INSTITUTE OF PHYSICS PUBLISHING and INTERNATIONAL ATOMIC ENERGY AGENCY NUCLEAR FUSION

Nucl. Fusion 43 (2003) 1570–1582 PII: S0029-5515(03)64081-0

Overview of ASDEX Upgrade resultsH. Zohm1, C. Angioni1, R. Arslanbekov1, C. Atanasiu2, G. Becker1, W. Becker1,K. Behler1, K. Behringer1, A. Bergmann1, R. Bilato1, V. Bobkov1, D. Bolshukhin1,T. Bolzonella3, K. Borrass1, M. Brambilla1, F. Braun1, A. Buhler1, A. Carlson1,G.D. Conway1, D.P. Coster1, R. Drube1, R. Dux1, S. Egorov4, T. Eich1,K. Engelhardt1, H.-U. Fahrbach1, U. Fantz5, H. Faugel1, K.H. Finken6, M. Foley7,P. Franzen1, J.C. Fuchs1, J. Gafert1, K.B. Fournier8, G. Gantenbein9, O. Gehre1,A. Geier1, J. Gernhardt1, T. Goodman10, O. Gruber1, A. Gude1, S. Gunter1,G. Haas1, D. Hartmann1, B. Heger5, B. Heinemann1, A. Herrmann1, J. Hobirk1,F. Hofmeister1, H. Hohenocker1, L.D. Horton1, V. Igochine1, A. Jacchia11,M. Jakobi1, F. Jenko1, A. Kallenbach1, O. Kardaun1, M. Kaufmann1, A. Keller1,A. Kendl1, J.-W. Kim1, K. Kirov1, R. Kochergov1, H. Kollotzek1, W. Kraus1,K. Krieger1, T. Kurki-Suonio12, B. Kurzan1, P.T. Lang1, C. Lasnier8, P. Lauber1,M. Laux1, A.W. Leonard13, F. Leuterer1, A. Lohs1, A. Lorenz1, R. Lorenzini3,C. Maggi1, H. Maier1, K. Mank1, M.-E. Manso14, P. Mantica12, M. Maraschek1,E. Martines3, K.-F. Mast1, P. McCarthy7, D. Meisel1, H. Meister1, F. Meo1,P. Merkel1, R. Merkel1, D. Merkl1, V. Mertens1, F. Monaco1, A. Muck1,H.W. Muller1, M. Munich1, H. Murmann1, Y.-S. Na1, G. Neu1, R. Neu1,J. Neuhauser1, F. Nguyen15, D. Nishijima1, Y. Nishimura1, J.-M. Noterdaeme1,I. Nunes14, G. Pautasso1, A.G. Peeters1, G. Pereverzev1, S.D. Pinches1, E. Poli1,M. Proschek16, R. Pugno1, E. Quigley7, G. Raupp1, M. Reich1, T. Ribeiro14,R. Riedl1, V. Rohde1, J. Roth1, F. Ryter1, S. Saarelma12, W. Sandmann1,A. Savtchkov6, O. Sauter10, S. Schade1, H.-B. Schilling1, W. Schneider1,G. Schramm1, E. Schwarz1, J. Schweinzer1, S. Schweizer1, B.D. Scott1, U. Seidel1,F. Serra14, S. Sesnic1, C. Sihler1, A. Silva14, A.C.C. Sips1, E. Speth1, A. Stabler1,K.-H. Steuer1, J. Stober1, B. Streibl1, E. Strumberger1, W. Suttrop1, A. Tabasso1,A. Tanga1, G. Tardini1, C. Tichmann1, W. Treutterer1, M. Troppmann1,H. Urano1, P. Varela14, O. Vollmer1, D. Wagner1, U. Wenzel1, F. Wesner1,E. Westerhof17, R. Wolf1, E. Wolfrum1, E. Wursching1, S.-W. Yoon1, Q. Yu1,D. Zasche1, T. Zehetbauer1 and H.-P. Zehrfeld1

1 Max-Planck-Institut fur Plasmaphysik, EURATOM Association, Garching, Germany2 Institute of Atomic Physics, EURATOM Association, Romania3 Consorzio RFX, EURATOM Association, Padova, Italy4 Plasma Physics Department, Technical University, St Petersburg, CIS, Russia5 University of Augsburg, Germany6 Institut für Plasmaphysik, EURATOM Association, FZ Julich, Germany7 Physics Department, University College Cork, Association EURATOM-DCU, UK8 Lawrence Livermore National Laboratory, Livermore, USA9 Institut fur Plasmaforschung, Stuttgart University, Germany10 CRPP, Ecole Polytechnique Federale de Lausanne, EURATOM Association, Switzerland11 IFP Milano, EURATOM Association, Italy12 Helsinki University of Technology, EURATOM Association, Finland13 General Atomics, San Diego, USA14 Centro de Fusao Nuclear, IST Lisbon, EURATOM Association, Portugal15 CEA Cadarache, EURATOM Association, France16 Institute for Applied Physics, EURATOM Association, Vienna, Austria17 FOM Rijnhuizen, EURATOM Association, The Netherlands

E-mail: zohm@ipp.mpg.de

Received 15 October 2002, accepted for publication 22 May 2003Published 1 December 2003Online at stacks.iop.org/NF/43/1570

0029-5515/03/121570+13$30.00 © 2003 IAEA, Vienna Printed in the UK 1570

Overview of ASDEX Upgrade results

AbstractRecent results from the ASDEX Upgrade experimental campaigns 2001 and 2002 are presented. An improvedunderstanding of energy and particle transport emerges in terms of a ‘critical gradient’ model for the temperaturegradients. Coupling this to particle diffusion explains most of the observed behaviour of the density profiles, inparticular, the finding that strong central heating reduces the tendency for density profile peaking. Internal transportbarriers (ITBs) with electron and ion temperatures in excess of 20 keV (but not simultaneously) have been achieved.By shaping the plasma, a regime with small type II edge localized modes (ELMs) has been established. Here, themaximum power deposited on the target plates was greatly reduced at constant average power. Also, an increase of theELM frequency by injection of shallow pellets was demonstrated. ELM free operation is possible in the quiescentH-mode regime previously found in DIII-D which has also been established on ASDEX Upgrade. Regardingstability, a regime with benign neoclassical tearing modes (NTMs) was found. During electron cyclotron currentdrive (ECCD) stabilization of NTMs, βN could be increased well above the usual onset level without a reappearanceof the NTM. Electron cyclotron resonance heating and ECCD have also been used to control the sawtooth repetitionfrequency at a moderate fraction of the total heating power. The inner wall of the ASDEX Upgrade vessel hasincreasingly been covered with tungsten without causing detrimental effects on the plasma performance. Regardingscenario integration, a scenario with a large fraction of noninductively driven current (�50%), but without ITB hasbeen established. It combines improved confinement (τE/τITER98 ≈ 1.2) and stability (βN � 3.5) at high Greenwaldfraction (ne/nGW ≈ 0.85) in steady state and with type II ELMy edge and would offer the possibility for long pulseswith high fusion power at reduced current in ITER.

