Presented by Rob La Haye. on behalf of Francesco Volpe. at the 4 th IAEA-TM on ECRH for ITER

Transcription

1 Locked Neoclassical Tearing Mode Control on DIII-D by ECCD and Magnetic Perturbations Presented by Rob La Haye General Atomics, San Diego (USA) on behalf of Francesco Volpe Max-Planck Gesellschaft (Germany) at the 4 th IAEA-TM on ECRH for ITER 6-8 June 2007, Vienna (Austria) 1

2 In ITER, NTM islands will be thin and tend to lock Neutral Beam Injectors (NBI) heat and spin present tokamak plasmas. ITER: few krad/s instead of present krad/s. Less rotation less shielding earlier locking 2/1 islands as small as w>5cm expected to lock [R. La Haye, Nucl.Fusion 2006] from ITER Techn. Phys. Basis 1999 Modest effect on performance, according to belt model but higher risk of disruption 2

3 Motivation ITER expected to have slow (few khz) NTMs high probability of locking. Locked 2/1 modes are a recurrent cause of disruptions. ECCD proved successful in suppressing rotating NTMs. Success of ECCD for locked modes not guaranteed, due to island locking in a position not accessible by gyrotrons. Resonant Magnetic Perturbations (RMPs) from I-coils used at DIII-D for RWM and ELM control and Error Field Correction. Although more challenging (modes lie deeper in the plasma and rotate faster), here RMPs are used to control NTM rotation and assist their ECCD Stabilization 3

4 Balanced NBI reproduces in DIII-D ITER-like conditions (slow rotation) Little or no momentum injection slow plasma rotation NTMs partly frozen in plasma NTMs rotating at 1-5kHz, nearly as slow as expected in ITER 4

5 Locked mode stabilization can pose difficulties in ITER, if only ECCD is used Upper launchers cover «25 o poloidally But 2/1 can lock anywhere in a 180 o poloidal range, and 3/2 can lock anywhere in a 120 o poloidal range Finite poloidal extent of O-point (90 o for 2/1, 60 o for 3/2) helps but not much Limited capability to shift resonance or plasma in ITER Multi-frequency gyrotrons horizontal adjustment 5

6 Locked mode stabilization can pose difficulties in ITER, if only ECCD is used Y(m) Top View X (m) Bad alignment Good alignment with 2/1 O-point 6

7 Magnetic perturbations can unlock and reposition or spin the mode and so assist its ECCD control in DIII-D and ITER Bootstrap deficit island wire carrying ctr-current wire can be moved by external B Approach 1 ( preferential locking ): island is dragged into a new position accessible by gyrotrons. ECCD: CW Approach 2 ( sustained rotation or entrainment ): externally applied rotating B unlocks the mode and forces it to rotate. ECCD: CW or modulated 7

8 Magnetic measurements were used to predict/detect locking and trigger response Mirnov Coils (poloidal field sensors) measure mode amplitude at >100Hz 20T/s threshold control Mirnov Coils+Frequency Counter measure angular frequency of mode, detect slowing down 1kHz threshold control Saddle loops (radial field sensors) measure mode amplitude at <100Hz. Suitable for born locked modes 5G threshold control 8

9 Toroidal phase and amplitude of NTM were controlled respectively by magnetic perturbations from internal (I) coils and by 2 gyrotrons 2x 110GHz, 600kW gyrotrons 9

10 FIRST TYPE OF EXPERIMENTS: Static EFC (to lock the mode with a preferential toroidal phase) + cw ECCD Slow travelling wave to find optimal tor. phase 10

11 Slowly rotating field used to drag plasma and toroidally align island to ECCD Density control NBI ECRH β N Overcorrection, rotating at 0.66Hz, I-coils Locked mode detector Static EFC, C-coils Slowdown detector D α Born-locked mode detector time (ms) 11

12 The island, rotated by EFC and illuminated by ECCD, changes amplitude B R as measured by pairs of Saddle Loops in 3 different toroidal locations n=1 toroidal phase n=1 absolute amplitude of island varies B R =4-7.5G 12

