Results from Alcator C-Mod ICRF Experiments

Transcription

1 Results from Alcator C-Mod ICRF Experiments 18 th Topical Conference on RF Power in Plasmas June 4-7, 009 S.J. Wukitch, Y.Lin and the Alcator C-Mod Team Key Results: 1. First demonstration of efficient RF flow drive by mode converted waves.. Observed direct fast wave heating on electrons with low single pass absorption. 3. RF sheaths were observed despite insulating protection tiles and were larger in H-mode than expected. 4. Demonstrated real time matching utilizing ferrite stubs at high power. All C-Mod presentations can be found at 18th RF Topical Ghent 1

2 ICRF Mode Conversion Flow Drive (MCFD) Background: RF plasma flow actuator may offer a unique means to externally:» Manipulate transport t via flow shear stabilization ti and» stabilize of certain high-beta MHD instabilities. Theoretical calculations suggest RF flow drive is possible. 1-3 Limited experimental evidence of flow drive with RF waves.» Enhanced confinement attributed to ion Bernstein wave shear flow in PBX-M. 4» Plasma flow observed with low frequency RF heating in Phaedrus-T. 5» Sheared poloidal flow observed with IBW in TFTR. 6» Poloidal flow observed between mode conversion layer and minority resonance in TFTR. 7 Experimental Description: Utilize mode conversion (MC) from long wavelength fast wave to two short wave length modes: ion Bernstein and ion cyclotron waves. Utilize L-mode discharges and Compare mode converted absorption scenario with minority absorption. 1. G.G. Craddock et al., Phys. Rev. Lett. 67, 1535 (1991).. L.A. Berry et al., Phys. Rev. Lett. 8, 1871 (1999). 3. E.F. Jaeger et al., Phys. Rev. Lett. 90,, (003). 4. B.P. LeBlanc et al., Phys. Plasmas, 741 (1995). 5. S.J. Wukitch et al., Phys. Rev. Lett. 77, 94 (1996). 6. B.P. LeBlanc et al., Phys. Rev. Lett. 8, 331 (1999). 7. C.K. Phillps et al., Nucl. Fus. 40, 461 (000). 18th RF Topical Ghent

3 Observe Toroidal and Poloidal Rotation with Mode Conversion Toroidal rotation, Vφ, profile is peaked and approximately twice that observed in D(H) 80 minority heated discharge. 60 For D(H) minority absorption the 40 rotation profile is approximately flat 0 and dominated by intrinsic rotation. Change in poloidal rotation profile is 0 peaked off-axis in the ion diamagnetic drift -0 direction. V θ is up to 1.5 km/s and localized: 0.3 < r/a < 0.7. For D(H) minority absorption, V θ less than 0.5 km/s and has no clear spatial structure. Discharge details: Upper single null, L-mode plasmas where the 3 He concentration was Toroidal Rotation V φ [k km/s] Mode Conversion Minority Ohmic r/a -ΔV [km/s] θ modest (n 3He /n e ~ 10%). 1.0 D( 3 He) mode conversion layer is near the magnetic axis for 50 MHz. 0.0 D(H) minority resonance is near the magnetic axis for 80 MHz. Injected power ~ 3 MW. Mode Conversion Minority r/a 18th RF Topical Ghent 3 Ion diamagnetic drift directi ion

4 MCFD Scales with RF Power D(3He) MC D(H) Minority ΔVφ [km m/sec] 60 ΔVφ [km m/sec] ΔW/Ip [kj/ma] P RF [MW]/<n e >[100 m-3] For D( 3 He) MC absorption, the change in toroidal rotation, ΔV φ, is approximately twice the empirical i intrinsic i i rotation ti scaling. 1 For D(H) minority heating, the ΔV φ scales with ΔW/I p 1 For D( 3 He) MC absorption, the change in toroidal rotation, ΔV φ, is scales as P RF / <n e > not with RF E-field. 1. J. Rice Nucl. Fus. 41, 77 (001). 18th RF Topical Ghent 4

5 MCFD is largely Independent of Antenna Phase The direction of rotation is independent of the antenna phase. V φ in co-current direction. V θ in ion diamagnetic drift direction. Phase scan showed only 10% variation between co- counter-and dheating phase. The rotation magnitudes are similar il for all antenna phases. 18th RF Topical Ghent 5

