by providing essential supplemental information. This strategy can be applied into any other machine learning-based disruption predictor.

Speaker: Giuseppe Ratta

Estimating Time to Disruption Using Transformer Models Ratta.pptx

36 DECAF update and study of the relationship between electron temperature collapses and disruption triggering through DECAF analysis

Rapid plasma dynamics preceding some disruptions in tokamak devices can be inferred through the electron temperature profile evolution due to the fast thermal transport along the field lines. In particular, local collapses in the electron temperature profile are the signature of nonlinear events such as flux surface tearing, observed due to the sudden thermal transport that follows changes in the confining field line topology. In some cases, such flux surface evolution can lead to the formation of Neoclassical Tearing Modes (NTMs), which can then lock and disrupt the plasma. In other cases, the reorganization of the magnetic field lines during an electron temperature collapse that precedes an accompanying current spike indicates that a fast reconnection event has occurred, which is often followed by a plasma disruption.

This work presents a general framework to identify and categorize electron temperature crashes, along with specific analysis applications. Identification of the crash time and radial location is performed by convolving a subset of electron temperature profile channels with a specialized kernel [1]. Once a collapse event has been identified, all of the available channels are used to reconstruct a ‘crash profile’ containing the features that allow the categorization of each event as a sawtooth crash, ELM, or a more general full electron temperature collapse. While all three events can seed NTMs, the required plasma state leading to NTM formation after a crash differs for each event type. It is shown that the plasma state before the thermal collapse, along with the characteristics of the electron temperature crash can give a significantly early prediction of plasma disruptions. The formalism presented in this work has been implemented in the DECAF* code [2], allowing the analysis of a wide range of shots across multiple tokamak devices.

Supported by US DOE Grants DE-SC0020415 and DE-SC0021311.
*U.S. and international patents pending.

References
[1] A Gude et al 2017 Plasma Phys. Control. Fusion 59 095009.
[2] S.A. Sabbagh, et al., Phys. Plasmas 30, 032506 (2023); https://doi.org/10.1063/5.0133825

Speaker: Guillermo Bustos Ramirez (Columbia University)

IAEA_2024_GuillermoBustosRamirez.pptx

Wednesday 4 September

Prediction & Avoidance

Convener: Gabriella Pautasso

37 Development and preliminary calibration of an off-normal warning system for SPARC

This work explores the development and preliminary calibration of an off-normal warning system for SPARC, the aim of which is to minimize disruption loads and maximize operation time via the detection, interpretation, and pacification (i.e. avoidance and mitigation) of anomalous events. Similar systems have been implemented for existing tokamaks like DIII-D [1], NSTX [2], and TCV [3], but the preliminary calibration of this system will have it ready on day 1 of operation, and it is being designed to quickly adapt to the addition of SPARC data in its initial campaigns. It is expected that the preliminary calibration and subsequent intended day 1 implementation of this system will also be useful for informing the initial implementation of ITER’s needed respective system. The detection and interpretation of this system will be facilitated via both physics-based warning thresholds as well as machine learning-based Proximity-to-Instability Algorithms, while the pacification will be handled by equilibrium steering, “soft-landings”, and the disruption mitigation system, the choice of which will depend on the severity and type of the anomaly. The implementation of the system will initially focus on developing physics-based warnings, which are expected to be more reliable than ML-based alternatives early in operation and can provide more interpretable results to use in pulse-planning. The preliminary calibration of these warnings will be performed using a novel technique that trains individual warning modules targeted at specific off-normal events on both simulated examples of these events in a SPARC-like environment as well as events from the C-Mod database. The validation of warning modules for impurity accumulation, vertical displacement events, locked modes, and density-limits on C-Mod will be presented here, along with the first implementation of an impurity accumulation warning module trained within the DEFUSE [4] event-monitoring framework and on simulated impurity injections within the POPSIM simulation framework.

Work funded by Commonwealth Fusion Systems.

[1] N.W. Eidietis et al 2018 Nucl. Fusion 58 056023
[2] S.P. Gerhardt et al 2013 Nucl. Fusion 53 063021
[3] T. Vu et al 2021 IEEE Transaction on Nuclear Science, vol. 68, no. 8
[4] Pau A et al 2023 A modern framework to support disruption studies: the EUROfusion disruption database 29th IAEA Int. Conf. on Fusion Energy (London, UK, 2023) (IAEA) p EX/4-1669

Speaker: Alex Saperstein (MIT - PSFC)

SPARCs off-normal warning system - Alex Saperstein - 4 Sept 2024.pdf

SPARCs off-normal warning system - Alex Saperstein - 4 Sept 2024.pptx

38 Machine Learning model for real-time SPARC vertical stability observers

Author: A. Kumar,
Co-authors: C. Clauser, T. Golfinopoulos, D.T. Garnier, J. Wai, D. Boyer, A. Saperstein, R. Granetz, C. Rea

Given the demanding requirements of the SPARC high-field tokamak (B_{0}=12.2 T) and its operation with high elongated plasma (\kappa_{sep}=1.97), robust real-time-compatible vertical stability observers are paramount. In this work, we present the fast surrogate-based modeling approach for observers (such as energy functional, VDE n=0 growth rate, stability margins ​​(m_s), inductive stability margins (m_i), max-Z, and frequency components), integrating advanced 2D electro-mechanical circuit and dynamic plasma response models [1]. These surrogate observers employ transformer-based machine learning techniques; trained to replicate and predict the results of the filamentary semi-rigid body MEQ-RZIp and deformable free boundary MEQ-FGE (and its linearized version FGElin) code suite, as detailed in Carpanese et al. [2]. The training dataset incorporates simulated SPARC primary reference discharge scenarios and the Alcator C-Mod (hot VDEs) 2012-2016 disruption warning database. To enhance robustness, these observers will also be trained over a range of simulated L- and H-mode SPARC plasma scenarios, including periods without and with ELM triggering (via artificial vertical kicks [2]). We will report on the assessment of observers’ sensitivity to the underlying RZIp & FGE models and their proximity to stability boundaries; thereby supporting disruption prediction and avoidance.

Acknowledgements
Work funded by Commonwealth Fusion Systems under grant number RPP2023.

References
1. M. L. Walker & D. A. Humphreys (2006), 50:4,473-489, DOI: 10.13182/FST06-A1271
2. Carpanese et al. 2020 EPFL PhD thesis no. 7914
3. Sartori F. et al Proc. 35th EPS Conf. on Plasma Physics (Hersonissos, Greece, 9–13 June 2008) vol 32D P5.045

Speaker: Arunav Kumar

IAEA_SPARC_vertical_compressed.pdf

39 P&A discussion

Speaker: Dr Carlo Sozzi (Istituto per la Scienza e Tecnologia dei Plasmi ISTP-CNR Milano Italy)

116-PandAdiscussion-Sozzi.pptx

10:35 Coffee

Mitigation

Convener: Eric Nardon (CEA)

40 Overview of the ASDEX Upgrade shattered pellet injection studies

In support of the ITER DMS development, a highly flexible SPI system [[1], [2]] was installed at ASDEX Upgrade (AUG). It offers the unique opportunity to investigate the effect of different fragment size and velocity distributions — which were characterised beforehand in extensive laboratory tests — on the disruption behaviour. The triple barrel setup with independent freezing cells, injection lines, and shatter heads allows to study dual and staggered injection schemes essential to support the finalisation of the ITER DMS.
The experiments are accompanied by modelling with the DREAM, INDEX and JOREK codes, showing good agreement with the experimental observations.
For pure deuterium (and low neon doping) SPI the largest effect of the shatter head geometry are observed in simulation and experiment: Hereby, large and fast fragments — as produced by the 12.5° rectangular shatter head [[3]] — lead to an increased material assimilation [[4]] and radiated energy fraction. This is in line with the current ITER SPI design [[5]] with a shatter angle of 15°.
The (toroidal) radiation asymmetries are studied via foil bolometers, located in 5 different toroidal sectors and matching viewing geometry. We observed a close connection of the radiation characteristics to the evolution of the disruption behaviour, where the asymmetry decreases and the total radiated energy increases with increasing neon content inside the pellet.
The radiated energy fraction was found to be a strong function of the neon content inside the pellet dominating over the shattering geometry — especially for high neon content.

