Optimizing beam-ion confinement in ITER by adjusting the toroidal phase of the 3D magnetic fields applied for ELM control

Optimizing beam-ion confinement in ITER by adjusting the toroidal phase of the 3-D magnetic fields applied for ELM control L. Sanchis1,2, M. Garcia-Munoz2,3, E. Viezzer2,3, A. Loarte4, L. Li5,6, Y.Q. Liu7, A. Snicker1, L. Chen8, F. Zonca9, S.D. Pinches4, D. Zarzoso10 1Dept. of Applied Physics, Aalto University, FI-00076, Aalto, Finland 2Dept. of Atomic, Molecular and Nuclear Physics, University of Seville, Avda. Reina Mercedes, 41012 Seville, Spain 3Centro Nacional de Aceleradores CNA (Universidad de Sevilla, Junta de Andaluc´ıa, CSIC), Avda. Thomas A. Edison 7, 41092 Seville, Spain 4ITER Organization, Route de Vinon-sur-Verdon, CS 90 046, 13067 St Paul-lez-Durance Cedex, France 5College of Science, Donghua University, Shanghai 201620, China 6Member of Magnetic Confinement Fusion Research Centre, Ministry of Education, China 7General Atomics, PO Box 85608, San Diego, CA 92186-5608, United States of America 8IFTS, Zhejiang University, 310027, 310027, Hangzhou, China 9ENEA C.R. Frascati, CP 65-00044, Frascati, Italy 10 Aix-Marseille Universit´e, CNRS, PIIM, UMR 7345, Marseille, France E-mail: [email protected] Abstract. The confinement of Neutral Beam Injection (NBI) particles in the presence of n=3 Resonant Magnetic Perturbations (RMPs) in 15 MA ITER DT plasmas has been studied using full orbit ASCOT simulations. Realistic NBI distribution functions, and 3D wall and equilibria, including the plasma response to the externally applied 3D fields calculated with MARS-F, have been employed. The observed total fastion losses depend on the poloidal spectra of the applied n=3 RMP as well as on the absolute toroidal phase of the applied perturbation with respect to the NBI birth distribution. The absolute toroidal phase of the RMP perturbation does not affect the ELM control capabilities, which makes it a key parameter in the confinement optimization. The physics mechanisms underlying the observed fast-ion losses induced by the applied 3D fields have been studied in terms of the variation of the particle canonical angular momentum (δPφ) induced by the applied 3D fields. The presented simulations indicate that the transport is located in an Edge Resonant Transport Layer (ERTL) as observed previously in ASDEX Upgrade studies. Similarly, our results indicate that an overlapping of several linear and nonlinear resonances at the edge of the plasma might be responsible for the observed fast-ion losses. The results presented here may help to optimize the RMP configuration with respect to the NBI confinement in future ITER discharges. Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 2 1. Introduction In tokamak plasmas operating in H-mode [1], magnetohydrodynamic instabilities can lead to a large release of particles and energy to the plasma facing components. In particular, type-I edge localized modes (ELMs) [2,3], caused by the steep gradients at the plasma edge, are one of the main concerns for ITER [4,5,6] and the associated power fluxes need to be controlled since they can lead to a significant reduction of plasma component lifetime and to the contamination of the plasma by the impurities produced during ELMs. Among different techniques to mitigate ELMs, the application of resonant magnetic perturbations (RMPs) has successfully mitigated or even suppressed ELM instabilities and has been adopted as the main base line technique for ITER [7,8,9,10,11,12]. However, the symmetry-breaking fields applied for ELM control can lead to a degradation of plasma energy and particle confinement, particularly of fast-ions [13,14,15,16]. This is a concern for ITER since an excessive loss of fast particles reduces the effective Neutral Beam Injection (NBI) power available to heat the plasma and can lead to localized power fluxes [17]. In most present tokamaks, RMPs are generated by two or three toroidal rows of windowframe coils where the current flowing through