6 Measurements Confirm Presence of MCICW Mode converted ion cyclotron wave (MCICW) detected by phase contrast imaging about ~ 4 cm away from the 3 He cyclotron resonance and on the HFS of magnetic axis. Wave number k R ~ 3-7 cm -1, consistent solution of dispersion equations and to previous MC experiments (Y. Lin et al, PPCF (005)). 18th RF Topical Ghent 6

7 MW W/m TORIC Simulation Indicates Significant Power to Ions ECE Fit Simulation r/a Using n( 3 He)/n e ~ 8-1% for TORIC simulation can reproduces measured MCICW profile from PCI And electron power deposition profile. The MC ICW is damped strongly onto 3 He ions through a substantially broadened IC resonance. Fast wave: k ~ 10 m -1 MCICW: k ~ m -1 ω = ω c + k th RF Topical Ghent 7 MW/m 3 v, 3 He

8 Power to Ions Appears to be Important Experimentally, no flow drive is observed when the power is absorbed by electrons. Flux averaged power deposition to 3 He from TORIC simulation is approximately at the same location where poloidal rotation is observed. [k km/s] V φ ΔV θ 3 MC ICW power to He (flux surface averaged) Suggests ICW - ion interaction is key to 0 (b) MCFD. r/a Thought experiment: Assume toroidal force is proportional to ICW power deposition profile to ions. Solve the momentum transport equation in cylindrical coordinates and select diffusion and pinch velocities to match the observed profiles. Assumed force profile is consistent with observed rotation profile with χ φ~ 0.1 m /s. Estimated total toroidal force ~ N per MW ICW or N per MW total injected power to match experimental data.» Injected fast wave momentum (P/v φ ) is ~0.03 N/MW.» Mode converted ICW momentum content is 0.15 N/MW. Sufficient wave momentum is injected but require asymmetry in plasma response. 18th RF Topical Ghent 8 MW/m [k km/s] (a)

9 Future Directions ) Time (s TFTR MC flow drive Experimental Examine dependence of flow drive on 3 He 4.5 concentration. Investigate importance of ion versus P RF electron absorption. MC Investigate flow drive scaling with plasma density. Ω 3He Appears to scale inversely with density Density also changes the ICW perpendicular wavelength, hence the R (m) radial wave pattern. Simulation Investigate MCFD in He and H plasmas. Seek scenarios for ITER and larger H mass density will change perpendicular devices where significant power wavelength and radial wave pattern. is absorbed b dby ions via mode Participate (pending approval) in JET converted waves. experiments investigating MCFD. Need to avoid direct electron Data mining has produced evidence and ion cyclotron absorption. consistent with analysis of JET plasmas. Simulate previous reported results Examine MCFD in H-mode plasmas. from TFTR. 1 Assess influence of plasma current on flow Will require increased computational di drive. resources. 1. C.K. Phillips et al, NF (000). km/s

10 Direct Fast Wave Absorption Experiments Motivation: Fast wave current drive is expected to provide central seed current for optimizing current profile in AT plasmas. Background: FW damping is dependent on electron β and electron temperature. 1 ω ImΔ A= 1 T = 1 e, ζ e =, vte = kvte π ω ω pi e k e ζ Im = βζ e e ω ω ci T T m k x e Estimates low single pass, 1-%, is accessible from existing discharges. Low single pass raised concerns since earlier work suggested 4% per pass edge loss. e Fundamental D resonance 1. M. Porkolab, AIP Conf. Proc. 314, 99 (1994).. C.C. Petty et al, Nucl Fus. 35, 773 (1995). 18th RF Topical Ghent

11 Initial Results from Fast Wave Absorption Successfully observed FW heating. Utilized 50 MHz and heating phase (n φ =±10) Used D(H) minority at 80 MHz to heat plasma to >4 kev at top of sawtooth. Discharges were USN, L-mode with an H-factor ~1.3. Assuming L-mode scaling, effective absorbed power fraction ~0.5.» Central T e increased ~ 1keV.» Significantly higher density production with FW compared to minority heating.» Increase in neutron rate during FWEH.» Significant impurity production but radiated power remains under control B T ~5. T, I p =1. MA, USN P RF (MW) FW H minority W MHD (MJ) T e0 (kev) n e (10 0 m -3 ) neutrons (x10 13 s -1 ) P RAD (MW) Time (s) th RF Topical Ghent 11

