Impact of the plasma response in threedimensional edge plasma transport modeling for RMP ELM control at ITER

Transcription

1 Impact of the plasma response in threedimensional edge plasma transport modeling for RMP ELM control at ITER O. Schmitz 1, H. Frerichs 1, M. Becoulet 2, P. Cahyna 3, T.E. Evans 4, Y. Feng 5, N. Ferraro 4, G. Huijsmans 6, M. Lanctot 4, L. Lao 4, A. Loarte 6, R. Pitts 6, D. Reiser 7, D. Reiter University of Wisconsin Madison, Department for Engineering Physics, Madison, WI, USA 2 - CEA/IRFM, Cadarache, St. Paul- lez- Durance Cedex, France 3 IPP AS CR, Za Slovanou 3, Prague 8, Czech Republik 4 - General Atomics, San Diego, California, USA 5 - Max- Planck InsNtute for Plasma Physics, Greifswald, Germany 6 - ITER OrganizaNon, Route de Vinon- sur- Verdon - CS St Paul Lez Durance France 7 - Forschungszentrum Juelich GmbH, IEK- 4, D Juelich, Germany Acknowledgement: This work was supported by ITER task IO/CT/11/ , F4E grant GRT-055(PMS-PE) and by Startup Funds of the College of Engineering at the University of Wisconsin - Madison. Computing resources were provided by the John von Neuman Institute for Computing (NIC) at the Juelich Supercomuting Centre (JSC), Germany. APS DPP Meeting, New Orleans, LA, October 2014

2 Motivation 1: Resonant Magnetic Perturbations (RMP) break the tokamak axisymmetry inducing a helical scrape-off layer Axisymmetric case Small amplitude RMP field DIII-D DIII-D 1

3 Motivation 1: Resonant Magnetic Perturbations (RMP) break the tokamak axisymmetry inducing a helical scrape-off layer Helical magnetic footprint on target Small amplitude RMP field DIII-D DIII-D Complex mesh of field lines with different connection length yield a strong asymmetry in scrape-off layer (SOL) transport to the divertor targets [O. Schmitz et al., PPCF 52 (2008) ] 1 [T.E.Evans et al., JPCS 7 (2005) 174] [A. Wingen et al. PoP 49 (2009) ] The consequences for ITER SOL physics and plasma wall interaction have to be understood

4 Motivation 2: Separatrix lobe shave been observed experimentally on several devices applying RMPs 2 RMP ELM suppressed H-mode plasmas show separatrix perturbation and strike line splitting

5 Motivation 3: A direct connection between plasma response and strike line splitting has been observed at DIII-D [O. Schmitz et al., Nuclear Fusion, 54 (2014) ] 3 Impact of shielding response on perturbed separatix has been found also on MAST, NSTX, Asdex-Upgrade

6 Talk Outline Assessing impact of RMP induced 3D transport and PMI physics requires understanding of edge transport and plasma response We are going to compare key features of RMP induced edge transport for two contrary plasma response cases Modeling setup and computational setup used Comparison of perturbed magnetic topologies Overview on 3D boundary structure and impact on energy and particle confinement in the perturbed plasma boundary ΔT Δτ P Sonic flow pattern Divertor fluxes Impact of plasma response case with screening and amplification Conclusions: what have we learned and which next steps do we have to go? 4

7 Talk Outline Assessing impact of RMP induced 3D transport and PMI physics requires understanding of edge transport and plasma response We are going to compare key features of RMP induced edge transport for two contrary plasma response cases Modeling setup and computational setup used Comparison of perturbed magnetic topologies Overview on 3D boundary structure and impact on energy and particle confinement in the perturbed plasma boundary ΔT Δτ P Sonic flow pattern Divertor fluxes Impact of plasma response case with screening and amplification Conclusions: what have we learned and which next steps do we have to go? 4

8 Modeling setup: focus on ITER standard Q=10 H-mode 5 Plasma equilibrium used at pedestal T e =4.4keV is a marginally MHD stable case just before ELM crash

9 Computational methods used in this initial survey study Edge transport modeling: EMC3-Eirene EMC3 Eirene Bragisnkii fluid plasma transport code (particle, energy & momentum transport) Kinetic neutral transport (particle, energy & momentum sources and sinks for EMC3) Actual recycling flux was scanned to explore recycling regime with RMP Plasma response scenarios considered 6 Strong screening Screening for Ψ N <0.96 Vacuum fields outside Based on cylindrical models RMHD and ATTEMPT [M. Becoulet et al., NF 52 (2012) ] [D. Reiser et al., PoP 17 (2009) ] Vacuum fields Coil Current 90 kat 45 kat Screening and amplification Linear, two fluid MHD (rotation/flows, diamagnetic effects) Non-linear, single fluid MHD (mode saturation, mode coupling) M3D-C1 in diverted, toroidal geometry [N. Ferraro et al., PoP 19 (2012) ]

