Experimental conditions for suppressing Edge Localised Modes by magnetic perturbations in ASDEX Upgrade

Experimental conditions for suppressing Edge Localised Modes by magnetic perturbations in ASDEX Upgrade W Suttrop1, A Kirk2, V Bobkov1, M Cavedon1, M Dunne1, N Leuthold1, R M McDermott1, H Meyer2, R Nazikian3, C Paz-Soldan4, D A Ryan2, E Viezzer1,5, M Willensdorfer1, the ASDEX Upgrade1and MST12Teams 1Max Planck Institute for Plasma Physics, Boltzmannstrasse 2, 85748 Garching, Germany 2CCFE Culham Science Centre, Abingdon, Oxon, OX14 3DB, UK 3Princeton Plasma Physics Laboratory, PO Box 451, Princeton, NJ 08543-0451, USA 4General Atomics, PO Box 85608, San Diego, CA 92186-5608, USA 5Dept. of Atomic, Molecular and Nuclear Physics, University of Seville, 41012 Seville, Spain The Edge Localised Mode (ELM) instability presents a significant challenge for the power handling in a next step fusion device, and mitigation of ELMs is a clear requirement for ITER [1]. One of the most promising techniques, the complete suppression of ELMs by resonant magnetic perturbations (MP), has been pioneered in DIII-D (see, e.g. the overview in Ref. [2]) and subsequently been reproduced, among other tokamaks, in ASDEX Upgrade (AUG) [3, 4] in plasmas with moderately elevated triangularity. In a current journal article [5], the empirical criteria to access ELM suppression in ASDEX Upgrade are investigated. They can be summarised as follows: 1. The poloidal spectrum of the MP must be aligned for best plasma response from weakly stable kink-modes, which amplify the perturbation [6], 2. The plasma edge density must be below a critical value, 3.3×1019 m−3. It is not clear to date how this condition can be expressed in dimensionless (and therefore extrapolatable) parameters, 3. The pedestal pressure must be kept sufficiently low to avoid the return of (small) ELMs. 4. The edge safety factor q95 lies within a certain window, q95 =3.57 −3.95 as has been identified so far. 5. Within the range of plasma rotation encountered so far, no apparent threshold of plasma rotation for ELM suppression is found. This includes cases with large cross field electron flow in the entire pedestal region. Here, we report additional recent experimental observations that pertain to (a) the role of plasma triangularity and (b) the role of resonant rational surfaces for accessing ELM suppression. The plasma scenario employed here is similar to that used in Refs. [3] and [5]. Plasmas are heated with 6 MW neutral beam (NB) power (fully co-Ipinjected) and up to 2.4 MW electron cyclotron heating power (ECRH) at 140 GHz, absorbed centrally at the third harmonic (X3) at the toroidal field Bt=−1.83 T. Central heating is successful to maintain peaked temperature profiles and prevent accumulation of heavy impurities from the all-metal wall of AUG in the absence of a strong gas puff. The plasma current is Ip=0.885 MA or Ip=0.9 MA for an edge safety factor about q95 =3.8 and 3.74, respectively. The AUG in-vessel saddle coils are used here exclusively to produce n=2 MPs with a differential phase ∆Φ =90◦between upper and lower coil rings, which is found optimal to reach ELM suppression in the above plasma scenario [5]. Figure 1 shows time traces of discharge 34835, which starts up like the pulses described in Refs [3] and [5]. ELM suppression is reached at about t=3.0 s, followed by a slow ramp of the ELM suppression plasma shape with upper triangularity δu∼0.2 towards a shape with triangularity of δu∼0.05 at t=5.0 s. This final shape is similar to the configuration in which ELMs were mitigated but not suppressed [3, 4]. 1See A Kallenbach et al, Nucl Fus 57 (2017) 102015 2See H Meyer et al, Nucl Fus 57 (2017) 102014 45th EPS Conference on Plasma Physics O2.107 W. Suttrop et al , 45th EPS Plasma Physics Conference, Prague 2018, O2.107 2 0.2 0.4 0.6 0.8 1.0 1.2 kA MP coil current Upper triangularity (x5) H98Py2 2 6 10 14 kA Outer divertor thermocurrent 2.5 3.0 3.5 4.0 10 m Peripheral line-averaged electron density 0.4 0.6 0.8 1.0 1.2 keV electronion 2.5 3.0 3.5 4.0 4.5 5.0 5.5 time [s] 0.1 1.0 nue: temperaturePedestal electronion collisionalityPedestal 19 -3 ASDEX Upgrade #34835 ELM suppression ELMs ELMs Figure 1: Discharge 34835, testing ELM suppression during a ramp of the upper triangularity. ELM suppression, as seen by the absence of ELM spikes in the outer divertor thermocurrent (second panel) is maintained until the end of the shape ramp. The end of ELM suppression is marked with a phase of infrequent ELMs at t= 5.0 s, before frequent ELMs take over after t=5.32 s. In this intermediate phase (t= 5.0−5.32 s), the baseline