Overview of ASDEX Upgrade Results - · PDF fileKantor13, A. Kappatou4, O. Kardaun, M. Kaufmann, ... Max-Planck-Institut fu¨r Plasmaphysik, EURATOM-Association, Boltzmannstr. 2, 85748

1 OV/2-2

Overview of ASDEX Upgrade Results

U. Stroth, J. Adamek1, L. Aho-Mantila2, S.Akaslompolo2, C. Amdor6, C. Angioni, M. Balden,S. Bardin3, L. Barrera Orte, K. Behler, E. Belonohy, A. Bergmann, M. Bernert, R. Bilato,G. Birkenmeier, V. Bobkov, J. Boom4, C. Bottereau20, A. Bottino, F. Braun, S. Brezinsek13, T.Brochard21, M. Brudgam, A. Buhler, A. Burckhart, A. Chankin, I. Chapman8, F. Clairet20, I.G.J.Classen4, J.W. Coenen13, G.D. Conway, D.P. Coster, D. Curran17, F. da Silva6, P. de Marne, R.D’Inca, M. Douai20, R. Drube, M. Dunne17, R. Dux, T. Eich, H. Eixenberger, N. Endstrasser,K. Engelhardt, B. Esposito5, E. Fable, R. Fischer, H. Funfgelder, J.C. Fuchs, K. Gal, M. GarcıaMunoz, B. Geiger, L. Giannone, T. Gorler, S. da Graca6, H. Greuner, O. Gruber, A. Gude, L.Guimarais6, S. Gunter, G. Haas, A.H. Hakola2, D. Hangan, T. Happel, T. Hartl, T. Hauff, B.Heinemann, A. Herrmann, J. Hobirk, H. Hohnle10, M. Holzl, C. Hopf, A. Houben, V. Igochine,C. Ionita12, A. Janzer, F. Jenko, M. Kantor, C.-P. Kasemann, A. Kallenbach, S. Kalvin7, M.Kantor13, A. Kappatou4, O. Kardaun, M. Kaufmann, A. Kirk8, H.-J.Klingshirn, M. Kocan, G.Kocsis7, C. Konz, R. Koslowski13, K. Krieger, M. Kubic20, T. Kurki-Suonio2, B. Kurzan, K.Lackner, P.T. Lang, P. Lauber, M. Laux, F. Leipold14, F. Leuterer, S. Lindig, S. Lisgo20, A.Lohs, T. Lunt, H. Maier, T. Makkonen, K. Mank, M.-E. Manso5, M. Maraschek, M. Mayer, P.J.McCarthy17, R. McDermott, F. Mehlmann12, H. Meister, L. Menchero, F. Meo14, P. Merkel, R.Merkel, V. Mertens, F. Merz, A. Mlynek, F. Monaco, S. Muller19, H.W. Muller, M. Munich,G. Neu, R. Neu, D. Neuwirth, M. Nocente15, B. Nold10, J.-M. Noterdaeme, G. Pautasso, G.Pereverzev, B. Plockl, Y. Podoba, F. Pompon, E. Poli, K. Polozhiy, S. Potzel, M.J. Puschel, T.Putterich, S.K. Rathgeber, G. Raupp, M. Reich, F. Reimold,T. Ribeiro, R. Riedl, V. Rohde, G. v.Rooij4, J. Roth, M. Rott, F. Ryter, M. Salewski14, J. Santos6, P. Sauter, A. Scarabosio, G. Schall,K. Schmid, P.A. Schneider, W. Schneider, R. Schrittwieser12, M. Schubert, J. Schweinzer, B.Scott, M. Sempf, M. Sertoli, M. Siccinio, B. Sieglin, A. Sigalov, A. Silva6, F. Sommer, A.Stabler, J. Stober, B. Streibl, E. Strumberger, K. Sugiyama, W. Suttrop, G. Tardini, M. Teschke,C. Tichmann, D. Told, W. Treutterer, M. Tsalas4, M. A. Van Zeeland9, P. Varela6, G. Veres7,J. Vincente6, N. Vianello16, T. Vierle, E. Viezzer, B. Viola16, C. Vorpahl, M. Wachowski22,D. Wagner, T. Wauters20, A. Weller, R. Wenninger, B. Wieland, M. Willensdorfer18, M. Wis-chmeier, E. Wolfrum, E. Wursching, Q. Yu, I. Zammuto, D. Zasche, T. Zehetbauer, Y. Zhang,M. Zilker, H. Zohm

