T. PÜTTERICH et al. 1

Transcription

1 T. PÜTTERICH et al. THE ITER BASELINE SCENARIO INVESTIGATED AT ASDEX UPGRADE T. PÜTTERICH, O. SAUTER, V. BOBKOV, M. CAVEDON, M.G. DUNNE, L. GUIMARAIS, A. KAPPATOU, P.T. LANG, M. MANTSINEN, R.M. MCDERMOTT, J. SCHWEINZER, J. STOBER, W. SUTTROP, M. WILLENSDORFER, THE EUROFUSION MST TEAM 5 AND THE ASDEX UPGRADE TEAM Max-Planck-Institut für Plasmaphysik, Boltzmannstr., 57 Garching, Germany Ecole Polytechnique Fédérale de Lausanne (EPFL), Swiss Plasma Center, CH-5 Lausanne, Switzerland Instituto de Plasmas e Fusão Nuclear, IST, Universidade de Lisboa, Portugal Barcelona Supercomputing Center (BSC), Barcelona, Spain & ICREA, Barcelona, Spain 5 see the author list H. Meyer et al, Nucl. Fusion 57, (7) see the author list A. Kallenbach et al, Nucl. Fusion 57, 5 (7) of corresponding author: thomas.puetterich@ipp.mpg.de Abstract A survey of recent investigations of the ITER baseline scenario (BLS) on the ASDEX Upgrade tokamak is presented. These investigations target key features and issues of the ITER baseline scenario and relate the observations to specific physics findings from other work. The investigations focus on plasmas featuring strong shaping, a safety factor of q 95. and a normalized plasma pressure of. β N., which implies very low ELM frequencies and thus, challenges for stable operation and impurity control. The high shaping allows for access to the desired Greenwald fractions of - at reasonable confinement. Under these conditions the performance and behavior of the plasmas during impurity seeding, ELM control, pellet fueling and low torque input are investigated. For the first time in AUG the ITER BLS investigations could be extended to low collisionality at q 95 =. and collisionalities only a factor - above the ITER collisionality were achieved. This was possible with the application of magnetic perturbations enabling a strong reduction of the deuterium gas puff while the plasma edge stayed stable. In these plasmas a decoupling of electrons and ions was possible and a clearly better confinement as compared to the high density plasmas was observed.. INTRODUCTION Investigations on the ITER baseline scenario (ITER BLS) are performed in several devices []. A fundamental drawback of all current experiments is that a perfect simulation of the ITER plasmas can never match all important parameters at once. Even the dimensionless parameters cannot all be matched at the same time due to practical and physics limitations. Partly this can be summarized in the problem that the Greenwald fraction is an important limit especially at the plasma edge, while in the plasma core collisionality is more relevant. It is impossible to match both parameters in devices with vastly different sizes. Nevertheless, it is worthwhile to try to match as many ITER parameters as possible in order to find, identify and solve issues, which might be relevant for ITER. The present work is a continuation of [] at ASDEX Upgrade (AUG), and it focuses on the scenarios developed therein, namely the ITER BLS at q 95 =. and a ITER-BL-like scenario with reduced plasma current at q 95 =.. The application of central wave heating is necessary for high-z impurity control and requires specific magnetic fields, i.e.. T for central ICRF minority heating at MHz or. T for central X ECRH heating using GHz gyrotrons. Other frequencies are available for ICRH and ECRH but do require either too large plasma current for the envisaged q 95 or do not propagate in the plasma at the relatively high densities of the ITER BLS. One issue identified in [], which is revisited in the present work, is depicted in figure. The energy confinement factor at the ITER relevant plasma pressure ofβ N =. (q 95 =.) andβ N =. (q 95 =.) is below the expected value of and, respectively. This observation is valid for discharges performed after AUG was converted to a full-w machine in 7 [], while corresponding discharges before the changeover to W allowed for sufficient confinement []. Thus, it has been speculated [] that the confinement reduction is related to measures of impurity control such as a drastic increase of the deuterium gas puff.. NITROGEN SEEDING FOR CONFINEMENT IMPROVEMENT In order to overcome the reduced confinement observed in AUG not only in the ITER BLS, but also in various scenarios after the conversion of the wall to W, a common recipe is the seeding of low-z impurities such as nitrogen (N) []. N-seeding is able to improve confinement via two distinct mechanisms [5, ]: First, an increased effective ion charge Z e f f reduces the bootstrap current at the pedestal and thus, increases the pedestal stability leading to slightly larger pedestal pressure and temperature. Second, confinement is recovered, which was lost before due to a changed poloidal fueling pattern, an observation made after the wall material was changed to W. In particular the fueling pattern using a W wall was influenced via a high-density region at the entrance to the inner

