P. A. Schneider

The effect of a metal wall on confinement in JET and ASDEX Upgrade

In both JET and ASDEX Upgrade (AUG) the plasma energy confinement has been affected by the presence of a metal wall by the requirement of increased gas fuelling to avoid tungsten pollution of the plasma. In JET with a beryllium/tungsten wall the high triangularity baseline H-mode scenario (i.e. similar to the ITER reference scenario) has been the strongest affected and the benefit of high shaping to give good normalized confinement of H98???1 at high Greenwald density fraction of fGW???0.8 has not been recovered to date. In AUG with a full tungsten wall, a good normalized confinement H98???1 could be achieved in the high triangularity baseline plasmas, albeit at elevated normalized pressure ?N?>?2. The confinement lost with respect to the carbon devices can be largely recovered by the seeding of nitrogen in both JET and AUG. This suggests that the absence of carbon in JET and AUG with a metal wall may have affected the achievable confinement. Three mechanisms have been tested that could explain the effect of carbon or nitrogen (and the absence thereof) on the plasma confinement. First it has been seen in experiments and by means of nonlinear gyrokinetic simulations (with the GENE code), that nitrogen seeding does not significantly change the core temperature profile peaking and does not affect the critical ion temperature gradient. Secondly, the dilution of the edge ion density by the injection of nitrogen is not sufficient to explain the plasma temperature and pressure rise. For this latter mechanism to explain the confinement improvement with nitrogen seeding, strongly hollow Zeff profiles would be required which is not supported by experimental observations. The confinement improvement with nitrogen seeding cannot be explained with these two mechanisms. Thirdly, detailed pedestal structure analysis in JET high triangularity baseline plasmas have shown that the fuelling of either deuterium or nitrogen widens the pressure pedestal. However, in JET-ILW this only leads to a confinement benefit in the case of nitrogen seeding where, as the pedestal widens, the obtained pedestal pressure gradient is conserved. In the case of deuterium fuelling in JET-ILW the pressure gradient is strongly degraded in the fuelling scan leading to no net confinement gain due to the pedestal widening. The pedestal code EPED correctly predicts the pedestal pressure of the unseeded plasmas in JET-ILW within ?5%, however it does not capture the complex variation of pedestal width and gradient with fuelling and impurity seeding. Also it does not predict the observed increase of pedestal pressure by nitrogen seeding in JET-ILW. Ideal peeling ballooning MHD stability analysis shows that the widening of the pedestal leads to a down shift of the marginal stability boundary by only 10?20%. However, the variations in the pressure gradient observed in the JET-ILW fuelling experiment is much larger and spans a factor of more than two. As a result the experimental points move from deeply unstable to deeply stable on the stability diagram in a deuterium fuelling scan. In AUG-W nitrogen seeded plasmas, a widening of the pedestal has also been observed, consistent with the JET observations. The absence of carbon can thus affect the pedestal structure, and mainly the achieved pedestal gradient, which can be recovered by seeding nitrogen. The underlying physics mechanism is still under investigation and requires further understanding of the role of impurities on the pedestal stability and pedestal structure formation.

H-mode pedestal scaling in DIII-D, ASDEX Upgrade, and JET

Multi device pedestal scaling experiments in the DIII-D, ASDEX Upgrade (AUG) and JET tokamaks are presented in order to test two plasma physics pedestal width models. The first model proposes a scaling of the pedestal width Δ/a ρ * 1/2 to ρ * based on the radial extent of the pedestal being set by the point where the linear turbulence growth rate exceeds the ExB velocity. In the multi device experiment where ρ * at the pedestal top was varied by a factor of four while other dimensionless parameters where kept fixed, it has been observed that the temperature pedestal width in real space coordinates scales with machine size, and that therefore the gyroradius scaling suggested by the model is not supported by the experiments. This density pedestal width is not invariant with ρ * which after comparison with a simple neutral fuelling model may be attributed to variations in the neutral fuelling patterns. The second model, EPED1, is based on kinetic ballooning modes setting the limit of the radial extent of the pedestal region and leads to Δ βp 1/2 . All three devices show a scaling of the pedestal width in normalised poloidal flux as Δ βp 1/2 , as described by the kinetic ballooning model, however on JET and AUG this could not be distinguished from an interpretation where the pedestal is fixed in real space. Pedestal data from all three devices have been compared with the predictive pedestal model EPED1 and the model produces pedestal height values that match the experimental data well.

