Professional Documents
Culture Documents
Measurement and Control of Turbulence Spreading in The Scrape-Off Layer of TJ-II Stellarator
Measurement and Control of Turbulence Spreading in The Scrape-Off Layer of TJ-II Stellarator
E-mail: gustavo.grenfell@igi.cnr.it
Abstract
In this paper, we report measurements of turbulence drive and spreading in the edge region
and scrape-off layer (SOL) of the TJ-II stellarator for plasmas heated by ECRH and NBI. In
addition, we show how these and various other relevant quantities can be modulated actively
via edge biasing, i.e. by establishing an edge flow shear layer. When the edge E × B shear
reaches values close to the inverse of the turbulence correlation time, turbulence spreading and
edge-SOL coupling are reduced significantly. The resulting control of SOL spreading via the
edge E × B shear is important in view of the issue of understanding and controlling the SOL
width in fusion-relevant devices.
1. Introduction stable regions, may play an important role in setting the SOL
width [4–9]. However, reported experimental data on turbu-
A critical issue for the development of viable fusion reactors lence spreading in the SOL are scarce [10].
based on magnetic confinement is the power exhaust from In the edge and SOL regions, propagating fluctuations are
the core plasma and the deposition of power on the divertor labelled ‘filaments’ or ‘blobs’. They are coherent structures
plates, which may not exceed the thresholds set by material generated by the nonlinear saturation of turbulence, predomi-
properties. The deposition area and hence the power density nantly traveling outward with a well-defined probability dis-
is determined to a large extent by the scrape-off layer (SOL) tribution function of the radial velocity [11, 12]. Stochastic
width [1, 2]. Therefore, it is of fundamental importance to dis- models suggesting that the SOL plasma is dominated by trave-
pose of a proper understanding of the mechanisms that set the ling filaments [13, 14] have recently been confirmed exper
SOL width, as well as techniques to influence it. imentally [15–17]. Gyrokinetic simulations indicate that the
To first order, the SOL width is determined by the com- SOL plasma in ITER might be dominated by blobs and/or
petition between cross-field and parallel transport, where streamers [18]. Filaments can be seen as agents of turbulence
parallel transport is typically treated classically, whereas spreading, carrying the fluctuations from their birth location to
cross-field transport in the SOL is dominated by turbulence other regions. The intermittent behavior of the ion saturation
[3]. Consequently, it has been suggested that turbulence current in different ranges, from the edge to the far SOL, is well
spreading, i.e. the propagation of turbulence from unstable to characterized in literature [19–21]. The reports are consistent
1741-4326/19/016018+9$33.00 1 © EURATOM 2018 Printed in the UK
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
2
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
3
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
Figure 3. Plasma parameters during the transition from ECRH to NBI; (a) the radial profile of floating potential, (b) ion saturation current,
(c ) turbulent particle flux and (d ) turbulence phase velocity in the TJ-II plasma boundary region. The vertical axis shows the radial position
(r ) relative to the the location of the Last Closed Flux Surface (LCFS, r0 ). Negative values of r − r0 correspond to the plasma edge, positive
values to the SOL. The horizontal black lines indicate the locations of probe pins.
of these and a few other quantities. The curves labelled ECRH To shed more light on the radial propagation of these
were calculated over 1100 < t < 1140 ms, while those effects, we computed the turbulence drive (ωD ) and spreading
labelled NBI were calculated over 1180 < t < 1220 ms. As (ωS) (figure 5). These parameters were evaluated by calcu-
noted, the edge profiles of φf and Is are less pronounced in lating the moving average and RMS values of the ion satur
ECRH than in NBI, and the base level of Is is higher in the ation current and floating potential from the two innermost
SOL (figure 4(b)), i.e. it has a longer ‘tail’ that extends to rows of probe pins. The figure suggests that the high base
the far SOL. Similarly, Γr is flatter and much higher during line of the turbulent particle flux in the SOL in the ECRH
ECRH in the SOL. As noted before, during NBI a shear layer phase is related to both the turbulence drive and spreading,
develops near the edge at r − r0 −15 mm, clearly visible as both parameters have comparable values in this range.