Max-Planck-Institut fur Plasmaphysik, EURATOM-Association, Boltzmannstr. 2, 85748 Garching, Ger-many,1 Institute of Plasma Physics, Praha, Czech Republic,2 Asscociation EURATOM-Tekes, Helsinki,Finland, 3 Institute of Atomic Physics, EURATOM Association-MEdC, Romania, 4 FOM-InstituteDIFFER, EURATOM Association, TEC, Nieuwegein, The Netherlands,5 C.R.E ENEA Frascati, EU-RATOM Association, CP 65, 00044 Frascati, (Rome), Italy ,6 CFN, EURATOM Association-IST Lis-bon, Portugal,7 KFKI, EURATOM Association-HAS, Budapest, Hungary,8 EURATOM/CCFE FusionAssociation, Culham Science Centre, UK,9 General Atomics, San Diego, California, 92186-5608, USA,10 Institut fur Plasmaforschung, Universitat Stuttgart, Germany,11 EFDA-JET, Culham, United King-dom, 12 University of Innsbruck, EURATOM Association-AW, Austria, 13 Forschungszentrum Julich,Germany,14 Riso, EURATOM Association-RISØ, Roskilde, Denmark,15 EURATOM Association-ENEA, IFP, CNR, Milano, Italy,16 Consorzio RFX, EURATOM Association-ENEA, Padova, Italy,17

Physics Department, University College Cork, AssociationEURATOM-DCU, Ireland18 IAP, TU Wien,EURATOM Association-AW, Austria,19 Dept. Mech. & Aerospace Eng. UCSD, 9500 Gilman Drive,La Jolla CA 92093, USA,20 CEA, Cadarache, France21Institut Jean Lamour, UMR 7198 CNRS, Van-doeuvre, France22 Warsaw Univ. of Technology, 00-661 Warsaw, Polande-mail: [email protected]

2 OV/2-2

Abstract. The medium size divertor tokamak ASDEX Upgrade possesses flexible shaping and versatileheating and current drive systems. Recently the technical capabilities were extended by increasing theECRH power [1], by installing 2×8 internal magnetic perturbation coils [2,3], and by improving theICRF compatibility with the tungsten wall [4]. Using these coils, reliable suppression of large type-IELMs could be demonstrated in a wide operational window, which opens up above a critical plasmapedestal density. The pellet fueling efficiency was observed to increase which opened a path to H-mode discharges with peaked density profiles at line densities clearly exceeding the empirical Greenwaldlimit. Owing to the increased ECRH power of 4 MW, H-mode discharges could be studied in regimeswith dominant electron heating and low plasma rotation velocities, i.e. under conditions particularlyrelevant for ITER. The ion-pressure gradient and the neoclassical radial electric field emerge as keyparameters for the transition. Using the total simultaneously available heating power of 23 MW, highperformance discharges have been carried out where feed-back controlled radiative cooling in the coreand the divertor allowed the divertor peak power loads to be maintained below 5 MW/m2. Under attacheddivertor conditions, a multi-device scaling expression for the power decay length was obtained which isindependent of major radius and decreases with magnetic field resulting in a decay length of 1 mm forITER. At higher densities, however, a broadening of the decay length is observed. In discharges withdensity ramps up to the density limit, the divertor plasma shows a complex behavior with a localizedhigh-density region in the inner divertor before the outer divertor detaches. Turbulent transport is studiedin the core and the scrape-off layer. Discharges over a wide parameter range exhibit a close link betweencore momentum and density transport. Consistent with a gyro-kinetic model, the density gradient at halfplasma radius determines the momentum transport through residual stress and thus the central toroidalrotation. In the scrape-off layer a close comparison of probe data with a gyro-fluid code showed excellentagreement and points to the dominance of drift waves. Intermittent structures from ELMs and fromturbulence are shown to have high ion temperatures even at large distances outside the separatrix.

1. Introduction and technical boundary conditions

The main objective of the ASDEX Upgrade programme is to develop integrated scenariosfor long-pulse operation of burning plasmas in ITER and DEMOwhich include solutions forplasma shaping, confinement and stability, divertor and power exhaust, as well as the choicefor wall materials. This effort includes advancing the physical understanding of related fun-damental problems in order to create reliable predicting capabilities and to discover new pathsto advanced plasma operation. To reach these goals, ASDEX Upgrade is realized as a flexibledevice with versatile heating systems and excellent diagnostics. Plasma shape and divertor con-figurations are close to those of ITER and in an path-breakingeffort tungsten has been qualifiedas a possible solution for divertor and first wall material [5,6]. In 2011 and 2012 systems forthe control of plasma stability and the mitigation of damages possibly caused by the plasmahave been improved. Systems for the real-time control of theplasma position by reflectometry[7], the divertor power load, of neoclassical tearing modes[8], and disruptions have been put inplace and 2×8 internal magnetic perturbation coils are now used to mitigate large edge local-ized modes (ELMs) [9]. Disruption mitigation studies usingmassive gas injection showed animproved fueling efficiency of up to a factor of 2, when the valve is located on the high-fieldside [10]. The ECRH power has been increased to 4 MW and it was demonstrated that replacingNBI by ECRH or ICRH power leads to comparable global plasma parameters [11,12] with thebenefit for transport studies to change momentum and particle sources and the ratio of electronto ion temperature in the core. The ICRH system was improved [4] by installing a modifiedbroad-limiter antenna, which reduced the rise in the tungsten concentration in the plasma dur-ing ICRH by up to 40 % and substantially lowered the tungsten sputtering yield at the antennalimiters, and by replacing tungsten-coated antenna side limiters by boron-coated ones on twoother antennas.

