Comparison study of particle transport and role of fluctuation in low collisionality regime in LHD and JT-60U K. Tanaka 1), H. Takenaga 2), K. Muraoka 3), C. Michael 4), L. N. Vyacheslavov 5), A. Mishchenko 6), M. Yokoyama 1), H. Yamada 1), N. Oyama 2), H. Urano 2), Y. Kamada 2), S. Murakami 7),A. Wakasa 8), T. Tokuzawa 1), T. Akiyama 1), K. Kawahata 1), M. Yoshinuma 1), K. Ida 1),I. Yamada 1), K. Narihara 1), N. Tamura 1) 1)National Institute for Fusion Science, Toki , Japan 2)Japan Atomic Energy Agency, Mukouyama, Naka, Ibaraki , Japan 3)Chubu University, 1200 Matsumoto, Kasugai, Aichi , Japan 4) EURATOM/UAKEA Fusion Association, Oxfordshire OX14 3 DB, United Kingdom 5) Budker Institute of Nuclear Physics, , Novosibirsk, Russia 6) Max-Planck-Institute fur Plasmaphysik, EURATOM-Association, D-17491, Greifswald, Germany 7) Department of Nuclear Engineering, Kyoto University, Kyoto , Japan 8) Graduate School of Engineering, Sapporo Hokkaido University, , Japan

Outline 1.General comparison of density profile in JT-60U and LHD 2.Density profiles and turbulence in JT-60U 3.Density profile and turbulence in LHD 4.Summary and Discussion (Possibility of curvature pinch)

Outline 1.General comparison of density profile in JT- 60Uand LHD 2.Density profiles and turbulence in JT-60U 3.Density profile and turbulence in LHD 4.Summary and Discussion (Possibility of curvature pinch)

The particularity of helical/stellarator is enhanced neoclassical transport in low collision regime Neoclassical Transport coefficient Banana regime ei Plateau regime 1/ regime Future operation regime of reactor Around one order Experimental D e,  e Around one order helical/stellarator tokamak Future operation regime of reactor Neoclassical Transport coefficient Plateau regime Experimental D e,  e 1/ regime Plateau regime ei

Magnetic axis position change magnetic helical ripple and higher ripple results in larger neoclassical transport Flux Surface Orbit of guiding center H C -I Plasma Helical coil Shifts by external vertical field and Shafranov shifts B contour

Density profile is Clear difference of density profiles were observed in JT-60U and LHD JT60U Elmy H mode LHD In both devices, the effect of particle fueling is small, the observed difference is due to the difference of particle transport Both NBI heated plasma

Similar b * dependence with tokamak at Rax=3.5m, opposite b * dependence at Rax=3.6m were observed Larger neoclassical Larger anomalous This is more important.

Outline 1.General comparison of density profile in JT60-U and LHD 2.Density profiles and turbulence in JT-60U 3.Density profile and turbulence in LHD 4.Summary and Discussion (Possibility of curvature pinch)

In tokamak, particle transport is dominated by anomalous one Ware pinch is negligible in the present data set of JT-60U. Linear Gyro kinetic theory (Angioni Nucl. Fusion 2004) and non linear gyro fluid theory (Garbet, P.R.L. 2003) predicts peaked density profile in ion dominant heating. These are caused by ion temperature gradient (ITG) mode. In the dataset of JT-60U,  i=Ln/Lt>1, this suggests ITG is lenaerly unstable. Role of turbulence was experimentally studied by using correlation reflectometry.

Omode correlation reflectmetry was used to measure radial correlation for different peaked density profiles Two frequency channels One is fixed at 47.3GHz (Cut off density (n c ) 2.78x10 19 m -3 ) The other is scanned from GHz (n c = x10 19 m -3 ) Six frequencies were scanned in step. Each frequency was 20msec duration. f2 ne2ne2 f3 ne3ne3 f1 ne1ne1 t Freq. 120ms

Radial correlation was analyzed from correlation reflectometry For quantitative estimation of radial correlation, coherence at  R=20mm was used for parameter dependence. 1.Representative values were averaged one for kHz. 2.Error was standard deviation for this frequency regime 3.2~5 scans were accumulated for constant N e and P NB duration. Accumulated

