Three-dimensional plasma transport in open chaotic magnetic fields [Elektronische Ressource] : a computational assessment for tokamak edge layers / vorgelegt von Heinke Gerd Frerichs. [Forschungszentrum Jülich, Energieforschung (IEF), Plasma Physics (IEF-4)]

Sujets

• Physik

Informations

• Publié par: rheinisch-westfalischen_technischen_hochschule_-rwth-_aachen
• Publié le: 01 janvier 2010
• Nombre de lectures: 12
• Langue: English
• Poids de l'ouvrage: 8 Mo

Extrait

Jül - 4321
Institute for Energy Research (IEF)
Plasma Physics (IEF-4)
Three-dimensional plasma transport
in open chaotic magnetic fields:
A computational assessment for
t okamak edge layers
Frerichs, Heinke Gerd
Mitglied der Helmholtz-GemeinschaftThree-dimensional plasma transport in open
chaotic magnetic elds: A computational
assessment for tokamak edge layers
VonderFakult atfur Mathematik,InformatikundNaturwissenschaften
derRWTHAachenUniversityzurErlangungdesakademischenGrades
eines Doktors der Naturwissenschaften genehmigte Dissertation
vorgelegt von
Dipl.-Phys. Heinke Gerd Frerichs
aus Julic h
Berichter: Univ.-Prof. Dr.rer.nat. Hans-J org Kull Detlev Reiter
Tag der mundlic hen Prufu ng: 24.02.2010
Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verfugbar.The major part of this thesis was carried out as member of the German Research
School for Simulation Sciences, a joint venture of RWTH Aachen University and
Forschungszentrum Julic h.Three-dimensional plasma transport
in open chaotic magnetic fields:
A computational assessment for
t okamak edge layers
Frerichs, Heinke GerdBerichte des Forschungszentrums Jülich; 4321
ISSN 0944-2952
Institut für Energieforschung (IEF)
Plasma Physics (IEF-4)
Jül-4321
D 82 (Diss., RWTH Aachen, Univ., 2010)
Vollständig frei verfügbar im Internet auf dem Jülicher Open Access Server (JUWEL)
unter http://www.fz-juelich.de/zb/juwel
Zu beziehen durch: Forschungszentrum Jülich GmbH · Zentralbibliothek, Verlag
D-52425 Jülich · Bundesrepublik Deutschland
Z 02461 61-5220 · Telefax: 02461 61-6103 · e-mail: [e-mail]
The development of nuclear fusion as an alternative energy source requires the research
on magnetically con ned, high temperature plasmas. In particular, the quanti cation
of plasma ows in the domain near exposed material surfaces of the plasma container
by computer simulations is of key importance, both for guiding interpretation of present
fusion experiments and for aiding the ongoing design activities for large future devices
such as ITER, W7-X or the DEMO reactor. There is a large number of computational
issues related to the physics of hot, fully ionized and magnetized plasmas near surfaces of
the vacuum chamber. This thesis is dedicated to one particular such challenge, namely
the numerical quanti cation of self-consistent kinetic neutral gas and plasma uid ows
in very complex 3D (partially chaotic) magnetic elds, in the absence of any common
symmetries for plasma and neutral gas dynamics.
In high-con nement (H-mode) plasmas, which are presently envisaged for ITER, an in-
stability occurs naturally at the plasma edge, leading to high transient heat and particle
loads to the rst wall and the divertor targets. Extrapolations to ITER relevant power
levels indicate that this instability, the so called edge localized mode (ELM), will lead
to a signi cant reduction of the ITER wall lifetime and additional impurity production
reducing the plasma performance. A promising technique to control ELMs is the appli-
cation of resonant magnetic perturbations (RMPs) at the plasma edge, as successfully
demonstrated at the DIII-D and JET tokamaks and recently under consideration for
ELM mitigation in ITER. Modeling of the perturbed magnetic eld structure and com-
parison to experimental observations suggests the formation of an open chaotic system
at the plasma edge with remnant magnetic island chains, chaotic domains and short
magnetic ux tubes. However, quantifying the impact of such a complex magnetic eld
structure on the edge plasma is still the subject of ongoing research. This is the topic
of the present thesis - and it requires at least 3D numerical transport models.
Distinct from conventional Eulerian 2D uid solvers, applied routinely to magnetic fusion
edge plasma studies, complex 3D magnetic topologies are currently treated by the geo-
metrically more exible Lagrangian schemes, supplemented by Monte Carlo procedures
for higher order derivatives (dissipative terms due to di usion processes) and sources.
These particle based algorithms are combined with a eld line reconstruction technique
for dealing with partially chaotic magnetic elds, involving eld aligned grids. A gener-
alization from regular-structured to block-structured grids is carried out in this thesis,
which greatly enhances the applicability range of present 3D fusion plasma edge codes,