in the graph of vph The local linear growth term is high in the edge and lower in
θ (figure 4( f )). The maximum shear in the
poloidal phase velocity (dvθ /dr) during NBI is 4 · 105 s−1, the SOL, consistent with the variation of the density gradient
while during ECRH it is 0.3 · 105 s−1 at the same position. The (high in the edge and low in the SOL) during ECRH. In the
NBI phase, ωD drops more strongly in the edge region (par
evolution of the phase velocity (vph θ ) can be explained in terms
ticularly at the location of the poloidal velocity shear layer)
of E × B drift [44], in agreement with previous results [45,
than in the SOL. On the other hand, the turbulence spreading
46]. This remarkable difference has a strong impact on the rate (ωS) is higher in the SOL than in the edge (especially
radial turbulent particle flux shown in figure 4(c ). The strong in the ECRH heating phase), while its value is comparable
shear is likely related to the reduction of the turbulence cor- to the local growth rate (ωD ) in this range in both heating
relation time (τturb —from the auto-correlation of the floating regimes.
potential) in the edge (figure 4(e )) and the reduction of the
effective radial turbulent velocity (veffr ) in the edge and SOL
(figure 4(d )). In conclusion, the spontaneous establishment of 4.2. NBI + biasing
the edge poloidal velocity shear layer affects all indicators of
turbulent transport and has a significant impact not only at The influence of edge biasing on various plasma profiles is
the location of the shear layer itself (in the edge, close to the shown in figure 6. Twelve reproducible shots were used to
LCFS), but also further outward (the SOL). measure the profiles of floating potential (φf ), ion saturation
4
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
Figure 4. Radial profiles of (a) floating potential, (b) ion saturation current, (c ) turbulent transport, (d ) radial effective velocity, (e )
turbulence correlation time and ( f ) phase velocity of the fluctuations during plasma heated by ECRH and NBI.
Figure 5. The effect of the transition from ECRH to NBI plasma heating on the local turbulence drive (a) and (c ) (ωD ) and the turbulence
spreading (b) and (d ) (ωS). The horizontal black lines indicate the locations of probe pins.
5
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
Figure 6. (a) Biasing voltage and its effect on (b) the radial profile of floating potential, (c ) ion saturation current, (d ) turbulent particle
flux and (e ) turbulence phase velocity in the TJ-II plasma boundary region. The vertical axis shows the radial position (r ) relative to the the
location of the Last Closed Flux Surface (LCFS, r0 ). Negative values of r − r0 correspond to the plasma edge, positive values to the SOL.
The horizontal black lines indicate the locations of probe pins.
current ( Is), radial turbulent particle flux (Γr ), effective radial floating potential profiles reaches values up to −8 kV m−1
velocity (veffr ), turbulence correlation time (τturb ) and phase (vθ = Er /B ≈ −∆φf /∆rB, where B ≈ 1 T) and the perpend
velocity of the fluctuations in the poloidal direction (vph θ )
icular velocity increases up to −9 km s−1, showing that the meas-
during dynamical edge biasing in the NBI phase. ured phase velocity is mainly due to the imposed Er × B drift
Figure 6 shows the modulation of (b) the floating poten- [46]. The maximum shear in the perpendicular velocity (dvθ /dr)
tial, (c ) the ion saturation current and (d ) the turbulent particle is of the order of 5 · 105 s−1, i.e. close to the inverse of the turbu-
flux (Γr ) in time. The analyses were performed in time sec- lence decorrelation time (figure 7(e )) [47]. The maximum shear
tions with a length of 1 ms. The fast rise and fall of the biasing occurs near r − r0 −10 mm. This result is in agreement with,
voltage (figure 6(a)) provokes a fast readjustment of the pro- e.g. conclusions reported at the LAPD device [48], where limiter
files, which are affected most inside the LCFS. biasing was shown to control the turbulent particle flux, dropping
These results are again summarized in figure 7, which to values near zero when the shearing rates became comparable
shows averages over the two biasing phases (0 and −350 V). to the inverse of correlation time at zero shearing.