2 EX/P- FIG.. Confinement properties of the ITER baseline scenario at AUG from earlier investigations, adapted from []. divertor at the high-field side (HFSHD region). This region with high electron density is created via the recycling of neutrals, which are then ionized via a power flux from the main plasma which travels from the low-field side midplane within the scrape-off layer towards the inner divertor entrance. Seeding nitrogen reduces this power flux, which lowers locally plasma fueling and leads to a changed separatrix density. In effect, the density pedestal is slightly shifted inwards via the N-seeding leading to higher pedestal pressures []. Note that for experiments with a carbon (C) first wall the residual C impurity was sufficient to mitigate the power towards the inner divertor. In figure some time traces of interest of discharge #, in which N-seeding was applied from.5 s, are depicted. In figure, the plasma current I p, the stored energy W mhd, the normalized confinement factor. # P NBI [MW] NDFD (Div.-Gauge) [E m - s - ] P ICRF [MW] β N line averaged n e [E9 m - ] : q 95 log(ν*/ν* ITER ) f GW T div [~ev] D-flux, valve [Ee/s] (d) N-flux, valve [Ee/s] FIG.. Time traces for an ITER BLS discharge (q 95 confinement is studied. =.) in which the effect of N-seeding on and the measurement of the divertor thermocurrent used as an ELM-monitor are depicted. In figure, the powers from neutral beam injection P NBI, ion cyclotron radio frequency heating P ICRF and radiated power P rad are depicted along the neutral divertor flux density (NDFD) as measured by a pressure gauge. In figure, the edge safety factor q 95, normalized plasma pressureβ N, the Greenwald density fraction f GW and the logarithm of the collisionalityν at mid-radius over the envisaged collisionality for ITERν ITER are presented. In figure (d), the divertor temperature T div derived from thermocurrents is presented along with the gas valve fluxes for deuterium and nitrogen and the line-averaged density measured by an interferometer, which views the plasma approximately across the midplane. Note, the dashed lines indicate the envisaged target values for the ITER BLS. The current and shaping flattop are reached at. s, when the heating power is reduced to the target level for the ITER BLS. In the following ms no changes are observed and the plasma is representative for other ITER BLS discharges with longer stationary phases. At.5 s N-seeding is applied, first at a low level and from.7 s a feedback mode of the N-valve is started, which tries to match the measured divertor temperature (T div ) to a value given to the control system. This leads to an almost stationary phase between. s and.5 s during which the N-seeding, T div, and

3 T. PÜTTERICH et al. the stored energy are almost constant. This phase is then stopped due to a planned reduction of heating. During this phase the confinement is following the same trend as observed for unseeded discharges (cf. figure ), while the Greenwald fraction is about. The exact confinement values of all presented discharges are summarized in figure in the last section. In order to identify the reason, why the N-seeding is not improving the performance as.5 HFS reflectometry z[m] LOS strike point #, s LOS, HFS baffle R[m] Strike Point[m]. Inner Divertor T div [~ev] ne [ m - ] N-flux, valve [Ee/s] D-flux, valve [Ee/s] AJ: point baffle log( ne[m-] ) FIG.. Spatially resolved analysis of the local electron density using spectroscopic measurements of the Stark broadening, before and during N-seeding. expected, the behavior of the HFSHD region is studied in figure. Indeed, it is observed that the N-seeding affects not only T div in the outer divertor, but also the electron density in the inner divertor, as measured spectroscopically via Stark broadening [7]. In figure the measurement geometry of two lines of sight (LOS) is presented. The analysis of LOS (cf. to [7]) allows to