in particular also to poloidally magnetic diverted con gurations, such as DIII-D and
ITER.
The EMC3-EIRENE code is such a 3D edge plasma and neutral particle transport code
and is applied in this thesis to two distinct con gurations of open chaotic magnetic
edge layers: at the TEXTOR and DIII-D tokamaks. The DIII-D tokamak has been
chosen because of its recent progress in ELM mitigation by application of RMPs. The
iadvancements of the code presented in this thesis have allowed for the rst time 3D
self-consistent plasma and neutral gas transport simulations for RMP scenarios at DIII-
D with ITER similar plasma shape. A strong 3D e ect of RMPs on the edge plasma
is found and analyzed in detail. It is found that a pronounced striation pattern of
target particle and heat uxes at DIII-D can only be obtained up to a certain upper
limiting level of anomalous cross- eld transport. Hence, in comparison to experimental
data, these ndings allow to narrow down the range of this model parameter. This
constraint to cross eld transport levels is more stringent in DIII-D than for TEXTOR
simulations, because of the shorter wall-to-wall connection length of magnetic eld lines
(and hence already geometrically reduced role of cross eld di usion) in the latter.
Furthermore, improvements of the edge transport model, such as the implementation
of an edge transport barrier, are carried out and discussed in order to allow realistic
simulations of H-mode plasmas.
Of particular interest is also the reduction of steady state heat uxes, even in absence of
the transient ones caused by ELMs. A favorable regime is the so called detached divertor
operation, in which plasma exhausts (particle and heat uxes towards the divertor tar-
get) are extinguished in a neutral gas cloud, resulting in low plasma power and ion uxes
to the material surfaces bounding the system. At TEXTOR, the achievement of a similar
favorable regime with a 3D perturbed boundary is studied in the RMP induced helical
divertor con guration. It is shown in this thesis that plasma states, which are both
consistent with the limited experimental data and show transition to reduced particle
and heat uxes, can indeed be observed in numerical simulations of the TEXTOR helical
divertor. This is, however, at the expense of also reduced upstream temperatures, which
might not be consistent with an advanced tokamak operation (in which the hottest possi-
ble con ned plasma periphery should be combined with the coldest possible plasma-wall
contact zone). The mechanism for the observed behavior is found to be a combination
of strong cross- eld diusion and consequently a damping of parallel counter- ows and
a sudden increase in momentum losses due to charge exchange with atomic hydrogen.
These ndings are similar to results from simulations for the W7-AS stellarator. The
presence of impurities in the plasma (released wall material due to plasma-wall inter-
action) is found to have a signi cant impact on the plasma edge. They introduce an
additional cooling of electrons by line radiation, which a ects plasma-neutral particle
interaction as well and eventually results in a strong reduction of the target ion ux.
The presented improvements of the EMC3-EIRENE code and the subsequent plasma
transport studies at TEXTOR and DIII-D allow future extrapolations to ITER relevant
scales. In particular the ongoing development of high-performance computers facilitates
the direct application of the EMC3-EIRENE code to RMP ELM-control scenarios at
ITER in the near future, at least from a point of view of the magnetic con guration
capabilities of the code and edge transport model, including transport barrier descrip-
tions. One critical remaining computational challenge is, however, the wall-clock time
of present supercomputers, when scaling CPU performance of the code from present
applications up to ITER size con gurations. This may require structural modi cations
of the EMC3 part of the code.
iiContents
1 Introduction 1
2 Chaotic magnetic edge layers in tokamaks 5
2.1 The concepts of magnetic con nement and plasma exhaust . . . . . . . . 5
2.1.1 The limiter con guration . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 The poloidal divertor con guration . . . . . . . . . . . . . . . . . . 8
2.1.3 Chaotic magnetic edge layers . . . . . . . . . . . . . . . . . . . . . 9
2.2 Experimental setup of resonant magnetic perturbations . . . . . . . . . . 10
2.2.1 The DIII-D tokamak . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.2 The TEXTOR tokamak . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Modeling of the magnetic eld structure . . . . . . . . . . . . . . . . . . . 12
2.3.1 Magnetic eld in vacuum approach . . . . . . . . . . . . . . . . . . 13
2.3.2 Field line tracing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4 Visualizat