12 FW Absorption is Appears Dominant Current drive phasing resulted in little or no heating. Current drive phasing has lower n φ =7 and faster wave phase speed. 0.1 FW absorption has nonlinear dependence on phase 0.05 speed. 4 Suggests fundamental deuterium absorption is not 1.5 significant. Lowering target tplasma temperature resulted in reduced heating efficiency for heating phase heating phase current drive phase P RF (MW) W MHD (MJ) T e0 (kev) n e (10 0 m -3 ) neutrons (x10 13 s -1 ) P RAD (MW) B T ~5. T, I p =1 MA, USN FW H minority Target discharge had lower T e (~3.6) which results in single pass <1% Time (s) 18th RF Topical Ghent ,3 10

13 ICRF Impurity Production with Metallic PFCs is Challenging A primary challenge to ICRF utilization in present experiments and future reactors is to reduce/eliminate impurity production specific to ICRF. Boronization required for high performance discharges in C-Mod. 1 Large gap, strong puffing necessary for ICRF H-modes in ASDEX-U. Metallic plasma facing components (PFCs) for fusion devices including ITER are being considered. Tritium retention is expected to be less than for carbon PFCs and Have significantly better erosion resistance. Allowable high Z metallic impurity concentration is very restrictive. Allowable W concentration is <10-4 for ITER. W will radiate in the plasma pedestal region and may detrimentally effect H-mode performance. High confinement modes have increased impurity confinement. Prescription 3 to ameliorate impurity production has been: Operate the antenna in dipole phasing, Align the Faraday screen with total magnetic field, and Utilize low Z film on antenna and plasma facing surfaces. 1. B. Lipschultz et al., Plasma Phys. (006).. Neu et al., PPCF 49, B59 (007) and Bobkov et al, 50 th IAEA (008). 3. Jacquinot et al., Fusion Eng. Design 1, 45 (1990). 18th RF Topical Ghent 13

14 Low Z Film Lifetime is Limited Following a boronization, successive discharges are heated by ICRF. Boron is eroded dwhen impurity i control is lost during the H-mode Estimated erosion rate is ~15-0 nm/s with RF heating JET utilizes beryllium coating on FS and limiter to control high Z impurity influx. 1 Estimated erosion rate from Faraday screen is 15 nm/80 MJ. Be layer is eroded from limiters faster but not specified. Assuming similar erosion rate for Be in ITER (400 s discharge) as B in C-Mod, lifetime of 5 mm layer in ITER is ~1000 discharges. As Be on JET FS, lifetime of 5 mm is ~15000 discharges P rad (MW) 1st Discharge nd d Discharge 4th Discharge H-mode Time (s) W MHD (MJ) MBures 1. M. et al., Fusion Engineering and Design 1, 51 (1990). 18th RF Topical Ghent

15 Can Plasma Performance be Improved by Insulating Limiters? Replaced molybdenum protection tiles with insulating BN tiles to Remove local antenna molybdenum source. Eliminate sheaths on field lines that intersect the antenna protection tiles. Remove RF enhanced plasma potential. Approach was similar to BN limited installed in the Phaedrus-T tokamak. * Successfully eliminated metal impurities with installation of BN tiles and removal of Faraday screen. *Majeski et al., 11 th RF Power in Plasmas, AIP Conf Proc 44, 3 (199). 18th RF Topical Ghent 15

16 Insulating Tiles did not Improve Performance Plasma performance was unimproved despite decrease in antenna limiter Mo source. Expected less RF power to obtain same stored energy. Suggests RF limiters are not the only source of core Mo. Significant RF enhanced plasma potential is observed despite BN tiles. BN tile impedance is 100x the plasma-pfc sheath impedance. RF voltage is dropped across insulating tile. Plasma potential ~100 V in H-mode and would have significant enhanced sputtering. Contrasts with previous results. In Phaedrus-T, little residual plasma potential increase was observed but power density is times lower than C-Mod. 1 In C-Mod, BN merely covered metallic surfaces whereas in Phaedrus-T the BN replaced the metallic antenna box. P RF (MW) RF power required to obtain 160 kj < W Tot < 180 kj BN installed Year RF Power [MW] H-mode th RF Topical Ghent 16 0 Plasma Potential [V] Time (s)