PACS numbers: 52.55.Fa, 52.55.Rk, 52.55.Tn, 52.55.Wq

1. Introduction

The ASDEX Upgrade tokamak programme [1] focuses onthe preparation of ITER. This includes the investigation ofthe physics of the baseline scenario, the edge localized mode(ELMy) H-mode, as well as the preparation of ‘advancedscenarii’ with performance exceeding that of the ELMyH-mode and a major fraction of noninductively driven current.The three underlying physics areas studied are transport ofenergy and particles, operational boundaries due to MHDinstabilities and exhaust of energy and particles by a poloidaldivertor. In addition, a number of technological issues,mostly related to the heating, fuelling and exhaust systemsare resolved. In particular, a strong emphasis is laid on thestudy of W as a wall material.

In this paper, we summarize the progress made with thesestudies in the 2001 and 2002 experimental campaigns. Theresults described confirm the viability of the conventionalscenario for ITER. The progress made with advanced scenariishows that also a significant improvement of performance withrespect to the standard scenario may be possible in ITER.

2. Upgrade of technical systems

In 2001 and 2002, several upgrades of the ASDEX Upgradetokamak and its auxiliary systems have been completed.A major change was the redirection of one of the two neutralbeam injection (NBI) boxes to provide off-axis depositionwith fairly tangential beams [2]. Figure 1(a) shows thegeometry. With 5 MW at 93 keV, it was possible to showthat a significant current could be driven by NBI. Figure 1(b)shows a comparison of two discharges: in #15884, off-axisbeams were used whereas in #15887, on-axis beams wereapplied. Both discharges have similar Te, ne and Zeff profiles,so that the electrical resistance is equal. However, with off-axis

NBI, the consumption of OH flux is reduced, as can be seenby the difference in OH current ramp needed to sustain theplasma current. While the magnitude of the driven current isin reasonable agreement with the theoretically expected value,the expected change of the current profile is not confirmed bythe MSE measurements. Such an effect could be due to ananomalous radial diffusion of the fast NBI particles. Furtherstudies will have to clarify the precise nature of this effect.

After a refurbishment of the ion cyclotron resonanceheating (ICRH) generator tubes and with the use of the 3 dBcouplers that minimize the effect of reflected ICRH poweronto the generators, reliable ICRH operation even in type IELMy H-mode was achieved with power levels up to 7.2 MW(90% of the generator power) at the antennae. Ion cyclotronresonance frequency (ICRF) power proved to be a viable toolto study transport physics since it complements NBI heatingby providing central heating without particle fuelling or directmomentum input (see section 3). Also, we have validatedion cyclotron heating at the 3

2 resonance frequency using atwo-species (D and H) plasma in which a fast wave excites anion Bernstein wave in the centre by mode conversion [3].

The outer divertor leg of the ASDEX Upgrade divertor wasrebuilt to accommodate all triangularities with good pumpingand power handling. This opened up the operational spaceat higher triangularity δ up to δ = 0.5, which had a positiveimpact on confinement and high densities as well as powerexhaust with small (type II) ELMs (see section 4). Also,more and more tiles on the inner heat shield were coveredsuccessively with a W-coating in order to advance towards aC-free interior of the machine (see section 5).

Improvements in the tokamak power supply led to anincrease in pulse length; now, 10 s Ip flattop are possible,exceeding the current redistribution time and thereforeproviding the possibility to study steady state issues even onthis timescale. In the coming campaign, even higher δ � 0.6will be possible.

1571

H. Zohm et al

≤60 kV

≤93 kV

CDPNBI

IOH

Ip

#15884: off-axis NBI #15887: on-axis NBI

0.8

5

0

0

-10

Time (s)5 10

same stored energy[MW]

[MA]

[kA]

0

Figure 1. Geometry of the ASDEX Upgrade NBI system (left) and comparison of two discharges with on-axis and off-axis NBI. The casewith off-axis NBI has lower OH flux consumption, proving the higher NBI driven current in this case.

ECH1 ECH20.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0 0.2 0.4 0.6 0.8 1

Te

[keV

]

ρtor

100 / 0

61 / 39

21 / 79

0 / 100

ECH1 ECH2

0

1

2

3

4

5

0 1 2 3 4 5 6

PBH PPBH P

gradTe/ Te [m-1]

Exp.

Model

κ = 2.3 qλ = 0.45I p = 800 kA

χ ePB/T

e3/2 o

rχeH

P /Te3/

2 [m2 /(

s ke

V

)]

3/2

Figure 2. Te profiles from a series of experiments with constant heating power, but varying power fraction between two different locations(left). The variation of local heat flux provides a series of points for analysis of χe from power balance (PB) and from modulationexperiments (HP) well described by the simple model from equation (1). The analysis is done at ρ = 0.5.

3. Transport studies

3.1. Energy transport in conventional scenarii

A major finding in the last years was the observation of‘stiff’ Te and Ti profiles, i.e. a fixed ratio between centraltemperature and edge temperature. This was explained by‘critical gradient’ models, that predict the onset of turbulence(ITG, ETG, TEM) below a critical value of the temperaturegradient length LT = T/∇T [4].

Using the flexible ASDEX Upgrade electron cyclotronresonance heating (ECRH) system, it was verified that anelectron heat conductivity of the generic form

χe = χe0 + qλT 3/2e

(1

LTe

−(

1

LTe

)crit

)

× H

(1

LTe

−(

1

LTe

)crit

)(1)

where H is the Heaviside function, q the local safety factorand λ a free parameter, explains the experimental observation

very well. This heuristic model is based on the idea that abovea critical gradient, transport increases strongly. Also, a gyro-Bohm dependence of anomalous transport was assumed. Thead hoc assumption of a q-dependence is needed to explainthe experimentally observed dependence on plasma current aswell as to reproduce the radial variation of transport. TheT 3/2-dependence was found to show good agreement withthe experimental data by varying T at constant q [5]. Thismodel also resolves the issue of different values of χe frompower balance and heat pulse analysis. Figure 2 shows theresults from an experiment where the location of the ECRHdeposition was varied over the minor radius, keeping the totalpower constant. This allows us to vary the local heat flux byup to an order of magnitude with unchanged edge Te. In theseexperiments, the ECRH power was also modulated to infer theheat pulse χe. The parameters (1/LTe

)crit and λ in equation (1)were fitted to the power balance data (χe0 is very small and setto zero in the fit). This set of parameters also describes verywell the results for the heat pulse χe,HP which can be derived

1572

Overview of ASDEX Upgrade results

from equation (1) by χe,HP = χe + (∂χe/∂∇Te)∇Te [6]. Itshould be noted that the factor q in equation (1) leads to agood agreement with the experiment for fixed λ even when Ip

is varied.These experiments demonstrate that the Te profiles are

moderately stiff: for finite central heating, the experimentallymeasured 1/LTe

is in general above (1/LTe)crit , i.e. 1/LTe

is usually situated somewhat above the marginal point forturbulence onset [7], in good agreement with the recentmodelling using the Weiland ITG/TEM model [8]. As shownin figure 2, in the off-axis case the Te profile is very close to thethreshold in the inner part. This allows us to measure directlythe threshold which is in good agreement with that derived forTEM whereas that for ETG is higher by a factor of about 2.