3 OV/2-2

2. High-performance discharges

Power exhaust is a key concern on the way to a fusion reactor and the demonstration of sta-ble high-performance discharges with acceptable divertorpower loads is an important task forpresent day devices [13]. In order to keep the power on divertor targets below the requiredlimit of 5 MW/m2, radiative cooling induced by injected impurities is used.In ASDEX Up-grade, the technique of feed-back controlled radiative cooling has been substantially advancedand applied to high-power discharges. By puffing argon into the main chamber, about 67 %of the heating power of 23 MW could be radiated in the outer core plasma without degradingthe confinement properties. In addition, nitrogen was injected from the divertor roof baffle inorder to further reduce the power load on the divertor platesleading to radiation losses of about5 MW from the divertor and X-point regions [13]. These discharges could be operated with apower load below 5 MW/m2 which has to be compared with a value of 110 MW/m2 as obtainedfrom the appliedP/R = 14 MW/m and a radial strike line width of about 2 cm at the target.

0

5

10

15

20

po

we

r /

MW

# 28371

0

1

2

3

H9

8,

bN

, n

1 2 3 4 5time (s)

0

2

4

6

8

10

pe

ak h

ea

t flu

x/

M

W/m

Prad,div

Prad,main

Prad,tot

Pheat

Ip / 10 A5

en / 10 m -320

H98

3/2 NTM

e

smoothed

a)

b)

c)

2

FIG. 1: High performance discharge with feed-back controlled double radiative cooling withargon and nitrogen (from Ref. [13]).

Figure 1 shows the time traces of such a dis-charge which became possible through a real-time feedback system to independently mon-itor core and divertor radiation in a sophisti-cated way [14] through bolometry and a tar-get temperature estimate, respectively. Themaximum heating power of 23 MW is a mixof 17.5 MW of NBI, 4.5 MW of ICRH and,to limit tungsten accumulation in the plasmacenter, 1.5 MW of ECRH, which is injectedin the second-harmonic ordinary (O2) mode,a heating scheme developed earlier [15]. Theline-averaged density was close to the ITERvalue of 1020 m−3 and a high confinementfactor of H98 = 1 and a normalized betaof βN = 3 were maintained stationary overmany energy confinement times. At the sametime, the tungsten and argon concentrationsstayed at the low values ofcW = 2×10

−5 andcAr = 3× 10

−3 with Zeff ≈ 2. The slight dropin βN at about 3 s is attributed to a 3/2 neo-classical tearing mode (NTM). These resultssuggest that the combination of high main-chamber radiation and high divertor radiationwill allow to control discharges at even highervalues ofP/R [13].In high performance plasmas, neoclassical tearing modes can limit βN . On ASDEX Upgrade,a closed-loop real-time feedback control system for NTM stabilization has been commissionedincluding mode detection, deposition calculation and deposition control using the steerableECRH mirror. For the experiments, a target plasma with 13 MW of external heating leadingto βN = 2.7 was used. On developingm = 3, n = 2 modes the functionality of the feed-back system was demonstrated and the mode amplitude could bereduced by means of localizedelectron-cyclotron current drive [8]. A complete stabilization was not yet achieved in all cases

4 OV/2-2

with only one gyrotron included in the feed-back loop. Demonstration of complete stabilizationusing several gyrotrons will be studied in the future.

3. ELM mitigation with perturbation coils

In order to study the mitigation of the high divertor power loads on the divertor plates caused byELMs, two rows of 8 saddle coils have been installed on ASDEX Upgrade. The coils allow formagnetic field perturbations with toroidal mode numbersn ≤ 4. In a first step, with a reducedset of 2×4 coils only, it was demonstrated that type-I ELMs could be replaced by smaller and be-nign MHD events which appear at higher frequency [2]. Although these events resemble type-III ELMs, they are probably not since they do not show a precursor and they appear at pedestaltemperatures well above 300 eV which is the upper limit for the appearance of type-III ELMs. Amore detailed characterization is ongoing. The suppression of the type-I ELMs appears abovea density threshold at about 65 % of the Greenwald density limit. With the full set of 2×8coils it became possible to study the influence of the toroidal mode number on ELM mitigation[9]. For these studies, NBI-heated type-I ELMy H-Mode discharges heated by NBI were used.

0

2

4

6

MW

NBI power

ECRH power total radiated power

0.300.35

0.40

0.45

0.500.55

MJ

MHD stored energy

0.00.51.01.52.02.53.0

1e22

D/s Deuterium gas puff rate

3.0

3.5

4.0

4.5

5.0

Peripheral electron line density

Empirical ELM mitigation threshold

-202468

10

kA

Outer divertor current

(shunt measurement)

1 2 3 4 5 6Time (s)

-1.0

-0.5

0.0

0.5

1.0

kA

upper row

lower row

Saddle coil current

-210

m19

FIG. 2: ELM mitigation with an = 2 per-turbation field comparing even (non-resonant)and odd (resonant) coil parity with single (up-per and lower) row operation. (From Ref. [9]).