[[1]] M. Dibon et al., Review of Scientific Instruments, 94 (4):043504 (2023).
[[2]] P. Heinrich et al. Fusion Engineering and Design, 206:114576 (2024).
[[3]] T. Peherstorfer, MSc Thesis, TU Wien, 2022.
[[4]] S. Jachmich et al., 49th EPS Conference on Plasma Physics, Bordeaux, France (2023).
[[5]] M. Lehnen et al., 29th IAEA FEC, TECH/1-1, London, UK (2023).

Speaker: Paul Heinrich (Max-Planck-Institut für Plasmaphysik)

2024-09-04_3rd_TM_Overview-AUG-SPI_handout.pdf

41 Thermal Energy Mitigation and Toroidal Peaking Effects in JET Disruptions

Previous investigations on JET suggest thermal stored energy ($W_{th}$) is poorly mitigated by either Massive Gas Injection (MGI) or Shattered Pellet Injection (SPI) Disruption Mitigation Systems (DMS), when measured by weighted averages of bolometer channels. A contrasting investigation on ASDEX-Upgrade found that thermal energy is well mitigated with MGI. We investigate whether the apparent poor thermal mitigation on JET is explained by radiation peaking near the injected impurity plume, combined with the limited toroidal resolution of bolometry diagnostics. High toroidal peaking in the pre-thermal quench (pre-TQ) is found in Ar/D2 MGI on JET, with $>$3x higher radiation near the injection location than elsewhere throughout the pre-TQ. A previously unexplained toroidal peaking measurement in neon SPI is also successfully reproduced with similar peaking. These observations agree with literature from Alcator C-Mod, ASDEX-Upgrade, and KSTAR. This peaking is not captured by the bolometry used in previous JET studies that found poor thermal mitigation. A set of 10% Ar / 90% D2 mix MGIs is analyzed using the Emis3D radiation modeling code. With injector-localized peaking, almost two thirds of $W_{th}$ is radiated, where estimates that do not account for injector peaking would indicate less than half. The toroidal spread of the injector peaking feature is poorly constrained, and up to 85% of the plasma's thermal energy may be radiated using the largest possible spread. Similar radiated fractions are seen in a second set of MGI discharges and in a set of neon SPI discharges. The improved thermal mitigation found here suggests a reduced divertor melt risk in high performance mitigated disruptions on ITER and SPARC. However, the higher peaking near DMS injectors could increase flash melting risk on nearby plasma facing components. Only single-injector mitigation is studied here. Multiple injector mitigation, as planned for ITER and SPARC, might further increase the radiated fraction and decrease pre-TQ peaking, but that is not addressed in this study.

Speaker: Benjamin Stein-Lubrano (Plasma Science and Fusion Center, Massachusetts Institute of Technology)

Ben_SL_TM_v3.pptx

42 Pellet Size Optimization for the ITER SPI

The ITER disruption mitigation system design using shattered pellet injection is reaching maturity, with the current leading strategy being a staggered pellet injection scheme. The first injection in this scheme uses pure hydrogen to reduce runaway electron (RE) generation by increasing plasma density and reducing electron temperature. The second injection consists of neon pellets to efficiently radiate the remaining thermal energy and prevent damaging thermal loads to plasma-facing components. To achieve these goals, the current barrel diameter has been set to 28.5 mm, delivering 1.5×1025 hydrogen atoms through 8 pellets, followed by 3 neon doped pellets delivering an additional 5.6×1024 atoms for a 15MA H-mode DT plasma. However, current modeling predicts that even with such large hydrogen pellets, it may not be possible to achieve RE generation suppression during the nuclear phase, necessitating an RE mitigation strategy.

The most promising RE mitigation strategy is currently low-Z benign termination. This approach requires a low-density companion plasma and a low q-edge to enable a fast-growing MHD instability that expels the confined REs, leading to a “benign” termination of the RE beam. Experiments on DIII-D and TCV have shown a clear upper limit in neutral pressure at approximately 0.5-1Pa, above which the ionisation cross-section between the REs and neutral gas is sufficient to increase the companion plasma density and reduce the mode growth rate. This is problematic as the currently envisaged staggered injection scheme on ITER will result in a neutral pressure of ~30Pa. Furthermore, the RE density on ITER is expected to be more than double that of DIII-D and TCV, further exacerbating this problem. This incompatibility with the RE mitigation scheme requires a re-evaluation of the pellet sizing.

Further motivation for assessing the staggered injection strategy comes from preliminary results of last year’s JET SPI campaign. Firstly, it was observed that the density rise in the core can take tens of milliseconds after pellet arrival due to density penetration timescales (Figure 1). Secondly, the pre-thermal quench duration was drastically reduced to below 5 ms if the hydrogen pellet was doped with 0.016% neon to prevent plasmoid drift (Figure 2). A similar reduction in pre-thermal quench duration was observed if the plasma was strongly seeded (Figure 3) 1. These factors suggest that there may not be sufficient time for the staggered pellet scheme to achieve its goals.

This talk will propose reduced pellet sizes to enable greater flexibility of the system and provide a contingency should the staggered injection scheme not meet its objectives. A new mitigation strategy addressing thermal load mitigation by leveraging seeded gases, further reducing reduced RE generation and providing a neutral pressure range more compatible with low-Z benign termination will also be presented1.

1 – Sheikh, U, et al., “Impact of Plasma Seeding on Shattered Pellet Injection Mitigations on JET”, submitted to Nuclear Fusion 2024

Speaker: Umar Sheikh (SPC-EPFL)

Presentation Sheikh IAEA TM 2024.pptx

12:25 Lunch

Posters

All posters will be displayed in both sessions on Tuesday and Wednesday.