3.2. Particle transport

Since the temperature profiles in general exhibit the ‘stiff’behaviour described above, a given pedestal temperatureimplies a certain central temperature. However, n profilesare in general not stiff and thus density peaking is a way toimprove confinement and has recently attracted attention. Itwas shown that on ASDEX Upgrade, density peaking occurson a timescale much longer than the energy confinement time.This can well be described by an inward pinch of the order ofthe neoclassical Ware pinch in combination with a fixed ratioof the particle diffusion coefficient D to the heat conductivityχ [9]. In fact, this model also explains the experimentalfinding that the density peaking observed with off-axis heatingvanishes when the heating power is applied in the centre.Figure 3 shows a comparison of two discharges where theICRH deposition is varied and a reference NBI discharge.

As can be seen, central deposition leads to a much flatterne profile, consistent with the interpretation that the centralheating power drives the turbulence stronger there (as isexpected from the critical gradient model) and thus, throughD ∝ χ , also the particle diffusion increases and dominatesover the inward pinch. This mechanism may also explainthe so-called ‘ECRH-pumpout’, i.e. strongly increased particletransport in the presence of ECRH [10].

Central heating by ICRH and ECRH has also successfullybeen applied to avoid impurity accumulation in the centre

0.0 0.2 0.4 0.6 0.8 1.0

200

2000

q≈1

0.0 0.2 0.4 0.6 0.8 1.0ρtor ρtor

10

12

14

8

q≈1

(10

m

)19

−3

Te

(eV

)

ICRH on-axisICRH off-axisNBI

ICRH on-axisICRH off-axisNBI

n e

(linear scale)

(logarithmic scale)

Figure 3. Te and ne profiles for discharges heated with on-axis ICRH, off-axis ICRH and NBI with off-axis deposition. Since Te profiles aresimilar, the centrally heated case has higher central χe and thus also higher D, leading to enhanced central particle transport.

[11, 12]: under circumstances with reduced central heating,impurity accumulation can be observed in ASDEX Upgradedischarges. This can be explained by neoclassical particletransport since a peaked main ion profile can lead to muchstronger peaked impurity profiles. A quantitative analysisshows good agreement between neoclassical predictions andexperimental results for ASDEX Upgrade [13]. If, underthese circumstances, some central heating is applied, impurityaccumulation can be completely suppressed. For example, theW-concentration in the centre can be reduced by two orders ofmagnitude with only 0.8 MW of ECRH in a discharge heatedwith 5 MW of NBI.

3.3. H-mode barrier and pedestal physics

The transport studies described above give a quantitativeunderstanding of the temperature and density profiles for givenpedestal values. However, the understanding of the H-modepedestal itself is less developed. Therefore, a major effortwas put into precisely determining of the pedestal parameters,including measurements of Te and ne with a special Thomsonscattering geometry that gives very high resolution in the edge.It had been found before that ∇p in the pedestal region isroughly consistent with the ideal ballooning limit. However,this is a limit to ∇p, but does not discriminate between ∇T

and ∇n. Recently, we have found that Te and ne profilesin the pedestal region are closely coupled by the relationηe = d(log Te)/d(log ne) = 2, i.e. Te = c × n2

e where c isspatially constant [14], but varies with plasma parameters suchas Ip, Bt or the neutral density, etc [15]. Figure 4 shows thisrelation for a set of ASDEX Upgrade discharges. For knownc, this relation, together with the ballooning constraint, woulddetermine the profiles in the pedestal region.

Concerning the L–H transition, a series of experimentswas conducted in ASDEX Upgrade and JET in order todetermine whether the L–H power threshold is invariantif expressed in dimensionless variables. This was clearlyconfirmed, ruling out a dominant role of atomic physics, whichwould introduce a different dependence in the physics of theL–H transition [16].

Finally, during a campaign with reversed Ip (i.e. counterNBI), it was possible to reproduce the completely ELM-free

1573

H. Zohm et al

1019 1020

n (m )

10

100

1000

T

(eV

) e

e-3

0 2×1019 4×1019

n (m )

0

1

2

3

4

e,sep-3

η

@ ∇

P,m

axe

Figure 4. Relation between edge Te and ne showing a relationship Te ∝ n2e (left, points of same colours relate to individual shots), i.e.

ηe = 2 as shown in the right figure.

quiescent H-mode (QH-mode) regime previously found inthe DIII-D tokamak [17]. Crucial elements to access theregime were counter NBI, low density and a ‘high clearance’(i.e. large plasma-wall distance) equilibrium. Figure 5 showsa typical example with an ELM-free period of more than 2 s insteady state [18].

Consistent with the DIII-D results, an edge harmonicoscillation (EHO) was observed instead of the ELM activity.Details on this mode can be found in section 4.1. However,due to the very low ne needed to access this regime, Zeff wasquite high in spite of the addition of ICRH which preventedaccumulation of impurities. Further experimental work has toclarify the relevance of this regime to reactor-type conditions.

3.4. Internal transport barriers

Recent studies on transport barriers have led to record values ofTe and Ti in ASDEX Upgrade. Based on existing scenarii [1],a major focus was on establishing reliable operational scenariito reproduce internal transport barriers (ITBs). Concerningion ITBs, we changed the scenario from an inner wall orupper single null scenario to the more usual lower single nulloperation. The previous scenarii had been chosen becauseunder these conditions, the discharge stayed in L-mode, whichwas necessary to avoid disruptions due to external kinks thatoccurred when an ITB discharge transited to H-mode. In therecent campaign, it was found that ITBs could be producedwith an H-mode edge without disruptions when the onset of theNBI heating was slightly delayed. A detailed analysis showsthat under these circumstances, the target density was lower,which seems to be a decisive parameter for ITB formation inASDEX Upgrade [19]. Also, the q-profile has more time torelax and may be less inverted than previously; this would bea possible explanation for the absence of the external kink inthe new scenario. Unfortunately, due to technical problems,no MSE measurements were available in the recent campaignto validate this hypothesis.

In ITB discharges with H-mode edge, the ITB phase isaccompanied by an ELM-free phase. In this phase, the storedenergy increases almost linearly without an apparent MHD

limit to the central pressure until the first ELM happens. Infact, the stored energy achieved before the first ELM is linearlyproportional to the heating power applied. The first ELM afterthis phase usually also affects the plasma centre and therebyterminates the ITB. The duration of the ELM free phase isalmost independent of heating power and the edge ∇p saturatesearly on in this phase; a possible explanation is that the edge∇p is limited by a more benign MHD instability and only afterthe build up of sufficient bootstrap current in the edge does thefirst ELM occur. Figure 6 shows an example in which βN ≈ 4is reached together with very good confinement (H89 = 3).