Figure 2 shows a typical discharge where thedensity was ramped up and with the switch-on of the perturbation coil current, the largeELMs disappear. The effect does not dependon the phasing of the coils. The suppressionholds even with only one of the coil rings ac-tive and with all coils phased in a resonantor non-resonant way. In Ref. [2] it was al-ready shown that forn = 2 configurations,the resonance condition is not important forELM suppression. The validity of this re-sults could now be extended ton = 1 andn = 4 magnetic-field perturbations [9]. Dueto low local shear at the outboard midplane,the choice of the resonance condition is aglobal one, which holds simultaneously in alarge radial range.The ELMs are replaced by repetitive smallscale MHD events, which cause lower en-ergy losses but are sufficient to keep the tung-sten concentration in the core plasma at a lowlevel. The temperature in the outer divertorrises moderately during ELM mitigation butthe inner divertor remains detached.These investigations show that in ASDEXUpgrade ELM mitigation can be obtained with perturbations of different toroidal mode numbersand does not require a resonant perturbation field component. ELM Mitigation was also suc-cessful in plasmas with different heating methods, different momentum input and thus differentplasma rotation velocities [16]. In all cases, ELM mitigation is found only at a relatively highpedestal density. At the same time, collisionality does notappear to be an ordering parameterfor the transition into the mitigated state.Although having a strong influence on the ELMs, the field perturbations do not substantially

5 OV/2-2

affect the H-mode pedestal profiles and do not cause modes to grow and lock; also existing tear-ing modes do not lock to the error field in H-mode plasmas. Thisindicates, that the perturbationfield is rather well screened by the plasma. Quantitative studies of error field penetration in theedge plasma are under way. Nevertheless, there is a mild sensitivity of the plasma density to thefield configuration. Perturbations withn = 1 cause a reduction or an increase of about 10 % inthe resonant and non-resonant configuration, respectively.In the SOL the calculated 3D vacuum field topology as calculated by the EMC3 code nicelyreproduces the patterns measured on the target plates by means of IR cameras and Langmuirprobes [17,18]. Hence although there is little sign of a penetration of the error fields into theedge plasma, the modification of the field in the SOL region is clearly present and consistentwith vacuum-field calculations. Further effects of the perturbation coils on the scrape-off layerare addressed Ref. [19].A further beneficial effect of the applied field perturbations is that pellets do not trigger ELMsas they do in normal H mode discharges. This opens again the possibility of pellet fueling withhigh efficiency. Injecting pellets from the high-field side of ASDEX Upgrade into ELM miti-gated H-mode discharges leads to centrally peaked density profiles and line-averaged densitieswell above the Greenwald limitnGW [20]. If maintained, this is a very attractive feature forburning plasmas.In discharges at medium densities (< 0.45nGW), the value of the power threshold for L-Htransitions is not influence byn = 2 magnetic perturbations. At intermediate densities (<0.65nGW) type-III ELMs develop right after the L-H transition and ateven higher density thefield perturbations lead to a threshold power which is at least a factor of 2 above the usual value.This increase is caused by a flattening of the ion and electronpressure gradients [21].

4. L to H-mode transitions

The studies of the power threshold for L to H-mode transitions and the search for the phys-ically relevant parameters for this transition have been continued. The L-H power thresholddependence on density is well-known to be non-monotonic andexhibits a minimum at a den-sity of about4 × 10

19 m−3 in AUG [22]. This behavior is schematically indicated in Fig.3by the orange region. As power is increased at constant density, the plasma often transitionsfrom L-mode to H-mode through an intermediate phase (I-phase) which exhibits an oscilla-tory behavior of the edge turbulence. The figure depicts previous and recent data where zonalflows and geodesic acoustic modes (GAMs) are observed in Ohmic and L-mode plasmas andphases where turbulence-flow oscillation have been observed [23]. At a line-averaged densityof n ≈ 4 × 10

19 m−3 the heating power required for the transition is minimal. The dependenceis not monotone and the threshold increases at higher and lower densities. In the low-densityrange strong zonal-flow activity was observed previously and the I phase with zonal-flow turbu-lence oscillations was limited to this region. This is an indication that the Reynolds stress couldplay an important role in providing the requiredE×B flow shear to stabilise the turbulence andtrigger the transition. In recent experimental campaigns the signatures of the I phase were alsoobserved at higher density in the regions marked by the coloured bars.With the upgraded ECRH power it is now possible to study L-H transitions at low density inmore detail. Due to strong heating of the electron channel the roles of the electron and iontemperatures in the transition could be disentangled [25].It was found that the ion pressuregradient plays the key role. This points to the neoclassicalradial electric field and the relatedflow shear as the important player in the L-H transition. The parameter which orders bestbetween L and H-mode phases is found to be related to the ion pressure gradient in the form∇pi/eni, whereni is the main ion density. If the maximum of this quantity in thepedestal is