construct a density profile for each time point as depicted in figure. A drastic change is observed at about.7 s, when the N-seeding is strongly increased and T div in the outer divertor is reduced to a few ev (cf. figure ). When only focusing on the two LOS depicted in figure, it becomes obvious that at the inner divertor baffle the electron density is strongly reduced, while at the strike point the electron density remains approximately constant (cf. figure (d)). Note that the HFS reflectometer (cf. figure ) sees the upward extension of the HFSHD regions as observed earlier []. However, the relative shift of the density pedestal profile, w.r.t. the temperature pedestal profile as expected from conclusions in [], is not observed here and consequently the confinement stays unchanged. This behavior is reminiscent of observations at Alcator C-Mod [9], where a change of the HFSHD also is not correlated with a change of the separatrix density. This effect may be related to the shaping of the plasma, which is puzzling because the shaping is applied to allow for good confinement at high densities [] as observed for devices with carbon walls. Further investigations of the underlying physics are ongoing (e.g. [9]).. ELM-CONTROL VIA MAGNETIC PERTURBATIONS, GAS FUELING AND PELLET INJECTION Another important issue for the ITER BLS which is tackled with multiple approaches, are the huge ELMs. These are a result of strong shaping, low q 95 and small heating power just above the LH power threshold such that the observed ELMs [] are clearly larger than expected from []. Various strategies are investigated at AUG in dedicated experiments. The first strategy aims for ELM mitigation via the application of magnetic perturbations (MP) [, ]. This was tried for the available ITER BLS at q 95 =. and q 95 =. and the effects on the edge pedestal and thus, on the type-i ELM size were minor, possibly due to the very large collisionality (ν ) when matching the ITER Greenwald fraction f GW = 5 (at q 95 =.) or f GW =.9 (at q 95 =.). In order to amplify the effects, the investigations have been extended to q 95 =., because at the reduced plasma current of ka (at B t =. T) the currents in the MP-coils can be raised to. ka resulting in larger absolute and even more so larger relative MPs. In figure, time traces of a ITER BLS discharges at a reduced plasma current of ka are presented. P ECRH is the heating power from electron cyclotron resonance heating (ECRH). The applied toroidal mode number of the MP is n=. The MP structure in the upper ring of coils is shifted toroidally w.r.t. that in the lower ring of coils by and during the two phases indicated via the gray shading. During the MPs with a shift angle of a slight density pump-out with a concomitant confinement degradation of about % is observed, while the ELM

4 EX/P- #7 shift shift shift shift NDFD (Div.-Gauge) [E m - s - ] P NBI [MW] P ECRH [MW] Time [s] 5 5 fgw: 5 q 95 log(ν*/ν* ITER ) f GW D-flux, valve [Ee/s] line averaged n e [E9 m - ] Time [s] FIG.. Time traces for the ITER BLS at an elevated q 95 =., in which the effect MP-coils on ELMs is studied. The plasma current is reduced in order apply larger currents in the MP-coils maximizing the perturbation. β N (d) type changes to a smaller one and no further large type-i ELMs occur. When the shift angle is moved to, the density and stored energy increases above the original level and both seem to saturate probably limited by a small ELM-type, while occasionally small type-i ELMs occur with an ELM loss of about 5 7%. More rarely, large type-i ELMs with an ELM loss of % occur. The effect of density pump-out for a shift angle of is consistent with the existence of a maximal resonant field component in the plasma edge. The latter was determined by calculating the resonant field component in a vacuum calculation and then applying a correction for the plasma response according to []. The interpretation of the observation is not straightforward. However, consistently with other MP experiments [5] at high density the edge pedestal is not strongly degraded via the MPs and the type-i ELM size remains approximately unchanged. The qualitative new observation is the different ELM type. Note that it was observed in earlier experiments [], that small ELMs can also be accessed without MPs via minor shape adjustments. During the first phase ( shift) the small ELMs manage to replace the type-i ELMs fully, while in the second phase ( shift) the closeness to the stability boundary still leads to occasional type-i ELMs. In summary, all ITER BLS experiments at the correct Greenwald fraction show little impact of the MP-coils on the type-i ELM behavior and size, unless a transition to small ELMs is achieved. An involvement of fueling physics is plausible also because other experiments [] demonstrated that small ELMs for the same plasma shape are accessible with fueling via gas puffing, while pellet fueling leads to type-i ELMs. An explanation for this behavior is presented in []. These observations raised the question, whether the detrimental type-i ELMs can be avoided by plain gas puffing without MPs. To that end, neutral divertor flux densities (NDFD) beyond 5 m s were envisaged corresponding to gas valve fluxes beyond electrons/s in standard pumping configuration. In several discharges, very large NDFD could be achieved for moderate gas valve fluxes by turning off the cryo pump. This was also the case for #, which is depicted in figure 5-(d). This experiment was performed in the ITER BLS at q 95 =.. The deuterium gas puff is in the range of electrons/s and below, while achieving NDFD in the range of 7 m s. Also MPs (n=, at a resonant configuration) were applied, but it is obvious that the type-i ELMs disappear independently as at. s the small ELM phase starts, while the MPs are turned on at.5 s. The edge is deeply into the small ELM regime, as pellets, which are injected from 5.-. s cannot trigger a single type-i ELM. In discharge #7,depicted in figure 5(e)-(h), the same mechanism is observed for the ITER BLS at q 95 =.. In this case the cryo-pump was turned on and thus, a very large deuterium gas puff was applied in order to achieve sufficient NDFD. The large ELMs disappear at the beginning of the flattop at.s and do not reappear. It should be noted that a small nitrogen gas puff is applied from.-. s, but it is too small to have any notable effect such as increased radiation or lower divertor temperatures. However, there are other ITER BLS discharges, in which only the addition of nitrogen on top of the considerable deuterium gas puff is also able to cause the transition towards small ELMs. Such a transition is observed at about. s in discharge # visible in figure. Therefore, the nitrogen applied in the ITER BLS does not seem to have beneficial effects on the confinement, but may facilitate a transition to small ELMs. In any case, the experiments demonstrate that the transitions towards the small ELM regime is possible without MPs and possibly the application of MPs leading to the small ELM regime as observed in figure may be just a special case of more general physics. How this scales

5 T. PÜTTERICH et al. 5. # resonant MP-coils line averaged n e [E9 m - ] Pellets D-flux, valve [Ee/s] T div [~ev] Time [s] 5 7. NDFD (Div.-Gauge) [E m - s - ] NDFD (Div.-Gauge) [E m - s - ] P NBI [MW] P NBI [MW] P ICRF [MW] P ICRF [MW] q q β N β N log(ν*/ν* ) ITER log(ν*/ν*.: ITER ) f f GW.: GW.: (d) 5 #7 T div [~ev] D-flux, valve [Ee/s] N-flux, valve [Ee/s] line averaged n e [E9 m - ] Time [s] 5 FIG. 5. Time traces for the ITER BLS at q 95 =. (part a) and q 95 =. (part b), in which the effects of extremely large neutral divertor flux densities (NDFD) on ELMs are studied. (e) (f) (g) (h) to ITER is yet another question, the answer of which requires further insights into the mechanism of confinement enhancement via nitrogen and transitions to small ELM regimes. This question is also complicated by the fact that MPs may also degrade the edge pedestal and consequently reduce the type-i ELM size or may lead to ELM suppression for quite specific conditions [7]. Additionally, a transition to small ELMs seems to be also possible at low collisionality [, 9]. Thus, a final answer of what option is best used in ITER seems at this stage unclear. Another possibility of ELM mitigation is ELM-pacing. This was investigated at AUG by using small pellets, which are suitable for triggering ELMs. When applying pellets, several observations have been made. The injected pellets triggered ELMs and as a consequence it was possible to reduce the fueling gas puff in the ITER BLS by almost a factor of and the limit to this reduction has not yet been encountered. Still, the confinement is also affected by pellets, such that the ELM-frequency cannot be increased indefinitely, if a remedy for the confinement reduction is not found. Further experiments are planned, also because