Data used for analysis Density peaking factor increased with decrease of collisionality as reported previously ITG induce density peaking for eff <1

Ln decreased with coherence. Coherence was higher for more peaked profile. Lt does not show any systematic trend with coherence L n and L t was estimated from two YAG points (  and 0.5) indicated by arrow. Peaked Longer correlation

Radial correlation shows clearer dependence on density peaking factor more than collisionality

Density peaking Smaller Ln Larger radial coherence Larger radial diffusion Particle flux is given by For steady state Enhancement of inward pinch

Outline 1.General comparison of density profile in JT-60U and in LHD 2.Density profiles and turbulence in JT-60U 3.Density profile and turbulence in LHD 4.Summary and Discussion (Possibility of curvature pinch)

These differences are not due to particle fueling but due to transport characteristics. Density profile of LHD changes from peaked to hollow. P NBI = Change of density profile in N-NBI heated plasma at Rax=3.6m Last closed flux surface

Magnetic axis position changes density profile as well. Inward shifted Small magnetic helical ripple and reduced neoclassical transport Outward shifted Large magnetic helical ripple and enhanced neoclassical transport

Density modulation experiments shows D core is anomalous, outward V core is comparable with neoclassical one (K.Tanaka FUSION SCIENCE AND TECHNOLOGY VOL. 51 JAN ) D core D edge 0.7  V core V edge 0.7  1.0 D neo * h Plateau 1/ V neo is dominated by thermo diffusion.

20 With increase of P NB, density pump out and increase of core turbulence were observed. 1MW 6MW NeNe NeNe TeTe TeTe

 =0.6 T.H. Watanabe, H.Sugama Non linear GKV simulation  i =3 Nucl. Fusion 47 (2007) 1383– 1390 Comparison between non linear GKV simulation and PCI measurement shows rough agreement. k y  i -0.3 k y  s -0.4 Preliminary Measurement by 2 dimensional phase contrast imaging (2D-PCI) T e /T i -2, so, ky  i-0.3 Further confirmation is necessary.  =0-0.7

Outline 1.General comparison of density profile in JT-60U and in LHD 2.Density profiles and turbulence in JT-60U 3.Density profile and turbulence in LHD 4.Summary and Discussion (Possibility of curvature pinch)

General formulation ( both for tokamak and stellarator/heliotron) of curvature pinch was developed (A. Mishchenko et al, POP 14, (2007) The calculation is based on Isichenko’s model (M,B, Isichenko et al., Pjys. Plasma 3,1916 (1995)) It was assumed that the anomalous pinch is only curvature pinch and profile is in the steady state. For large aspect ratio (r/R<<1) approximation, canonical density profile is given by For larger aspect ration, curvature pinch becomes smaller. Sign of lnq (positive for tokamak, negative for stellarator/heliotron ) change the profile peaked or hollow. Only trapped electron was taken into account (  b >>  )

Comparison between JT60-U density profile and curvature pinch model (data set of density scan) Both exp. and model profile are peaked for higher density The modeled curvature pinch is not large enough to account for observed profile. high density, low density Exp. Model

Comparison between LHD density profile and curvature pinch model (data set of  scan)  =0.11%,  =1.0% Both exp. and model profiles are more peaked for lower . Curvature pinch is not large enough. Neoclassical effect may be necessary. Exp. Model

Density Peaking factor *dependence Particle diffusion Turbulence behavior JT-60UIncrease with a decrease of b * AnomalousRadial correlation increases with increase of density peaking LHDIncrease with a decrease of b *at Rax=3.5m, increase with a increase of b * at Rax=3.6m Anomalous, but neoclassical contribution increases with a decrease of b *. Fluctuation power increase with decrease of density peaking Summary I

Direction of particle convection Origin of particle convection Thermo diffusion Curvature pinch JT-60UInwardAnomalous?Induce inward but not strong enough for observation LHDInward at Rax=3.5m. increase outward with a decrease of b * at Rax=3.6m Neoclassical at hollowed profile, anomalous at peaked profile Strong and induces outward due to neoclassical effect Induce outward but not strong enough for observation Summary II

