ELM transport in the JET scrape-off layer

1
ELM transport in the JET scrape-off layer
R. A. Pitts, P. Andrew, G. Arnoux, T.Eich, W.
Fundamenski, E. Gauthier, A. Huber, S. Jachmich,
C. Silva, D. Tskhakaya and JET EFDA Contributors
18 October 2006
2
OUTLINE

• ELM divertor energy asymmetries
• ELM filamentary structure
• Modelling the ELM transport
• Particle-in-cell (PIC) simulations
• Transient modelling of ELM filament parallel
  losses
• Main wall particle energies
• Main wall power deposition
• Conclusions

3
Brief diagnostic overview
Wide angle main chamber IR
Fast reciprocating probes TTP, RFA
Diagnostic Optimised Configuration (DOC)
Divertor IR and tile thermocouples
4
Divertor target ELM energy asymmetry (I)

• ELM resolved target heat flux (IR)
• Type I ELM energy deposition strongly favours
  INNER target for FWD-Bj
• For REV-B, some evidence for more balanced
  deposition,
• Consistent with similar analysis from AUG (WELM lt
  20 kJ) and linked to passage of net current
  through target plates

• Favourable trend for ITER target power loading
  (since always more energy to OUTER target
  inter-ELM)

T. Eich et al., PSI 2006
5
ELM filaments main chamber IR

• Filamentary power deposition detected with new
  wide angle IR
• 100 Hz frame-rate, but 300 ms snapshot ? catches
  an occasional ELM
• Seen by substracting pre-ELM and ELM frames

P. Andrew, G. Arnoux
Ip 2 MA, Bj 3TWELM 150 kJ
Two discharges with different contact point of
first limiting flux surface
Coord. Transformation (x,y) ? (q,j)
6
ELM filaments in the far SOL
TTP
r - rsep 80 mm at the probe

• Clear filamentary structure in the particle flux,
  Te and radial velocity
• WELM 100 kJ
• Te (pedestal) 500 eV
• TeELM(limiter) 30 eV
• vrELM 500 ? 1000 ms-1
• Electrons cool rapidly in the filament as it
  crosses the SOL
• ELM duration at the probe 10x higher than tELM
  seen on MHD activity etc.

C. Silva et al., J. Nucl. Mater. 337-339 (2005)
722
7
Modelling the ELM transient
Losses along B
WALL
Present understanding MHD perturbs pedestal ?
radial expulsion of plasma ? parallel loss along
field lines to divertor until filament hits wall
Two separate approaches being followed at JET to
modelling the 1D SOL parallel transport.
Particle-in-Cell (PIC) simulations CPU
intensiveInject ELM energy kinetically via
particle source at Tped, nped for time tELM and
follow particles to targets including full target
sheath dynamics
Transient model Fluid and kinetic
versions.Simpler to solve, captures many effects
of PIC simulationsIntroduces 2D nature of
filament propagation by relating loss times to
radial velocities
8
PIC simulations of parallel losses

• More realistic description of the ELMy JET SOL
  using improved PIC simulations (BIT1 code)
• Scan in Tped, nped to vary WELM
• Most of the heat flux arrives with ions on the
  acoustic timescale
• BUT, only 30 of ELM energy deposited when
  qtarget peaks
• Electrons account for 30 of target energy
  deposition
• Strong transient increase over Maxwellian
  sheath transmission factors during the ELM
• Fluid code assumption of fixed g underestimates
  qtarget at high WELM

Example Tped 1.5 keV, nped 1.5x1019m-3WELM
120 kJ, tELM 200 ms
D. Tskhakaya
9
Transient model of ELM parallel losses

• Key elements of model
• Temporal evolution of n, Te and Ti in the
  filament frame of reference
• Time and radius related by filament propagation
  velocity
• Parallel loss treated as conductive and
  convective removal times
• Radial expansion included
• Filament cools faster than it dilutes, electrons
  cooled more rapidly than ions? in the far
  SOL,Ti gt Te in the filament at wall impact

Example with Ti,ped Te,ped 400 eV nped
1.5x1019m-3, H ions
W. Fundamenski, Plasma Phys. Control. Fusion 48
(2006) 109
10
Model consistent with RFA hot ion data
RFA
r - rsep 80 mm at the probe

• Filaments on plasma and hot ion fluxes
• WELM 50 kJ,Ti,ped 400 eV
• Lower ion energy in successive filaments
• Net flow to inboard side!

? ELM enters SOL mainly on the outboard side
Current of ions with energy gt 400 eV

• Good agreement with transient model for i-side
  peak fluxes
• Predicts Ti,RFA/Ti,ped 0.3?0.5
• Te,RFA/Te,ped 0.13?0.25
• ne,RFA/ne,ped 0.3?0.4
• Consistent with low Te on TTP probe

R. A. Pitts et al., Nucl. Fusion 46 (2006) 82W.
Fundamenski, PPCF 48 (2006) 109
11
ELM-wall power loads

• Fraction of ELM energy in the divertor decreases
  with increasing ELM size
• Up to 60 missing from divertor at high WELM

• Dedicated plasma-wall gap expts. give far SOL
  power widths of lW,ELM 35 mm for WELM/Wped
  12
• Agrees well with transient model prediction
• Use this lW,ELM as reference for empirical
  scaling lW,ELM ? 35(WELM/0.12Wped)1/2
• Factor 1/2 consistent with recent ELM amplitude
  scaling due to interchange motion

WELM,wall?WELMexp(-D/lW,ELM)f 1 -
WELM,wall/WELM
T. Eich et al., subm. to Plasma Phys. Control.
FusionW. Fundamenski et al. PSI 2006O. E.
Garcia et al., Phys. Plasmas 13 (2006) 082309
12
CONCLUSIONS

• Significant progress at JET in the measurement
  and modelling of ELM SOL transport
• Strong asymmetry in divertor Type I ELM energy
  deposition favouring inner target
• ELM filaments seen on several diagnostics
• Sophisticated 1D PIC modelling now providing
  scalings of target heat flux with ELM energy
• Available data in good agreement with new
  transient parallel energy loss model
• Implies that filaments detached from pedestal
  plasma
• ELM ions can reach limiters with high energies
• See poster by A. Loarte (IT/P1-14) for more
  applications of the transient model to ITER wall
  power loads