pellet fueling might be the only way of fueling ITER discharges reliably, due to a small penetration probability of thermal neutrals into the confined region. Again the extrapolation to ITER yields uncertainties, as pellets and gas puffing might be needed simultaneously depending on the fueling efficiency, the exact physics of entering a small ELM regime and exhaust physics for which the neutral distribution in the divertor is of importance.. PURE-WAVE HEATING IN ITER BLS In ITER the auxiliary heating characteristics provide a relatively large electron heating fraction and very limited torque as compared to its moment of inertia. Experiments at DIII-D [] suggest that weakly rotating plasmas at q 95 =. are unstable to neoclassical tearing modes (NTM). For AUG, weakly rotating plasmas can only be prepared via a large fraction of wave heating, i.e. a low fraction of neutral beam injection. Concomitantly, also the electron heating fraction is increased. However, this has little implications as most experiments in the ITER BLS are performed at ITER relevant Greenwald fractions leading to a strong coupling between electrons and ions such that the original heating fractions do not matter for most of the plasma radii. For the ITER BLS at q 95 =., pure-wave heating was demonstrated at ITER relevant shaping and Greenwald fraction. No NTMs formed and the flattop was stable. The ELM size did not change drastically and rather changed accordingly with the pedestal-top pressure. For extrapolation to ITER further insights and experiments are necessary, as the plasma β was slightly below ITER valuesβ N.7 and the collisionality was about a factor of 5 larger than is envisaged for ITER. Note that reduced plasma pressure, larger q 95 and possibly larger collsionality [, ] are all in favor of NTM stability and thus, the NTM stability might be more critical for the ITER BLS at q 95 =.. Experiments at q 95 =. with full shaping resulted in an increased density, due to the increased plasma current but also because pure-wave heating implies relatively small total heating power and thus, a small ELM frequency. As a consequence, the EC-waves could not be transmitted to their resonance positions in the core of the plasma due to a cut-off. As a remedy, experiments with less shaping were performed, such that the density of the discharges decreased staying slightly below the ITER value with values of <f GW <.7. At the same time the

6 EX/P- ELM frequency increased which is connected to a smaller ELM size. A comparison of the two plasma shapes is presented in figure. The ramp-up of the weakly shaped discharge #5 is still performed with neutral beam injection (NBI), but then in the course of the flattop the NBI is switched off and only used for short blips to provide rotation and ion temperature measurements via CXRS. As can be seen in figure the rotation is indeed strongly reduced. Again no modes are triggered, while the plasma pressure was too small withβ N =.5 and the collisionality is approximately a factor too large w.r.t. to the ITER values. #5 #9@.5s NBI dominated z [m] #5@.5s Toroidal rotation [krad/s] pure wave.5..5 R [m] - ρ pol FIG.. Comparison between the strongly shaped ITER BLS at AUG (#9) and the weakly shaped case from which the profiles in originate. Toroidal rotation profiles for the weakly shaped case during a mostly beam heated and a pure-wave heated phase. 5. ITER BLS IN AUG CLOSE TO ITER COLLISIONALITY At AUG the main focus has been put on matching f GW of ITER, however, many aspects of the core plasma physics are rather governed by collisionality. The mentioned aspects comprise mode stability, density peaking in the core, fast particle populations, coupling of electrons and ions and physics items related to these topics such as energy confinement. Matching collisionality corresponds to much lower f GW in the range of.