the coils is modulated toroidally with a given symmetry and the relative toroidal phase between the waveforms can be adjusted to vary its effect on the plasma. The relative toroidal phase between toroidal rows is thus used to modify the poloidal mode spectrum of the perturbation, which has been observed to affect the plasma stability in several experiments such as MAST [18], ASDEX Upgrade [19] and DIII-D [20,21]. In the presence of these RMP fields, energetic particles are especially sensitive to this complex 3D magnetic field structure due to their relatively long mean free path and slowing down times [22,23,24,25,26]. However, as mentioned above, a good confinement of the energetic ions is crucial to ensure efficient heating of burning fusion plasmas since fast particles are the main source of plasma heating (NBI, alpha heating and ICRH heating) and of momentum (NBIs) into the plasma. ITER is designed on the basis of the H-mode scenario at 15 MA/5.3 T to achieve its Q = 10 goal and thus it is equipped with a set of in-vessel coils to achieve RMP fields for ELM control [11,12] and thus it is essential to analyze in detail the impact of RMP perturbations on the energetic particle population and to miminize it as far as possible. Previous work has shown that linear and nonlinear resonant interactions between fastions and the RMP fields in the region near the separatrix lead to an edge resonant transport layer (ERTL) that is responsible for the observed fast-ion transport in ASDEX Upgrade [27]. To characterize the RMP-induced transport, the variation of the particle toroidal canonical momentum (Pφ) caused by the broken magnetic symmetry was used as a figure of merit. The model used to calculate the particle transport was then compared and validated using the fast-ion loss measurements of the Fast-Ion Loss Detector (FILD) [28]. In this work, the methods used to understand the fast-ion losses Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 3 induced by RMPs in ASDEX Upgrade are applied to analyze ITER plasmas. By applying these models, several RMP configurations are studied in order to optimize the operational point that should ensure ELM suppression in ITER while maintaining fastion NBI losses acceptable. The RMP optimization was carried out considering two main parameters, the poloidal mode spectra of the RMP perturbed fields and the absolute toroidal phase of the perturbation. While the RMP poloidal spectra can affect both the ELM control capabilities and the fast-ion confinement, the absolute toroidal phase modifies the fast-ion behaviour without altering the ELM control performance of the configuration. This paper is organized as follows: Section 2presents the magnetic field structure for several RMP configurations with different poloidal mode spectra, through relative toroidal phase adjustments and thus effect on the plasma. In section 3, the impact of the poloidal mode spectra of the RMP perturbation and of its absolute toroidal phase on fast-ion NBI losses is evaluated using the variation of the toroidal canonical momentum as quantifying variable. In section 4, the wave-particle resonant interaction is presented as the transport mechanism responsible for the particle losses observed. Finally, section 5summarizes this work. 2. Perturbed magnetic fields calculated with the MARS-F code In this work, the 3D plasma equilibria were generated by the combination of the axisymmetric field of ITER, and the 3D perturbation induced by the RMP coils for a range of current waveforms including the response of the plasma as evaluated with the MARS-F code. In ITER, the RMPs are generated by 3 toroidal rows of 9 in-vessel ELM coils, which are individually powered (Figures 2a) and b)). This allows changes to the absolute toroidal phase of the perturbation applied as well the relative toroidal phase of the current waveforms between the upper and equatorial and lower and equatorial rows 0 5 10 15 20 25 T [keV] 2 4 6 8 10 12 ne [m-3] 1019 1.0 1.5 2.0 2.5 3.0 3.5 ρpol a) 0.2 0.4 0.6 0.8 q 0 1.0 ρpol 0.2 0.4 0.6 0.80 1.0 ρpol 0.2 0.4 0.6 0.80 1.0 b) c) Te Ti Figure 1: ITER DT baseline scenario: a) Safety factor, b) electron density and c) electron (blue) and ion (red) temperature profiles as a function of the normalized poloidal flux coordinate ρpol. Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 4 Upper coils Mid coils Lower coils ΔϕUM=φU- φM=180º ΔϕUM=φL- φM=180º a) b) c) R [m] -5 -4 -3 -2 -1 0 1 2 3 4 5 z [m] -4 -3 -2 -1 0 1 2 3 4 10 -6 4 6 8 δBR [T] Upper RMP coils Mid RMP coils Lower RMP coils -I +I -I ITER ΔϕLM=180º ΔϕUM=180º Figure 2: a) Schematic representation of the ITER RMP coils set-up. b) Radial component of the perturbed magnetic field generated by positive (yellow) and negative (blue) 1 A currents flowing through the RMP coils as a function of the radial and z coordinates. c) Phase shifts between the current waveforms flowing through the upper, mid and lower set of coils. of coils (Figure 2c)). For the analysis in this paper, we consider current waveforms with n = 3 symmetry and a maximum current level of 90 kAt in the coils, which is its maximum design value [12]. The poloidal mode spectra of the perturbation applied is determined by the phase shift between the upper and middle row (∆ϕUM ) and the lower and middle row (∆ϕLM ), while the toroidal spectrum is determined by that of the toroidal current waveform. The plasma response to the external perturbation has been generated using the MARS-F code [29], which calculates the linear response to the RMP fields considering a single fluid model and full 3D toroidal geometry. The H-mode plasma scenario considered here corresponds to that of the Q = 10 baseline scenario to be demonstrated during the Fusion Power Operation (FPO) phase of the ITER Research Plan [30]. Figure 1a) shows the safety factor profile corresponding to the H-mode scenario with Ip= 15 MA,Bφ= 5.3Tand Q= 10. Figures 1b) and c) show the radial profiles of the predicted electron density and electron and ion temperature. To vary the poloidal mode spectra of the perturbation, different relative toroidal phases have been considered. The set of RMP configurations analyzed are characterized in figure 3by the relative phase shifts between the three current waveforms and the associated plasma surface displacement at the X-point, which is a robust parameter correlated with the achievement of ELM mitigation/control by RMP fields [31]. Case 1 Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 5 a) b) φL-φM [deg] 300 200 100 0300200100 0φU-φM [deg] 8 6 4 2 |ξX| (r=a) [mm] (30,305) (200,270) (315,180) (180,6) (80,100) (200,140) Figure 3: a) Differential phase shift of the upper and lower set of coils with respect to the the coils in the middle row. b) The MARS-F computed plasma surface displacement near the X-point [31]. The relative toroidal phase shifts correspond to the ones shown in a). corresponds to the case with maximum X-point displacement while case 2 corresponds to the minimum one for which no effect on ELM behaviour is expected from RMPs in ITER (e.g. a non-resonant case). Case 3-6 are cases for which the X-point displacement is the same but have different poloidal spectra. The poloidal and toroidal Poincar´e maps of the magnetic field lines associated to cases 1 and 2 are shown in figures 4-5in vacuum and including plasma response. Note that, in the poloidal map, the apparent lower density of magnetic field lines at θ= 0 is a consequence of this direction being aligned with the shortest distance between the magnetic axis and the separatrix. These two configurations show the two extreme cases in terms of RMP perturbation impact on the X-point displacement. In case 1, the resulting vacuum magnetic field shows a strongly chaotic region at the plasma edge. The toroidal Poincar´e map in figure 4a) shows n= 3 symmetry modulated by the toroidal symmetry of the perturbation. In this case, the inclusion of plasma response decreases the effect of the perturbation reducing the width of the