6 OV/2-2

FIG. 3: Shaded area indicates power threshold for L-H transitions as function of line-averageddensity. Symbols indicate where GAM oscillation were measured by Doppler reflectometry andin which regime they were. Adapted from Ref. [23,24].

plotted as function of the density a horizontal line separates L and H-mode phases [25], as itcan be seen in Fig. 4. Since the neoclassical radial electricfield in the simplest approximationfor a tokamak plasma is given by this term,Eneo

r = ∇pi/eni, this finding revives the interest inthe role of the neoclassicalE×B flow shear for the H-mode transition, as it was also stressed inRef. [26]. In developed H-modes, charge-exchange spectra of different impurities (He2+, B5+,C6+, Ne10+) were analyzed and yielded consistent results for the radial electric field whichalso agreed with the simple neoclassical prediction [27]. Whether the neoclassicalE×B flowprovides the seed shear flow needed to initiate Reynolds stress drive, which then causes thetransition, or whether it is itself sufficient to suppress turbulence remains an important questionto be addressed in the future.

FIG. 4: Simple estimated of the neoclassical radial electric field as function of the edge densityfor different confinement regimes. Adapted from [25]

At medium densities and at different heating powers, the edge plasma parameters of L-H and

7 OV/2-2

H-L transitions were compared [25,28,29]. Although L-H andH-L transitions happen at differ-ent densities, no strong sign of a hysteresis in the electronpressure was found. Both transitionshappen at very similar values of the electron pressure at theplasma edge. In these transientphases radial electric field measurements do not have sufficient time resolution. But since atedge densities above2 × 10

19 m−3 electron and ion temperatures are closely coupled, the elec-tron parameters were used to derive the ion-pressure gradient. In both directions of the transitionthe estimated ion-pressure gradient and, therefore, also the neoclassical radial electric field haveagain the values in the ranges identified above as the the critical ones for the transition.Although the transitions happen at similar values of the ionpressure gradient, a significantdifference was found in the temporal development of electron density and temperature profilesin all phases across the L-H-L cycle [30]. Another interesting observation is that if the externalgas feed is switched off, the final density to which the plasmadevelops after the transition intoH mode is closely linked to the neutral pressure in the divertor and hence to the neutral particlereservoir stored in the divertor [29].

5. Plasma-wall interaction

Studies related to the divertor and to plasma-wall interaction have been emphasized. The lim-itation of the divertor heat load is a main concern for futuredevices. It is closely linked to thepower-decay length in the scrape-off layer (SOL) and an accurate prediction of it for ITER andDEMO is of great importance. Since first principle modeling of power and particle exhaust isnot available yet, for predictions one has to rely on scalingexpressions.Using improved infra-red camera systems with high spatial and temporal resolution, it wasfound that previous ELM-averaged measurements substantially overestimate the power decaylength. This is due to two reasons: (i) ELM and inter-ELM phases correspond to differentphysical processes which cannot be scaled in the same way and(ii) the strike line is foundto move between ELMs. Resolving these issues on ASDEX Upgrade and JET discharges, animproved scaling expression for the power decay length at the divertor entrance for ELM andinter-ELM phases became available [31]:

λq = 0.73×B−0.78tor q1.20cycl P

0.10SOLR

0.02geo (1)

ITER will be operated at similar values of the safety factorqcycl as the discharges used forthe scaling. For the prediction to ITER, the important dependence is the one on the magneticfield strengthBtor which leads to a reduction inλq compared to the values found in present-day devices. Only a weak dependence on the power flux into the SOL, PSOL, and virtuallyno dependence on the major plasma radiusRgeo is found. The experimental scaling showsvery similar parameter dependencies as a heuristic model where drifts are used to explain thebroadening of the decay length in the SOL [31]. For ITER parameters the regression yields arather short power decay length ofλq ≈ 1 mm. Other than for confinement scalings, wherethe ITER prediction lies about a factor of 10 away from the underlying data, the ITER powerdecay length is only a factor of two shorter than that in JET. It is important to note, however,that expression (1) is obtained for attached divertor conditions. In more realistic scenarios fora burning plasma closer to the density limit with a partiallydetached divertor, broader decaylengths can be expected. Also in the high-performance discharges presented in Sec. 1, the powerdecay length widens by a factor of two with respect to the scaling value.The understanding of the processes leading to power and particle detachment of the divertortarget plates in current devices is rather incomplete and a reliable prediction for the divertorbehavior of future large scale devices is presently out of reach. For example, the fundamentalobservation that the detachment of the inner divertor appears much earlier than in the outer