Concerning ITB physics, a clear reduction of turbulencewithin the ITB and at its foot point has been found usingreflectometry, although the turbulence level does not go downto the diagnostic noise level at low frequencies, indicatingthat not all modes are suppressed [20]. A further point ofinterest is the observation that the foot point of the ITB isusually located just outside the shear reversal zone and the steepgradient zone then extends across the shear reversal zone [19].This is of particular importance concerning the prospects forstationarity of an ITB discharge with large bootstrap current,since a location of the foot point inside the shear reversal zonewould point to a tendency for the ITB to shrink in time.

While the results mentioned so far mainly concernpronounced ion ITBs, we have also made progress inestablishing electron ITBs with ctr-electron cyclotron currentdrive (ECCD). Previously, it had been found that ctr-ECCDcould lead to strong electron ITBs, but the ITBs had a tendencyto collapse. By using higher Ip and ne (i.e. a smaller fractionof driven current), it has been possible to sustain the durationof the ITB phase and reach Te values in excess of 20 keV [19].

4. MHD stability

4.1. Edge localized modes

A major programmatic point on ASDEX Upgrade is thepreparation of H-mode scenarii with tolerable ELMs, sinceextrapolations for the transient power loads onto the ITERdivertor due to regular type I ELMs give rise to concerns. One

1574

Overview of ASDEX Upgrade results

Figure 5. Typical QH mode discharge in ASDEX Upgrade with an ELM-free phase of more than 2 s. Note that between 3.85 and 4.1 s, noECE data are available due to signal cut-off (- - - -).

0

1

2

3

4

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Time (s)

Confinement factor H89Normalized pressure βN

H light fromdivertor(a.u.)

α

0

2

4

6

q95Stored energy (100 kJ)Density (10 )19

Plasma current (100 kA)Heating power (MW)

β HN

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Time (s)

0

4

8

12

0

5

10

15

Figure 6. ITB discharge with H-mode edge reaching transiently βN = 4 and H89 = 3. The first ELM ends the high performance phase.

way to accomplish this is the operation with type II ELMs [21].The operational space for type II ELMs in ASDEX Upgradeis characterized by high δ, closeness to double null and highdensity. Recent studies indicate that closeness to double nullis required, whereas the constraint of high δ, which is to someextent coupled to the closeness to double null in ASDEXUpgrade, has recently been reduced to 〈δ〉 � 0.35. Also, thetype II operational space is characterized by somewhat elevatedq95 with respect to the ITER value, but has been extendeddown to q95 = 3.5 when operated close to double null [22].MHD stability analysis using carefully reconstructed equilibriaincluding an estimate of the edge current has shown that in thetype II ELM configuration, the peeling–ballooning stabilityfor lower mode numbers is improved and also, the width ofthe eigenfunction for given n decreases due to the higher edgeshear [23,24]. Both effects have a tendency to lead to smallerELMs, consistent with the observation of type II ELMs underthese conditions.

Another regime with small ELMs is the EDA-regimefound on ALCATOR C-Mod [25]. Experiments to establishdimensionless similar discharges between ASDEX Upgradeand C-Mod have been performed; however, they arecomplicated by the fact that the H-mode is not accessible inthe dimensionless similar ASDEX Upgrade discharge and,

at the higher heating power required in ASDEX Upgrade,ELM-free phases or type I ELMs occur [26]. This willrequire further studies, in particular because, as mentionedabove, in JET-ASDEX Upgrade similarity experiments, theL–H threshold has been found to scale in dimensionless similardischarges.

Experiments were also performed to increase the type IELM frequency by injection of shallow pellets that have littleeffect on ne, but instantaneously trigger a type I ELM. It couldbe shown that, by injecting pellets at a frequency higher thanthe natural type I ELM frequency, the ELM frequency could besignificantly increased [27]. An example is shown in figure 7.Since the ELM energy loss associated with these ELMs isconsistent with the scaling derived from variation of the naturalELM frequency (νELM × δW ≈ const. at constant Pheat), thismay open a route to mitigation of the peak heat loads for givenheating power.

As mentioned above, the ELM-free (QH)-mode regimewas established on ASDEX Upgrade. The EHO that was foundto be characteristic for this regime in DIII-D is also observed inASDEX Upgrade during the ELM-free phase. Figure 8 showsan example where up to the 11th harmonic can be seen in thespectrum of a Mirnov coil. A feature previously not seen isshown on the right-hand side of figure 8. Here, data acquired

1575

H. Zohm et al

from magnetic coils with a very high frequency response upto 1 MHz and sampled at 2 MHz are shown together withthe envelope of the EHO. Note the strongly nonsinusoidalcharacter of the radial magnetic field oscillation which pointsto a poloidally localized structure, consistent with the existenceof many harmonics. As can be seen, each cycle of the EHOcoincides with a high-frequency burst at 350–400 kHz. Theburst frequency actually decreases in time and therefore pointsto a fast particle drive of these oscillations. It is at presentnot clear, how this high frequency oscillation is linked to theEHO, but its high frequency seems to be related to the EHOfrequency [18]. However, a role of fast particles in the EHOphysics might explain why it is found only at low density andwith ctr-NBI, where first orbit losses are much higher than inthe co-case.

The fast magnetic measurements shown above alsoshowed interesting new features about MHD connected toELMs [28]. While type III ELMs did not exhibit any other

#15420 #155203.2 4 4.8 5.6Time [s]

n (10 m )e20 -3

W (MJ)MHD0.5

1

1.1

0.9

0.4

D outer divα

3 Hz

20 Hz

Figure 7. Comparison of two discharges with same external controlparameters. While #15420 shows long ELM free phases with largeELM events in between, injection of shallow pellets at 20 Hz in#15520 provokes ELMs at this frequency.

2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 5

200

time [s]

100

20

10

50

f [kH

z]

quiescentphase

FB

FB2

n=1

n=2

n=3

n=4n=5n=6

n=7n=8

n=9n=10

n=11

ASDEX Upgrade #16111ELMyH-mode

FB3

-2-1012

-20

0

2040

120

130

140

150

2.0812 2.0814 2.0816 2.0818 2.0820 2.0822time [s]

0.100.12

0.14

0.16

2.0810

Soft X-ray intensity (peripheral chord)

Radial magnetic field pickup coil (f > 100 kHz)

Radial magnetic field pick-up coil (filtered and integrated)

H-alpha intensity (outer divertor)

-40

[a.u

.][a

.u.]