During the negative biasing phases (−350 V), the floating In view of the above results, we conclude that the enhance-
potential profiles (figure 7(a)) and edge ion saturation cur ment of the velocity shear at −350 V biasing (figure 7( f ))
rent (figure 7(b)) steepen in the plasma edge region while the leads to the reduction of the base level of the ion saturation
turbulent particle flux (figure 7(c )) and effective radial tur- current in the SOL (figure 7(b)), the amplitude of turbulent
bulence velocity (figure 7(d )) are reduced in both edge and transport (figure 7(c )) and the effective radial velocity of fluc-
SOL (r − r0 > −30 mm), concomitant with a reduction in the tuations in the SOL region (figure 7(d )). Thus, the increase in
magnitude of the ion saturation current in the SOL. Er × B shear in the vicinity of the LCFS has a strong impact
The phase velocity of the fluctuations vph θ is shown in on SOL density ( Is), and turbulent transport profiles.
figure 7( f ). During the negative biasing phase, the plasma The impact of edge biasing on the local turbulence drive
edge potential becomes more negative, the radial gradient in (ωD ) and turbulence spreading (ωS) is shown in figure 8. At
6
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
Figure 7. Radial profiles of (a) floating potential, (b) ion saturation current, (c ) turbulent transport, (d ) radial effective velocity,
(e ) turbulence correlation time and ( f ) phase velocity of the fluctuations during biasing at 0 V and −350 V.
Figure 8. The impact of edge biasing on the local turbulence drive (a) and (c ) (ωD ) and the turbulence spreading (b) and (d ) (ωS).
The horizontal black lines indicate the locations of probe pins.
7
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
0 V biasing, the local growth rate term (ωD ) is high in the edge This work has been carried out within the framework of the
and low in the SOL, roughly consistent with the variation of EUROfusion Consortium and has received funding from
the density gradient (high in the edge and low in the SOL, the Euratom research and training programme 2014–2018
figure 7(b)). At −350 V biasing, ωD is strongly suppressed at under grant agreement No 633053 and from Erasmus
the shear layer position, see figure 7( f ), consistent with the Mundus International Doctoral College in Fusion Science
idea of turbulence suppression by the E × B sheared flow. On and Engineering (FUSION-DC). The views and opinions
the other hand, the rate of turbulence spreading (ωS) is much expressed herein do not necessarily reflect those of the
higher in the SOL than in the edge at 0 V biasing, where its European Commission.
amplitude is comparable to the local growth rate, while it is
reduced at −350 V biasing, more strongly in the SOL region
ORCID iDs
(figures 7(b) and (d )).
It seems that the velocity shear layer that forms in the edge
G. Grenfell https://orcid.org/0000-0003-0107-5787
region at negative biasing (figure 7( f )) constitutes a transport B.Ph. van Milligen https://orcid.org/0000-0001-5344-6274
barrier that partially decouples the edge and SOL regions M. Spolaore https://orcid.org/0000-0002-2350-2033
[49, 50], both by suppressing edge turbulence (local turbu-
lence drive—ωD ) and by reducing outward turbulent transport
(spreading—ωS) [21]. A similar effect, though less intense, References
is observed during the transition from ECRH to NBI (fig-
ures 4 and 5). One would expect such a decoupling to lead to
[1] Eich T., Sieglin B., Scarabosio A., Fundamenski W.,
a reduction of the SOL width (the SOL density scale length), Goldston R. and Herrmann A. 2011 Inter-ELM power
assuming parallel transport remains constant. Indeed, this decay length for JET and ASDEX upgrade: measurement
seems to be consistent with the Is curves shown in figure 7(b). and comparison with heuristic drift-based model Phys. Rev.
