1/1318 th PSI conference – Toledo, May 2008P. Tamain Association EURATOM-CEA 3D modelling of edge...
-
Upload
mervyn-sharp -
Category
Documents
-
view
214 -
download
0
Transcript of 1/1318 th PSI conference – Toledo, May 2008P. Tamain Association EURATOM-CEA 3D modelling of edge...
1/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
3D modelling of edge 3D modelling of edge parallel flow asymmetriesparallel flow asymmetries
P. Tamainab, Ph. Ghendriha, E. Tsitronea, Y. Sarazina, X. Garbeta, V. Grandgirarda, J. Gunna, E. Serrec,
G. Ciraoloc, G. Chiavassac
aAssociation Euratom-CEA, CEA Cadarache, FrancebEuratom-UKAEA Fusion Association, Culham Science Centre, UK
cUniversity of Provence/CNRS, France
2/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Strong poloidal asymmetries in parallel Mach Strong poloidal asymmetries in parallel Mach number measured on all machinesnumber measured on all machines
Expected situation:
symmetrical parallel flow
parallel Mach number at the top: M = u// /cs ~ 0
Experimental results:
asymmetrical flow
M~0.5 at the top universal phenomenon
(X-point and limiter)
[Asakura, JNM (2007)]
expected M measured M
[courtesy J. Gunn]
M~0
M=+1 M=-1
3/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Hard to reproduce with transport codes Hard to reproduce with transport codes without any ad-hoc hypothesiswithout any ad-hoc hypothesis
[Zagorski et al., 2007]
DLFS / D
HFS ~ 200
Modelling attempts with 2D transport codes [Erents, Pitts et al., 2004]
[Gunn & al., JNM (2007)]
M=+1 M=-1
M~0.5
HFS LFS
ad-hoc strong ballooning of radial transport necessary to recover experimental amplitudes:
evidence for influence of drifts (ExB and diamagnetic)
4/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
TOKAM-3D: a new numerical tool able to TOKAM-3D: a new numerical tool able to tackle consistently this kind of issuetackle consistently this kind of issue
centre
edgeSOL
limiter
simulated region
3D - full torus
closed + open field lines
3D fluid drift equations [B. Scott, IPP 5/92, 2001]
density, potential, parallel current and Mach number M Bohm boundary conditions in the SOL flux driven, no scale separation => Turbulence + Transport
5/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
3D drift fluid equations3D drift fluid equations
continuity
charge
parallel momentum
parallel current
vorticity definition
ExB advection
curvature
parallel dynamics
diffusive transport
6/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
““Neoclassical” vs turbulent regime Neoclassical” vs turbulent regime
D/DBohm ~ *
t
1
8.10-2
high D: “neoclassical” equilibrium with drifts => only large scale effects
low D: turbulent regime => impact of small scales
7/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
neoclassical equilibrium with ExB and diamagnetic drifts
non-zero Mach number at the top: M~0.25
uniform D not linked to poloidal distribution of radial transport
LFSHFS
Parallel Mach number M
M=0
M>0
M<0
0.5
0
-0.5
1
-1
Neoclassical regime: growth of poloidal Neoclassical regime: growth of poloidal asymmetries even without turbulenceasymmetries even without turbulence
mechanism: combination of global ExB drift and curvature
8/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Par
alle
l Mac
h nu
mbe
r
θ
1
0
LFSHFS Top
-1P
aral
lel M
ach
num
ber
0.25
0
Minor radius r/ρL
10040 160
LCFS
Is that enough to recover experimental Is that enough to recover experimental results?results?