-. By applying MP-coils to discharges right after boronization a density pump-out is obtained which then enables also ELM-suppression. As a reference, the discharges for ELM suppression studies [7] were used, but with a slight adjustment of the plasma shape in order to match that used in the ITER BLS investigations. Additionally, the q 95 value of the ITER BLS was attempted to be matched, which implies a further increase of the plasma current w.r.t. the reference discharge. In figure 7, the time traces of one of the development discharges are presented. From.5 s to 7. s the MP-coils were used in a most resonant configuration in order to pump-out density. The absence of large, low-frequent ELMs allows for the drastic reduction of the gas fueling, w.r.t. the ITER BLS investigations at high density, such that values of electrons/s lead to a stable discharge. As a result, a collisionality at mid-radius of only factor - larger than envisaged for ITER is achieved. The ramp-up before. s is unchanged from the reference discharge, but the plasma current is ramped again after. s in order to achieve q 95 =.. Note that from.5 s to. s MHD activity occurs which is visible in the stored energy and related signals and settles at about s. During the plasma current flattop from.5 s the heating mix is varied using NBI, ECRH and ICRH. For the phase at.-.5 s,. MW of ECRH and. MW of NBI heating is applied, resulting inβ N =. at =. The -factor seems too low, however, when judging it in the versusβ N plane (cf. figure the performance seems to be better than in comparable discharges at highν, even though the application of magnetic perturbations leads to density pump-out. This even clearer, when focusing at the short phase with ICRH heating around.5 s the energy confinement performs much better featuringβ N =. and =. A comparison of electron and ion temperature demonstrates that in the NBI+ECRH phase the ion temperature is clearly lower than the electron temperature (T i,. kev, T e,.5 kev), while during the phase including ICRH both temperatures are about equal (T i, T e, =.5 kev). This may have to do with the heating fractions, or confinement properties as has been observed in []. Still, the change in T e /T i demonstrates that the collisionality is low enough to decouple electrons and ions. The studies on the ITER BLS at low collisionality have just started at AUG and the next steps will be to further reduce q 95, study the confinement effects for various heating fractions and to move towards pure wave heating in order to investigate NTM stability.

7 T. PÜTTERICH et al. 7. P NBI [MW] resonant MP-coils # P ECRH [MW] P ICRF [MW] NDFD (Div.-Gauge) [E m - s - ] f GW resonant MP-coils q 95 β N log(ν*/ν* ITER ) line averaged n e [E9 m - ] (d) D-flux, valve [Ee/s] 5 7 FIG. 7. First experiments accessing the lowν regime at for the ITER BLS at q 95 =.. SUMMARY AND CONCLUSIONS The ITER BLS investigations at AUG have focussed on various aspects and were performed in strongly shaped plasmas at low q 95, namely q 95 =. and q 95 =.. These conditions provoke ELMs with very large energy drops (up to % of W mhd ) and low frequency (down to Hz). In AUG, earlier investigations revealed that at the ITER-relevant plasma pressureβ N =. and Greenwald fraction, the confinement factor 5 is considerably lower than aimed for. This behavior is used as a guideline for judging the plasma performance for the present investigations. Further aspects of the ITER BLS have also been investigated in the present work. First, the effect of N-seeding on the confinement has been studied. While clear effects of the N-seeding on the poloidal fueling distribution are observed, the expected confinement improvement is not observed. Thus, the rigid connection between a high density region at the baffle at the inner divertor to a shifted density pedestal seems to be broken for the ITER BLS discharges and a clear explanation for this observation is yet missing. Second, various methods of ELM mitigation have been studied at the ITER relevant Greenwald fractions, i.e. at core plasma collisionalities which are about -5 times larger than in ITER. The application of magnetic perturbations (MP) has minor effect on particle confinement and only mildly affects the size of type-i ELMs. However, a transition towards small ELMs either fully or partly replacing type-i ELMs is triggered in experiments with reduced plasma current (at q 95 =.) allowing for a % increase in current inside the perturbation coils. Similarly, small ELMs can be provoked with an extreme increase of the neutral density in the divertor without any MP, either via raising the flux through gas valves by about a factor of to or via reducing the effective pumping speed for deuterium by about a factor of via switching-off the cryo pump. Independently, the effectiveness of pellets for fueling and ELM-pacing was investigated and it was observed that most of the deuterium gas puff, which is needed for keeping up a stable ELM frequency, can be omitted as the pellets also keep triggering ELMs. As a