Mitigation

Convener: Nicholas Eidietis (General Atomics)

43 The Effect of Externally Applied and Self-Excited Waves on Relativistic Electrons

A series of experiments is underway to explore the effect of both self-excited and externally launched plasma waves on relativistic electrons (REs) across a wide range of geometries and plasma parameters. While O and X-mode waves are routinely used for heating and current-drive in tokamaks they are incapable of directly resonating with REs since their phase velocity is much greater than the speed of light. However, on the DIII-D tokamak it has been observed that when EC waves are launched into the low density quiescent runaway electron (QRE) regime the RE population is expelled on a 100ms time-scale. It is hypothesized that this flushing is due to a transition to a new turbulence regime and therefore increased RE transport. Recent experiments exploring this effect measured for the first time an increase in low frequency (<200kHz) density and magnetic fluctuations whose amplitude correlates with increased RE flushing as the ECH power is increased (Figure 1). A series of experiments is also planned through the Frontier Science Program to study the effect of both externally injected and self-excited plasma waves on REs. Experiments are planned to explore the effect of externally launched Whistler waves on REs in the LArge Plasma Device (LAPD) linear device. Experiments are also planned on the Madison Symmetric Torus (MST) which is equipped with a high frequency probe capable of measuring waves up to 6GHz. This provides a unique opportunity to measure self-excited waves up to and approaching the EC frequency as well as measure the RE-driven slow-X EC wave due to the ability to insert this probe directly into the plasma.

Work supported by US DOE under DE-FC02-04ER54698 (D3D), DE-SC0021622 (Frontiers), DE-SC0022270 (CU), DE-SC0019003 (Wisc.), DE-SC0021624 (UCI), and DE-FG02-07ER54917 (UCSD)

This work was supported by the US Department of Energy Office of Fusion Energy Sciences at Los Alamos National Laboratory (LANL) under contract No. 89233218CNA000001.

This work was supported by the US Department of Energy Office of Fusion Energy Sciences at Lawrence Livermore National Laboratory
(LLNL) under contract No. DE-AC52-07NA27344.

This work was supported by the US Department of Energy Office of Fusion Energy Sciences at Oak Ridge National Laboratory (ORNL) under contract No. DE-AC05-00OR22725.

Speaker: Alexander Battey (Columbia University)

batteyA_IAEAinvited_fin.pptx

44 Effect of Electron Cyclotron Waves on Plasma with Runaway Electrons

Runaway electrons generated during tokamak disruptions are a major concern for the safe operation of future tokamaks. These energetic electrons can carry significant current and cause severe damage to a tokamak. Therefore, mitigating runaway electrons is essential for the safety and efficiency of fusion devices. Interaction of runaway electrons with waves is one of the potential mechanisms for their mitigation.

This study investigates the effect of electron-cyclotron (EC) waves on post-disruption plasma with runaway electrons. "Free space" O- and X-mode EC waves are routinely used for plasma heating (ECH) and current drive. However, these modes do not interact directly with relativistic electrons and cannot be injected into plasma with a density above the corresponding cutoff density. In contrast, the internal slow X-mode is capable of resonant interaction with runaway electrons.

We report on experiments conducted on the DIII-D tokamak, where the internal slow X-mode was generated via the so-called OXB (Ordinary- to eXtraordinary- to Bernstein-) mode conversion, a process previously explored in space plasmas to understand planetary radio emissions, auroral kilometric radiation, and particle acceleration. The OXB process has also been considered for heating and current drive in overdense plasmas of tokamaks and stellarators.

In this experiment, the application of ECH during the runaway electron plateau resulted in a significant increase in background plasma density, along with a doubling of the loop voltage and runaway electron synchrotron emission. These observations indicate effective mitigation of the runaway electron beam. It is noteworthy that when pre-calculated OXB conditions are met, a characteristic density spike forms on the density profile.

Kinetic modeling and 1D impurity and neutral transport modeling suggest that collisional dynamics play a significant role in the observed effects. The modeling indicates that a further increase in background density and temperature is prevented by strong line emission resulting from high argon (Ar) concentration. A reduction in Ar content is expected to allow for higher background temperatures, potentially permitting kinetic instabilities of the runaway electron beam.

The study highlights the potential of ECH/OXB heating as a novel approach to mitigate runaway electrons.

Work supported by US DOE under DE-FC02-04ER54698 (D3D), DE-SC0021622 (Frontiers), DE-SC0022270 (CU), and DE-FG02-07ER54917 (UCSD)

This work was supported by the US Department of Energy Office of Fusion Energy Sciences at Lawrence Livermore National Laboratory
(LLNL) under contract No. DE-AC52-07NA27344.

This work was supported by the US Department of Energy Office of Fusion Energy Sciences at Oak Ridge National Laboratory (ORNL) under contract No. DE-AC05-00OR22725.

Speaker: Pavel Aleynikov (Max-Planck-Institut für Plasmaphysik)

Aleynikov_IAEA_TM_ITER_2024.pdf

16:05 Coffee

Mitigation

Convener: Nicholas Eidietis (General Atomics)

45 Plasmoid drift and first wall heat deposition during ITER H-mode dual-SPIs in JOREK simulations

The heat flux mitigation during the Thermal Quench (TQ) by the Shattered Pellet Injection (SPI) is one of the major elements of disruption mitigation strategy for ITER. It's efficiency greatly depends on the SPI and the target plasma parameters, and is ultimately characterised by the heat deposition on to the Plasma Facing Components (PFCs). To investigate such heat deposition, JOREK simulations of neon-mixed dual-SPIs into ITER baseline H-mode and a ``degraded H-mode'' with and without good injector synchronization are performed with focus on the first wall heat flux and its energy impact. It is found that low neon fraction SPIs into the baseline H-mode plasmas exhibit strong major radial plasmoid drift as the fragments arrive at the pedestal, accompanied by edge stochasticity. Significant density expulsion and outgoing heat flux occurs as a result, reducing the mitigation efficiency. Such drift motion could be mitigated by injecting higher neon fraction pellets', or by considering the pre-disruption confinement degradation, thus improving the radiation fraction. The radiation heat flux is found to peak in the vicinity of the fragment injection location in the early injection phase, while it relaxes later on due to parallel impurity transport. The overall radiation asymmetry could be significantly mitigated by good synchronization. Time integration of the local heat flux is carried out to provide its energy impact for wall heat damage assessment. For the baseline H-mode case with full pellet injection, melting of the stainless steel armour of the diagnostic port could occur near the injection port, which is acceptable, without any melting of the first wall tungsten tiles. For the degraded H-mode cases with quarter-pellet SPIs, which have $1/4$ total volume of a full pellet, the maximum energy impact approaches the tolerable limit of the stainless steel with un-synchronized SPIs, and stays well below such limit for the perfectly synchronized ones.

Speaker: Di Hu (Beihang University)

20240905.pptx

46 Pellet fragmentation process in the context of the SPI technology for the ITER DMS: analysis of the fragment characteristics supported by numerical simulations and image diagnostics of shatter tests

The disruption mitigation system (DMS) for ITER is based on the shattered pellet injection (SPI) technology. The principle of operation is to form cylindrical cm-sized cryogenic pellets and accelerate them to high speeds towards a shattering chamber, where the pellets disintegrate into a plume of fragments of different sizes and velocities, which then enter the plasma for the mitigation process. The effectiveness of this mechanism is governed by the material assimilation, which strongly depends on the fragment size and velocity distributions resulting from the shattering process. In order to optimize this, it is important to know how the impact characteristics, namely the pellet material, velocity, impact angle, etc., influence the fragmentation properties. The ITER DMS task force has launched a program to characterize and study the fragmentation both experimentally and via numerical simulation. In parallel, many modelling activities are ongoing to study the effect of the injected fragmented pellet material on the disruption dynamics. In order to provide predictions for ITER with a high degree of confidence, it is essential to utilize a fragment size and velocity distribution that is as realistic as possible as an input to these models.