Lett. 107 215001
[2] Horacek J. et al (JET Contributors) 2016 Multi-machine
5. Conclusions scaling of the main SOL parallel heat flux width in tokamak
limiter plasmas Plasma Phys. Control. Fusion 58 074005
In this paper, we report experimental measurements of the tur- [3] Myra J., Russell D., D’Ippolito D., Ahn J.-W., Maingi R.,
Maqueda R., Lundberg D., Stotler D., Zweben S. and
bulence drive and turbulent spreading in the SOL region of the Umansky M. 2011 Turbulent transport and the scrape-off-
TJ-II stellarator. The experimental data are consistent with the layer width J. Nucl. Mater. 415 S605
results obtained from a three dimensional gyro fluid turbulence [4] Mattor N. and Diamond P. 1994 Drift wave propagation as a
model [37], in which spreading is also shown to be mainly active source of plasma edge turbulence Phys. Rev. Lett. 72 486
in the SOL region. This effect was demonstrated for plasma [5] Garbet X., Laurent L., Samain A. and Chinardet J. 1994
Radial propagation of turbulence in tokamaks Nucl. Fusion
heated by both ECRH and NBI. For the latter, the effect is less 34 963
pronounced due to a strong shear close to the LCFS, which [6] Gürcan Ö., Diamond P., Hahm T. and Lin Z. 2005 Dynamics
reduces the local turbulence and the spreading in the SOL. In of turbulence spreading in magnetically confined plasmas
addition, we show, for the first time, how the turbulence drive Phys. Plasmas 12 032303
and spreading terms can be modulated via edge biasing and [7] Wang W., Hahm T., Lee W., Rewoldt G., Manickam J.
and Tang W. 2007 Nonlocal properties of gyrokinetic
the generation of a velocity shear layer in the edge, with pos- turbulence and the role of E × B flow shear Phys. Plasmas
sible consequences for the SOL width. When the Er × B shear 14 072306
reaches values close to the inverse of the turbulence correlation [8] Isoardi L., Ciraolo G., Chiavassa G., Haldenwang P., Serre E.,
time in the vicinity of the LCFS, edge-SOL coupling is reduced Ghendrih P., Sarazin Y., Schwander F., Garbet X. and
significantly [51] as well as turbulence spreading. The observed Tamain P. 2009 Modelling SOL flow pattern spreading in
the edge plasma J. Nucl. Mater. 390–1 388
relation between the edge shear layer and turbulence in the SOL [9] Estrada T., Hidalgo C. and Happel T. 2011 Signatures of
is relevant both for tokamaks and stellarators when edge shear turbulence spreading during the H-L back-transition in TJ-II
flows are established near the LCFS. plasmas Nucl. Fusion 51 032001
Consequently, we find that the edge radial electric field shear [10] Pan O. et al (The HL-2A Team) 2016 Investigation of
is an important tool for suppressing turbulence, decoupling the mechanisms for the generation of blobs/holes at the
boundary of the HL-2A tokamak Preprint: 2016 IAEA
edge and SOL regions and reducing non-local effects (spreading). Fusion Energy Conf. (Kyoto, Japan) pp EX/P7–29 (https://
As turbulent drive and spreading are likely important factors nucleus.iaea.org/sites/fusionportal/Shared%20Documents/
in setting the SOL width, together with parallel dynamics, we FEC%202016/fec2016-preprints/preprint0580.pdf)
believe the approach reported here provides a route to a better [11] Yu G. and Krasheninnikov S. 2003 Dynamics of blobs in
understanding of power exhaust in fusion plasmas. scrape-off-layer/shadow regions of tokamaks and linear
devices Phys. Plasmas 10 4413–8
[12] D’Ippolito D., Myra J. and Zweben S. 2011 Convective
Acknowledgments transport by intermittent blob-filaments: comparison of
theory and experiment Phys. Plasmas 18 060501
[13] Garcia O., Kube R., Theodorsen A. and Pécseli H. 2016
The authors acknowledge useful discussions with P. Manz. Stochastic modelling of intermittent fluctuations in the
Research sponsored in part by the Ministerio de Economía y scrape-off layer: correlations, distributions, level crossings,
Competitividad of Spain under project Nr. ENE2015-68206-P. and moment estimation Phys. Plasmas 23 052308
