Turbulence and transport reduction with innovative plasma shapes in TCV - correlation ECE measurements and gyrokinetic simulations A. Pochelon, and the TCV team 1 Ecole Polytechnique de Lausanne (EPFL) Centre de Recherches en Physique des Plasmas Association EURATOM-Confédération Suisse Lausanne, Switzerland 1 see IAEA FEC 2010 list A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 1
MOTIVATION Why study plasma shapes different from ITER? Shaping: a tool for the test and validation of plasma modeling, in particular transport modeling Usual τ E -scaling laws exhibit no triangularity dependence HFS δ=+0.4 δ=-0.4 LFS Confinement in TCV core plasma is found to improve towards negative triangularity δ < 0 (in L-mode) A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 2
TCV facility TCV: shaping flexibility R = 0.88 m, a = 0.25 m, R/a ~ 3.5 B 1.5T, I p 1MA 16 independent pol. field coils elongation 0.9 < κ < 2.8 triangularity - 0.7 < δ < 1 ECRH: matched by flexible heating 4.5 MW at 2 nd and 3 rd harmonic with 7 independent launchers allowing local power deposition A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 3
Example: H-mode Snowflake divertor plasmas on TCV L SN SF+ Piras PPCF09 Piras PRL10 Piras PPCF10 F.Piras : PP9.136 Wednesday p.m. Snowflake Divertor H-mode A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 4
OUTLINE 1. Confinement and transport versus triangularity 2. Gyrokinetic simulations, local and global 3. Initial turbulence measurements versus triangularity 4. Conclusions and Outlook A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 5
1. Confinement and transport versus triangularity A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 6
Energy confinement improves towards δ < 0 (at low collisionality) 15 P tot [MW] 10 τ Ee [ms] 5 0.3 0.9 OH EC Plasma conditions: low density ECH plasmas, low collisionality ν eff ~0.2-1 high R/L Te >7 and T e /T i ~3-5 0 0.6 0.4 0.2 0.0 0.2 0.4 δ TEM dominated regime Coda 98, Pochelon NF99 & EPS99 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 7
Transport depends on both triangularity and collisionality HFS LFS δ= 0.4 δ= 0.2 δ=+0.2 δ=+0.4 δ = -0.4-0.2 +0.2 +0.4 Negative triangularity δ shows reduced heat diffusivity χ e over a wide range of collisionality ν eff χ e [m 2 /s] 10 8 6 4 2 0 δ=+0.4 δ=+0.2 δ= 0.2 δ= 0.4 OH δ=+0.4 EC δ= 0.4 ρ=0.55 0 1 2 3 4 5 6 2 1/ν eff T /(ne e Z eff ) Camenen NF07 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 8
Puzzle: limited triangularity penetration, but strong effect δ δ 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 δ LCFS = 0.4 δ LCFS =0.4 0.4 0 0.2 0.4 0.6 0.8 1 ρ ρ v δ=+0.4 δ=-0.4 Large transport variations with δ measured at mid-radius, but triangularity at mid-radius only represents ~25% of δ edge A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 9
Same profiles, different powers at δ = ± 0.4 power balance δ=+0.4 δ=+0.4 δ= -0.4 δ= -0.4 Adjusting power at each triangularity δ = ± 0.4 to reach the same T e, n e, (i.e. same plasma energy), and q-profiles : - only half power needed at δ = - 0.4, - resulting in χ e halved over all radia outside heat deposition (from pow bal) Camenen NF07 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 10
2. Gyrokinetic simulations, local and global A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 11
local GK simulations: GS2, local (flux-tube), linear γ/k 2 δ=+0.4 ρ v =0.70 Mixing length estimate of heat diffusivity decreases towards δ<0: in qualitative ageement with experiment δ=-0.4 Most TEM transport occurs in the range k θ ρ i ~ 0.3 k θ ρ i this same range is the one mostly affected by changing triangularity All simulations based on TCV equilibria Marinoni PPCF09 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 12
Penetration of δ-effect on χ e (for non-linear flux-tube & power balance ) e,δ=+0.4 e,δ= 0.4 χ e,δ=+0.4 χ e,δ= -0.4 2.4 2.2 2 1.8 1.6 1.4 1.2 1 expt. power balance ρ v ρ v NL GS2 0.4 0.5 0.6 0.7 power balance δ=+0.4 δ= -0.4 Marinoni PPCF09 Triangularity effect on χ e from local GK simul. vanishes towards the centre, unlike in experiment, where the triangularity effect does penetrate A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 13
GS2: local non-linear, with collisions GS2 simulations experiment 16 14 12 δ LCFS = +0.4 δ LCFS = 0.4 δ=+0.4 10 8 δ=+0.4 δ=+0.2 δ= 0.2 δ= 0.4 δ=+0.4 χ e [m 2 /s] 10 8 6 δ= 0.4 χ e [m 2 /s] 6 4 δ= 0.4 4 2 0 1 2 3 4 1/ν eff ρ v =0.70 ρ = 0.70 2 0 ρ v =0.55 ρ=0.55 0 1 2 3 4 5 6 2 1/ν T /(ne Z ) eff e eff From flux-tube calculations, triangularity effect on χ e remains confined to the edge Marinoni PPCF09 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 14