Differences in the H-mode pedestal width of temperature and density

A pedestal database was built using data from type-I ELMy H-modes of ASDEX Upgrade, DIII-D and JET. ELM synchronized pedestal data were analysed with the two-line method. The two-line method is a bilinear fit which shows better reproducibility of pedestal parameters than a modified hyperbolic tangent fit. This was tested with simulated and experimental data. The influence of the equilibrium reconstruction on pedestal parameters was investigated with sophisticated reconstructions from CLISTE and EFIT including edge kinetic profiles. No systematic deviation between the codes could be observed. The flux coordinate system is influenced by machine size, poloidal field and plasma shape. This will change the representation of the width in different coordinates, in particular, the two normalized coordinates ΨN and r/a show a very different dependence on the plasma shape. The scalings derived for the pedestal width, Δ, of all machines suggest a different scaling for the electron temperature and the electron density. Both cases show similar dependence with machine size, poloidal magnetic field and pedestal electron temperature and density. The influence of ion temperature and toroidal magnetic field is different on each of and . In dimensionless form the density pedestal width in ΨN scales with , the temperature pedestal width with . Both widths also show a strong correlation with the plasma shape. The shape dependence originates from the coordinate transformation and is not visible in real space. The presented scalings predict that in ITER the temperature pedestal will be appreciably wider than the density pedestal.

High-density H-mode operation by pellet injection and ELM mitigation with the new active in-vessel saddle coils in ASDEX Upgrade

Recent experiments at ASDEX Upgrade demonstrate the compatibility of ELM mitigation by magnetic perturbations with efficient particle fuelling by inboard pellet injection. ELM mitigation persists in a high-density, high-collisionality regime even with the strongest applied pellet perturbations. Pellets injected into mitigation phases trigger no type-I ELM-like events unlike when launched into unmitigated type-I ELMy plasmas. Furthermore, the absence of ELMs results in improved fuelling efficiency and persistent density build-up. Pellet injection is helpful to access the ELM-mitigation regime by raising the edge density beyond the required threshold level, mostly eliminating the need for strong gas puff. Finally, strong pellet fuelling can be applied to access high densities beyond the density limit encountered with pure gas puffing. Core densities of up to 1.6 times the Greenwald density have been reached while maintaining ELM mitigation. No upper density limit for the ELM-mitigated regime has been encountered so far; limitations were set solely by technical restrictions of the pellet launcher. Reliable and reproducible operation at line-averaged densities from 0.75 up to 1.5 times the Greenwald density is demonstrated using pellets. However, in this density range there is no indication of the positive confinement dependence on density implied by the ITERH98P(y,2) scaling.

Quantification of the impact of large and small-scale instabilities on the fast-ion confinement in ASDEX Upgrade

The confinement fast ions, generated by neutral beam injection (NBI), has been investigated at the ASDEX Upgrade tokamak. In plasmas that exhibit strong sawtooth crashes, a significant sawtooth-induced internal redistribution of mainly passing fast ions is observed, which is in very good agreement with the theoretical predictions based on the Kadomtsev model. Between the sawtooth crashes, the fishbone modes are excited which, however, do not cause measurable changes in the global fast-ion population. During experiments with on- and off-axis NBI and without strong magnetohydrodynamic (MHD) modes, the fast-ion measurements agree very well with the neo-classical predictions. This shows that the MHD-induced (large-scale), as well as a possible turbulence-induced (small-scale) fast-ion transport is negligible under these conditions. However, in discharges performed to study the off-axis NBI current drive efficiency with up to 10 MW of heating power, the fast-ion measurements agree best with the theoretical predictions that assume a weak level anomalous fast-ion transport. This is also in agreement with measurements of the internal inductance, a Motional Stark Effect diagnostic and a novel polarimetry diagnostic: the fast-ion driven current profile is clearly modified when changing the NBI injection geometry and the measurements agree best with the predictions that assume weak anomalous fast-ion diffusion.

Fast-ion transport and neutral beam current drive in ASDEX upgrade

The neutral beam current drive efficiency has been investigated in the ASDEX Upgrade tokamak by replacing on-axis neutral beams with tangential off-axis beams. A clear modification of the radial fast-ion profiles is observed with a fast-ion D-alpha diagnostic that measures centrally peaked profiles during on-axis injection and outwards shifted profiles during off-axis injection. Due to this change of the fast-ion population, a clear modification of the plasma current profile is predicted but not observed by a motional Stark effect diagnostic.The fast-ion transport caused by MHD activity has been studied in low collisionality discharges that exhibit strong modes. In particular due to sawtooth crashes, significant radial redistribution of co-rotating fast-ions is observed which can very well be described by the Kadomtsev model. In addition, first tomographic reconstructions of the central 2D fast-ion velocity space in the presence of sawtooth crashes allow the investigation of the pitch dependence of the mode-imposed redistribution: a stronger redistribution of mainly co-rotating fast ions is observed than of those with smaller pitch values.

