Review of Recent Experimental and Modeling Progress in the Lower Hybrid Range of Frequencies at ITER Relevant Parameters

Review of Recent Experimental and Modeling Progress in the Lower Hybrid Range of Frequencies at ITER Relevant Parameters P. T. Bonoli PSFC MIT, Cambridge, MA 02139 (USA) 20 th Topical Conference on Radio Frequency Power in Plasmas Sorrento, Italy June 25-28, 2013

Outline for review Short review on LH wave physics and motivation for using LHCD in ITER and beyond. Experimental demonstrations of LH actuator capabilities needed for ITER (core ITB control). Progress in diagnostic measurements and validating core simulation capability (combined ray tracing/fokker Planck, spatial diffusion effects, full-wave effects, spectral gap). Role of the scrape off layer (SOL) linear and nonlinear absorption processes, wave scattering, long distance coupling, modifications of the SOL by LH waves. New beneficial applications of LHRF toroidal rotation changes, modifications of the pedestal and ELM s. Looking to the future role of integrated modeling and high field side (HFS) launch.

Lower Hybrid (LH) Wave Coupling and Propagation RF power at ~ ( ce ci ) 1/2 is coupled from a waveguide launcher or Grill with E rf = E // (slow wave) and tunnels through an evanescent layer where > pe (Brambilla, 1976): Accessibility of slow wave to the plasma core is determined by n // : D Pn Pn P 4 2 4 2 0 0 n 2 2 pe pe pi n acc 1 ce ce Slow wave is electrostatic in the core with pe / >>1: n 2 2 pe 2 n 2

LH Wave Absorption and Current Drive Waves undergo electron Landau damping at v // ~ 3 v te forming a quasilinear plateau in the electron distribution function f e : CD efficiency is high for narrow spectrum of damped LH waves (Fisch, PRL, 1980; Karney, PoP, 1985): Plateau electrons are relatively collisionless since fast ~ 0 /v 3. Plateau electrons also pitch scatter from to direction further reducing the collisionality.

Lower Hybrid Current Drive is Attractive for Tokamak Reactor Applications in Devices Ranging From ITER to Power Plant (ARIES-RS) Lower hybrid (LH) waves damp strongly on tail electrons at ~3 v te and can therefore be used to drive current off-axis at r/a 0.6. High v // ~ 3 v te gives efficient current drive with reduced trapping effect. Localized J LH (r) can be used to control and move the shear reversal point associated with an internal transport barrier (ITB) to larger radii, with q min > 2. Improved confinement broadens the pressure profile, thus increasing Q and f BS = I BS /I p.

ITER steady state scenario studies found that adding 20 MW of LH can provide control of li(3) down to < 0.7 and raise q min while keeping ρ(q min ) about the same (Kessel, IAEA, 2010; Poli, RF Conf, Invited Talk, 2013) Ip = 7.75 MA I BS = 3.3 MA I NB = 3.2 MA I EC = 0.7 MA I FW = 0.25 P NB = 33 MW P EC = 20 MW P IC = 20 MW (feedback) Q = 3.4 li(3) = 0.9 n/n Gr = 1.0 β N = 2.2 H 98 = 1.57 Z eff = 2.4 T ped = 3.7 kev n(0)/<n> = 1.5 Ip = 9.4 MA I BS = 4.1 MA I NB = 3.5 MA I EC = 0.71 MA I LH = 0.93 MA P NB = 33 MW P EC = 20 MW P LH = 20 MW Q = 4.1 li(3) = 0.69 n/n Gr = 0.82 β N = 2.45 H 98 = 1.60 Z eff = 2.4 T ped = 3.75 kev n(0)/<n> = 1.5

ARIES-AT design utilized 37 MW of LHCD power in high f BS operating mode (Najmabadi, FED, 2006) Major Parameters for ARIES-AT Major radius (m) 5.2 Minor radius (m) 1.3 Plasma current (MA) 12.8 On-axis field (T) 5.8 Plasma elongation 2.20 Plasma triangularity 0.90 Normalized beta - N (%) 5.4 On-axis safety factor (q 0 ) 3.50 Minimum safety factor (q min ) 2.40 Edge safety factor q e 3.70 Internal inductance l i (3) 0.29 Bootstrap current fraction 0.91 Current drive power (MW) 42 T e (0), T i (0) (kev) 18 Electron density (10 20 m -3 ) 2.15 ITER89-P multiplier 2.0 ARIES-AT LHCD System (f 0 = 3.6 GHz) N// Power Driven Current (MW) (MA) 1.65 3.1 0.15 2.0 4.4 0.2 2.5 8.2 0.3 3.5 8.9 0.2 5.0 12.4 0.15