8 OV/2-2

divertor is not reproduced by the most sophisticated simulation codes, such as SOLPS5.0. Inorder to overcome this deficiency, experimental observations with 2D information on the plasmaparameters in the divertor are required. To this end, new diagnostics have been installed onASDEX Upgrade. From 25 lines of sight and with a time resolution of 2.65 ms, the electrondensity in the inner divertor volume was spectroscopicallydetermined for the first time fromStark broadening of the Dǫ line. In addition, radiative fluctuations were measured with a newarray of fast diode bolometers which cover the inner divertor volume also with a grid of crossedlines of sight [32].The divertor detachment on ASDEX Upgrade has been studied inOhmic and L-mode dis-charges using density ramps up to the density limit, but similar observations are also made indischarges where detachment is initiated by radiative cooling using nitrogen injection. In the Lmode discharges between 400 and 900 kW of ECRH power was applied. With increasing den-sity it was observed that the divertor undergoes different distinct states and the behaviors of theinner and the outer divertor were found to be strongly coupled. Prior to the detachment of theouter divertor, strong fluctuations in the radiated power appear in the SOL of the inner divertorclose to the X-point. The frequency is in the kHz range and scales with the ion mass asm−1/2

i .Simultaneously a high-density region appears in the inner far SOL and around the X-point.During this phase, the experimentally measured particle flux in the inner divertor remains wellbelow the prediction of a two-point model. A high degree of detachment at the inner divertorappears already in an early phase of the density ramp. On the other hand the roll-over in theparticle flux at the inner and outer divertor appear at similar densities. After the disappearanceof the fluctuations, detachment occurs along the entire inner target plate.With tungsten as the plasma facing material, in the divertorvolume high density is correlatedwith high total radiation. Therefore the tomographic reconstruction of the emission measuredalong the line of sight of the foil bolometers can be used to study the 2D temporal densityevolution. Thus it was found that after detachment of the inner strike point a high density regionis located at the target plate next to the X-point. This region then moves radially inwards andthen closer to the separatrix above the X-point. Simultaneously, the outer divertor completelydetaches [32]. The modeling of these observation is ongoing[33].

6. SOL-turbulence studies

For a fundamental description of the power decay length, theunderstanding of turbulent trans-port in the scrape-off layer has to be improved. The fluctuations in the SOL are known to bestrongly intermittent and large events, called blobs (in 2D) or filaments (in 3D), radially trans-port plasma as far as to the plasma facing components. The erosion of the wall material by theplasma blobs is of great concern for ITER [34].Advance turbulence simulation codes for the plasma edge andalso for the SOL are available.In order to validate these codes, the characteristics of theturbulence needs to be measured indetail and in order to estimate the erosion rate on the first wall, the plasma parameters inside theblobs need to be known. These topics have been addressed in ASDEX Upgrade using differentkinds of electric probes.Fluctuation measurements have been carried out using Langmuir probes close to the separa-trix of L-mode discharges [35]. Since the plasma potential and its cross-phase to the den-sity fluctuations is of key importance for a comparison with theory, the plasma potentialwas directly measured with an emissive probe. In addition, aconditional sampling tech-nique was used to compile current-voltage probe characteristics from which the full set ofelectron plasma parameters could be deduced inside of blobs. Both methods yielded con-sistent information on the plasma-potential fluctuations.They were found to be in phase

9 OV/2-2

with the density fluctuations as deduced from the ion-saturation current in good approxi-mation (see Fig. 5a). The plasma-potential fluctuations are, however, out of phase withthe directly measured floating-potential fluctuations by 180 ◦ (Fig. 5c). The difference be-tween plasma and floating potential fluctuations is caused byelectron-temperature fluctua-tions (Fig. 5b). The temperature fluctuations are also in phase with the density fluctuationsan they reverse the amplitude of the floating potential fluctuations with respect to the plasmapotential. This emphasizes the known problem related to themeasurement of cross-phase re-lated quantities such as turbulent transport or Reynolds stress with Langmuir probe arrays.

-20 -10 0 10 20240

260

280

300

320

340

1.2

1.25

1.3

1.35

1.4

-20 -10 0 10 2032

34

36

38

40

42

44

46

Dt (ms)-20 -10 0 10 20

50

60

70

80

-58

-56

-54

-52

-50

ff(V)

Te

(eV)

fp

(V)

Isat (mA)

#26530

~

~

~ ~

n

a)

b)

c)

e (1019

m-3

)~

FIG. 5: Fluctuating plasma parameters in ablob from conditional sampling. Adapted fromRef. [35].