of the divertor thermocurrent is still elevated and similar to that of the ELM suppression phase, indicating that there is significant inter-ELM transport. The inter-ELM losses slow down the pedestal recovery after each ELM and are thus the likely reason for the observed reduced ELM frequency. The transport mechanism that governs the low plasma density and the stationarity during ELM suppression is obviously not precluding the occurrence of ELMs. At t=3.2 s, the MP coil current is stepped down from 1.2 kA to 1.0 kA, which results in a temporary increase of peripheral plasma density ne,p, pedestal ion and electron temperatures (Ti,ped,Te,ped) and hence pedestal pressure. The confinement factor is also increasing. In the course of the plasma shape ramp, ne,pas well as Ti,ped,Te,ped are decreasing, most strongly between t=3.3 s and 4.0 s, when δu=0.16 −0.12. Clearly, the pedestal pressure is governed by both plasma shaping and the MP strength which are main parameters to optimise confinement of ELM suppression plasmas. 0.70 0.75 0.80 0.85 0.90 0.95 1.00 1.05 normalised poloidal flux Ψn 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 electron temperature [keV] q=5/2 6/2 7/2 8/2 Te ASDEX Upgrade #34398 t=3.5-3.6 s ELM suppression t=4.05-4.15 s ELMy Figure 2: Edge temperature profiles in ELMy H-mode and ELM suppression. We now consider the role of resonant rational surfaces for ELM suppression, i.e. surfaces where q=m/nwhere mis the poloidal mode number of a non-vanishing spectral component of the MP. Resonant surfaces are special because of the possibility of a resistive (tearing) response to the applied MP which can cause a modification of the magnetic topology and lead to braking torque and enhanced radial transport along the magnetic field around islands or in an ergodised volume. In a model proposed in Refs. [7], profile flattening at a resonant surface near the top of the H-mode barrier can prevent the edge gradient region from expanding towards the core plasma and potentially keep the plasma edge stable against an ELM crash if the resonant surface is located at a suitable position near the top of the gradient region. Such a model implies a sensitivity of ELM suppression to the value of the edge safety factor q95 which is indeed observed in DIII-D [8]. Also in ASDEX Upgrade, the loss of ELM suppression can be provoked by small variations of q95. Figure 2 shows edge electron temper- 45th EPS Conference on Plasma Physics O2.107 W. Suttrop et al , 45th EPS Plasma Physics Conference, Prague 2018, O2.107 3 ature (Te) profiles for one case of Ref. [5], discharge 34398, taken during two time intervals, t=3.5−3.6 (q95 =3.88 −3.94, ELM suppression) and t=4.05 −4.15 (q95 =4.01 −4.02, ELMing). The q=8/2 surface is near the top of the gradient region in the suppressed case, es expected in the model. The radial shift that corresponds to the qchange which causes the transition to ELMing H-mode is very small. 35 40 45 50 55 a.u. Inner divertor AXUV diode bolometer 2.5 2.6 2.7 2.8 2.9 3.0 1e19 m^-3 Peripheral line-averaged density 0.2 0.4 0.6 0.8 keV Pedestal top electron temperature ECE Trad (scrape-off-layer) 3.2 3.4 3.6 3.8 4.0 time [s] -30 -20 -10 0 10 km/s ASDEX Upgrade #34398 Impurity ion (B ) toroidal rotation 5+ ψ = 0.93 ψ = 0.97 n n ELM suppression sawtooth crashes (ECE) Figure 3: Time traces during backtransition from ELM suppression to ELMing H-mode. Times traces of this discharge around the backtransition are shown in Fig. 3. The coloured areas correspond to the time intervals of the Te profiles in Fig. 2. ELM activity is well diagnosed by several measurements, notably inner divertor radiation measured with diode bolometers (top panel) and spikes in the electron cyclotron emission (ECE) in a channel with cold ECE resonance in the scrape-off layer (third panel). These spikes are believed to be partly due to downshifted emission from energetic electrons in the confined plasma caused by reconnection events in sawtooth and ELM crashes. The transition from ELM suppressed to ELMing at t= 3.6 s is sharp and occurs within a few ms or less. After the transition, the inner divertor bolometer signal drops and Teand nebegin to rise, indicating a loss of the transport mechanism that maintains ELM suppression. Soon after (from t=3.65 s), small ELMs appear which grow as the pedestal pressure increases. The bottom panel shows the toroidal impurity (B5+) rotation velocity vtor, taken at Ψn=0.93 (pedestal top, at q=7/2) and Ψn=0.97 (gradient region, at q=8/2). A sharp change of the rate-of-change of vtor is detected at the backtransition, t=3.6 s near q=7/2. At q=8/2, vtor