17 In L-mode, Sheath Voltage Scales as P 1/ only with Boronized Walls. Sheath voltage is expected to scale with antenna power as 150 RF Power Density [MW/m ] 6 10 unboronized boronized Vplasma = VantCn PRF where V ant is the antenna voltage, C n is constant depending on antenna loading, and P rf is the coupled power. 1 Plasma a Potential [V] ITER Power Density,7, , , Post-boronization, find the plasma potential is proportional to square root of power. 150 In a poorly conditioned machine, plasma potential scales with RF power. 100 tential [V] RF Power [MW] unboronized boronized Plasma Pot 50 0, ,7,9 1. D.A. D'Ippolito et al., Nuclear Fusion 38, 1543 (1998) Sqrt(RF Power) [MW 1/ ] 18th RF Topical Ghent

18 Sheath Potentials are Higher in H-mode For similar power, the sheath voltage is expected to increase with antenna voltage. Measured antenna voltage in H-mode is ~0% than comparible L-mode. Expect RF enhanced plasma potentials to be ~0% higher in H-mode Observe that plasma potentials are about twice as large in H-mode than L-mode. Caveats: RF power is affecting antenna loading.» Density profile is modified during power power ramp.» Antenna loading is constant above 100 kw. RF sheath model needs to allow for cross-field currents.,3» Ion flows across the magnetic field along the length of the flux tube are balanced by electron currents in and out of one end of the flux tube.» Integrated flux tube impedance is less than or equal to the impedance through the sheath. RF is creating an energetic edge electron population.» Small energetic electron population, ~0.1% can double the sheath voltage. 4» Possible mechanisms are Fermi 5 and near field 6 acceleration.» No evidence of energetic edge electrons from TS measurements. 7 Plasm ma Potential [V] P1/ H-mode L-mode H-mode with BN tiles RF Power [MW] 1 Sorensen et al., Nucl. Fusion 36 (1996). D.A. D Ippolito et al., Nuclear Fusion 4, 1357 (00). 3 E. Faudot et al., Phys. Plasmas 13, 0451 (006). 4 D. Tskhakaya et al., Phys. Plasmas 9, 486 (00). 5 M. D. Carter et al, Phys. Fluid B 4, 1081 (199). 6 V. Petržílka et al., in Proc. 3nd EPS Conf. on Plasma Phys. 9C, P-.095 (005). 7 J. Gunn et al., 50 th IAEA (008). 18th RF Topical Ghent ,,

19 Future Research: Proposed ICRF 4-Strap Antenna Reduce coupled E by making antenna symmetric along a field line. Integrated E along a field line would be reduced. Entire structure is perpendicular to field line. Estimated reduction due to rotation is ~10-30x reduction in E dl along a field line. Compared tilted versus horizontal antenna load via lossy dielectric ect c load antenna. a. Feedthrus are 5 diameter (present are 4.5 ). Power density at MW (3MW) is 9.8 MW/m (14.8 MW/m ) Strip line impedance is 30 Ω (J is 50 Ω and D/E is 30 Ω) Screen is aligned to B-field and is 50% transparent (same as J) Peak n φ = 14 (0,π,0,π); 11 (0,π,π,0); and 8 (0,π/,π,3π/) in vacuum spectrum (bit higher than J antenna).

20 Reconciling C-Mod and AUG Results C-Mod and AUG both have insufficient H-mode performance with ICRF and metallic PFCs. C-Mod ddata indicates the primary RF impurity i source is away from the antenna. AUG concludes the local ICRF limiters are primary source.» Larger gap and high h puffing reduce associated impurity i production. JET in the 90s thought it was the Faraday Screen. C-Mod ICRF limiters are behind plasma limiters by ~0.8 cm whereas AUG ICRF limiters it are the first point of contact t with the plasma. C-Mod antenna generates less E per MW than AUG. Straps are out of the antenna box in C-Mod and inside the antenna box in AUG. Greater image currents with straps inside the antenna box. For sputtering: C-Mod is dominated by deuterium on Mo when plasma potential is >90 V (typical of H-mode). AUG is dominate impurity sputtering from C and O on W. If dominated by impurity sputtering of high Z, practically c no safe E and lifetime of low Z materials will be shortened. AUG data from Bobkov et al., 50 th IAEA (008). 18th RF Topical Ghent 0