magnetic islands in both the poloidal and toroidal dimensions. The analysis of these Poincar´e maps shows the chaotic regions in the magnetic field generated by the RMP perturbations and how this chaotic areas are modified by the plasma response. However, the chaotic behaviour of fast-ions in the proximity of the RMP coils is different from the magnetic field lines and so are the thresholds between the regular and chaotic regions. Figure 5shows similar Poincar´e maps for case 2 (the non-resonant case) presenting a configuration with a considerable smaller chaotic region at the edge and magnetic island structures that extend to ρpol = 0.65. The effect of the plasma response in this case leads to the suppression of the island structures and strongly reduces the chaotic behaviour at the edge. In figure 6, the component of the magnetic perturbation perpendicular to the separatrix including the plasma response is presented for cases 1-3, with 3 being a case with large X-point displacement but smaller than case 1. This figure shows that the resulting perturbed field generated using the same coil current intensity depends strongly on the relative toroidal phases of the three current waveforms in ITER. The current waveform configuration of case 1 leads to a strong perturbed field at the separatrix and Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 6 Case 1 (plasma response) Case 1 (plasma response) Case 1 (vacuum) Case 1 (vacuum) a) b) c) d) ΔϕLM=200º ΔϕUM=140º ρpol ρpol ρpol ρpol Figure 4: Toroidal and poloidal Poincar´e maps of the magnetic field lines for case 1 in vacuum approach (a) and b)) and including plasma response (c) and d)). has maximum X-point respond and thus effectiveness for ELM control, while case 2 has a much more reduced perturbation (reduced by a factor of 3), but also a small X-point displacement and thus no ELM control effect. The configuration associated to case 3 presents an intermediate state where ELM control should be expected. 3. Fast-ion loss optimization in the presence of RMPs to provide ELM control The optimal RMP configuration to reduce ion losses while ensuring ELM control is analyzed in terms of the variation of the toroidal canonical momentum (Pφ) of the fastions generated by the ITER neutral beam injectors. Figure 7shows the surface density of energetic particles from NBI injector#1 in the poloidal and toroidal cross sections calculated with the ASCOT module BBNBI [32] from the kinetic profiles shown in figure 1. The NBI system in ITER consists of up to 3 injectors (2 NBI injectors are part of the baseline with a possibility of a third one as an upgrade) located at consecutive equatorial ports, each one can be injected with on and off-axis injection geometries [33]. The NBI#1 on-axis beam has the maximum ionization at z= 0.40, φ= 58◦with pitch λ=vk/v = 0.63, while the off-axis beam ionization is located at z=−0.1, φ= 58◦ Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 7 Case 2 (plasma response) Case 2 (plasma response) )muucav( 2 esaC)muucav( 2 esaC b) c) d) ΔϕLM=30º ΔϕUM=305º a) ρpol ρpol ρpol ρpol Figure 5: Toroidal and poloidal Poincar´e maps of the magnetic field lines for case 2 in vacuum approach (a) and b)) and including plasma response (c) and d)). Case 1 Case 2 Case 3 −0.05 0 0.05 Br [T] Figure 6: Perpendicular component of the RMP perturbed magnetic field at the separatrix including the plasma response. with pitch λ= 0.58. 