A detailed comparison of the fluctuation mea-surements with gyro-fluid simulations usingthe GEMR code [36,37] was carried out [35].The simulated volume encompasses the tran-sition from close to open field lines includ-ing a sheath model in the SOL. In the ana-lyzed region close to the separatrix, simula-tion and experiment are in excellent agree-ment. Both consistently find in-phase fluctu-ations in density, plasma potential and elec-tron temperature which is in agreement witha mixing-length approach. Also in the code,where synthetic Langmuir probes have beenincluded, the ion-saturation current measure-ments turn out to reproduce density fluctua-tions quite well. As in the experiment, thefluctuations in the floating potential, however,are strongly influenced by temperature fluctu-ations and, hence, are strongly distorted com-pared to the actual plasma-potential fluctua-tions. The fact that both experiment and sim-ulation shows that plasma-potential and den-sity fluctuations are almost in phase clearlypoints to drift waves as the dominant turbu-lence mechanism in the L-mode edge andnear-SOL plasma. Details of the blob gen-eration close to the separatrix are studied andit is observed that an admixture of interchange characteristics increases as the blob propagatesradially outward [38].In order to estimate sputter yields related to the interaction of turbulent filaments or blobs withthe plasma facing components and to develop models for the blob dynamics with predictivecapabilities, the ion density and temperature need to be measured. Using a radially movableretarding-field analyser, RFA, and again conditional sampling techniques, a systematic study ofthe ion energy in turbulent events has been carried out in theSOL of ASDEX Upgrade [19,39].Inside of plasma filaments which were created by ELM crashes,the RFA measured rather highion temperatures of up to 200 eV in the far scrape-off layer. The measured ion temperaturesamount to values between 5 and 50 % of the ion temperature at the pedestal top. The temperaturewas found to scale with the total energy drop induced by the ELM. Large ELMs seem to carry,on average, ions with higher energy into the far SOL. This might suggest that filaments in larger

10 OV/2-2

ELMs propagate faster radially.Radial propagation velocities of 500 – 2000 m/s were estimated from the temperature decaywith the distance from the separatrix using a simple blob model [40]. From the measuredion temperature together with densities deduced from the ion-saturation current, parallel powerfluxes could be estimated and they were found to agree quite well with thermographic measure-ments using an IR camera viewing the RFA.Also the blobs occurring during inter-ELM phases were foundto transport high ion tempera-tures over large radial distances into the SOL. The temperature decreases with a radial decaylength of about 2 cm and blobs with higher density show also a somewhat higher ion temper-ature. With increasing distance from the separatrix, the temperature decays faster with radiuspointing to lower radial propagation velocities.

7. Core transport studies

R/Lne

0 2 4 6

LH

-1

-0.5

0.5

0

0.5

0

-0.5

-10.50-0.5-1

a)

b)

measured u

pre

dic

ted

u

L

u

FIG. 6: Top: measured density gradient lengthvs. normalized rotation gradient, bottom: pre-dicted vs. measured rotation gradient. Adaptedfrom Ref. [41].

Core momentum and particle transport havebeen studied over a wide range of param-eters. Taking advantage of the enhancedECRH capabilities, core transport was stud-ied in discharges without an external sourceof particles or momentum [41]. Usingcharge-exchange spectroscopy, a comprehen-sive database of toroidal flow measurementsin Ohmic, ECR and ICR-heated L and H-mode discharges could be assembled. Inspite of the absence of an external momen-tum source, a large variation of the flow ve-locity in the plasma core from co to ctr. di-rection was measured. In addition it was ob-served that in all scenarios the central toroidalMach number closely correlates with the nor-malized velocity gradient calculated at aboutmid plasma radius. The observed variationsin the intrinsic rotation velocity at zero exter-nal momentum input clearly point to substan-tial changes in the the terms governing the ra-dial momentum transport. Assuming a fixedvalue for the diffusive component of the trans-port equation entrain the existence of a termleading to transport in direction of the flowgradient. This could be a convective pinchor/and a Reynolds stress term.Furthermore, the database also exhibits a cor-relation between the normalized density gra-dient at about half radius with the velocitygradient and thus to the central rotation ve-locity. This is shown in Fig. 6a, which alsoincludes some H-mode discharges. In a previ-ous study, core density peaking in the absence of a particle source was successfully described bytransport coefficients derived from linear calculations with the gyro-kinetic model GS2 [42,43].

11 OV/2-2

Using the same code and an appropriate assumption on the tilting angle of the turbulent ed-dies with the experimental parameters as input, the momentum transport and thus the toroidalflow profile can be predicted. Figure 6b shows, that the agreement between the experiment andthe model is excellent. According to that model, all points in the database which fall into thetrapped-electron-mode (TEM) regime are well described by the modification of the internal mo-mentum transport which is caused by changes in the residual stress through a density gradientdependence. The pinch term plays a minor role only.A related behavior was found in NBI-heated discharges, too.With NBI central particle and mo-mentum sources are present and due to almost equal electron and ion temperature profiles, thedischarge core is in the ion-temperature-gradient (ITG) driven turbulence regime. The additionof 2 MW of ECRH to the 2.5 MW of NBI leads to an increase in electron temperature and to atransition into the TEM regime. According to the GS2 linear model this transition enhances theturbulent particle pinch and indeed a central peaking of thedensity profile was found. At thesame time the rotation profile flattens qualitatively consistently with the observations from theinternal-rotation database [44].Turbulence can also influence the radial fast particle distribution in the plasma core. This is alsoof interest for future devices whenα-particle heating becomes important or neutral beams willbe used for plasma current profile control. Turbulent fast ion transport was one candidate toexplain the observation in ASDEX Upgrade that for the off-axis neutral beams the current-driveefficiency is below the theoretical prediction [45]. Recently, fast-ion D-alpha (FIDA) spec-troscopy, which analyses the Doppler shifted Balmer-α radiation from neutralised deuteriumions, was used to measure the radially resolved slowing-down distribution function of fast ionsoriginating from NBI [46]. The fast-ion profiles from on and off-axis beam sources measuredwith the FIDA diagnostic were compared with slowing-down ion distribution functions whichwere calculated with the TRANSP transport code. For the 93 keV beam, good agreement of thedistribution of ions with energies in the range 30 – 60 keV wasfound when classical slowingdown was used in the calculation. In contrast, assuming an anomalous diffusion of 1 m2/s forthe fast ions in the TRANSP simulations yields fast ion profiles which do not agree at all withthe experimental result. Therefore, diffusion of fast ionscannot explain the low current driveefficiency of the off-axis case [46].