13 Question: are we observing mode shrinking? Or are we just measuring, with diagnostic coils, the effect of perturbing coils? Magnetic dipole of the rotating mode Sta ti c Erro r Fi e l d Rotating magnetic perturbation 3 angularly equally spaced measuring coils 13

14 Answer: Vacuum shot allows measuring I-coil and C-coil rotating field at saddle loop locations. This is small and constant. B R as measured by pairs of Saddle Loops in 3 different toroidal locations n=1 toroidal phase n=1 absolute amplitude small and constant, B R =1G 14

15 Subtraction of Vacuum Field confirms that locked mode changes amplitude when toroidally steered in presence of ECCD. Tor.angle (deg) B R (G) phase (deg) B R contours from saddle loops. Color range: -8.5/8.5G Strong mode strong magn.braking slow rotation Weak mode less braking faster rotation time (s) B R (G) 15

16 Non-uniform mode rotation and strength in previous shot appears more evident after comparing with vacuum shot Tor.angle (deg) B R (G) phase (deg) B R contours from saddle loops. Color range: -1.5/1.5G Fit at t=2.48s shows nice n=1 structure time (s) B R (G) 16

17 Radial jog of plasma: mode amplitude decreases when ECCD is in the right place, increases when ECCD is off. R=2cm R(m) where q=2 ECCD (n=1) (n=1) Mode recovers when ECCD off 17

18 SECOND TYPE OF EXPERIMENTS Dynamic EFC fast travelling wave to sustain Mode Rotation 18

19 I-coil travelling wave entrains mode up to 60Hz if gently accelerated. Strong mitigation at 10Hz Tor.angle (deg) B R (G) phase (deg) B R Contours from saddle loops time (s) Color range: -10/10G n=1 from Mirnov toroidal array below marginal island width? m=2 from Mirnov poloidal array 19

20 Summary and Conclusions Demonstrated Preferential Locking of NTM to a (static) toroidal phase such that it can be accessed by ECCD Optimal toroidal phase was found by slowly steering the mode (0.66Hz) while applying cw ECCD ECCD radially aligned to locked island as to rotating ones Mode Mitigation observed with 1.2MW ECCD Sustained Mode Rotation up to 60Hz by means of I-coil travelling wave Travelling wave needs to be applied gently (0-60Hz ramp in 1s) 20

21 Ongoing & Future Works Preferential locking experiment to be repeated with 2.4MW ECCD complete stabilization? Entrainment to be improved from 60 to 100Hz (-3dB from image currents; at 1kHz attenuation would be -10dB) Several applications of sustained mode rotation: Prevents further locking Rotationally mitigates the mode Brings the locked mode case in the well-studied, easy-to-stabilize rotating NTM case Opens up the possibility to synchronize and phase-lock the mode rotation to the ECCD modulation. Easier than adapting ECCD to natural mode rotation Post-locking control was shown here, but detectors of rotating precursors will allow pre-emptive application of magnetic perturbations ( locked mode avoidance, rather than control) 21

22 Backup Material 22

23 ECCD not deposited in the island (by lowering I p and B T by 3%). Same heating, β, n e, shape, same mode-error field interaction but no ECCD mitigation. As expected, mode is stronger. Tor.angle (deg) B R (G) phase (deg) B R contours from saddle loops Color range: -19/19G time (s) Other instabilities not related to magnetic braking or (de)stabilization of 2/1. Top view: node 23

24 Clear Difference in Phase Consistent with the fact that two different phases are relevant: Between island and ECCD in ECCD case (top) Between I-coil and Saddle-Loops in no-eccd case (bottom) 24

25 I-coil travelling wave entrains mode if gently accelerated. Ramp 1-100Hz NBI 25

26 I-coil Travelling Wave less effective at high frequencies Current I I-coil delivered by power supplies falls off with f Besides, for the same I I-coil, the magnetic perturbation exerted on the plasma decreases due to: Partial compensation from image currents in the wall More Shielding associated with (faster) rotation Furthermore, the same B I-coil couples less effectively with a faster, rotationally mitigated, weaker mode (= compass of reduced µ immersed in the same B B imparts reduced µxb torque) Phase delays in power supplies (SPAs) SPAs=Switching Power Amplifiers discrete steps 26