Since 18 of December 2019 conferences.iaea.org uses Nucleus credentials. Visit our help pages for information on how to Register and Sign-in using Nucleus.
10-15 May 2021
Nice, France
Europe/Vienna timezone
The Conference will be held virtually from 10-15 May 2021

European effort towards an edge plasma code for reactors

13 May 2021, 14:00
4h 45m
Nice, France

Nice, France

Regular Poster Magnetic Fusion Theory and Modelling P6 Posters 6

Speaker

Dr Patrick Tamain (CEA Cadarache)

Description

This presentation reports on the status of a recently launched European effort towards the development of a novel Boundary plasma code for reactor relevant applications. The targets of the project, its road-map and key physical and technical choices already made are presented. The first milestone of the project, under the form of a prototype code is presented as well as physical applications dedicated to the impact of neutral recycling and divertor closure on 2D interchange turbulence.

Heat and particle exhaust is expected to become an increasingly difficult issue for magnetic fusion devices, already in ITER and even more in DEMO or future reactors. As the machine size and fusion power increase, constraints imposed by engineering limits will reduce drastically operational margins, making the reliability of numerical simulations critical for the successful design of a reactor.
2D mean-field transport codes such as SOLPS-ITER [1] or SOLEDGE2D [2] have been and will remain for upcoming years the main work-horse for divertor design and interpretative studies. They provide an integrated framework to model the edge plasma, including its interaction with the divertor targets, neutrals particles and seeded or sputtered impurities. In the fusion numerical tool box, they have been for long the only tools able to treat self-consistently multi-species plasmas and recycling in realistic magnetic and divertor geometry and thus provide plasma conditions for design studies. Nevertheless, these codes suffer from inherent limits which curb their predictive capabilities, especially for reactor relevant applications. On the physics side, key physics is hidden in free parameters (e.g., turbulent transport in perpendicular transport coefficients or kinetic effects in flux limiters) or simply ignored (e.g., fluctuations-induced non-linearities); on the geometry side, the vast majority of mean-field codes cannot treat the full wall geometry or non-axisymmetric magnetic field such as Resonant Magnetic Perturbations (RMPs) expected to be used for the control of edge MHD instabilities; on the numerical side, the design of these codes is little adapted to the latest evolution of HPCs which prevents making full usage of the available computing power. The latter point becomes extremely critical when considering the numerical effort needed to model large machine such as DEMO.
Significant progress has been made in the last decade regarding going beyond these limits. Turbulence codes have now reached a stage where most can address complex magnetic geometries and/or include neutrals physics. Novel numerical approaches have been developed or applied for the first time in fusion, allowing for example simulations up to the first wall in any geometrical configuration, and HPC compliance and optimization is at the heart of modern code development. The expertise gathered by the edge plasma community now opens the door for the development of a new edge plasma code solving the main limits of existing tools.
Motivated by this observation, a European effort has been launched within the ETASC program in order to develop such tool with DEMO relevant applications in sight for 2025. The project involves expertise from 10 associations and 35 contributors. This contribution reports on the status and perspectives of the project 16 months after its kick-off.
Following a careful review of the state of the art in terms of physics and numerics, we present the road-map of the project identified as the path of optimum reward/risk ratio. On the physics side, the 5-years target of the project is the implementation of a multi-species N-moment electromagnetic gyro-fluid model, which appears as the best compromise to capture key kinetic effects (e.g., Landau damping, parallel flux limits) at tractable numerical cost for DEMO scale simulations. This model will constitute the high end of a hierarchy of models offered in the code, starting from 2D mean-field drift Braginskii and going through 3D drift fluid turbulence models. This will allow users to arbitrate the fidelity / cost ratio of their simulations. Such approach will also be followed for neutrals, with the code initially featuring a simple fluid model but eventually offering also hybrid or full kinetic approaches via a coupling to EIRENE. On the numerics side, constraints imposed by the necessity to capture accurately both the magnetic and the wall geometry led us to opt for using a Discontinuous Galerkin discretization which has shown its capability to treat complex geometries in both turbulent and mean-field applications (Fig. 1).
Such choices raise several open issues, each of which is addressed in the presentation. We in particular report on the progress made towards proposing closures for N-moment gyro-fluid model, showing that kinetic limits of the ITG instability growth rate can be recovered with a finite number of moments. Special focus is also given to the development of relevant sheath boundary conditions for these models via comparisons to PIC codes and full-f gyro-kinetic closures. The impact of boundary conditions in the SOL on the convergence of the moment hierarchy is analyzed. In sight of the modelling of plasmas with species mix including D-T-He plasmas, the implementation of multi-fluid solvers in 2 turbulence codes is also presented. In the 2D miHESEL code [5], we show that a careful numerical implementation of the equations system allows one to preserve required stability and conservation properties. In the SOLEDGE3X code, a full implementation of the Zhdanov closure [6] has been achieved using a specific numerical approach at reasonable numerical cost. In both cases, first application cases to D-T plasmas are presented.
On the numerical side, the efficient treatment of parallel dynamics with generic wall geometry will be discussed with a quantitative comparison between field-aligned and high-order non-aligned approaches in existing edge plasma codes. We will see that advanced schemes can be designed to cope with apparently contradicting demands from turbulence and plasma-wall interaction physics. A specific study is also dedicated to the numerical implementation of elliptic solvers, which have been identified as the main potential bottleneck of the code. A careful comparison of various approaches (including a range of direct and iterative solvers) demonstrates the potential of iterative methods combined with physics based or multi-grid preconditionners.
Illustration of the capability of the Discontinuous Galerkin approach to capture complex geometries for edge plasma modelling. Left: 3D turbulence modelling in diverted geometry (FELTOR code, ref. 3); right: 2D mean-field modelling in WEST geometry with real wall geometry (SOLEDGE-HDG code, ref. 4).
Finally, we present a prototype code as the first development milestone of the project. The code is able to treat 2D fluid interchange turbulence in arbitrary geometry, including neutrals dynamics. First applications are devoted to the impact of neutral recycling and divertor closure on turbulent transport in the edge plasma of tokamaks and demonstrate the capability of the chosen numerical approach to capture complex geometries and scale efficiently to large cases.