lags behind by a few ms. If the backtransition is caused by a torque change at a rational surface then this observation indicates that the torque acts more likely at q=7/2 rather than at q=8/2. In ASDEX Upgrade, repetitive transitions between ELM suppression and ELMing H- mode can be observed, although rarely. Figure 4 shows time traces with three such cycles. During ELM suppression, the plasma density and impurity ion rotation at the pedestal and in the core drop, while in the ELMing phases they recover. It is interesting to note that in the gradient region, Ψn=0.95, the vtor reverses into the direction opposite to the plasma current only during ELMy phases. This reversal occurs very quickly at the times of the backtransition. We may speculate that the nature of the dominant torque input onto the plasma by the MP changes. Strong braking towards zero rotation is consistent with a resonant plasma response to the MP, while the reversal of the rotation direction may be indicative of approaching intrinsic rotation associated with neoclassical toroidal viscosity (NTV). NTV torque is usually much weaker than resonant torque and would be noted only in the absence of the latter. The repetitive transitions are reminiscent of limit cycle oscillations such as dithering H-mode transitions. However, the state variable governing the transitions is unknown to date. 45th EPS Conference on Plasma Physics O2.107 W. Suttrop et al , 45th EPS Plasma Physics Conference, Prague 2018, O2.107 4 3.8 3.9 2.6 2.8 3.0 3.2 10 m 2 4 6 8 10 12 14 kA 4.6 4.7 4.8 4.9 Time (s) 0 20 40 60 km/s Edge safety factor q Peripheral line-averaged density MP coil current (x0.1) Outer divertor thermocurrent Impurity ion (B ) toroidal rotation 5+ ψ = 0.74 ψ = 0.89 ψ = 0.95 n n 95 19 -3 2.4 3.7 suppression suppressionsuppression ELMy ELMy ELMy n ASDEX Upgrade #33120 Figure 4: A discharge with repeated transition between ELMy and ELM-suppressed H-mode. The B5+impurity ion rotation changes sign in early ELMy phases. We observe a change of q95 (top panel of Fig. 4) which is probably caused by current profile changes due to the transitions and hence implies that the (edge) current diffusion time scale governs the time between transitions. However, the maximum value of q95 ≈3.85, reached at the backtransition, is not near a bound of the n=2 suppression window found in Ref. [5]. In summary, we can compare our observations in ASDEX Upgrade with the model of Refs. [7]. The existence of q95 access windows for ELM suppression, the position of a resonant surface at q=8/2 near the gradient top, and the braking torque during ELM suppression support the existence of a resonant response during ELM suppression. However, the resonant response can be shielded by cross-field flows. In ASDEX Upgrade, surprising insensitivity of ELM suppression access to flow variations is found, including cases with strong electron rotation ωe,⊥everywhere in the edge region [5], which is predicted in two-fluid MHD to be the cause of strong shielding. Theoretical work on the plasma response is ongoing and new insight can be expected from non-linear and kinetic modelling. Experimentally, a wideband density fluctuation in the gradient region is found in ELM suppression phases which might play a role for the transport across the edge barrier in the absence of ELMs [9]. Continued experimental work in ASDEX Upgrade will be needed to support and check models for ELM suppression as well as to identify critical parameters for access of and performance with ELM suppression as a basis for extrapolations to fusion reactor plasmas. This work has been carried out within the framework of the EUROfusion Consortium and has received funding from the Euratom research and training programme 2014-2018 under grant agreement No 633053. The views and opinions expressed herein do not necessarily reflect those of the European Commission. References [1] A Loarte et al Nucl. Fusion 54 (2014) 033007 [2] T Evans et al Plasma Fusion Res. 7(2012) 2402046. [3] W Suttrop et al, Plasma Phys. Control. Fus. 59 (2017) 014050 [4] R Nazikian et al, IAEA FEC 2016, post-deadline. [5] W Suttrop et al, accepted for Nucl. Fusion, doi = 10.1088/1741-4326/aace93, http://arxiv.org/abs/1804.00933 [6] Y Q Liu et al, APS-DPP 2016; Physics of Plasmas 24, 056111 (2017) [7] M R Wade et al, IAEA FEC 2012, EX/3-1; P B Snyder et al, Phys. Plas. 19 (2012) 056115 [8] M E Fenstermacher et al, Phys. Plas. 15 (2008) 056122; O Schmitz et al, Nucl. Fus. 52 (2012) 043005; M J Lanctot et al, Nucl. Fus. 53 (2013) 083019 [9] N Leuthold et al, this conference, P1.1109 45th EPS Conference on Plasma Physics O2.107