Experiments on electrostatic turbulence in the TCV edge

1
Experiments on electrostatic turbulence in the
TCV edge
Jan Horacek with help from R.A. Pitts, J. Graves

• Outline
• Introduction to SOL turbulence
• SRAP the diagnostic
• Statistical analysis of density fluctuations
• Radial particle flux
• Study of ELMs
• Parallel flows

Head of Sonde Langmuir RAPide
B-field
4mm
Grand merci a P.Conti, P.Gorgerat, X. Llobet
2
Why ELM and bursty studies on TCV?

• ELMy H-mode is a standard scenario for ITER
• Recent indications that SOL radial power flux is
  too large for ITER first wall which is not
  designed for such high power flux.
• Thats not only during ELMs but also continuously
  by radial propagating plasma bursty events.
• Why such study on TCV? Excellent edge diagnostic
  is available ?

3
Edge turbulence diagnostic in TCV

• Reciprocating Langmuir probe SRAP, very fast
  6-10MHz sampling
• Pins measure
• Te (flt3kHz)
• floating potential Vflf-3Te (plasma potential
  f, temperature Te)
• ion saturation current IsatneTe1/2 (density ne)
• Since fast Te (ETEF project) is still not
  available we must assume dTe/Teltlt 2dne/ne
  3dTe/Teltlt df/f to get plasma physics relevant
  quantities
• Different settings on 5 pins allows also Eq,
  vpol, Gr
• 90 of presented data are from SRAP. Analysis of
  ELMy shot is done in the inter-ELM period only.

Wall shadow
SOL
4
Main plasma parameters

• ohmic L/H-mode, ELMs, density limit, change in
  Btor direction, Ip260-400kA, density ramp

5
DC edge profiles

• Edge Thomson scattering, confirmed SRAP profiles
  of Te, ne.
• Excellent match of absolute values and space
  position
• density ramp discharges gt typically 1st/2nd rcp.
  at 22 and 45-55 of Greenwald density limit,
  respectively

6
SOL plasma is bursty
Plasma in wall shadow SOL LCFS Confined
plasma

• Positive density bursts become dominant (larger
  and less frequent) far away from LCFS.

7
Density fluctuation analysis

• Autocorrelation time ltn(t)n(ttAC)gtt 1/e
  characterizes time scale of turbulent structures
• inside LCFS the PDF of density is Gaussian,
  however, nearby wall not at all.

8
Density fluctuations

• tAC constant inside LCFS but strongly increases
  (longer memory) towards the wall
• High S indication that bursty behavior is more
  important approaching wall. Also shows that
  bursts are not created in the wall shadow.

9
Stochastic modeling of density fluctuations

• In progress with J. Graves gt J.Horacek, PPCF
  2004.
• Statistical approach based on Gamma process in
  analogy with the sandpile model
• Time evolution and PDF of skewed
  Gamma-distribution PGm,s(ne) of no free parameter
  fits well neSOL in any TCV Ohmic discharge

10
Model predictions
11
Poloidal velocity and turbulence dimension on TCV

• Cross-correlation of Isat and Vfl results in the
  same vpol ?
• VSL position confirms well the LCFS (Liuque) wrt.
  absolute SRAP position ?
• Direction change in reversed B-field but not in
  the whole SOL!
• Turbulent structure size lpol-dpol/log(C) to do!

Vfl1
B-field
Vfl4
12
Radial particle flux

• Frequencies fgt300Hz only
• GrvExBn (df1-df4)dne/(Bd14)
• ltGrgt increases with density, not due to different
  PDF (S equal), not due to correlation C but due
  to SOL s(ne) increase
• When structure size is smaller than pin distance
  (dmidlt10mm) the Gr has no more meaning
• Reversed B-field reverses C(Epol,ne) field gt Gr
  stays positive!
• VSL moves outwards when ltnegt increases. Has it
  ever been published?

13
(No Transcript)
14
Average ELM temporal behavior
Da peak defines tELM0

• Conditional average (wrt. Da peak) of SRAP
  density and potential. Very well reproducible
• ne rises quickly and decays slowly its really
  IsatneTe1/2 (Te changes during an ELM)
• Potential starts to oscillate 1ms before its
  time-symmetric

Isat
15
ELM details
Da

• Detail time evolution of Isat during an ELM shows
  a complex structure, explosive growth

Isat (mA)
dmid
16
Post ELM oscillation
600kHz gt -80dB!

• Found post ELM oscillations of 600kHz, visible
  regularly on floating potential (Vfl) but never
  on particle flux (Isat)
• Link to sawtooth ST releases free energy into
  the edge which is later triggered by an ELM that
  starts electrostatic and MHD fluctuations with
  frequencies 14,15,16,x37kHz.
• 600kHz is too fast to be visible by MHD probes ?.
  They see, however, the fundamental 40kHz,
  correlated in time and localized on HFS (in
  contrary to ELM being localized on LFS).