[1] S. Wiesen et al., J. Nucl. Mater. 463, 480 (2015)
[2] H. Bufferand et al., Nucl. Fusion 55, 053025 (2015)
[3] https://feltor-dev.github.io
[4] G. Giorgiani et al., J. Comp. Phys. 374, 515 (2018)
[5] A. Poulsen et al., Phys. Plasmas 27, 032305 (2020)
[6] V. Zhdanov, Transport Processes in Multicomponent Plasma, CRC press (2002)

Country or International Organization France
Affiliation CEA-IRFM

Primary authors

Mr Alexander Thrysoe (DTU) Anders Henry Nielsen (Technical University of Denmark, Physics Department) Andreas Stegmeir (Max-Planck-Institute for Plasma Physics) Mr André Coroado (SPC) Mr Baptiste Frei (SPC) Benjamin Dudson (York University) Dr Chris Ridgers (University of York) David Coster (Max Planck Institute for Plasma Physics) David Moulton (CCFE) Dr David Tskhakaya (Institute of Plasma Physics of the Czech Academy of Sciences, Prague, Czech Republic) Dr Derek Harting (Forschungszentrum Jülich GmbH, Institute for Energy and Climate Research Plasma Physics) Dr Eric Serre (M2P2, CNRS, Aix-Marseille Univ., ECM) Dr Fabio Riva (UKAEA/CCFE) Dr Fabio Subba (NEMO Group, Politecnico di Torino, Torino, Italy) Prof. Felix Parra (University of Oxford) Dr Frédéric Schwander (M2P2 - CNRS) Fulvio Militello (Culham Centre for Fusion Energy) GUIDO CIRAOLO (CEA, IRFM) Dr Gilles Fourestey (SCITAS) Dr Giorgio Giorgiani (M2P2, CNRS/Aix-Marseille Univ., ECM) Dr Giovanni Montani (ENEA) Guido Huijsmans (ITER Organization) Hugo Bufferand (CEA) JJ Rasmussen (Department of Physics, Technical University of Denmark) Dr Jeppe Olsen (DTU) John Omotani (Department of Physics, Chalmers University of Technology) Dr Lucian Anton (UKAEA-CCFE) Dr Markus Held (Chalmers University) Prof. Martine Baelmans (KU Leuven) Mathias Groth (Aalto University) Dr Matthias Wiesenberger (DTU) Nakia Carlevaro (ENEA) Dr Nicola Varini (SCITAS) Nicolas Fedorczak (CEA, IRFM, Saint Paul Lez Durance, France) Dr Omar Maj (Max-Planck Institute Für Plasmaphysik) Dr Paolo Innocente (RFX) Prof. Paolo Ricci (Ecole Polytechnique Federale de Lausanne) Dr Patrick Tamain (CEA Cadarache) Philippe Ghendrih (CEA-IRFM) Roman Zagorski (Institute of Plasma Physics and Laser Microfusion) Dr Stefano Carli (KU Leuven) Dr Thomas Body (IPP Garching) Prof. Volker Naulin (DTU Physik) Dr Wayne Arter (UKAEA-CCFE) Wladimir Zholobenko (Max Planck Institute for Plasma Physics) Dr Wouter Dekeyser (KU Leuven) Yannick Marandet (PIIM, CNRS/Aix-Marseille Univ., Marseille, France, EU)

Presentation Materials

There are no materials yet.