3.1. Fast-ion transport analysis The fast-ion transport was evaluated through the variation of Pφinduced by the symmetry-breaking RMP fields. In order to calculate δPφ, 40000 markers are followed in full orbit motion using the ASCOT code [34]. The markers are distributed in a 200×200 homogeneous grid varying the initial radial and toroidal location, which ensures Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 8 R [m] z [m] 4 6 8 −5 −4 −3 −2 −1 0 1 2 3 4 5 0 0.5 1.0 1.5 2.0 x 10 20 #ions [m-2] ITER NBI#1 on-axis beam NBI#1 off-axis beam y [m] x [m] −5 0 5 −8 −6 −4 −2 0 2 4 6 8 0.5 1.0 1.5 2.0 2.5 3.0 3.5 x 1020 #ions [m-2] BBNBI4 NBI#1 on-axis + off-axis NBI#3 on-axis + off-axis NBI#2 on-axis + off-axis Bφ Ip b) ITER BBNBI4 a) Upper RMP coils Mid RMP coils Lower RMP coils Mid RMP coils Figure 7: 2D histogram showing the surface density of the initial NBI#1 on and off-axis distribution projected onto the poloidal a) and toroidal b) cross sections. Green lines indicate the locations of the mid RMP coils. a sufficient resolution in the initial coordinates to observe the δPφstructures according to a convergence study. The parameters of this distribution are set to the ones fixed by the injection geometry of the NBI#1 on-axis. In this case, deuterium particles are started at z= 0.40 m, with energy E= 1 MeV and pitch angle λ= 0.63, which lead to a maximum of trapped ions at R= 8.15 mwith the trapped/passing boundary located at R= 7.9m. The calculation of the variation of Pφ(i) is done by averaging the evolution of Pφ(i) along the ion trajectory for approximately 10 poloidal turns or until the ion hits the wall using the expression: hδPφi= N X i=1 Pφ(i)−Pφ(0) N(1) Here, Pφ(i) is the value of Pφat each time point considered to evaluate hδPφiand Nis the total number time points along the trajectory. Pφis a constant of motion in 2D fields, but varies when a 3D perturbation is applied. This variation can be related to a radial transport of the particle where the direction of the drift depends on the sign of hδPφi. By considering the sign of the particle toroidal velocity vφand poloidal flux ψpositive in the direction of the plasma current (as shown in figure 7b)), a positive variation of hδPφileads to an inwards transport while a negative variation implies an outwards transport. Most of the fast NBI ions that are lost come from the edge as shown in figure 8. In this region, the typical ion trajectory in the presence of RMPs is lost within 4-8 poloidal turns. This ensures that the following time used to calculate hδPφi, which covers approximately 10 poloidal turns, represents a good approximation to the variation of Pφcaused by the RMP fields. Using this variation, the global toroidal phase of the perturbation was scanned (note Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 9 <δPφ> [a.u.] e) f) d) h) i) g) δP > 0 φδP < 0 φ δP > 0 φδP < 0 φ δP > 0 φδP < 0 φ 8 Case 1 NBI #1 ΔϕLM=200º ΔϕUM=140º a) Separatrix δP < 0 φ δP > 0 φ c) Separatrix Case 3 ΔϕLM=200º ΔϕUM=270º δP < 0 φ NBI #1 b) Case 2 ΔϕLM=30º ΔϕUM=305º Separatrix δP > 0 φ δP < 0 φ δP > 0 φ −1.0−0.500.51.0 7.2 7.4 7.6 7.8 8.0 8.2 R [m] 7.2 7.4 7.6 7.8 8.0 8.2 R [m] 7.2 7.4 7.6 7.8 8.0 8.2 R [m] φ [deg] 50 100 150 200 250 300 350 1 2 345678 9 10 11 #ions/m 1019 φ [deg] 50 100 150 200 250 300 350 φ [deg] 50 100 150 200 250 300 350 1 2 345678 9 10 11 #ions/m 1019 1 2 345678 9 10 11 1019 1 2 345678 9 10 11 #ions/m 1019 1 2 345678 9 10 11 #ions/m 1019 1 2 345678 9 10 11 #ions/m 1019 Initial FI distribution Lost FI Initial FI distribution Lost FI Initial FI distribution Lost FI Figure 8: a)-c) hδPφias a function of the particle toroidal angle and major radius for cases 1- 3. d)-i) Radial histogram of the initial radial distribution of the beam NBI#1 on axis (blue) and lost fast-ions (red) for the initial toroidal angle leading to a minimum (d)-f)) and maximum (g)-i)) fast-ion losses. that maximum deposition of the NBI injector# 1 is located at a toroidal angle of 58 degrees) and the maximum radial transport evaluated for all 6 cases. In figures 8a)-c), hδPφiis presented as a function of the particle initial radial position and starting toroidal angle at fixed initial z= 0.40 m,E= 1 MeV and λ= 0.63 for three representative cases. Case 1 associated to the largest X-point displacement also leads to the maximum variation of hδPφi. For case 2, with minimum X-point displacement, there is a smaller Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 16 7.8 8.0 8.2 R [m] 0.3 0.4 0.5 0.6 0.7 0.8 0.9 pitch Case 2 NBI #1 on axis δP < 0 φ 7.8 8.0 8.2 R [m] 7.8 8.0 8.2 R [m] Case 3 NBI #1 on axis -2 -1 