forced high ELM-frequency also has detrimental effects on confinement, further investigations are proposed in order to find the optimal setting in terms of pellets size and frequency. Third, first investigations in a low collisionallity regime, i.e. close to the ITER relevant collisionality, have been undertaken. Using the MPs during the discharges right after boronization, the stability of the plasma edge does not need to be kept via a considerable deuterium gas puff. This allows for a reduction of the deuterium gas puff while also density pump out occurs both reducing collisionality. The resulting core plasma collisionality at mid-radius is only a factor - above that of the ITER core plasma. This enables new investigations at AUG, among others are NTM stability, importance of collisional coupling between ions and electrons, larger fast particle fractions and related physics. As observed in earlier studies on the ITER BLS for various machines including JET-ILW (e.g. []) matching the ITER collisionality instead of Greenwald fraction leads to better normalized confinement for the ITER BLS also at AUG. In principle this behavior was known for various scenarios at AUG with the W wall, but not for the ITER BLS shaping and q 95. In figure, data points from recent ITER BLS investigations are presented within the versusβ N plane and the two data points at lowν, i.e. close to ITER collisionality clearly break the trend of all other data points including the data from earlier investigations indicated with the gray shaded area. Looking at the two green data points at q 95 =., it seems straightforward to obtain the appropriate confinement factor = atβ N =. via more heating power. The trend that low collisionallity is beneficial for confinement has been also observed before in JET-ILW [] and it is also a general observation across machines

8 EX/P- H9(y,) alternative ITER BL (q 95 =.) low ν* ITER BL (q 95 =.),. 5,.5,. 7,. 7,5. pure-wave heating,.5 77,5.9 MP-coils 7,. small ELMs, large puff 9,.9 75,. alternative ITER BL (q95=.),. 7,. N-seeding H 9 / β N = const. 9,5.7 desired improvement q 95 =. q 95 =. q 95 = βn FIG.. Parameter Space for the ITER BLS for both q 95 =. and q 95 =.. All above described experiments are indicated with colored data points, the gray shaded points are from experiments before 5. in []. Known mechanisms of the confinement improvement at low collisionality range from core physics to pedestal physics and are not pinpointed in the present work. However, it should be noted that even though the MP coils lead to density pump-out, sufficient confinement seems to be obtained. As future work it will be attempted to extend the low collisionality regime also towards q 95 =., in order to match the full current ITER BLS. ACKNOWLEDGMENTS This work was partly performed within the framework of the EUROfusion Consortium and has received funding from the Euratom research and training programme - under grant agreement No 5. The views and opinions expressed herein do not necessarily reflect those of the European Commission. REFERENCES [] A. Sips et al., Nuclear Fusion 5, (). [] J. Schweinzer et al., Nuclear Fusion 5, 7 (). [] R. Neu et al., Physica Scripta T, (pp) (9). [] J. Schweinzer et al., Nuclear Fusion 5, (). [5] M. G. Dunne et al., Plasma Physics and Controlled Fusion 59, 7 (7). [] M. G. Dunne et al., Plasma Physics and Controlled Fusion 59, 5 (7). [7] S. Potzel et al., Plasma Physics and Controlled Fusion 5, 5 (). [] L. Guimarais et al., Nuclear Fusion 5, 5 (). [9] M. Dunne et al., this conference (EX/P-). [] ITER Physics Expert Groups, Nuclear Fusion 7, (7). [] A. Loarte et al., Plasma Physics and Controlled Fusion 5, 59 (). [] W. Suttrop et al., Fusion Engineering and Design, 9 (9). [] M. Teschke et al., Fusion Engineering and Design 9-97, 7 (5). [] D. A. Ryan et al., Plasma Physics and Controlled Fusion 59, 5 (7). [5] N. Leuthold et al., Plasma Physics and Controlled Fusion 59, 55 (7). [] G. Harrer et al., Nuclear Fusion 5, (). [7] W. Suttrop et al., Nuclear Fusion 5, 9 (). [] W. Suttrop et al., Plasma Physics and Controlled Fusion 59, 9 (7). [9] R. Nazikian et al., Nuclear Fusion 5, (). [] F. Turco et al., Nuclear Fusion 5, (). [] H. Zohm et al., Plasma Physics and Controlled Fusion 9, B7 (997). [] M. Maraschek, S. Günter, H. Zohm, and ASDEX Upgrade Team, Plasma Physics and Controlled Fusion, L (999). [] F. Sommer et al., Nuclear Fusion 55, (5). [] L. Frassinetti et al., Nuclear Fusion 57, (7).