As part of this ongoing program, Fraunhofer EMI is developing numerical models and computer codes to simulate and analyze the complex fragmentation process. A broad range of impact scenarios are simulated for different pellet materials and impact velocities, as well as for different pellet shattering chamber designs, based on calibrated and validated models. The resultant fragment characteristics are systematically analyzed in terms of fragment size and velocity distributions. The goal is to optimize the pellet shattering chamber design as well as to derive guidelines for optimized impact conditions in order to get the desired fragment characteristics.

Our presentation will begin with a short overview of the modeling approach as well as of its calibration and validation process. Subsequently, we present the results of simulations for various impact scenarios, which illustrate the fragmentation process, and analyze and discuss the distributions of fragment size and velocity. The statistical fragmentation model, as described by Parks [1], will be applied to the pellet shattering process and compared with direct results from numerical simulation as well as analyzed shattering videos from either experiments and simulations. Our simulation results show deviations from Parks’ model, which might influence the ablation process of the fragments in the plasma.

[1] P. Parks, “Modeling dynamic fracture of cryogenic pellets”, Tech. Rep. GA-A28325, June 2016.

Speakers: Pascal Matura (Fraunhofer Institute for High-Speed Dynamics, Ernst-Mach-Institut, EMI), Stefano Signetti (Fraunhofer Institute for High-Speed Dynamics, Ernst-Mach-Institut, EMI)

2024-09-04_DMS_Technical_Meeting_EMI_final.pdf

47 Shattered Pellet Technology and related developments at the ITER DMS support laboratory

The ITER disruption mitigation support laboratory is part of the ITER Disruption Mitigation System (DMS) Task Force programme to establish the physics and technology basis for the ITER DMS. The laboratory is located at the HUN-REN Centre for Energy Research (CER), Budapest Hungary. The aims include production, launching and shattering of 28.5x57 mm (d x L) H, D, Ne and mixture pellets, testing various propellant suppressor setups, testing the ITER Optical Pellet Diagnostic (OPD) concept, and diagnosing the fragment plumes resulting from shattering. The OPD prototype, a fast shutter, propellant valve and additional technologies are being developed in the framework of additional contracts.

The pellet production process was studied in detail experimentally and via modelling. Pellet production recipes were developed where a loose (cryogenic snow) layer is formed at the pellet-barrel interface, enabling intact launch of H, D and Ne pellets. The conditions for snow formation are qualitatively understood. After launch the pellets fly thorugh a propellant suppressor volume similar to the ITER design. Various inner structures are studied and gas flow modelled to select the most suitable solution for ITER. After the suppressor and OPD pellets fly the same distance to the shattering head as they would do on ITER. Large diameter flight tubes are installed so as the pellet free flight directions and rotation can be diagnosed by fast cameras before shattering. The shattering head reproduces the ITER geometry, but enables camera view into it, therefore details of the shattering process can be observed. The fragments resulting from shattering are diagnosed by a laser curtain diagnostic. Spatial, temporal, velocity, size distributions are measured. Altogether about 800 pellets have been launched in the past two year of operation, therefore extensive experimental database has been accumulated.

Additional components and technologies are being developed at CER for SPI, and more broadly for cryogenic pellet injection: fast shutter, propellant valve, 3D printed cryogenic components, parahydrogen pellet technology. These are partly funded by ITER, partly by other sources. A general purpose cryogenic test infrastructure has been set up recently to enable testing of these technologies with both cryocooler and liquid Helium cooling.

Speaker: Sandor Zoletnik (HUN-REN Centre for Energy Reserach)

IAEA_disruption_2024_zoletnik.pptx

Thursday 5 September

Mitigation

Convener: long zeng

48 Assimilation of deuterium into relativistic runaway electron beams and the implications for benign terminations in present devices, ITER, and future devices

Localized wall damage from post-disruption runaway electron (RE) wall impact is a significant concern for future large tokamaks. One possible method for reducing this wall damage in the event of an unavoidable RE-wall impact is massive injection of low-Z (H2 or D2) gas. This injection can have the effect of partially recombining the cold thermal background plasma, resulting in a greatly increased RE final loss instability MHD amplitude, larger RE wetted area, and reduced local RE heat flux and damage. Experimental trends in thermal plasma partial recombination resulting from massive D_2 injection into high-Z (Ar) containing runaway electron (RE) plateaus in DIII-D and JET were studied with the goal of understanding the parameters needed to achieve sufficiently low electron density (n_e≈10^18/m^3) to increase the RE final loss MHD levels. In both DIII-D and JET, thermal electron density n_e is found to drop by ~ 100× when the thermal plasma partially recombines, with a minimum at a vacuum vessel-averaged D_2 density in the range 10^20-10^21/m^3. RE effective resistivity also drops after partial recombination, indicating expulsion of the Ar content. Achieving partial recombination is found to become more difficult as RE current is increased. The amount of initial Ar in the RE plateau is not observed to have a strong effect on partial recombination. Partial recombination timescales of order 5 ms in DIII-D and 15 ms in JET are observed. These basic trends and timescales are matched with a 1D diffusion model, which is then used to extrapolate to ITER and SPARC tokamaks. It is predicted that ITER will be able to achieve sufficiently low n_e values on time scales faster than expected RE plateau vertical drift timescales (of order 100 ms), provided sufficient D_2 or H_2 is injected. In SPARC, it is predicted that achieving significant n_e recombination will be challenging, due to the very high RE current density. In both ITER and SPARC, it is predicted that achieving low n_e will be easier with Ar as a background impurity (rather than Ne).

Speaker: Eric Hollmann (UC San Diego)

EMH IAEA talk 2024 v1.pdf

49 Runaway mitigation and safe termination during startup and disruptions

Runaway electrons of MeV and higher energies can dominate the plasma
current during ITER startup and the current quench phase of a major
disruption. The plasma regime spans from reasonably low-density and
high-temperature (startup) to high-density and low-temperature
(disruption mitigated by high-Z impurities), and somewhere in between
(disruption mitigation by low-Z injection). Here we describe the
recent physics findings informing how runaways can be managed in these
diverse situations. The first result is on how runaway-induced wave
instabilities, particularly the slow-X modes as opposed to the
whistler waves, can efficiently transfer the plasma current from
high-energy runaways to suprathermal electrons, during which
relativistic runaways can even reverse their direction with respect to
the magnetic field. These findings come from fully-kinetic simulations
that remove the shackle of quasilinear formulation previously reported
in the literature, and the physics is of interest to warm plasmas
during ITER startup or in reheated plasmas during a current quench.
The second result is on the standard ITER scenario of high-Z impurity
mitigated disruption in which runaway dissipation and transport loss
are greatly enhanced by high-Z impurities while plasma column scrapes
off against the first wall during a vertical displacement event
(VDE). The runaway loss pattern on the first wall, as the result of
both collisional transport and VDE scrape-off, is of particular
interest in assessing the wall tolerance for the runaway impact. A
hybrid model that couples quasi-static MHD with drift-kinetic runaway
electrons has been simulated to account for the ITER VDE dynamics with
the full kinetic physics of runaway dissipation and transport loss, as
well as the scrape-off process. The third result is on the use of
solid tungsten particulates for standoff termination of the
relativistic runaway electrons. The idea is that instead of having the
runaways scrape-off against the wall, one can place a cloud of
tungsten particulates in front of the first wall to be impacted by the
VDE, so runaways can terminate on these solid particulates. This is
similar to the previous dust shield concept for divertors, but the new
twist is that the tungsten particulates can facilitate safe
termination by both runaway energy attenuation and effective pitch
angle scattering, which can alter the runaway orbits (e.g. from
passing to trapped) for broader deposition pattern on the first
wall. Here we will show both effects by the tungsten particulates via
MCNP calculations. The fourth result is on the feasibility of doing
away with thermal quench mitigation by radiative cooling. The idea is
to inject enough amount of hydrogen that the plasma would be
dilutionally cooled to be collisional for open field line transport,
but still warm enough that the inductive electric field from
$E_\parallel = \eta j_\parallel$ stays below the avalanche threshold
electric field if not the Connor-Hastie critical field. Interestingly,
MHD simulations with Braginskii transport coefficients are supposed to
be theoretically sound for such a mitigated collisional plasma. Here
we will show the most up-to-date PIXIE3D simulations that establish
comparable TQ and CQ time scales in such mitigation scenario.