JT60-U demonstrated pure non-inductive plasmas (BSCD+LHCD) in quasi-steady state with reverse shear sustaining and controlling an ITB (Ide, NF, 2000) f BS =23% B 0 = 2 T I LH /I p = 77% n e ~ 2 10 19 m -3 I p = 0.9 MA P NB =2.5 MW P LH =2.5 MW V l 0 f 0 = 2 GHz n // (0) =[1.65, 2.25] q( ) during LHRF q( ) after LHRF J LH = J(MSE) J BS (ACCOME) q( ) from MSE

Core ITB s have been sustained with LHCD in ELM y H-mode plasmas in JET heated by NBCD and ICRF (Litaudon, PPCF, 2002) ~2.7 MW of LHCD at 3.7 GHz [n // (0) = 1.8] is coupled using gas injection (CD 4 ) technique with P NBI = 15 MW and P ICRF = 4 MW at n e ~ 5-6 10 19 m -3. B 0 = 3.45 T I p = 2 MA N = 1.7 I BS ~ 1.0 MA I LH ~ 0.4-0.8 MA I NB ~ 0.2-0.6 MA ITB ~ res

Electron ITB has been triggered with LHCD in Alcator C-Mod (Shiraiwa, IAEA, 2012) [MA] [V] 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1.0 0.5 0.0-0.5-1.0-1.5-2.0 5 4 [kev] 3 2 1 0 1101119004 I p & P LH LH Loop Voltage T e0 0.0 0.5 1.0 1.5 2.0 P LH = 800 kw, n // (0) = 1.9 I p = 0.5 MA, B 0 = 5.4 T 0.3 s after LH injection T e (0) increases spontaneously from 2.5 4 kev.

Steady state regimes have been achieved with LHCD in diverted plasmas on Alcator C-Mod at the ITER n e, B 0 pe / ce P LH [kw] V loop [V] 1101019026 1000 500 n e [10 19 m 3 ] T e0 [kev] x 10 19 0 8 6 4 2 0 1.5 1 0.5 0 4 2 0 0.8 1 1.2 1.4 1.6 Time [s] B 0 = 5.4 T n e ~ 5 10 19 m -3 pe / ce ~ 0.42 f 0 = 4.6 GHz n // (0) = 1.6 V loop ~ 0 for 0.2 s ~ R CD ~ 0.2-0.25 (10 20 A/W/m 2 )

As density increases, the effect of nonthermal contribution to current from LH can be seen to decrease as E // / E dreicer 0 (data analyzed following Giruzzi, NF, 1997) 0 0.5 0.85 10 19 A/W/m 2 0 2.5 10 19 A/W/m 2 EAST: n e = (1-1.3) 10 19 m -3, I P = 0.25 MA, n // (0) = 2.1 (B. Ding, FST, 2011) C-Mod: n e = (3.5-7) 10 19 m -3, I P = (0.5-1.0) MA, 120 kw < P LH < 830 kw) (Bonoli, PoP, 2008)

Steady state discharges in Tore-Supra have made it possible to test launcher cooling concepts for long pulse (Hillairet, NF, 2013) C3 - FAM is an actively cooled fully active multijunction launcher C4 PAM is a passive-active multi-junction launcher concept. f 0 = 3.7 GHz

What are the issues then for LHRF applications in ITER and beyond? Core wave propagation and absorption: Will spatial diffusion of fast electrons be important. What is the underlying mechanism(s) for bridging the spectral-gap and should we worry about this for ITER? For SOL propagation: How will LH waves interact with the SOL as density is increased in single pass and multi-pass regimes importance of wave scattering, and nonlinear effects (PDI), modifications to the SOL by LH power. Can waves be coupled over long distances (~10-15 cm). Complications from divertor geometry and separatrix Can we utilize un-anticipated benefits of LHRF power: LH-induced rotation profile changes. Modifications of edge plasma at higher density (ELM & pedestal changes). What innovative solutions will be needed to solve problems for LHRF applications to ITER and beyond.

Progress in theory, simulation, and model validation will help to answer key questions for future applications Extensive use of synthetic diagnostic techniques for hard x-ray (HXR) emission and Motional Stark Effect (MSE) have been a major advance: Provides stringent tests of models for core wave propagation and Fokker Planck physics. Can be used to understand experimental trends. This talk will focus primarily on HXR measurements. Advances in modeling capability: Development of full-wave EM field solvers for LH waves. Routine use of 3D (v, v //, r) Fokker Planck codes. Inclusion of 2D SOL plasma profiles, 3D vessel geometry, and magnetic geometry in diverted plasmas (separatrix) New attempts to assess nonlinear processes such as parametric decay instability (PDI) and wave scattering from density fluctuations. Routine use of advanced models for LH actuators in whole device modeling frameworks. Some of this work has been done as part of an ITPA-IOS Joint Activity IOS5.3.