0 1 2 Separatrix Separatrix b) c) NBI #1 off axis δP < 0 φ <δPφ> [a.u.] Trapped/passing bounrdary 5.50 6.50 0.30 0.33 0.38 0.42 0.50 0.60 0.75 0.75 Case 1 ωb/ωd δP < 0 φ a) 3 Separatrix Figure 14: hδPφias a function of the particle major radius and pitch angle for cases 1-3 at z= 0.40 m, toroidal angle φ= 58◦and energy E= 1 MeV . White contours in a) indicate the constant ratio of ωb/¯ωdand the white solid line in b) and c) shows the position of the boundary between passing and trapped orbits. The NBI#1 on-axis birth distribution is indicated with a solid green line and and the FWHM with a dashed green line. The solid black line in c) indicates the NBI#1 off-axis birth distribution. the perturbation geometry and p= 1. This resonance is observed in all cases shown in figures 12 and 13 for the hδPφi<0 configuration, excluding case 2. The other resonant structures can be explained by an overlapping of nonlinear resonances in the range ωb/¯ωd= 4 −7, which shows a good agreement with the hδPφipatterns (figure 12 a)). 5. Summary and Conclusions Using the variation of the toroidal canonical momentum, the fast-ion transport in the presence of RMP fields was analyzed for the ITER Q = 10 H-mode baseline scenario with 3-D fields applied for ELM control by the in-vessel ELM control coils. This approach has been previously applied and validated in ASDEX Upgrade against experimental data, revealing the existence of an ERTL responsible for the radial particle transport. In this work, the same method was used to identify the fast-ion transport and to optimize the absolute toroidal phase of the RMP perturbation as well as its poloidal mode spectra for ITER. The analysis of the ion transport in terms of hδPφi>0 for ITER showed an ERTL similar to the one previously observed in AUG. The results presented here indicate that, in this edge layer, a resonant interaction between the energetic particles and the perturbed magnetic fields may lead to the fast-ion transport observed in these ASCOT full orbit simulations. The ERTL is located within 10 cm around the separatrix and has a strong dependency on the RMP perturbation applied. First, the absolute toroidal phase of the applied RMP perturbation has been studied. This was done by aligning the absolute phase of the perturbation with the toroidal angle at which the ITER NBI injectors are located so as to provide a maximum variation of Pφ. The alignment with the negative variation of Pφcorresponded to the maximum of the Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 17 outward transport while the positive variation of Pφmaximized the inward transport. By comparing the total fast-ion losses in each case, it has been found that the optimization of the toroidal angle can decrease the ion losses down to less than 8% for the largest X- point displacement configuration studied, which corresponds to more than 50% decrease compared to the de-optimized phase. The analysis to make this first optimization is fast in terms of CPU-time and the results can be directly applied by changing the absolute toroidal angle of the RMP perturbation since the NBI location is fixed. This optimization has no impact on the capabilities of the RMP perturbation applied to provide ELM control. In the second step of this analysis, the poloidal mode spectra was optimized using the variation of Pφto scan different configurations that have the same potential for ELM control, i.e. the same X-point displacement. The analysis showed that fast-ion losses are not necessarily correlated with X-point displacement and the associated losses for cases 3-6 showed a relative variation of up to ±40%, once the absolute RMP phase is optimized to minimize losses, potentially down to 4.2% for case 5, which is a very low value for the ELM suppressed H-modes required in ITER. Finally, by comparing hδPφiwith the particle resonances (ωb/¯ωd), it was shown that the observed particle transport in our simulations could be explained by an overlapping of multiple linear and nonlinear