8
Nucl. Fusion 59 (2019) 016018 G. Grenfell et al
[14] Militello F. and Omotani J. 2016 Scrape off layer profiles [33] Nascimento I. et al 2005 Plasma confinement using
interpreted with filament dynamics Nucl. Fusion 56 104004 biased electrode in the TCABR tokamak Nucl. Fusion
[15] Walkden N., Wynn A., Militello F., Lipschultz B., 45 796–803
Matthews G., Guillemaut C., Harrison J., Moulton D. and [34] Spolaore M. et al 2017 H-mode achievement and edge
JET Contributors 2017 Interpretation of scrape-off layer features in RFX-mod tokamak operation Nucl. Fusion
profile evolution and first-wall ion flux statistics on JET 57 116039
using a stochastic framework based on fillamentary motion [35] Grenfell G.G. et al 2018 H-mode access and the role of
Plasma Phys. Control. Fusion 59 085009 spectral shift with electrode biasing in the TCABR tokamak
[16] Kube R., Theodorsen A., Garcia O., LaBombard B. and Phys. Plasmas 25 072301
Terry J. 2016 Fluctuation statistics in the scrape-off layer of [36] Alejaldre C. et al 1999 First plasmas in the TJ-II flexible
Alcator C-Mod Plasma Phys. Control. Fusion 58 054001 heliac Plasma Phys. Control. Fusion 41 A539
[17] Kube R., Garcia O., Theodorsen A., Brunner D., Kuang A., [37] Manz P., Ribeiro T., Scott B., Birkenmeier G., Carralero D.,
LaBombard B. and Terry J. 2018 Intermittent electron Fuchert G., Müller S., Müller H., Stroth U. and Wolfrum E.
density and temperature fluctuations and associated fluxes 2015 Origin and turbulence spreading of plasma blobs
in the Alcator C-Mod scrape-off layer Plasma Phys. Phys. Plasmas 22 022308
Control. Fusion 60 065002 [38] Ritz C. et al 1988 Advanced plasma fluctuation analysis
[18] Chang C. et al 2017 Gyrokinetic projection of the divertor techniques and their impact on fusion research Rev. Sci.
heat-flux width from present tokamaks to ITER Nucl. Instrum. 59 1739
Fusion 57 116023 [39] Hidalgo C. et al 2004 Improved confinement regimes induced
[19] Boedo J. et al 2001 Transport by intermittent convection in the by limiter biasing in the TJ-II stellarator Plasma Phys.
boundary of the DIII-D tokamak Phys. Plasmas 8 4826–33 Control. Fusion 46 287
[20] Cheng J. et al 2010 Statistical characterization of blob [40] Silva C., Goncalves B., Pedrosa M., Hidalgo C., McCarthy K.,
turbulence across the separatrix in HL-2A tokamak Plasma Calderon E., Herranz J., Pastor I. and Orozco O. 2005
Phys. Control. Fusion 52 055003 Transport and fluctuations during electrode biasing on TJ-II
[21] Xu G. et al (JET EFDA Contributors) 2009 Blob/hole Czech. J. Phys. 55 1589–96
formation and zonal-flow generation in the edge plasma of [41] Melnikov A. et al 2011 Plasma potential and turbulence
the JET tokamak Nucl. Fusion 49 092002 dynamics in toroidal devices (survey of T-10 and TJ-II
[22] Bisai N., Das A., Deshpande S., Jha R., Kaw P., Sen A. and experiments) Nucl. Fusion 51 083043
Singh R. 2005 Edge and scrape-off layer tokamak plasma [42] Ascasíbar E. et al 2010 Global energy confinement studies in
turbulence simulation using two-field fluid model Phys. TJ-II NBI plasmas Contrib. Plasma Phys. 50 594–9
Plasmas 12 072520 [43] Velasco J., Alonso J., Calvo I. and Arévalo J. 2012 Vanishing
[23] Agostini M., Zweben S., Cavazzana R., Scarin P., Serianni G., neoclassical viscosity and physics of the shear layer in
Maqueda R. and Stotler D. 2007 Study of statistical stellarators Phys. Rev. Lett. 109 135003
properties of edge turbulence in the National Spherical [44] Hidalgo C., Pedrosa M.A., García L. and Ware A. 2004
Torus Experiment with the gas puff imaging diagnostic Experimental evidence of coupling between sheared-flow
Phys. Plasmas 14 102305 development and an increase in the level of turbulence in
[24] Fuchert G., Birkenmeier G., Nold B., Ramisch M. and the TJ-II stellarator Phys. Rev. E 70 067402
Stroth U. 2013 The influence of plasma edge dynamics [45] Ritz C.P., Bengtson R.D., Levinson S.J. and Powers E.J.