[a.u

.][W

/ m

]

2

ASDEX Upgrade #16112

Figure 8. EHO in QH-mode. The low frequency mode shows harmonics up to n = 11 (left part, fishbone activity and its harmonics are alsoseen, marked ‘FB’). The right part shows that this is correlated with high frequency bursts transporting particles across the separatrix.

significant precursor activity than the already well-known70–100 kHz precursor, type I ELMs, which show pronouncedlow-frequency precursors usually only with ctr-NBI, exhibit apreviously unresolved high frequency activity. This is shownin figure 9, where two examples from co- and ctr-NBI heatedH-mode discharges are shown.

A high frequency oscillation can clearly be seen for 1–2 msbefore the actual ELM. The frequency is different betweenco- and ctr-NBI, pointing towards a contribution of both ω∗

e andvE×B due to the beams, which add up for ctr-NBI, but have tobe subtracted for co-NBI. In fact, the same local m number of16 is found in both cases. Note that this is a local m measuredin the midplane outside, and the actual m may be much highersince the local poloidal wavelength is highest on the midplaneoutside. It should be noted that this oscillation precedes type IELMs, but actually does not always grow explosively beforeELMs, so that it may be related to the saturation of edgepressure before an ELM occurs, but not necessarily to the MHDmode(s) responsible for the ELM itself.

4.2. Neoclassical tearing modes

ASDEX Upgrade continues to study ways to avoid, removeor mitigate the effects of neoclassical tearing modes (NTMs).Previous experiments demonstrated the possibility to stabilizeNTMs with DC co-ECCD, but without full recovery ofconfinement and β if the NBI power was kept constant [29].Now, with higher NBI power, βN could be significantlyincreased above the NTM onset level without reappearance ofthe mode [30]. In addition, the confinement properties of thedischarge recover to the value before NTM onset. However,this may be due to the use of the off-axis NBI sources whichhave a tendency to produce peaked density profiles [2] andtherefore counteract the confinement degradation by ECRH.

An example is shown in figure 10. At even higher βN ,the NTM reappeared in spite of the ECCD; however, this may

1576

Overview of ASDEX Upgrade results

0.20.40.60.81.01.2

f [M

Hz]

0

0.20.6 #16054

3.0225 3.0235 3.0245 3.0255 3.0265 3.0275time (s)

ctr-inj.,5 MW NI

Dα

magnetic coil

#15875

4.0115 4.0125 4.0135 4.0145 4.0155 4.0165time (s)

co-inj.,5 MW NI

Dα

magnetic coil

0.20.40.60.81.01.2

f [M

Hz]

0

0.20.6

Figure 9. High frequency magnetic oscillations prior to type I ELMs with co-NBI (left) and ctr-NBI (right).

NBI Power (MW) #14989 510

ECRH Power (MW*10)

Toroidal Field -2.25-2.15

Normalized Beta 13

Neutron Rate (1e15/s) 0.51.5

1.6 2 2.4 2.8 3.2 3.6 4 4.4time (s)

H-Factor ITER98(p,y)0.81

n=2 amplitude( a.u) 0

Figure 10. NTM stabilization by DC ECCD on the high field side. The phases in which the (3, 2) NTM exists are marked by the shadedareas. With increased NBI power, βN can be raised significantly above the onset value and confinement recovers.

be due to a mismatch between deposition and island due tothe higher Shafranov shift and calls for feedback controlleddeposition which will become possible in the near future, whena fast steerable ECRH launcher will be installed in ASDEXUpgrade.

An interesting finding concerning NTMs was thediscovery of the so-called frequently interrupted regime (FIR)NTMs, in which, for sufficiently high β, an (m, n) NTM neverreaches its saturated island width because it is periodicallyreduced due to interaction with a (m + 1, n + 1) mode [31,32].In this regime, the confinement reduction due to the NTM issignificantly reduced and it may be a possible operating regimewith tolerable NTMs.

It was found that in discharges with peaked ne profile, theNTM onset β is unusually low, which may be explained by thefact that ∇n contributes more strongly to the bootstrap currentthan ∇T [33]. This effect has now been included in the NTMonset scalings by using ( 2

3T ∇n+ 13n∇T )/∇p rather than βN in

the evaluation of the perturbed bootstrap that drives the mode.Also, experiments were carried out where the heating powerwas ramped down in order to determine β when the mode isturned off [34, 35]. This scaling is shown in figure 11. It can

0.00 0.04 0.08 0.120

2

4

6

8

ρ*

NTM onset

NTM decay

j /

j [

a.u.

]bs

p

l

Figure 11. Scaling of the perturbed bootstrap current at NTM onsetand switch off versus ρ∗. Both curves show a linear dependence.

be seen that a marked hysteresis exists between the onset andthe switch-off, but the previously found ρ∗-dependence stillholds.

4.3. Sawtooth tailoring with ECRH

Sawteeth play a key role in determining the central confinementof energy and particles [36]. On the one hand, a loss ofsawteeth leads to improved confinement, on the other hand, itmay lead to intolerably high central impurity concentrations.

1577

H. Zohm et al

inversion radius

completestabilization

HFS LFSctr-ECCD

ECR deposition in ρ-0.6 -0.4 -0.2 0 0.2

0

1

2

3

4

5

6

7

τ

/ τS

T w

ith E

CC

DS

T w

/o E

CC

D

inversion radius

complete stabilization

HFS LFSco-ECCD

ECR deposition in ρ0.4-0.4 -0.2 0 0.2

0

1

2

3

4

5

6

7

τ

/ τS

T w

ith E

CC

DS

T w

/o E

CC

D

Figure 12. Ratio of sawtooth repetition period with and without ECCD as a function of the ECCD deposition (negative values of ρ indicatedeposition on the HFS). Complete stabilization can be achieved by co-ECCD outside the inversion radius or central ctr-ECCD.

Therefore, a method to tailor the sawtooth amplitude andrepetition frequency is highly desirable. This has specialimportance in the presence of fast particles, that act on sawtoothstability themselves. Experiments were conducted in ASDEXUpgrade in which 0.8 MW of ECRH power was applied to anH-mode discharge heated with 5 MW of NBI. The location ofthe ECRH deposition was varied by slow ramps of the toroidalfield. The results are summarized in figure 12.

It can be seen that co-ECCD leads to an increase in thesawtooth period when deposited just outside the sawtoothinversion radius. With 800 kW of ECRH/ECCD, it is actuallypossible to completely stabilize the sawteeth under thesecircumstances. For central deposition, a reduction in sawtoothperiod occurs. Similar behaviour is observed with pure ECRH,which is consistent with a local decrease of resistivity dueto ECRH that also generates co-current. Conversely, forctr-ECCD, central deposition stabilizes the sawteeth, clearlydemonstrating the importance of the local current generatedby ECCD. These findings are consistent with the hypothesisthat a decrease in current density gradient at the q = 1 surfacetends to stabilize sawteeth. Note that, contrary to stabilizationof sawteeth by fast particles, the amplitude does not increasehere. This method has already been used to create sawtooth-free discharges for impurity transport investigations; futurestudies will also examine the effect of sawtooth tailoring onNTM onset by affecting the seed island size.