Speaker: Xianzhu Tang (Los Alamos National Laboratory)

LANL-DisruptionMitigation.pptx

50 Preparing disruption solutions for tokamak power plants

Realizing tokamak power plants requires reducing the frequency and impact of disruptions sufficiently to accept them as a part of operations. SPARC is a high field tokamak [1] ($B_o=12.2$ T, $I_p=8.7$ MA) designed to demonstrate Q>1 and to explore divertor and disruption solutions for the ARC power plant. Significant disruption work is ongoing in hardware, software, operational planning, and science to prepare for SPARC and to realize the disruption strategy for ARC. Much of the physics work is common to all tokamak concepts and Commonwealth Fusion Systems (CFS) seeks to engage the community on these topics.

The SPARC Disruptions Team is developing and publishing physics-based and machine learning off-normal warnings (ONWs) for the plasma control system (PCS) and off-normal simulations (ONSIMs) for stress-testing the PCS and calibrating the predictors. The soon-to-be open-source MOSAIC framework as well as the DEFUSE code [2] facilitate sharing ONWs and ONSIMs to pool efforts and enable cross-validation. Tokamak power plants will naturally accrue a large database of repeat pulses, or many samples of the steady-state, both of which provide an opportunity for detecting variation from nominal which is a conceptually simpler problem than detecting disruption boundaries. Pulsed power plant prediction algorithms and disruptivity might be assessed by running repeat discharges in relevant plasmas.

The Runaway Electron Mitigation Coil (REMC) will be operated on SPARC providing a critical test of this runaway prevention technique. To increase the likelihood of runaway solutions for future machines and to provide optionality, the community is encouraged to continue to explore alternative runaway prevention techniques. Better quantifying the runaway risk is also important, including modeling runaway impacts and further benchmarking runaway models on empirical data where runaways are and are not observed.

Next generation machines, including SPARC and ITER, will access power plant relevant disruption heat fluxes on tungsten, steel, and other in-vessel materials. Material testing facilities might be leveraged to complement and accelerate our understanding of the degradation of tungsten-based materials from unmitigated thermal quenches, halo current heat fluxes, and runaway impacts.

Important physics questions remain unresolved in the ITPA scalings including the effect of the plasma-vessel mutual inductance on the shortest current quench duration [3], a scaling for the longest current quench duration, and the expected correlation of the poloidal arc length on the maximum halo current fraction.

Collaboratively identifying the greatest risks and motivating the greatest opportunities in disruptions is expected to accelerate the realization of fusion power plants. This talk will discuss how CFS intends for SPARC to be a member of this ecosystem and solicit active support from the community in preparing for operations and beyond [4].

Funded by Commonwealth Fusion Systems.
[1] A.J. Creely, et al. J. Plasma Phys. 86.5 (2020) 865860502
[2] A. Pau, et al. IAEA FEC (2023)
[3] T. Yokoyama, et al. Nucl. Fusion 63 (2023) 126049
[4] A.J. Creely, et al. Phys. Plasmas 30.9 (2023)
*rsweeney@cfs.energy

Speaker: Ryan Sweeney (Commonwealth Fusion Systems)

sweeney_IAEA_TM_2024_Live (2).pdf

10:30 Coffee

Mitigation

Convener: long zeng

51 Disruption mitigation in tokamak by Fast Gas and Macroparticles Injection

Safe termination of the plasma discharge using injection of intense gas flows and macroparticles (pellets) is considered as the main system for preventing development of the runaway electron beams in tokamak-reactor (ITER) [1]. One of the main limitations in the use of these systems in large-scale tokamaks is the weak penetration of injected gas and particles into the central zones of the high temperature plasma discharge. This reduces reliability of the disruption stabilization and leads to the need to develop additional methods for safe discharge termination.
To minimize consequences of the plasma disruptions in tokamak, “alternative” methods of gas flow injection are considered, including initiation of fast chemical combustion reactions, the injection of neutral particles from targets when a potential is applied, injection of impurities sprayed using powerful microwave waves-target interaction, and superheated liquid bubbling. Present report represents analysis of the experiments carried out with “alternative” methods at the laboratory stand and in T-10 tokamak (target voltage U°=°0°-°450°V, battery capacity C°=°0.4°F, maximum energy reserve W°~°40°kJ, microwave power up to 1°MW) [2]. Analysis indicated that “alternative” methods could provide effective penetration of fast gas and macroparticles flows into the central zones of the plasma and demonstrated the possibility of quickly stopping a plasma discharge with current decay rate of up to 35°-°40°MA/sec and control system response time less than 0.1°ms.
Preliminary analysis considers possibility of using “alternative” methods in tokamak reactor for generation of directed gas and macroparticles flows, including fast chemical combustion reactions with initiating substances [3] (increased radiation resistance to fast neutron flows, E°>°1°MeV, F°~°7.5°e12°neutron/cm2/sec, temperature stability T°~°280-300C, and stability under high vacuum conditions).
One of the key issues of the disruption mitigation is trigger condition of the control system. Experiments in the T 10 tokamak with all-metal (tungsten, lithium) in-vessel elements indicated that transition from quasi-stationary discharge to a fast phase of disruption can be associated with development of arc plasma discharges at the plasma-facing components [4]. In-vessel movable electric and magnetic probes installed in the T-10 tokamak allows monitoring of the arc discharges and could provide reliable trigger parameters for disruption stabilization systems and/or for safe discharge termination.

[1] M. Lehnen, K. Aleynikova, P.B. Aleynikov, et al., Disruptions in ITER and strategies for their control and mitigation, Journal of Nuclear Materials, 2015, 463, 39-48, DOI:10.1016/j.jnucmat.2014.10.075.
[2] P.V. Savrukhin, E.A.Shestakov E.A., A.V.Khramenkov, Influence of fast gas flow injection on the plasma discharge shutdown in tokamak, Int. Conf. Plasma Physics and Controlled Fusion, Zvenigorod, Moscow reg., 16 - 20 March, 2020, p.270.
[3] W.E.Voreck, M.E.Downs, E.I.Lindberg, AEROJET GENERAL Corporation, California, Report No. RN-S-0368, 1967 (https://www.osti.gov/servlets/purl/4221497).
[4] P.V. Savrukhin, E.A. Shestakov, A.I.Ermolaeva and R.Yu.Solomatin, Plasma arcs formation in the plasma periphery during disruptions in the T-10 tokamak plasma, J. Phys.: Conf. Ser. 2017, 907, 012006, DOI 10.1088/1742-6596/907/1/012006.