17
Oscillation appears only if a sawtooth anticipates
sound
SRAP floating potential fluctuations
Frequency (kHz)
Time-correlation with 37kHz oscillation on HFS
MHD probe
Help from A.Scarabosio, D.Raju, F.M.Poli
Time (s) during SRAP reciprocation
18
Mach number

• 2003 Mach probe installed on TCV gt
• parallel plasma flow Mach number
  Mv/csClog(Isat,5/Isat,2)
• Radial density (potential) gradients
  (Isat3-Isat5)/dr and velocities

Magnetic field
Isat2
Isat5
Isat3
19
Parallel flow
Density increase

• Deuterium profile is much flatter (and values
  lower) than the Helium one. As density approaches
  density limit, the flow slows down and flattens
  up to the wall (in D)!
• Confirmed classical drifts prediction that M
  decreases with density
• Strong outside midplane wall sink action
• Too high for classical flow mechanisms gt
  ballooning instability plays role
• Prediction of important wall sputtered material
  migration for ITER!

ELMy H-mode
Density increase
To present at PSI-16 gt Horacek, JNM 2005
20
Parallel flows during ELMs

• ELMs do perturb M in the wall shadow only. Why
  not in SOL?

SRAP close to LCFS at wall
at wall shadow
21
ELM poloidal radial velocity

• At LCFS the vpol is strongly accelerated upwards
  (surprise!), however vpol/cs may not change (Te
  fast required)
• vrad shows no significant change during an ELM

Velocity (km/s)
poloidal radial, radial
22
Probable 2D flow

• In total a 2D flow is present, assuming vpol
  corresponds to flow velocity (i.e. turbulence is
  frozen in), together with vcsM
• Inside VSL vpol Bz/Btor csM gt rather
  parallel flow
• Outside VSL vpol ? -vpol gt strong flow ? B

B
23
4-parameter cross-correlation fit
Allows finally structure size l, life time tlife
and velocity v proper calculation

• Assuming correlation in form
• Cezcos(z), where
• z-(t/tlife)2d-vt/l)2H
• tAC-2 tlife-2 (l/v) -2

24
Radial velocity results from oblique turbulent
structures
25
Impossible to derive radial velocity from
cross-correlation

• Radial velocity shear stretches out turbulent
  structures gt oblique gt virtual radial
  propagation
• vradvpol/(dvpol/drtlife).
• 4 space points allows to reconstruct a blob
  described by vpol,vrad, lpol, lrad

26

• Summary
• 2003 1st systematic edge turbulence
  investigation on TCV
• Evaluation of Eq, ne, f, M, vpol, Gr during
  L/H-mode, ELMs, density limit, deuterium/helium,
  change in Btor direction, very high
  time-resolution
• Clear evidence of non-diffusive transport
  intermittent bursts dominate density far away
  from LCFS
• SOL profile flattening of ne, M while
  approaching density limit as well as strong
  increase of radial particle transport
• Found post-ELM-oscillations, its link to MHD and
  saw tooth
• Small effect of ELMs on M and vrad but large on
  vpol
• Outlook
• Discussion on Post-ELM-oscillation is it Alfven
  wave? important?
• Compute radial propagation of ELMs, IPOs, link
  between M and vrad
• Continue Ne and start Gr stochastic modeling with
  J. Graves, apply on other tokamaks
• Experiments in Btorlt0 and with new probe head at
  another poloidal positions
• This talk is at http//crppwww.epfl.ch/horacek/pu
  blications/
• Thank you for your attention ?

27

• The END

28
Density fluctuations
29
Density profile

• For very different shots the mean mltnegt and std
  s2 lt(ne -m)2gt are different, however, the
  relative density fluctuation 1/A is a quite
  universal quantity
• m and s are the only parameters of the ne-PDF
  description used by J. Graves

30
Edge turbulence description I

• SOL electrostatic (fluctuations in ne, Fpl, Te)
  is dominant for turbulent transport (magnetic
  turb. plays minor role)
• ?rne ?rTe provides the free energy to drive the
  turbulence
• Transport on microscopic scale is non-diffusive
  even though on macroscopic scale it looks like
  diffusive GRD?dn/dr gt nn0e-r/l.

31
Edge turbulence description II

• Turbulent structures characterized by
• PDF, (autocorrelation) life-time t
• Poloidal lpol and radial lrad scale lengths
  toroidally very long
• Velocities vpol,vrad,vtor
• Universality SOL turbulence character is rather
  independent on device size magnetic
  configuration tokamaks, stellarators,
  limiter/divertor, e.g. C.Hidalgo, PPCF 2002 1557