on blob properties in the stellarator TJ-K Plasma Phys. 1984 Turbulent structure in the edge plasma of the TEXT
Control. Fusion 55 125002 tokamak Phys. Fluids 27 2956–9
[25] Nold B., Manz P., Ribeiro T., Fuchert G., Birkenmeier G., [46] Hidalgo C., Goncalves B., Pedrosa M., Silva C., Balbín R.,
Müller H., Ramisch M., Scott B. and Stroth U. 2014 Hron M., Loarte A., Erents K., Matthews G. and Pitts R.
Turbulent transport across shear layers in magnetically 2003 Experimental evidence of fluctuations and flows
confined plasmas Phys. Plasmas 21 102304 near marginal stability and dynamical interplay between
[26] Diamond P., Itoh S.-I., Itoh K. and Hahm T. 2005 Zonal flows gradients and transport in the JET plasma boundary region
in plasma-a review Plasma Phys. Control. Fusion 47 R35 J. Nucl. Mater. 313–6 863–7
[27] Wagner F. 2007 A quarter-century of H-mode studies Plasma [47] Losada U. et al 2016 Influence of long-scale length radial
Phys. Control. Fusion 49 B1 electric field components on zonal flow-like structures
[28] Myra J., Davis W., D’Ippolito D., LaBombard B., Russell D., in the TJ-II stellarator Plasma Phys. Control. Fusion
Terry J. and Zweben S. 2013 Edge sheared flows and the 58 084005
dynamics of blob-filaments Nucl. Fusion 53 073013 [48] Schaffner D.A., Carter T.A., Rossi G.D., Guice D.S.,
[29] Zweben S., Maqueda R., Hager R., Hallatschek K., Kaye S., Maggs J.E., Vincena S. and Friedman B. 2012 Modification
Munsat T., Poli F., Roquemore A., Sechrest Y. and of turbulent transport with continuous variation of
Stotler D. 2010 Quiet periods in edge turbulence preceding flow shear in the large plasma device Phys. Rev. Lett.
the LH transition in the national spherical Torus experiment 109 135002
Phys. Plasmas 17 102502 [49] Beyer P., Benkadda S., Garbet X. and Diamond P. 2000
[30] Taylor R.J., Brown M.L., Fried B.D., Grote H., Liberati J.R., Nondiffusive transport in tokamaks: three-dimensional
Morales G.J., Pribyl P., Darrow D. and Ono M. 1989 structure of bursts and the role of zonal flows Phys. Rev.
H-mode behavior induced by cross-field currents in a Lett. 85 4892
tokamak Phys. Rev. Lett. 63 2365–8 [50] Yi S., Kwon J., Diamond P. and Hahm T. 2015 Turbulence
[31] Weynants R.R. et al 1992 Confinement and profile changes spreading as a non-local mechanism of global confinement
induced by the presence of positive or negative radial degradation and ion temperature profile stiffness Nucl.
electric fields in the edge of the TEXTOR tokamak Nucl. Fusion 55 092002
Fusion 32 837 [51] Ghendrih P., Sarazin Y., Ciraolo G., Darmet G., Garbet X.,
[32] Hidalgo C. et al 2003 Improved confinement regimes induced Grangirard V., Tamain P., Benkadda S. and Beyer P. 2007
by limiter biasing in the TJ-II stellarator Plasma Phys. Transport barrier fluctuations governed by SOL turbulence
Control. Fusion 46 287 spreading J. Nucl. Mater. 363 581–5