Speaker: Petr Savrukhin (Institution “Project center ITER”, Moscow, 123060, Russia)

2024 IAEA Workshop savrukhin v1.pdf

52 MIT Discussion

Speaker: Eric Nardon (CEA)

Discussion_on_Mitigation_IAEA_TM_2024_v2.pptx

12:00 Lunch

ITER Site Tour

Consequences

Convener: Fabio Villone (DIETI, Università degli Studi di Napoli Federico II)

53 Modelling of runaway electron – induced PFC damage

The potential of localized heat loads under disruptions to cause considerable melting of plasma-facing components (PFC) has been extensively investigated. Two distinctive regimes exist, which lead to different types of PFC damage and necessitate different modelling approaches; surface loading and volumetric loading.
Surface heating is caused by electrons and ions with energies in the keV range with depth ranges of the order of a few nm. Dedicated EUROfusion and ITPA coordinated experimental activities have provided a wealth of empirical data on PFC melting induced by transient surface loads, which have guided the development of physics models that enable high-fidelity simulations. This led to the reduction of the incompressible resistive thermoelectric MHD description under the magnetostatic shallow water approximation. Coupling with heat transfer including phase transitions supplemented with free interface boundary conditions, as implemented in the MEMENTO code, yields a very accurate description of melt dynamics and deformation. Successful validation against multiple tokamak experiments has lent confidence in the predictive power of such tools.
Volumetric heating is caused by electrons with relativistic energies far into the MeV range with depth ranges of the order of mm even inside high-Z metals like W. Runaway electrons (REs) constitute the final frontier in the context of PFC damage. RE incidence can lead to deep melting beyond the shallow water approximation, induce material explosions driven by uneven thermal expansion and cause loss-of-coolant accidents due to heat deposition close to the cooling pipe. Unique evidence of explosive RE-PFC interaction accompanied by the expulsion of fast solid debris have been obtained in FTU (accidental, TZM) and DIII-D (deliberate, graphite). Extensive RE-induced damage has also been reported in JET, WEST and COMPASS.
RE velocity distributions at the PFC surface constitute the initial conditions for Monte-Carlo (MC) modelling of electron transport inside metals to obtain volumetric heat maps. Determination of the heat map gradients comprises an essential part of the work-flow, since these control how the internal thermal stresses build-up and whether explosive detachment will occur. However, these are highly sensitive not only to the RE energies but also to their impact angles. The depth range refers to normal incidence and is thus indicative of energy loss along the path, which lies just beneath the surface at grazing incidence. The magnetic field presence, the large PFC area / wetted area and need for large particle statistics, makes the MC simulations costly. The situation is also complex regarding the PFC response, since pure thermal or even linear thermoelastic modelling does not suffice for ductile metals with stable liquid phase like W. Finally, high power densities (grazing incidence and fast ms deposition within a thin surface layer) yield temperatures in excess of 10000K at which thermophysical W properties are unknown.
A complete work-flow will be presented that has been developed to model the thermal response of ITER W first wall panels under RE impacts. The results on short (relevant for PFC damage) and long (relevant for coolant failure) temporal scales will be critically assessed.

Speaker: Svetlana Ratynskaia (Royal Institute of Technology KTH)

Ratynskaia_Thursday.pdf

54 Characterization of transient heat flux induced damages of tungsten PFCs during plasma disruption in EAST

Extreme high transient heat flux up to thousands of MW/m2 in a short pulse (~ms) during disruption in future large scale tokamak imposes great challenge on plasma facing components (PFCs), which is very concerned and worried by the ITER. Currently, understanding the consequence of thermal damage behaviors on W PFC by is an critical issue. EAST, as a superconducting tokamak, is installed metal PFCs, i.e., W PFCs for divertor and limiter, and TZM tiles for first wall, which have the similar castellated structure with ITER. After recent plasma campaigns, various damages in form of cracking and melting were post mortem inspected on divertor dome and baffle plates, first wall and limiter, which are finally distinguished to be induced by transient heat flux. Such phenomena were successfully caught by real time monitored CCD and IR cameras in plasma start-up phase of each plasma campaign, and thus these damages were generally attributed to the runaway electron(~10MeV) loss during plasma disruption. And, the circular configuration plasma tended to hit the PFCs closest to plasma if disruption occurred. For the castellated PFC structures, the melting and cracking damages often occur at the leading edges with obvious misalignment (on divertor and first wall) or the protruded parts (on limiters). In view point of melting, there were always three layers of grain from the surface to the deep region, namely columnar grain, equiaxed grain (recrystallization region) and original grain in the melting regain. The depth of columnar grain is normally 100-300µm and the depth of recrystallization region is about the same magnitude. Such grain distribution indicates steep temperature distribution from surface to the deep region when melting evens occurred. All the melting PFCs’ surface morphology was similar, undulated with melting waves. The slight migration of melting layer can be observed along the toroidal direction with the in-situ melting PFCs shown here. The direction of plasma pressure and Marangoni flow was along the toroidal direction, which means they might be the dominant forces here. And, the influence of J×B force might not be obvious since the limited melting pool life which would result in the limited acceleration time and expected bulk melt displacements even J×B force was the dominant force. Meanwhile, both macrocracks and microcracks are observed on melting region on tungsten surface in some times. Moreover, macrocracks can be even observed in the region which is far away from the melting zone. There are some columnar grain even exfoliated from the material indicating severe cracking. In addition, on the leading edges without melting, visible dense cracks can be also found. Such transient heat flux induced melting and cracking by runaway electron loss during disruption in EAST provide important reference for ITER.

Speaker: Mr Dahuan Zhu (Institute of Plasma Physics，Chinese Academy of Sciences)

IAEA- 3rd Disrutpion confernece France-2024-09-05V2.pdf

16:20 Coffee

Consequences

Convener: Fabio Villone (DIETI, Università degli Studi di Napoli Federico II)