28 t=4.0sec Before pump out Neo classical ambipola condition predict weak positive Er field in core (  <0.5) before and after density pumping out. If this is correct, phase velocity increases to i-dia direction in plasma frame. This is against GLF23 prediction in tokamak. 1.Are there different role between tokamak and helical plasma? 2.Are there different propagation direction between linear and non linear status? 3.Does larger contribution of neoclassical cause different role of turbulence? 4.Plasma poloidal rotation should be measured. t=4.5sec After pump out

1.Different collisionality dependence of density peaking was observed in JT-60U and LHD. 2.At Rax=3.5m with larger anomalous contribution, collisionality dependence of density peaking is similar to tokamak one (peaking factor increases with decrease of collisionality). 3.At Rax=3.6m with smaller anomalous contribution, collisionality dependence was opposite to Rax=3.5m and JT-60U. 4.With increase of NB power, density profile becomes more peaked in JT-60U. 5.Radial correlation becomes higher for more peaked density profile 6.With increase of PNB, density profile become hollow in Rax=3.6m of LHD. 7. Peak wavenumber did not change with low and high PNB. This suggests correlation length did not change. 8.Fluctuation power increased with increase of PNB. 9.Both in JT-60U and LHD, fluctuation characteristics changed for different density profiles. 10.In JT-60U, fluctuation may induce both diffusion and convection. 11.In LHD, fluctuation may induce diffusion, convection may be due to neoclassical process. Summary

Additional considerations are necessary. 1. Qualitative tendencies were agreed both in JT-60U and LHD i) Peaked profile for positive lnq in JT-60U and hollow profile for negative lnq. ii) JT-60U; more peaked profile in smaller density (lower collisionality) iii) LHD; more hollow profile in higher beta 2. In both case, curvature pinch is not strong enough. 3. Is situation really stationary? 4.Anomalous thermo diffusion should be considered in JT-60U 5.Neoclassical thermo diffusion should be considered I LHD.

Larger fluctuation power induce larger diffusion according to quasi linear theory Density flattening Larger Ln Larger fluctuation power Larger radial diffusion Smaller V (V is reversed. This may casused by neoclassical process

<> a is averaged for trapped electron Shape of density profile does not require saturated level of fluctuation. D 0 is canceled out.

What can we measure from correlation reflectometer? 1.The interpretation of the reflectometer signal is not simple. 2.The change of signal power does not represent fluctuation power (may indicate change of fluctuation power in some stage but sensitive to density gradient and curvature of the magnetic surface). 3.In the present system, we looked correlation of the signal assuming signal correlation represent correlation of fluctuation.

Density peaking factor increases with decrease of Ln. Cleaer tred is seen between peaking factor and eff(0.5) more than between peaking facotr and eff(0.35)

In both tokamak and helical/stellarator, turbulence driven transport is important for density profiles. Therefore, it is essential to investigate characteristics of turbulence for different density profiles. from the measurements. Comparison study between turbulence and density profiles will give a god guidance for numerical simulation.

The particularity of helical/stellarator is enhanced neoclassical transport in low collision regime Neoclassical Transport coefficient Banana regime ei Plateau regime 1/ regime Future operation regime of reactor Around one order Experimental D e,  e Around one order helical/stellarator tokamak Future operation regime of reactor Neoclassical Transport coefficient Plateau regime Experimental D e,  e S. Murakami Nucl. Fusion 42 (2002) L19–L22 Axis Position D neo /D tokamak plateu 1/ regime Plateau regime In 1/, neoclassical transport is minimum at Rax=3.53m In Plateau, neoclassical transport is smaller at more inward axis. Inward shifted Outward shifted Inward shifted Outward shifted ei

Role of neoclassical transport in tokamak and heliotron/stellarator in low collisionality regime 1. In tokamak, neoclassical effect (Ware pinch) is negligible in low collsionality regime 2.In LHD, diffusion is anomalous but convection is comparable with neoclassical thermo-diffusion in certain configuration. ( K.Tanaka 2007 Fusion Sci. Tech., K.Tanaka 2008 Plasma Fusion Research )

More systematic trend was observed between L n and coherence more than eff and coherence Radial coherence might be more essential parameter for density peaking more than eff.