larger amplitude than that found in previous studies but still lower than experiments
radial extension in the SOL not deep enough
The answer seems to be NO: there must be something else…
θ = π/2 (top)
r/a = 1.07
9/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Edge turbulent transport can generate large Edge turbulent transport can generate large amplitude poloidal asymmetriesamplitude poloidal asymmetries
low diffusion: D /DBohm=0.02
=> turbulent transport
poloidal up/down (m=1,n=0) asymmetry in spite of strongly aligned structures
M<r
turb >t,,
<rdiff
>t,,
r0
1
Density n no limiter => previous large scale drifts effect not included
10/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
0
0.4
-0.4
HFS LFS
90% of the flux at the LFS90% of the flux at the LFS
Mach number at the top: Mtop~0.35
LFS r / tot r ~ 0.9
M
<r>
HFS LFS
asymmetry due to ballooned transport:
90% of total flux
11/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
turbulent regime in closed + open flux surfaces
filament-like structures generated in the vicinity of the LCFS
and propagate in the SOL
How do these two mechanisms overlap in a How do these two mechanisms overlap in a complete edge simulation?complete edge simulation?
N
HFS LFS
M
HFS LFS
12/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
experimental large amplitudes Mtop~0.5 recovered even in far SOL
The superposition of both mechanisms The superposition of both mechanisms allows a recovery of experimental featuresallows a recovery of experimental features
good qualitative and quantitative agreement with experimental data
13/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
SUMMARY / CONCLUSIONSUMMARY / CONCLUSION
Parallel flows poloidal asymmetries: confirmation of the existence of 2 distinct mechanisms:
- large scale drifts => coupling between ExB, curvature and the limiter
- ballooning of turbulent radial transport => coupling between turbulent scales and curvature
superposition of both mechanisms leads to good agreement with experimental data without requiring ad-hoc hypotheses
TOKAM-3D: new generation of edge codes: 3D transport & turbulence across LCFS success of fully consistent approach on-going development for model and code improvements
14/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
The TOKAM-3D ModelThe TOKAM-3D Model
Fluid modeling with no scale separation 3D fluid drift equations : matter, charge, parallel momentum,
parallel current (generalized Ohm’s law) [Scott, IPP 5/92, 2001]
isothermal closure in current version
periodic
buffer zone
buffer zone
15/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
3D drift fluid equations3D drift fluid equations
continuity
charge
parallel momentum
parallel current
vorticity definition
ExB advection
curvature
parallel dynamics
diffusive transport
16/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Geometrical dependancesGeometrical dependances
3 identical cases: limiter poloidal shift and B reversal (B drift sign)
Impact of field inversion and of limiter position
Bottom, normal B
LFSHFS
M=0
(BxB)ions
Equatorial, normal B
LFSHFS
M=0
(BxB)ions
Bottom, reverse B
LFSHFS
M=0(BxB)ions
17/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
‘‘Return parallel flow’ mechanism driven by Return parallel flow’ mechanism driven by ExB and curvatureExB and curvature
r
LCFS
step 1: establishment of large poloidal ExB drift at LCFS
LFSHFS
step 2: curvature effects trigger symmetric inhomogeneities
step 3: ExB drift advects the density and breaks the symmetry
M//ExB
N
top LFS
N ExB
LFStop
18/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Origin of the Mach number asymmetryOrigin of the Mach number asymmetry
ExB drift + curvature plays a major role step 1: establishment of large poloidal ExB drift at LCFS
Bottom, normal B
Bottom, reverse B Equatorial, normal B
r r r
LCFS
r sign changes => drift direction does not change with field direction
19/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Origin of the Mach number asymmetryOrigin of the Mach number asymmetry
ExB drift + curvature plays a major role
Bottom, normal B Bottom, reverse B Equatorial, normal B
N N N
top LFS
step 1: establishment of large poloidal ExB drift at LCFS step 2: curvature effects trigger symmetric inhomogeneities
No curvature effect at equatorial plane
20/1318th PSI conference – Toledo, May 2008 P. TamainAssociationEURATOM-CEA
Origin of the Mach number asymmetryOrigin of the Mach number asymmetry
ExB drift + curvature plays a major role
Bottom, normal B Bottom, reverse B Equatorial, normal B
N N N
top LFS
step 1: establishment of large poloidal ExB drift at LCFS step 2: curvature effects trigger symmetric inhomogeneities step 3: ExB drift advects the density and breaks the symmetry
ExBExB