55 Electric field effects during disruptions}

During tokamak disruptions, the magnetic surfaces are broken creating large regions of chaotic magnetic field lines. The physics associated with post-disruption chaotic magnetic fields needs be understood to address force, heat, and runaway electron loading on the walls. Direct simulations are too challenging to allow parameter scans and have uncertainties that can only be addressed by a reliable physics understanding. Even an ideal instability that grows on a timescale $\tau_I$ and does not saturate at a small amplitude will lead to a breakup of the magnetic surfaces on a timescale $\sim 10\tau_I$. The ratio of the closest to the average separation between two neighboring magnetic surfaces drops until resistive diffusion across the locations of their closest approach competes with $\tau_I$. As surfaces break, regions of chaos are created. With chaos, each magnetic field line will have neighboring lines that exponentially separate from it with the distance along the line. When followed long enough, a single line will come arbitrarily close to every point in a single chaotic region. The annuli of magnetic surfaces between chaotic regions break by forming Cantori, which are toroidal surfaces punctured by pairs of inward and outward tubes of magnetic flux called turnstiles. In each chaotic region, the parallel current density divided by $B$ relaxes toward a spatial constant by shear Alfv\'en waves. The electric potential $\Phi_q$ required for quasi-neutrality produces both a diffusion coefficient that is Bohm-like, $D_q\approx T_e/eB$, and a large scale flow $\approx T_e/eB a_T$ across the magnetic field lines, where $a_T$ is the scale of the large scale difference in the electron temperature $T_e$. This diffusion and flow are important for sweeping impurities into the core of a disrupting tokamak plasma. These results follow from from general magnetic-evolution properties and from the separation of the electric field in the plasma into the sum of a divergence-free, $\vec{E}_B$, and a curl-free, $\vec{E}_q$, part. The divergence-free part of $\vec{E}$ determines the evolution of the magnetic field. The curl-free part enforces quasi-neutrality. This separation is given by a Helmholtz decomposition, which is unique if a boundary condition is given on the enclosing chamber wall. A deeper understanding of disruption experiments and simulations will clarify the roles of chaos, Alfv\'en waves, quasi-neutrality potentials, and helicity conservation not only in tokamak disruptions but also in magnetic reconnection in general---whether in the laboratory or in space. For more details, see $<$ https://arxiv.org/pdf/2404.09744 $>$. U.S. Department of Energy grants DE-FG02-03ER54696, DE-SC0018424, and DE-FG02-95ER54333 provided support.

Speaker: Prof. Allen Boozer (Columbia University)

76-Boozer-E-disrupt-T.pdf

56 Fluid and kinetic modeling of runaway electron seed generation during disruptions

Since the RE generation during tokamak disruption is exponentially sensitive to initial plasma current, highly energetic RE beams pose a critical challenge for future tokamaks. Accurate simulations of tokamak disruptions are therefore essential for the development of successful mitigation strategies and safe operation. However, when simulating such disruptions, fluid plasma models are often preferred due to their low numerical cost, even though they generally are less accurate than kinetic models.

We have compared simulations using both fluid and kinetic modeling of the RE seed generation for a diverse set of disruption cases in ITER and SPARC. The kinetic model is simplified by assuming that pitch angle scattering dominates the electron dynamics, enabling the distribution function to be analytically averaged over pitch angle. Furthermore, the distribution function has only been evolved for electrons within the mildly superthermal energy range, while ions, thermal electrons and REs are evolved as fluids. We have considered both non-activated and activated scenarios; for the latter we have derived and implemented kinetic sources for the Compton scattering and tritium beta decay RE generation mechanisms [1] in the simulation tool DREAM [2].

We find that fluid and kinetic disruption simulations of non-activated scenarios can have significantly different RE dynamics, due to an overestimation of the RE seed generation by the fluid model. The primary cause of this is that the fluid hot-tail generation model neglects superthermal electron transport losses during the thermal quench. In the activated scenarios the fluid and kinetic models give more similar predictions, which can be explained by the activated sources' significant influence on the RE dynamics and the seed.

[1] J.R. Martín-Solís et al., Nucl. Fusion. 57 066025 (2017).
[2] M. Hoppe et al., Comp. Phys. Comm. 268 108098 (2021).

Supported in part by Commonwealth Fusion Systems.

Speaker: Ida Ekmark (Department of Physics, Chalmers University of Technology, Gothenburg, Sweden)

IAEA2024.pptx

57 Preliminary Design of Halo Machine

In 1996, during a disruption, 3 hundred tonnes of JET vacuum vessel moved by 7 mm sideways. In spite of significant efforts to understand the phenomena, the horizontal force on the tokamak wall during plasma disruptions still remains poorly understood. For example, the predictions for ITER vary greatly, from 2 to 60 MN, with the upper estimate exceeding the design margin of 48 MN. To resolve this uncertainty, in 2018, the ITPA community organized a dedicated joint experiment MDC25. While significant progress has been made [1 – 3], it has also become apparent that for reproducible studies of the related physics a specialized linear plasma device is needed [4]. Accordingly, here we present the preliminary design of such machine. The main focus of new apparatus is the interaction between halo currents and kink modes during a plasma-wall contact. The currents in plasma facing components (PFCs) will be measured with recently developed probes [5]. The design of PFC arrangement is supported by SPICE3 modelling [6]. In many aspects Halo Machine [4] is similar to RSX from LANL [7], but it has 10 times larger magnetic field (up to 1 T) to mimic the ITER conditions [8]. Apart from disruptions, the parameters of new device are also appropriate to study magnetic reconnection [7] and stability of magneto-dynamic-plasma thrusters [9]. The synergy between these closely interconnected research topics will be exploited in the Halo Machine.

References:
[1] V. V. Yanovskiy et al., “Magnetic Measurements of Disruption Forces on COMPASS”, 49th EPS, O2.104 (2023)
[2] V. V. Yanovskiy et al., Nucl. Fusion 62 086001 (2022)
[3] M. Hron et al., Nucl. Fusion 62 042021 (2022)
[4] V. V. Yanovskiy et al., “Conceptual Design of Halo Machine”, 20th EFTC, O.15 (2023)
[5] F. Villone et al., “Design and experimental validation of an eddy currents probe”, 44th EPS Conference on Plasma Physics (Belfast, Northern Ireland), P1.106 (2017)
[6] M. Komm et al., Plasma Phys. Control. Fusion 55 025006 (2013)
[7] T. P. Intrator et al., Nature Physics 5 521 (2009)
[8] F. J. Artola et al., Nucl. Fusion 62 056023 (2022)
[9] M. Zuin et al., Phys. Rev. Lett. 92 225003 (2004)

Speaker: Vadim Yanovskiy (Institute of Plasma Physics of the Czech Academy of Sciences)

Preliminary Design of Halo Machine.pdf

Friday 6 September

Consequences

58 Practical model for AVDE loads calculations

The talk begins with an overview of several consequent steps in evaluation of the pulsed EM loads in tokamaks and concentrates on calculation of AVDE-induced loads in ITER. The presented practical EM model uses the superposition of two patterns of the halo current: one perfectly symmetric and another perfectly anti-symmetric. It combines the following features of two recent trial models: (a) helically distorted halo layer wrapping around core plasma, and (b) halo-to-wall interception belts slipping along plasma-facing walls. This combination, not tested before, almost doubles the lateral net forces in comparison with ones found with trial models. Any AVDE creates not only lateral net forces but also significant lateral net moments. Being negligible at VDEs, these moments become dominant at AVDEs. The model carefully compensates any numerically accumulated imbalance of net EM forces and moments between the VV and the Magnets (zero total for the tokamak), as a necessary condition for the consequent simulation of tokamak’s dynamic response and for the testing of tokamak monitoring algorithms and simulators. To decouple from the current uncertainties in the interpretation, prediction, and numerical simulation of AVDE physics, the model does not attempt to simulate plasma evolution with AVDE distortion but takes it as input assumption based on the existing interpretations of AVDE physics. This means the model is to be used in a manner of parametric study, at widely varied input assumptions on AVDE evolution and severity. Parametric results will gradually fill a library of ready-for-use waveforms of asymmetric loads (distributed and total), and then the physics community may point out specific cases for subsequent engineering analysis. This talk shows the first practical contribution to the AVDE library.