resonances close to the separatrix. Both linear and nonlinear resonances are involved in the ERTL, however, the linear resonance ωb/¯ωd= 3 (n= 3, p= 1, l= 0) is dominant. This suggests that the toroidal symmetry of the RMP perturbation also entails a significant impact on the intensity and location of the fast-ion transport and thus other RMP configurations that are planned to be used in ITER for ELM control (n = 4 and, potentially, n = 2) should also be analyzed as well as other plasma scenarios. From the results in our studies, the effects on alpha particles are expected to be much smaller given that their source is in the plasma central area. The procedure developed here to select the optimal RMP configuration can be applied to any plasma equilibrium and profiles and particle species to define the best ELM control scheme to target ELM suppression while avoiding the degradation of the fastion confinement. Our study also highlights the need to include appropriate diagnostics for fast particle losses in ITER since they are likely to be required for the optimization of the operational scenarios and RMP characteristics to ensure ELM suppression, good fast-ion confinement and acceptable fast-ion power fluxes to the main wall and divertor in ITER. Acknowledgments This work has received funding from the Spanish Ministry of Science, Innovation and Universities (grant BES-2013-065501). The simulations were partly performed on the MARCONI supercomputer (CINECA) under project reference FUA33 ASCOT-US. This work has been partially carried out within the framework of the EUROfusion Consortium and has partially received funding from the Euratom research and training Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 18 programme 2014–2018 and 2019–2020 under grant agreement No. 633053. The views and opinions expressed herein do not necessarily reflect those of the European Commission. The support from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 805162) is gratefully acknowledged. The views and opinions expressed herein do not necessarily reflect those of the European Commission or of the ITER Organization. This work was partially funded by the Academy of Finland project No. 324759. We acknowledge the computational resources provided by Aalto Science-IT project. References [1] F Wagner et al. Regime of improved confinement and high beta in Neutral-Beam-Heated discharges of the ASDEX Tokamak. Phys. Rev. Lett, 49:1408, 1982. [2] H Zohm et al. Edge localized modes (ELMs). Plasma Phys. Control. Fusion, 38:105, 1996. [3] E Viezzer. Access and sustainment of naturally ELM-free andsmall-ELM regimes. Nucl. Fusion, 58:115002, 2018. [4] A Loarte et al. Characteristics of type I ELM energy and particle losses in existing devices and their extrapolation to ITER. Plasma Phys. Control. Fusion, 45:1549–1569, 2003. [5] K Tani et al. Effects of ELM mitigation coils onenergetic particle confinement in ITER steady-state operation. Nucl. Fusion, 52:013012, 2011. [6] E J Doyle et al. Chapter 2: Plasma confinement and transport. Nucl. Fusion, 47:013012, 2007. [7] T E Evans et al. Suppression of Large Edge-Localized Modes in High-Confinement DIII-D Plasmas with a Stochastic Magnetic Boundary. Phys. Rev. Lett, 92(23):235003, 2004. [8] T C Hender et al. Effect of resonant magnetic perturbations on COMPASS-C tokamak discharges. Nucl. Fusion, 32:2091, 1992. [9] Y Liang et al. Active control of type-I edge-localized modes with n=1 perturbation fields in the JET tokamak. Phys. Rev. Lett, 98(26):265004, 2007. [10] W Suttrop et al. First Observation of Edge Localized Modes Mitigation with Resonant and Nonresonant Magnetic Perturbations in ASDEX Upgrade. Phys. Rev. Lett, 106:225004, 2011. [11] A Loarte et al. Chaos cuts ELMs down to size. Nat Phys., 2:369, 2006. [12] A Loarte et al. Progress on the application of ELM control schemes to ITER scenarios from the non-active phase to DT operation. Nucl. Fusion, 54:033007, 2014. [13] D A Spong. 