Core fluctuation may play role on density profile shaping. Most of fluctuation components exists in ITG/TEM unstable region Tokamak like. Turbulence transport produce peaked profile Helical particular. Inward turbulence driven flux can be balanced with outward neoclassical

Propagation direction tells local field angle. Local field angle tells local position. Decomposed by FFT and MEM Measured by 2D detector Strong magnetic shear helps as well. Shear techniques was done by Truc on Tore Supra (RSI (1992)) and Kado on Heliotron E (JJAP (1996)) toroidal

Rax=3.5m →Peaked density profile Rax=3.6m→Peaked ~ hollow density profile Rax>3,75m→hollow density profile Anomalous contribution becomes larger K. Tanaka et al., Journal of Plasma Fusion and Research 2008 Minimum Rax is different between experimental (~anomalous) and neoclassical diffusion coefficient→It is a contrast to energy confinement  eff ∝ 1/  e is minimum at neoclassical minimum. V exp is comparable with V neo. Both are minimum at Rax=3.5m

D core D edge 0.7  Rax=3.5m →Peaked density profile→stronger Te dependence of D Rax=3.6m→Peaked ~ hollow density profile→weaker Te dependence of D

 V vv Vcore Rax=3.5m →Peaked density profile→negative Te dependence of V core Rax=3.6m→Peaked ~ hollow density profile→positive Te dependence of V core

In LHD, anomalous and neoclassical contribution depends on magnetic configuration. The biggest configuration effect is magnetic ripple. Inward-> Smaller ripple, Outward -> larger ripple Diffusion; Anomalous contribution is larger at more inward shifted configuration. Convection; More or less comparable with neoclassical, but direction becomes inward against neoclassical prediction at more inward shifted configuration. Rax=3.6m was compared with JT-60U data

D neo * h Plateau 1/ Larger neoclassical Smaller neoclassical At more outwardly R ax, neoclassical transport becomes larger and density profile becomes more hollow.

1.Different collisionality dependence of density peaking was observed in JT-60U and LHD. 2.At Rax=3.5m with larger anomalous contribution, collisionality dependence of density peaking is similar to tokamak one (peaking factor increases with decrease of collisionality). 3.At Rax=3.6m with smaller anomalous contribution, collisionality dependence was opposite to Rax=3.5m and JT-60U. 4.With increase of NB power, density profile becomes more peaked in JT-60U. 5.Radial correlation becomes higher for more peaked density profile 6.With increase of PNB, density profile become hollow in Rax=3.6m of LHD. 7. Peak wavenumber did not change with low and high PNB. This suggests correlation length did not change. 8.Fluctuation power increased with increase of PNB. 9.Both in JT-60U and LHD, fluctuation characteristics changed for different density profiles. 10.In JT-60U, fluctuation may induce both diffusion and convection. 11.In LHD, fluctuation may induce diffusion, convection may be due to neoclassical process. Summary

Density profiles were scanned from scan of NB power Change of power spectrum suggest the change of fluctuation property.

Ln decreased with coherence. Coherence was higher for more peaked profile. Lt does not show any systematic trend with coherence L n and L t was estimated from two YAG points (  and 0.5) indicated by arrow.

Ne scan for constant power profile time s profile time s profile time s profile time s profile time profile time, s Power scan for constant Ne case profile time 9.45s profile time 9.556s profile time 9.282s Power scan for constant Ne case profile time s profile time s profile time s Ne scan and Power scan

14 cases for constant density (0.5~1s) and P NB from 5 discharges were used fro analysis. Both profile and coherence was accumulated for each cases.

Ne scan and Power scan It is difficult to say systematic trend for Power scan for constant Ne because of limited number of data

The density peaking factor increases with increase of radial coherence

Density modulation experiments shows D core is anomalous, outward V core is comparable with neoclassical one (K.Tanaka FUSION SCIENCE AND TECHNOLOGY VOL. 51 JAN ) Blank; Experiment, Colored; Neoclassical Rax=3.6n, Bt=2.75, 2.8T Rax=3.6n, Bt=1.49T Rax=3.75n, Bt=1.5T Rax=3.9n, Bt=1.54T D core D edge 0.7  V core V edge 0.7  1.0 At lower collisionality D core is close toD neo. D neo * h Inward V core is not neoclassical. Plateau 1/