10 Computational methods used in this initial survey study Edge transport modeling: EMC3-Eirene EMC3 Eirene Bragisnkii fluid plasma transport code (particle, energy & momentum transport) Kinetic neutral transport (particle, energy & momentum sources and sinks for EMC3) Actual recycling flux was scanned to explore recycling regime with RMP Plasma response scenarios considered 6 Strong screening Screening for Ψ N <0.96 Vacuum fields outside Based on cylindrical models RMHD and ATTEMPT [M. Becoulet et al., NF 52 (2012) ] [D. Reiser et al., PoP 17 (2009) ] Vacuum fields Coil Current 90 kat 45 kat Screening and amplification Linear, two fluid MHD (rotation/flows, diamagnetic effects) Non-linear, single fluid MHD (mode saturation, mode coupling) M3D-C1 in diverted, toroidal geometry [N. Ferraro et al., PoP 19 (2012) ]

11 Talk Outline Assessing impact of RMP induced 3D transport and PMI physics requires understanding of edge transport and plasma response We are going to compare key features of RMP induced edge transport for two contrary plasma response cases Modeling setup and computational setup used Comparison of perturbed magnetic topologies Overview on 3D boundary structure and impact on energy and particle confinement in the perturbed plasma boundary ΔT Δτ P Sonic flow pattern Divertor fluxes Impact of plasma response case with screening and amplification Conclusions: what have we learned and which next steps do we have to go?

12 Strong screening reduces stochasticity in magnetic field inside of ideal surface Reduce coil current Screening Vacuum, 90kAt Vacuum, 45kAt Strong screening Vacuum, 45kAt, q95=4.2 Significant variations of effective perpendicular vs. parallel transport ratios is expected 7

14 Strong thermal confinement loss is obtained only for high RMP amplitude vacuum field 8 Thermal confinement loss and radial spread of helical fingers depend on resonant field amplitudes

15 Reduction of current and plasma response yield reduced finger width Only vacuum case with high vacuum RMP amplitude yields beneficial heat flux spreading 9

16 Recycling scan suggests transition into high recycling regime approaching a low temperature, high density divertor solution screened- RMP Upstream density vs. divertor recycling flux at constant external source level (Q n =1.9 x at/s) Same density requires less recycling flux at fixed external source No- RMP 10

17 Recycling scan suggests transition into high recycling regime approaching a low temperature, high density divertor solution screened- RMP Upstream density vs. divertor recycling flux at constant external source level (Q n =1.9 x at/s) Same density requires less recycling flux at fixed external source Particle pump out No- RMP 90kAt vacuum case screened case 10

18 Helical SOL flux bundles establish helical sonic flows which are likely to be one driver for particle pump out by direct sonic particle transport 3D flow channels FlaEening of density profile Increase upstream pressure in 3D SOL Downstream n up Downstream T down Candidate for enhanced outward parmcle flux and reduced fueling 11 [O. Schmitz et al., JNM 415 (2011)] See poster GP by H. Frerichs et al. on Formation of Counter Flows by Magnetic Perturbations in Edge Transport Modeling

19 Recycling scan suggests transition into high recycling regime approaching a low temperature, high density divertor solution RMP case with screening Temperature No- RMP case Density Temperature 11 ev Density 14 ev Steep recycling increase for density regime approached yields divertor cooling by ionization losses with similar characteristics for RMP and normp case Divertor scenario equivalent to high recycling divertor from axis-symmetric modeling with B2-Eirene [Y. Feng et al., EPS conference 2011, ECA P1.071] 12 Detached divertor solution for ITER not accessible with EMC3-Eirene yet

21 M3D-C1 predicts mixture of screening and amplifying plasma response along radius Courtesy of N. Ferraro, from final report of IO task IO/CT/11/ Linear, two-fluid modeling Underlying rotation profiles Screen Screen Amplify vacuum Linear response shows strong screening close to separatrix Resonant field amplification in plasma edge Moderate screening radially deeper inside Non-linear 13

22 Linear response from M3D-C1: X-point structure features additional fine scale structures in helical fingers due to amplified field Vacuum ﬁeld as used in M3D- C1 ISP 14 Linear response from M3D- C1 ISP Linear plasma response alters magnetic field structure significantly but predicts that radial flux channels are thin

23 Electron temperature iso-surfaces are marginally affected by both plasma response fields from M3D-C1 Vacuum field at 45kAt Linear, two- fluid response Linear, two-fluid response does not alter perturbed separatrix structure significantly 15