From flux-tube to global GK: ORB5, linear runs ρ* = ρ L /a ~ 1 / 70 thus finite ρ* effects expected, i.e. differences between local and global simulations global simulations required to account for the full radial profiles the global gyrokinetic code ORB5 enables to carry out linear and non-linear, collisional simulations currently, for the present triangularity study, linear global ORB5 runs have been carried out so far A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 15
k larger at negative triangularity -> lower mixing length γ/k 2 30 γ [10 4 s 1 ] 20 10 0 0.6 δ=+0.5 δ=+0.2 δ= 0.3 2 2 γ/k k ρ [m /s] s 0.4 0.2 30 15 δ=+0.5 δ = -0.3 mixing length transport Mixing length transport estimate reduced for δ < 0, through reduction of k at low n 0 5 10 15 20 25 n Camenen NF07 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 16
Larger k r at negative triangularity φ, n=4 1 k r at the equatorial midplane θ=0 0.9 k r [cm 1 ] 0.8 0.7 0.6 δ= 0.4 TCV28008 δ=+0.4 TCV28014 0.5 5 10 15 20 Triangularity predominantly affects k r at the low n s, which are the important modes for TEM transport k r useful for comparison with radial correlation length measurements in experiment A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 17 n
3. Initial turbulence measurements versus triangularity A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 18
Turbulence measurements using ECE ECE view lens polarizer waveguide entrance equatorial LFS with focusing lens A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 19
New correlation ECE diagnostic setup, geometry ρ q=1 ρ v ~ 0-0.7 ρ=1 single Gaussian-beam line-of-sight measurement from LFS 2 tunable, narrowband (100-160MHz) YIG filters range 69-81GHz, 0.0<ρ v <0.7 accessible range k θ 0.5-0.8 cm -1 (k θ ρ s 0.3-0.5) optical thickness τ~3-4, i.e. blackbody radiation, thus only T e fluct. A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 20
T e -fluctuation cross-spectra (δ T/T) 2 (khz 1 ) 14 x 10 5 12 10 8 6 4 2 n e =0.9 10 19 m 3, T e =1.8 2.0 kev δ=+0.45 δ= 0.45 - Low frequency peak < 20kHz - TEM turb range up to ~130kHz - T e fluctuations from integral on spectral range 20-100kHz 0 50 100 150 200 Frequency (khz) A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 21
Corr ECE allows a clear correlation length measurement at δ =+0.4 ρ v =0.42 δ=0.45 shot-intensive experiments to overcome two limitations: - two channels - long integration times for statistics L c = 2.7 cm ~ 4 ρ s A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 22
Correlation length decreases with density (collisionality) L c = 2.7cm ~ 4ρ s L c = 1.1cm ~ 2ρ s at given power, L c very sensitive to density, which needs to be well controlled A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 23
Comparison between positive and negative triangularity measuring location profiles δ=+0.45 δ=-0.45 0.33 < ρ v < 0.53 on LFS equator matching n e and T e profiles at δ=±0.45, by adding more ECH at δ > 0 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 24
Strong reduction of turbulence level for δ < 0 δ =+0.45 EC δ = -0.45 EC going from δ=+0.45 to δ=-0.45, results in a strong reduction of fluctuations; especially below 80kHz A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 25
4. Conclusions and Outlook A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 26
Conclusions from experiment Confinement and transport confinement time τ Ee is increased towards δ<0 up to factor 2 heat diffusivity at mid-radius χ e PB is reduced up to factor 2 Turbulence correlation lengths have been measured at δ>0 : L c /ρ s ~ 2-4 turbulence amplitude at mid-radius, is strongly reduced at δ<0, over the entire spectral range L c is very sensitive to density/collisionality (TEM regime) Challenges for experiment measure correlation length at δ<0, in spite of reduced turb. signal a shorter correlation length is expected, as inferred from simulated k r A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 27
Conclusions from gyrokinetic simulations Local simulations linear: triangularity affects the low k modes non-linear: the effect of triangularity does not penetrate radially as found in experiment Global simulations (linear runs) low n, large structures, responsible for the transport change with δ radial wave vector k r is larger at δ<0, particularly in the low n s Challenges for GK simulations develop a synthetic diagnostic for non-linear global code, mimicking T e -fluctuations from correlation ECE reproduce the lower turbulence amplitude and different spectral shape found at δ<0 A.Pochelon, 52 rd APS DPP, 8-12 Nov. 2010, Chicago, NI2 6 28