4.4. Disruptions

Disruptions continue to be an important point of investigationon ASDEX Upgrade. A major concern for ITER are the forcesconnected with halo currents induced by the movement of avertically unstable plasma during a disruption. On JET, it hadbeen found that a large asymmetric radial force componentcan occur during disruptions that can considerably shift thevacuum vessel horizontally. On ASDEX Upgrade, carefulmeasurements were performed to see if this also occurs, butthe lateral forces were found to be only a minor fraction of thevertical forces occuring during a disruption [37]. Thus, it hasnot been possible yet to establish a connection between lateral

force and the toroidal asymmetry in halo current that is usuallymeasured in ASDEX Upgrade.

Experiments on disruption mitigation focused on theapplication of an impurity gas puff at 1–3 times atmosphericpressure as an alternative to the previously used ‘killer’ pellet.From figure 13, it can be seen that with 3 × 1021 particles ofNe injected into a disruption, the forces were reduced up toroughly a factor of 3 due to the faster Ip decay and the smallervertical displacement which reduce the halo currents. Thus,Ne gas injection is an attractive alternative to a killer pellet.

5. Exhaust and plasma wall interaction

5.1. Operation with Divertor IIb

As mentioned above, the outer leg of the ASDEX Upgradedivertor was redesigned to accommodate higher δ plasmaswith good pumping and power handling capability. The designprinciple that had led to the previous Divertor II design waskept: vertical targets with close baffling ensure high divertordensity and thus enable strong radiative losses from this region,mainly due to C radiation. Consequently, it was found that thepower handling capability of Divertor IIb is as good as that of itspredecessor [38]: although a somewhat higher heat flux to thetarget plates is observed in Divertor IIb under otherwise similarmain plasma conditions, this can be attributed to the changedgeometry that leads to a reduction of the wetted surface dueto a steeper field line intersection angle. In fact, by correctingthese effects and calculating the parallel heat flux, we arrive atvery similar values for Divertor II and Divertor IIb. Figure 14shows the geometry and a comparison of heat fluxes in bothdivertors.

Another important property of the divertor is its pumpingcapability. Comparison between the old and new configurationshow only a modest change in the He compression, i.e. thefavourable properties of Divertor II were preserved [38]. Thenew Divertor IIb exhibits the same favourable properties as itspredecessor, but allows for more flexibility in plasma shapingwhich has become increasingly important for many aspects.

1578

Overview of ASDEX Upgrade results

time (s)time (s) time (s) time (s)

no gas

#14503#15434

-0.4

-0.2

0.2

0.4

0.6

0.80

-0.4

-0.2

m

#15462 #15484

1 1.2 1.43.35 3.45 3.55 3.65 3.75 3.55 3.65 3.75 3.85 3.95 3.75 3.85 3.95 4.05 4.15

a.u.

Ip

zcurr

100 ms100 ms100 ms100 ms

straingauge

3 x 10 particles

211.5 x 10 particles

210.75 x 10 particles

21

Figure 13. Disruption forces for disruptions in which Ne gas was puffed to mitigate the loads. With 3 × 1021 particles, the forces are greatlyreduced.

1.10 1.15 1.20 1.25 1.30length across surface (m)

0

1

2

3

4

5

6

pow

er d

ensi

ty (

MW

/m ) (shifted on new

strike-point position)

# 14330 4.10 s

# 13920 4.20 s

transitionmoduleheat fluxnot calcu-lated

1.00 1.01 1.02

0

20

40

60

80

100

para

llel h

eat f

lux

(M

W/m

)2

2

ρpol

0.99

# 13920 4.20 s

# 14330 4.10 s DIV IIbDIV II

Figure 14. Geometry of Divertor II (grey) and Divetor IIb (black) showing the better fit of high δ equilibria into Divertor IIb. The rightfigures show power density and parallel heat flux under similar conditions (ELMy H-mode with 5 MW of NBI) for both divertors.

5.2. Tungsten as first wall material

Recent results on the co-deposition of T in C plasma facingcomponents as well as the observation of T-enriched C-flakesin the louvers of the JET divertor have drawn increasingattention to alternative wall materials. In 1996, ASDEXUpgrade had been successfully operated with a W-divertor.However, C as the main impurity prevailed, pointing towardsa significant role of the C-covered inner heat shield as animpurity source. Thus, we started to successively cover theinner heat shield tiles with W. Operation with up to 7.1 ofthe 8.2 m2 of the inner heat shield covered with W has, underroutine operation conditions, not shown any detrimental effect.In particular, the W-concentration rarely exceeded 10−5, whichis the critical number for ITER. Conditions in which thisnumber was exceeded were either related to direct contactof the plasma with the inner wall or conditions in which theelectron density showed a strong tendency to peak, such aswith low central heating. As has been mentioned above, inthese cases, addition of central heating led to a large reductionof the central W-concentration [11].

Erosion of W from the inner heat shield showed apronounced asymmetry over the tiles that led to the conclusionthat it is mainly due to ion bombardment (i.e. D and impurities)which, due to the guiding of the particles by the magneticfield, can lead to shadowing of certain areas [39]. This waseven found when the plasma was relatively far away from theinner wall, which demonstrates that plasma reaches far acrossthe separatrix, a fact that also manifests in the observation oftwo different decay lengths of the SOL density [14]. Thismay be due to the bursty nature of SOL transport which canexpel particles further than would be expected from ordinarydiffusion.

Figure 15 shows a view of the inner heat shield coveredwith W-coated tiles. Also shown are traces from an experimentwhere the plasma was moved towards the central column andthen away from it. From the observed increase in W-influxand the concomitant increase in the W-concentration in themain plasma, a penetration probability for W eroded from theinner wall to enter the main plasma can be derived [11]. Thisnumber is found to be of the order of 1%, consistent withmodelling using the DIVIMP [40]. The analysis of migration

1579

H. Zohm et al

double-tiles

divertor-baffle

2000

2001

2000

1999 inne

r ga

p (c

m)

8

4

12

2.0 3.0 4.0 5.0t (sec)

Γ (

10

at./

m s

)W

182

0

2

4

c (

10

)W

-6

0

2

1

2.0 3.0 4.0 5.0t (sec)

n (

10

/m )

20e

3_

T (keV

)e0

0

2

4

0.0

0.5

1.0

Figure 15. Inner wall with W-coated tiles (left) and traces from an experiment where the plasma was brought near the inner wall todetermine the relation between W-influx and W-content. The position scan does not influence global plasma parameters such as ne, Te.

and redeposition patterns is less clear due to the fact that themain W-erosion actually comes from the plasma ramp-downphase at the end of the discharge when the hot plasma columnleans on the inner wall [41]. It should be pointed out that C isstill present in these discharge and significantly contributes tothe radiation; radiation control in a C-free environment remainsa challenge to be addressed in the future, when full W coverageis implemented.