3D toroidal physics: Testing the boundaries of symmetry breaking. Phys. Plasmas, 22:055602, 2015. [14] K C Shaing. Symmetry-Breaking Induced Transport in the Vicinity of a Magnetic Island. Fi, 8725(52), 2001. [15] K C Shaing et al. Transport theory for energetic alpha particles and tolerable magnitude of error fields in tokamaks with broken symmetry. Nucl. Fusion, 54(2):033012, 2014. [16] W Suttrop et al. Experimental studies of high-confinement mode plasma response to nonaxisymmetric magnetic perturbations in ASDEX Upgrade. Plasma Phys. Control. Fusion, 59:014049, 2016. [17] R J Akers et al. High fidelity simulations of fast ion power flux driven by 3D field perturbations on ITER. 2016 IAEA Fusion Energy Conference (Kyoto, Japan, 17–22 October 2016) TH/4-1, 2016. [18] A Kirk et al. Resonant magnetic perturbation experiments on MAST using external and internal coils for ELM control. Nucl. Fusion, 50:034008, 2011. [19] M Willensdorfer et al. Field-Line Localized Destabilization of Ballooning Modes in Three- Dimensional Tokamaks. Phys. Rev. Lett, 119(8):085002, 2017. Optimizing beam-ion confinement in ITER by adjusting the ∆ϕof the RMPs 19 [20] C Paz-Soldan et al. Observation of a multimode plasma response and its relationship to density pumpout and edge-localized mode suppression. Phys. Rev. Lett, 114(10):105001, 2015. [21] R Nazikian et al. Pedestal Bifurcation and Resonant Field Penetration at the Threshold of Edge- Localized Mode Suppression in the DIII-D Tokamak. Phys. Rev. Lett, 114(5):105002, 2015. [22] M Garcia-Munoz et al. Fast-ion losses induced by ELMs and externally applied magnetic perturbations in the ASDEX Upgrade tokamak. Plasma Phys. Control. Fusion, 55:124014, 2013. [23] T Kurki-Suonio et al. Protecting ITER walls: fast ion power loads in 3D magnetic field. Plasma Phys. Control. Fusion, 59:014013, 2017. [24] K Shinohara et al. Effects of rippled fields due to ferritic inserts and ELM mitigation coils on energetic ion losses in a 15 MA inductive scenario in ITER. Nucl. Fusion, 52:094008, 2012. [25] M A Van Zeeland et al. Fast ion transport during applied 3D magnetic perturbations on DIII-D. Nucl. Fusion, 55:073028, 2015. [26] J Varje et al. Effect of plasma response on the fast ion losses due to ELM control coils in ITER. Nucl. Fusion, 56:046014, 2016. [27] L Sanchis et al. Characterisation of the fast-ion edge resonant transport layer induced by 3D perturbative fields in the ASDEX Upgrade tokamak through full orbit simulations. Plasma Phys. Control. Fusion, 61:014038, 2019. [28] M Garcia-Munoz et al. Scintillator based detector for fast-ion losses induced by magnetohydrodynamic instabilities in the ASDEX upgrade tokamak. Rev. Sci. Instrum., 80(81):053503, 2009. [29] Y Q Liu et al. Feedback stabilization of nonaxisymmetric resistive wall modes in tokamaks. I. Electromagnetic model. Phys. Plasmas, 7:3681, 2000. [30] ITER Organization. ITER Research Plan within the Staged Approach. pages ITR–18–003, 2018. [31] L Li et al. Modeling 3D plasma boundary corrugation and tailoring toroidal torque profiles with resonant magnetic perturbation fields in ITER. Nucl. Fusion, 59:096038, 2019. [32] O Asunta et al. Modelling neutral beams in fusion devices: Beamlet-based model for fast particle simulations. Comput. Phys. Commun., 188:33–46, 2015. [33] R S Hemsworth et al. Overview of the design of the ITER heating neutral beam injectors. New J. Phys., 19:025005, 2017. [34] E Hirvijoki et al. ASCOT: Solving the kinetic equation of minority particle species in tokamak plasmas. Comput. Phys. Commun, 185:1310, 2014. [35] A Snicker et al. Realistic Simulations of Fast-Ion Wall Distribution Including Effects Due to Finite Larmor Radius. IEEE Trans. Plasma Sci, 38:2177–84, 2010. [36] F Zonca, L Chen, S Briguglio, G Fogaccia, G Vlad, and X Wang. Nonlinear dynamics of phase space zonal structures and energetic particle physics in fusion plasmas. New J. Phys., 17(17):013052, 2015. [37] L Chen and F Zonca. Self-consistent kinetic theory with nonlinear wave-particle resonances. Plasma Sci. Technol., 21:125101, 2019.