References

[1] D. Wagneret al., Nucl. Fusion52, 54006 (2008).[2] W. Suttropet al., Phys. Rev. Lett.106, 225004 (2011).[3] W. Suttropet al., sub. Fusion Eng Des. (SOFT 2012).[4] V. Bobkov et al., this conference.[5] R. Neuet al., Physica ScriptaT138, 14038 (2009).[6] R. Neuet al., , sub. J. Nuclear Materials (PSI 2012).[7] J. Santoset al., Nucl. Fusion52, 32003 (2012).[8] M. Reich et al., First results of closed loop feedback control of NTMs at ASDEX Upgrade, 39th EPS Con-

ference & 16th Int. Congress on Plasma Physics, Stockholm, 2012.[9] W. A. Suttropet al., Europhysics Conference Abstracts36F, P2.092 (2012).

[10] G. Pautassoet al., Massive gas injection from the high field side of ASDEX Upgrade, 38th EPS Conferenceon Plasma Physics, Strasbourg, France, 2011).

[11] F. Sommeret al., Nucl. Fusion (2012), in press.[12] J. Stoberet al., this conference.[13] A. Kallenbachet al., Nucl. Fusion (2012), submitted.[14] A. Kallenbachet al., Plasma Phys. Controll. Fusion52, 055002 (2010).[15] H. Hohnleet al., Nucl. Fusion51, 083013 (2011).[16] W. Suttropet al., this conference.[17] T. Lunt et al., Nucl. Fusion52, 054013 (2012).[18] H. W. Muller et al., Europhysics Conference Abstracts35G, P4.122 (2011).

12 OV/2-2

[19] M. Kocanet al., this conference.[20] P. T. Langet al., Nucl. Fusion52, 11 (2012).[21] F. Ryteret al., Nucl. Fusion (2012), in press.[22] F. Ryteret al., Nucl. Fusion49, 062003 (2009).[23] G. Conwayet al., Phys. Rev. Lett.106, 65001 (2011).[24] G. Ryteret al., presented at TTF Workshop, Padua 2012.[25] P. Sauteret al., Nucl. Fusion52, 5 (2011).[26] U. Stroth, P. Manz, and M. Ramisch, Plasma Phys. Controll. Fusion53, 24006 (2011).[27] E. Viezzeret al., to be published in Nucl. Fusion.[28] F. Ryteret al., this conference.[29] M. Willensdorferet al., Nucl. Fusion (2012), in press.[30] E. Wolfrumet al., Plasma Phys. Controll. Fusion (2012), in press.[31] T. Eichet al., Phys. Rev. Lett.107, 215001 (2011).[32] S. Potzelet al., submitted to J. Nucl. Materials (PSI 2012).[33] M. Wischmeieret al., this conference.[34] R. Pittset al., J. Nucl. Mater.415, S957 (2011).[35] B. Nold et al., New J. Phys.14, 063022 (2012).[36] S. J. Zwebenet al., Phys. Plasmas16, 82505 (2009).[37] A. Kendl, B. Scott, and T. Ribeiro, Phys. Plasmas17, 072302 (2010).[38] B. Nold et al., phD thesis, to be published (unpublished).[39] M. Kocanet al., Nucl. Fusion52, 023016 (2012).[40] W. Fundamenskiet al., Plasma Phys. Controll. Fusion48, 109 (2006).[41] R. McDermottet al., this conference.[42] M. Kotschenreuther, G. Rewoldt, and W. Tang, Comp. Phys. Comm.88, 128 (1995).[43] E. Fable, C. Angioni, and O. Sauter, Plasma Phys. Controll. Fusion52, 015007 (2010).[44] R. M. McDermottet al., Plasma Phys. Controll. Fusion53, 124013 (2011).[45] S. Gunteret al., Nucl. Fusion47, 920 (2007).[46] B. Geigeret al., Europhysics Conference Abstracts36F, P4.068 (2012).