6. Scenario integration

The physics elements described above are all importantsteps towards an integrated scenario for ITER. However, theintegration itself, i.e. the combination of good confinement,high stability and benign power exhaust at good pumpingcapability, becomes more and more important, since it isclear that some of the solutions for individual points arenot mutually compatible. Therefore, the development ofintegrated scenarii is a major focus of the ASDEX Upgradeexperimental programme.

For the conventional H-mode scenario, the remainingissues are the NTM stability (see section 4.2) and themitigation of ELM effects on the divertor as addressed insection 4.1. However, recent developments aim at improvingthe performance above the standard H-mode in view ofsteady state, i.e. noninductively driven tokamak discharges,the so-called ‘advanced tokamak’ (AT) line. The route ofdevelopment here is to maximize the fraction of bootstrapcurrent, which requires high βpol and to reduce the totalcurrent which requires enhanced confinement, expressed, e.g.as H89 = τE/τE,ITER89.

Figure 16 shows the commonly used figure of merit,βNH89, for various ASDEX Upgrade scenarii. As can be seen,ITB scenarii reach very high values, but only for short durationdue to the stability limitations discussed in section 3.4. In theso-called ‘improved H-mode’ and the ‘high βN H-mode’, highperformance has been achieved in stationary state over manyconfinement times. In fact, although the highest neutron ratein ASDEX Upgrade is reached in ITB discharges, the highestQequiv is actually achieved in the ‘improved H-mode’ [42].

The improved H-mode is characterized by flat centralshear and the absence of sawteeth while fishbones maintainthe stationarity of the q-profile [43]. The confinementimprovement mainly comes from density peaking while thetemperature profiles are still similar to usual H-mode profiles,but with a relatively high pedestal temperature as boundarycondition. A major drawback of this scenario was thelimitation to relatively low densities ne/nGW � 0.5. Asnoted before, operation at higher δ allows us to obtain goodconfinement at high fractions of nGW. Thus, by usinghigher δ, it was possible to extend the performance of the‘improved H-mode’ regime to ne/nGW ≈ 0.8–0.9. Again,these discharges do not have sawteeth and the ne profile isslightly peaked. Actually, higher δ also proved to be beneficialto lower q95 in the improved H-mode regime, which was notpossible before due to the onset of sawteeth with lower q95.The degree of peaking can be controlled by central ICRHheating as described in section 3.2. This regime also showsimproved stability properties with respect to the improvedH-mode regime (with the main limitation being NTM onset)and is called ‘high βN -regime’ [44, 45]. It can be obtainedin a stationary state at values up to βN = 3.5 [46]. Thismay partly be explained by to the relatively high collisionality,but stability also profits from shaping: increased shaping isexpected to affect the stability index �′, and it allows higherβN for given βpol, which is the actual quantity limited byNTMs. The beneficial effect of δ is also clearly documentedin βNH89n/nGW in the right graph of figure 16, where at lowδ ≈ 0.1–0.2, the value of βNH89n/nGW does not exceed 4, butcan be in excess of 5.5 for high δ ≈ 0.35–0.45, which reflectsthe above mentioned possibility to achieve good confinement athigh Greenwald fraction. Note that the region of intermediateδ ≈ 0.2–0.35 was not studied in detail so far, and is thus notfully populated in the figure.

Figure 17 shows an example of a high βN discharge. Inaddition to the favourable performance in steady state, it canalso be seen that the conditions for this scenario, i.e. high δ

and closeness to double null in combination with high ne arealso compatible with type II ELMs, which can be seen fromthe plots of deposited power on the target plates: although bothframes are obtained in phases with equal heating power, the

1580

Overview of ASDEX Upgrade results

Figure 16. Performance parameter βN × H89 for various high performance scenarii in ASDEX Upgrade (left). The right figuredemonstrates the beneficial effect of increasing δ in achieving higher densities at high performance.

βN

Dα

H92-P

# 14521

4

3

2

1

00 1 2 3 4 5 6 7

Time (s)

n /neGW

2.2-2.25 s 3.9-3.95 s

βN

Dα

H92-P

# 14521

4

3

2

1

00 1 2 3 4 5 6 7

Time (s)

n /ne eGW

2.2-2.25 s 3.9-3.95 s

e

Figure 17. High βN -discharge with type II ELMy edge. The power flux to the divertor plates as shown by the inserts exhibits type I peaksearly in the discharge (left), but is continuous later on in the type II phase (right).

distinct peaks due to type I ELMs exceed 10 MW m−2 whereasin the later phase with type II ELMs, the heat flux is continuousand below 5 MW m−2.

For ITER, this scenario therefore shows main advantagescompared to the usual H-mode scenario, i.e. the possibility torun at lower Ip and higher bootstrap fraction at similar fusionpower (note that Pfus/Ploss only depends on the ratio H/q),but reduced peak heat flux on the divertor, and therefore toextend the duration of the discharge. However, the bootstrapfraction is still smaller than in an ITB discharge with same β.Therefore, ITB discharges are still the prime candidate forsteady state tokamak operation, but the high βN scenario maybe considered an intermediate step on this route. It is thereforeof great importance to understand if the main ingredients, i.e.the density peaking, improved NTM stability and type II ELMedge, scale in a consistent way to ITER. This will be the subjectof further work.

7. Summary and conclusions

Recent ASDEX Upgrade results have enabled progress in theunderstanding of transport, stability and exhaust. An improved

understanding of heat and particle transport in the plasma corehas been gained using a stiff temperature model and coupling ofheat and particle transport coefficients. Concerning stability,active control strategies for both ELMs and NTMs start toemerge as well as regimes in which the detrimental effects ofthese instabilities are mitigated. It has also been shown thatW can be considered a serious candidate for a suitable firstwall material in reactor-grade devices. A scenario has beendemonstrated in which many of the ITER requirements aresimultaneously met or even exceeded.

Future work will focus on improved understanding ofunderlying physics as well as integration of the differentphysics elements to a consistent scenario. The future hardwareupgrades, i.e. an extension of the ECRH system to higherpower and longer pulse length at variable frequency and anincreasing W coverage of the plasma facing components willenable further progress in this direction.