24 Electron temperature iso-surfaces are marginally affected by both plasma response fields from M3D-C1 Vacuum field at 45kAt Linear, two- fluid response T e (Ψ N ) mid- plane Linear, two-fluid response does not alter perturbed separatrix structure significantly AmplificaMon Screening 15 Significant reduction of thermal confinement is seen even beyond vacuum field impact

25 Heat flux maxima are shifted outward in helical lobe trajectory including linear, two-fluid response Vacuum field at 45kAt Linear, two-fluid response 16

26 Heat flux maxima are shifted outward in helical lobe trajectory including linear, two-fluid response Vacuum field at 45kAt Linear, two-fluid response No reduction of heat flux splitting No heat flux peaking 3 General trends are opposite to case with strong screening

27 Plasma response enhances radial inward extension of sonic SOL flows with correlated increases in particle confinement loss No-RMP Vacuum field, 45kAt Linear M3D-C1 response 17 Stronger decrease of particle confinement time with linear response Particle losses are compatible with but are at the margin of the ITER pellet fueling capacity (a total core particle source strength of x s -1 is required)

28 What have we learned and what are the next steps? Resonant magnetic perturbations induce a 3-D plasma boundary with a 3-D scrape-off layer and helical divertor target heat and particle flux pattern The shape of the 3-D boundary and of the helical divertor target pattern is highly sensitive to the actual level of internal plasma response Thermal confinement as well as particle confinement is reduced by RMP fields and the actual reduction is dependent on internal resonant amplitudes Amplification effects can cancel screening response leaving only small effects on confinement and divertor flux properties compared to vacuum fields This is in contrast with results using strong screening of the RMP fields 18 Edge transport prediction depends vitally on accurate modeling of internal plasma response. A thorough validation of combined solutions against experiments is of pivotal importance for reliable extrapolation towards ITER.

29 Appendix

30 Motivation 1: Resonant Magnetic Perturbations (RMP) break the tokamak axisymmetry inducing a helical scrape-off layer 3D flux tubes are established Small amplitude RMP field Target strike point Start Point DIII-D [H. Frerichs et al., NF 50 (2010) ] See poster GP by H. Frerichs et al. on Formation of Counter Flows by Magnetic Perturbations in Edge Transport Modeling 1

31 Motivation for paper Recent results from DIII-D have shown a direct correlation between strike point splitting and the n=3 plasma response [O. Schmitz et al., Nucl. Fusion, 54 (2014) ] Low power L-mode! To which extend is this a reference case and which other experimental background do we have? 1

32 DIII-D L-mode results only ones with direct link to SP splitting LSN L-mode discharge yields decay of n=3 response Toridal rotation profiles 3 [O. Schmitz et al., Nucl. Fusion, 54 (2014) ] Braking Spin-Up

33 MARS-F modeling predicts resonant screening response Local resonant sheet currents reduce n=3 resonant mode amplitude [O. Schmitz et al., Nucl. Fusion, 54 (2014) ] Plasma response commences with intuitive understanding and in agreement with several modeling Ideal MHD type response with RFA not detected [M. Lanctot et al., PoP 18 (2011) ] 4

34 Strike line splitting is strongly enhanced after response decayed Lobe location agree (for this case) with vacuum prediction 6 This is a result particular to this case! We also saw cases with much wider striation widths!

35 In contrast: strike point splitting at DIII-D in high collisionality plasma deviates strongly from vacuum prediction T.E.Evans et al., JPCS 7 (2005) 174 M. Jakubowski et al., NF 49 (2009)

36 EMC3-Eirene used as transport solver on various magnetic grids Radial spread of fingers depend on resonant field amplitudes Radial extension of flow tube structures is very sensitive to plasma response 12

37 Reduction of current and plasma response yield reduced finger width Distance along target Distance along target 13 Toroidal direcmon

38 Toroidal averaged heat and particle flux profiles show narrow channeling for case with screening and reduced RMP amplitude Power flux decay length [cm] Toroidal averaged power flux profiles at outer strike point Beneficial spreading of divertor fluxes only seen for high current vacuum case Peak power flux [MW m -2 ] Striated footprint yields steeper power flux profiles for screened case and at reduced RMP amplitude 10

39 Recycling scan suggests transition into high recycling regime approaching a low temperature, high density divertor solution Downstream densimes Downstream temperatures 18 ev 12 ev 14 ev 10 ev 17 IonizaMon energy losses Detached divertor solution for ITER can not be approached with EMC3-Eirene model yet

40 Plasma response from RMHD (cylindrical) requires translation to toroidal geometry Toroidal q-profile Cylindrical q-profile! 22 Representation as toroidal q-profile requires adaptation of current density profile with negative current density in very edge (this region is at the end treated as vacuum field)