21 Proposed ICRF Antenna

22 Challenge: Maintain Reliable Power Transfer Edge plasma density and density profile determines the antenna coupling efficiency. i Sets the distance to propagation and propagation characteristics. Edge density and its profile is dependent upon : Plasma current, target density, magnetic field, confinement mode, wall conditions, and RF power. Transient phenomena modify the coupling on fast time scales: ELMs, Monster sawteeth, and Confinement mode transitions n e [m-3 ] 10 1 LCFS Main Plasma Limiter Near SOL Cut-off density Far SOL Scrape-off Layer Limiter Shadow Antenna Limiter Rmajor [cm] 18th RF Topical Ghent

23 Utilize Triple Stub System Ferrite Tuner #1 Ferrite Tun ner # Fixed Stub Generator Antenna DC1 DCC DC DC3 System is installed in the E antenna matching network and operates at 80 MHz. Fixed stub is used for pre-matching to prevent excessive voltages in the ferrite stubs. Ferrite stubs were originally designed for 60 MHz for hybrid ring system at DIII-D and tested at ASDEX-U. Have directional couplers for forward power, reverse power, and phase at four locations to enable analysis. 18th RF Topical Ghent 3

24 Ferrite System Characteristics Electrical length is 35 cm for +/- 150 A. Swept end to end is 4 msec. Computational time is 00 μs and presently limited by computational speed. Calculating tuner settings from transmission line relations. Comprehensive arc protection to prevent coupling to an arc. Optical arc detection in the ferrite tuners. Reflected to forward power threshold set to 5%. Tuner #1 Tuner # Arc detection fiber Phase balance monitor utilizes current probes in the antenna strap near the strap ground.» Detects loss of current in the current strap.» Primary defense against arcs at low voltage locations. 18th RF Topical Ghent 4

25 Smith Chart Representation of Matching Range To XMTR Admittance Smith Chart Y = G+jB 80% 0% 3A 10% B 5% 40% 3B To oxmtr Example Admittance Smith Chart 40% -Im(Γ) 0.0 A 1A 1% -Im(Γ) 0.0 G=0.5 G= To AN 1B ANT Re(Γ) <1% reflected <5% reflected -0.5 T -1.0 To ANT Re(Γ) Design range covers wide range of plasma conditions in both L and H- mode. 18th RF Topical Ghent 5

26 Overall FFT Performance is Good With pre-matching, obtained 37 kv and 1.85 MW coupled into H- mode. Found voltage limit to be ~5 kv in one of the tuners. 0.0 mode 0.15 Other tuner is higher by at least 30%. tio Powe er Loss Ra Observed increased losses at high power Related to net magnetic field in particular tuner rather than nonlinearity in losses due to power. total loss in double stub loss through first FFT short ttransmission i line loss P circulating cu g (MW) FFT characteristics show that for coil current >100 A will result in increased losses. Pre-matching avoids: High voltage in FFTs and Avoids driving the net magnetic field to zero and increasing losses. 18th RF Topical Ghent 6

27 FFT Load Following is Reasonable P forw (M MW) Coil Current (A) Ant. Loading (Ω) D α (A.U.) L-mode H-mode Tuner Tuner Time (s) P relf (MW) 15 Ant. Loading (Ω) D α (A.U.) 0.15 Γ = P refl /P forw Time (s) FFT follows load transitions associated with discharge evolution and confinement transitions very well. Response is insufficient to maintain match <5% during entire ELM because it is <50 μs but sufficient to reduce mismatch < fault level. Currently limited by computational time but ultimately will be the FFT power supply. 18th RF Topical Ghent 7

28 Summary Observed significant toroidal and poloidal flow with D( 3 He) mode conversion. Toroidal rotation exceeds empirical i scaling with hinstrinsic/spontaneous i i rotation. Scales with RF power. Majority of power is absorbed by ions. First observation of Fast wave electron heating in C-Mod with low single pass absorption. RF sheaths were: Still present for an antenna with insulating protection tiles and Observed increase with H-mode compared to L-mode was larger than expected. A triple stub, fast ferrite matching network reliably maintains power transfer to the plasma with low reflected power over wide range of conditions. To date, 1.85 MW coupled into H-mode. Further improvements are possible in the voltage limit and time response of these ferrite tuners. 18th RF Topical Ghent 8