Analysis of temperature and density pedestal gradients in AUG, DIII-D and JET

A comparison of the AUG and DIII-D temperature pedestals showed significant differences between electrons and ions. For high collision rates the ions are coupled to the electrons and show very similar pedestal top values and gradients. For lower collision rates both decouple and the ion pedestal becomes less steep. The electron temperature gradient scales linearly with its pedestal top value. This trend is independent of collisionality and plasma shape. The normalized total pressure gradient α shows strong correlations with the plasma shape in a way expected by peeling–ballooning theory. The different behaviours of the electron temperature gradient only and the total pedestal pressure gradient suggests a limit for the electron temperature pedestal different from linear edge magnetohydrodynamic stability.

Pedestal and edge localized mode characteristics with different first wall materials and nitrogen seeding in ASDEX Upgrade

A comparison of ASDEX Upgrade (AUG) discharges performed with carbon and the full tungsten wall shows that the pedestal performance at low triangularity is not altered without gas puffing. The pedestal electron pressure is the same for both wall materials as is the confinement. With the tungsten wall the natural density is higher even without an additional gas puff. In typical operation with gas puffing the density is again higher in tungsten. This results in a higher collisionality with the tungsten wall. Pedestal pressure and plasma confinement, however, are not degraded until very large amounts of deuterium are puffed.The edge localized mode (ELM) crash in typical AUG discharges is observed to be composed of two independent phases. This is observed for both the carbon and the tungsten wall. The 1st phase of the crash is unaffected by scans of the plasma parameters as long as the pedestal pressure remains constant. The duration of the 2nd phase is strongly anti-correlated with the separatrix density and can be suppressed by the application of nitrogen seeding for divertor cooling. A consistent explanation for the two phases of the ELM crash does not seem possible when considering only the pre-ELM pedestal profiles. The scrape off layer (SOL) plasma provides the necessary free parameter for a consistent explanation, indicating the importance of the SOL in understanding the ELM crash evolution.

Overview of recent pedestal studies at ASDEX Upgrade

New or upgraded diagnostics of the edge transport barrier allow investigations of the dominant transport mechanisms in the pedestal. The density build-up after the L–H transition can be explained with a mainly diffusive edge transport barrier. A small inward convection term improves the agreement between modelling and experiment, but its existence cannot be confirmed due to the uncertainty in the neutral sources. Measurements of the impurity ion flow asymmetry as well as the edge current density are in agreement with neoclassical modelling. The inter-ELM pedestal recovery was traced with ideal peeling–ballooning modelling, which shows that the stability boundary moves closer to the operational point as the pedestal becomes wider. Gyrokinetic modelling of the different phases reveal that density gradient driven trapped electron modes are dominant during the early recovery, while electron temperature gradient modes or kinetic ballooning modes determine the temperature gradient in the final phase. Microtearing modes are modelled and also experimentally determined at the top of the pedestal. Non-linear coupling between modes could explain the failure of ideal linear MHD modelling.

Impact of Te and Ne on edge current density profiles in ELM mitigated regimes on ASDEX Upgrade

ELM resolved edge current density profiles are reconstructed using the CLISTE equilibrium code. As input, highly spatially and temporally resolved edge electron temperature and density profiles are used in addition to data from the extensive set of external poloidal field measurements available at ASDEX Upgrade, flux loop difference measurements, and current measurements in the scrape-off layer. Both the local and flux surface averaged current density profiles are analysed for several ELM mitigation regimes. The focus throughout is on the impact of altered temperature and density profiles on the current density. In particular, many ELM mitigation regimes rely on operation at high density. Two reference plasmas with type-I ELMs are analysed, one with a deuterium gas puff and one without, in order to provide a reference for the behaviour in type-II ELMy regimes and high density ELM mitigation with external magnetic perturbations at ASDEX Upgrade. For type-II ELMs it is found that while a similar pedestal top pressure is sustained at the higher density, the temperature gradient decreases in the pedestal. This results in lower local and flux surface averaged current densities in these phases, which reduces the drive for the peeling mode. No significant differences between the current density measured in the type-I phase and ELM mitigated phase is seen when external perturbations are applied, though the pedestal top density was increased. Finally, ELMs during the nitrogen seeded phase of a high performance discharge are analysed and compared to ELMs in the reference phase. An increased pedestal pressure gradient, which is the source of confinement improvement in impurity seeded discharges, causes a local current density increase. However, the increased Zeff in the pedestal acts to reduce the flux surface averaged current density. This dichotomy, which is not observed in other mitigation regimes, could act to stabilize both the ballooning mode and the peeling mode at the same time.