Speaker: Sergei Sadakov (ITER IO (Emeritus Fellow))

AVDE_loads_Sadakov_r.05+.pdf

AVDE_loads_Sadakov_r05++.pdf

AVDE_loads_Sadakov_r.05++.pptx

AVDE_loads_Sadakov_r.05+.pptx

AVDE_VV_motion.gif

IAEA_AVDE_loads_Sadakov.pdf

IAEA_AVDE_TALK_Sadakov_r.05.pptx

59 Validating Hot and Cold VDEs in C-Mod with M3D-C1

Electromagnetic (and thermal) loads during vertical displacement events (VDEs) are of major concerns in tokamaks, and in future pilot plants that are based on this concept. In particular, Alcator C-Mod has been used to study disruptions from planned as well as unexpected VDEs, and where halo currents were well analyzed in the disrupting plasma [1,2]. Therefore, it offers comprehensive hot and cold VDE data that could be used to validate numerical codes. On the modeling side, M3D-C1 [3] is becoming a valuable tool to help inform SPARC and ARC physics/design. Therefore, validation of M3D-C1 with C-Mod VDEs data is critical to improve our predictive modeling capabilities. In this work, we present our efforts in modeling hot and cold VDEs with M3D-C1 on different C-Mod discharges. We generated different C-Mod models, including the massive dome and cylinder structures that surrounds the vacuum vessel, as well as ports. The figure below shows on the left panel an example of the M3D-C1 mesh model. Using the actual wall toroidal resistivities, we observe trends that qualitatively agree with experiments and that the massive dome and cylinder do not seem to play a role in the VDE dynamics. We also employed different assumptions on the poloidal wall resistivities scanning over different parameters of interest to better understand halo current magnitude and general VDE behavior on C-Mod. An example of simulated halo current patterns is shown in the right panel of the figure.

Acknowledgements
Work supported by Commonwealth Fusion Systems. The authors are also grateful to the support from J. Chen (PPPL) and E. S. Seol (RPI) on the installation of M3D-C1 in the new MIT cluster used for this work.

References

1. R. Granetz et al., Nucl. Fusion 36 (1996) 545
2. R.A. Tinguely et al., Nucl. Fusion 58 (2018) 016005
3. S. Jardin et al., Comput. Sci. Disc. 5 (2012) 014002

Speaker: Cesar Clauser (Massachusetts Institute of Technology)

Clauser-IAEA-TM-Disruption_v3.pdf

60 Prediction and validation of disruption-induced eddy currents and forces within engineering design cycles using ThinCurr and TokaMaker

In tokamaks, eddy currents and associated forces, driven by rapid current quenches during disruptions are important drivers for structural engineering requirements. Additionally, recent interest in disruption-driven 3D currents, such as the Runaway Electron Mitigation Coil (REMC) concept, further motivates the need to capture currents in passive conducting structures early in and throughout the design process. To support this the ThinCurr [1] and TokaMaker [2] tools, part of the broader open source Open FUSION Toolkit [3], are being developed, validated, and applied to provide analysis of such currents and their effects as part of the engineering design cycle for future devices (eg. SPARC, ARC, NT, etc.). TokaMaker is a time-dependent Grad-Shafranov tool that can be used to model current quenches, Vertical Displacement Events (VDE) and other relevant events with accurate (2D) vessel geometry, including thick-wall effects, and the effect of control systems, realistic power supplies, and other relevant features. Currents, forces and other impacts can be directly evaluated in TokaMaker or used as input to ThinCurr, a fully 3D thin-wall electromagnetic modeling code, for more accurate simulations and assessment of 3D effects from asymmetries in the vessel and plasma. The use of Hierarchical Off-Diagonal Low-Rank (HODLR) approximation within ThinCurr enables scalability to large models that can capture complete devices, including both large (eg. VV) and small (eg. first wall tiles) features in a single model. Both tools utilize unstructured mesh approaches, common in commercial analysis software, that enable tight coupling to design and engineering workflows (eg. directly from CAD). This talk will present application of ThinCurr and TokaMaker to the prediction of eddy currents and forces in present and future devices as well as plans for, and results from, validation using a newly-installed REMC coil on the HBT-EP tokamak [4] at Columbia University.

[1] A. Battey et al., Nucl. Fusion 64 016010 (2024)
[2] C. Hansen et al., Comput. Phys. Commun. 298 109111 (2017)
[3] https://github.com/hansec/OpenFUSIONToolkit
[4] D. Maurer et al., Plasma Phys. Control. Fusion 53 074016 (2011)

Work supported by US DOE awards DE-SC0019239, DE-SC0021325, DE-FG02-86ER53222, and DE-SC0022270, Commonwealth Fusion Systems, and Next Step Fusion

Speaker: Chris Hansen (Columbia University)

CJH-IAEA_disruption-9_6_24.pptx

10:30 Coffee

Consequences

61 Characteristics of the thermal-quench process in the EAST disruptions and its interpretive MHD modelling with JOREK

During the thermal quench (TQ), the stored thermal energy is released with a short timescale and might cause serious damage to plasma-facing components (PFCs), especially in future large-scale tokamaks. Here presents the detailed description of the TQ database consisting of 164 disruption discharges, including both major disruptions (MDs) and hot vertical displacement events (VDEs), on EAST [1]. The dependence of the TQ time on plasma parameters has been statistically analysed. Besides, the TQ process triggered by neon massive gas injection (MGI) in EAST is simulated with the JOREK 3D non-linear MHD code [2].
On EAST, the TQ duration of MDs is within 60~800 μs, and the value of VDEs is approximately in the range of 100~3000 μs. In particular, for MDs, the lower bound of TQ duration decreases as the plasma current increases. This decrease is due to the connection length shortening and the plasma temperature increasing. For MDs, two typical TQ processes, single-stage TQ and double-stage TQ, are characterized by different magnetic perturbations. For hot VDEs, the plasma temperature collapses step by step from the edge to the core, and every progressive collapse corresponds to a magnetic perturbation, the growth rate of which is approximately equal to or less than double-stage TQ.
First simulations of neon MGI into an EAST L-mode plasma with the JOREK MHD code are presented and compared in detail to experimental data. The effect of several parameters on MHD activity and TQ dynamics is studied and MHD influence on ablation is shown. The MGI creates a local density deposit that rapidly expands in the direction parallel to the magnetic field. TQs are obtained quickly after injection in most simulations with a typical duration of 3 ms. Although the n = 1 magnetic perturbation dominates in the simulations, toroidal harmonics up to n = 5 contribute to stochastization and stochastic transport in the plasma core. With 1020 atoms injected, TQ is typically incomplete. At a larger number of injected atoms, TQ can set in when the local density deposit is close to the q = 2 rational surface and substantial (up to 90%) ‘thermal collapse’ energy is lost during TQ.
[1] W Xia et al Plasma Phys. Control. Fusion 65 (2023) 085011;
[2] D Hu et al Nucl. Fusion 61 (2021) 026015

*See Hoelzl et al 2021 (https://doi.org/10.1088/1741-4326/abf99f) for the JOREK Team.

Speaker: Dr Long Zeng (Tsinghua University)

Characteristics of the thermal-quench process in the EAST disruptions.pdf

62 Cons Discussion

118-Consequences_discussion_v03.pptx

General Discussion & Closure

Convener: Stefan Jachmich (ITER Organization)

99-Jachmich-IAEA-TM2024-Closing.pdf

12:40 Lunch (optional)