References

[1] Gruber O. et al 2001 Overview of ASDEX Upgrade resultsNucl. Fusion 41 1369

1581

H. Zohm et al

[2] Stabler A. et al 2002 The role of neutral beam injectiongeometry in advanced discharge scenarios on ASDEXUpgrade 29th EPS Conf. on Plasma Phys. and Control.Fusion (Montreux, Switzerland) vol 26B (ECA) O-5.03

[3] Nguyen F. et al 2002 ICRF ion heating with mode conversionon the tokamak ASDEX Upgrade 29th EPS Conf. onPlasma Phys. and Control. Fusion (Montreux, Switzerland)vol 26B (ECA) P-1.045

[4] Tardini G. et al 2002 Comparison of theory based transportmodels with ASDEX Upgrade data Nucl. Fusion 42 258

[5] Ryter F. et al 2001 Experimental characterisation of theelectron heat transport in low density ASDEX Upgradeplasmas Phys. Rev. Lett. 86 5498

[6] Lopez-Cardoso N.N. et al 1995 Plasma Phys. Control. Fusion37 799

[7] Ryter F. et al Electron Heat Transport in ASDEX Upgrade:Transport and Modelling this conference, IAEA-CN-94EX/C4-2Ra

[8] Tardini et al 2002 Theory-based modelling of ASDEXUpgrade discharges with ECH modulation Nucl. Fusion42 L11

[9] Stober J. et al 2001 Dependence of the density shape on theheat flux profile in ASDEX Upgrade high density H modesNucl. Fusion 41 1535

[10] Stober J. et al Dependence of Particle Transport on theHeating Profile on ASDEX Upgrade this conference,IAEA-CN-94 EX/C3-7Rb

[11] Neu R. et al 2002 Impurity behaviour in the ASDEX Upgradedivertor tokamak with large area tungsten walls PlasmaPhys. Control. Fusion 44 811

[12] Neu R. et al 2003 J. Nucl. Mater. 313–316 116–26[13] Dux R. et al 2003 J. Nucl. Mater. 313–316 1150–55[14] Neuhauser J. et al 2002 Transport into and across the

scrape-off layer in the ASDEX Upgrade divertor tokamakPlasma Phys. Control. Fusion 44 855

[15] Kallenbach A. et al Edge Transport and its Interconnectionwith Main Chamber Recycling in ASDEX Upgrade thisconference, IAEA-CN-94 EX/P4-05

[16] Suttrop W. et al 2002 Testing H-mode parameter similarity inJET and ASDEX Upgrade 29th EPS Conf. on Plasma Phys.and Control. Fusion (Montreux, Switzerland) vol 26B(ECA) P-1.030

[17] Greenfield C.M. 2001 Quiescent double barrier regime in theDIII-D tokamak Phys. Rev. Lett. 86 4544

[18] Suttrop W. et al 2003 Plasma Phys. Control. Fusion 45 1399[19] Peeters A.G. et al Internal Transport Barriers in ASDEX

Upgrade this conference, IAEA-CN-94 EX/P4-03[20] Conway G.D. et al 2001 Turbulence reduction in internal

transport barriers on ASDEX Upgrade Plasma Phys.Control. Fusion 43 1239

[21] Stober J. et al 2001 Type II ELMy H modes on ASDEXUpgrade with good confinement at high density Nucl.Fusion 41 1123

[22] Gruber O. et al Tolerable ELMs in Conventional andAdvanced Scenarios et ASDEX Upgrade this conference,IAEA-CN-94 EX/C2-1

[23] Horton L.D. et al 2002 Pedestal physics at ASDEX Upgrade29th EPS Conf. on Plasma Phys. and Control. Fusion(Montreux, Switzerland) vol 26B (ECA) P-2.047

[24] Saarelma S et al 2003 MHD stability analysis of type II ELMsin ASDEX Upgrade Nucl. Fusion 43 262

[25] Greenwald M. et al Characterization of enhanced Dα

high-confinement modes in Alcator C-Mod Phys. Plasmas6 1943

[26] Suttrop W. et al Parameter Similarity Studies in JET, ASDEXUpgrade and ALCATOR C-Mod this conference,IAEA-CN-94 EX/P5-07

[27] Lang P.T. et al 2002 Advanced high field side pellet refuellingand mitigation of ELM effects in ASDEX Upgrade 29thEPS Conf. on Plasma Phys. and Control. Fusion (Montreux,Switzerland) vol 26B (ECA) P-1.044

[28] Martines E. et al 2002 ELM-related high frequency MHDactivity in ASDEX Upgrade 29th EPS Conf. on PlasmaPhys. and Control. Fusion (Montreux, Switzerland) vol 26B(ECA) P-1.038

[29] Zohm H. et al 2001 Neoclassical tearing modes and theirstabilization by electron cyclotron current drive in ASDEXUpgrade Phys. Plasmas 8 2009

[30] Gantenbein G. et al 2002 On the stabilisation ofneoclassical tearing modes with ECRH at high βN inASDEX Upgrade 29th EPS Conf. on Plasma Phys. andControl. Fusion (Montreux, Switzerland) vol 26B (ECA)P-1.036

[31] Gunter S. et al 2001 High-confinement regime at high βN

values due to a changed behaviour of the neoclassicaltearing modes Phys. Rev. Lett. 87 275001

[32] Gude A. et al 2002 Temporal evolution of neoclassical tearingmodes and its effect on confinement reduction in ASDEXUpgrade Nucl. Fusion 43 833

[33] Stober J. et al 2001 Optimization of confinement, stability andpower exhaust of the ELMy H-mode in ASDEX UpgradePlasma Phys. Control. Fusion 43 A39

[34] Maraschek M. et al Stabilisation scaling of neoclassicaltearing modes during power ramp-downs in ASDEXUpgrade Plasma Phys. Control. Fusionsubmitted

[35] Gunter S. et al Neoclassical Tearing Modes on ASDEXUpgrade: Improved Scaling Laws, High Confinement atHigh βN and New Stabilization Experiments thisconference, IAEA-CN-94 EX/S1-4

[36] Muck A. et al 2002 Sawtooth tailoring experiments withECRH in ASDEX Upgrade 29th EPS Conf. on PlasmaPhys. and Control. Fusion (Montreux, Switzerland) vol 26B(ECA) P-1.037

[37] Pautasso G. et al Disruption Studies on ASDEX Upgrade thisconference, IAEA-CN-94 EX/P4-14

[38] Neu R. et al 2002 Properties of the new divertor IIb inASDEX Upgrade Plasma Phys. Control. Fusion 44 1021

[39] Krieger K. et al 2003 J. Nucl. Mater. 313–316 327–32[40] Geier A. et al 2003 J. Nucl. Mater. 313–316 1216–20[41] Rohde V. et al Operation of ASDEX Upgrade with Tungsten

Coated Walls this conference, IAEA-CN-94 EX/D1-4[42] Zohm H. et al 2002 Neutron production in high performance

scenarios in ASDEX Upgrade 29th EPS Conf. on PlasmaPhys. and Control. Fusion (Montreux, Switzerland) vol 26B(ECA) P-1.043

[43] Gruber O. et al 1999 Stationary H-mode discharges withinternal transport barrier on ASDEX Upgrade Phys. Rev.Lett. 83 1787

[44] Sips A.C.C. et al 2002 Progress towards steady-state advancedscenarios in ASDEX Upgrade Plasma Phys. Control.Fusion 44 A151

[45] Na Y.-S. et al 2002 Transport in high normalized betadischarges on ASDEX Upgrade Plasma Phys. Control.Fusion 44 1285

[46] Sips A.C.C. et al 2002 Steady state advanced scenarios atASDEX Upgrade Plasma Phys. Control. Fusion44 B69

1582