Fokker Planck solution via a response function ray tracing approach is computationally fast and thus widely used Define and solve an Adjoint Problem for the Spitzer - Harm function ( ): [Karney, NF 1985] J rf d 3 p p rf rf D QL f e p The response function contains all the physics effects already in the numerical 2D and 3D FP solvers such as particle trapping, DC electric field effect, and momentum conserving corrections in C(f e ) (J rf / S rf ) can be found accurately, but computation of J rf requires separate knowledge of rf and f e. rf and f e are evaluated from a 1-D (p // ) solution of the FPE. Codes using this approach FRTC, LSC, ACCOME

Solution via combined 3-D(p, p //, r) Fokker Planck ray tracing models are now more routine Codes compute the steady state solution of FPE, with and without the radial diffusion operator: p // D rf ( p // f ) p s e // + ( C( p // f e, p //, p ) ee 1 fe ) r F r r r // f p f t e e // Ray tracing and FP solver iterate until a self-consistent D rf and f e are obtained. FP codes employ numerical bounce averages for C(f e ) and D rf. Codes using this approach GENRAY/CQL3D, C3PO/LUKE

Understanding fast electron diffusion is critical for current profile control applications Early work by Baranov (NF, 1996) on JET showed that FAST could be inferred by matching the measured FEB signal with a synthetic diagnostic: f e (r, v, v // ) simulated with a ray tracing / 2D (r, v // ) Fokker Planck code. Found FAST ~ 0.5 m 2 /s reproduced FEB signal. n e (0) = 1.7 10 19 m -3, n // = 1.85, I P =0.4 MA Fast electron diffusion important in this regime.

Recent measurements / analysis on C-Mod have shown that FAST is negligible at higher density and current (Schmidt, PoP, 2011) Measured HXR profiles on C-Mod (solid lines) could be simulated (dashed lines) using GENRAY- CQL3D with FAST ~ 0.01 (v // / (v te 3 ) m 2 /s : n e (0) = 9.0 10 19 m -3, n // = 3.1, I P =0.8 MA Fast electron diffusion negligible in this regime since slow << FAST Fast electron diffusivity inferred from HXR analysis of LH power modulation experiments consistent with FAST needed in simulations: D = FAST = 0.01 m 2 / s n e (0) = 9.0 10 19 m -3, n // = 3.1, I P =0.8 MA

Tore Supra: Trends in measured HXR consistent with simulated HXR from ray tracing / Fokker Planck (Goniche, NF, 2013) Decline in HXR at n e ~ 5 10 19 m -3 consistent with loss of wave accessibility (f 0 =3.7Ghz, B 0 =3.85T): HXR successfully simulated using C3PO/LUKE and synthetic code R5- X2. Measured profiles always peaked at r/a ~ 0.2 No conclusive data to link decline to SOL interactions (except for one discharge where PDI was observed).

Tore Supra: Measured HXR simulated using ray tracing / Fokker Planck model (Peysson, PoP, 2008) Simulated HXR (solid line) from ray tracing/fokker Planck model (C3PO/LUKE) and RX-52 HXR synthetic diagnostic code: I P =0.7 MA, P LH =4 MW, f 0 =3.7 GHz, n // (0) = 2.0. Included effect of helical winding of magnetic field to explain the LFS/HFS asymmetry. Have developed simplified models for D QL in multi-pass regime and tested them using the synthetic HXR technique (EPS, 2007). Have recently interpreted HXR results in terms of LH wave scattering from density fluctuations as a way to close spectral gap in LHCD (Decker Invited Talk I3.3).

1 ] Count Rate (Chords 9 24, 40 200 kev) [s C-Mod: Measured HXR found to be consistent with ray tracing / Fokker Planck if 2D SOL is added to GENRAY ray tracing code (Wallace, PoP, 2010; NF, 2011) 10 7 Line Integrated HXR Count Rate 10 6 10 5 10 4 n =1.9, 5.4T, 800kA n =2.3, 5.4T, 800kA n =1.9, 7.0T, 800kA n =2.3, 7.0T, 800kA n =1.9, 5.4T, 1.1MA n =2.3, 5.4T, 1.1MA 10 3 0.4 0.6 0.8 1 1.2 1.4 1.6 x 10 20 Line Averaged n e [m 3 ] GENRAY-CQL3D w/o SOL GENRAY-CQL3D with SOL n e-av = 1.4 10 20 m -3, I P =0.8 MA, B 0 =5.4 T Decline in HXR at n e > 1.1 10 19 m -3 consistent with collisional damping of LH waves in SOL (f 0 =4.6Ghz, B 0 =5.4T): HXR simulated using GENRAY-CQL3D and a synthetic code Caveat: PDI also observed in these discharges.

FTU: Measured HXR behavior at high density is also consistent with ray tracing / Fokker Planck if collisional damping is included (Barbato, NF, 2011) Standard Regime (SR) plasmas exhibit dominant collisional damping at edge while New Regime (NR) discharges produced with pellet injection exhibit reduced collisional damping and recover HXR: Ray tracing / CD simulated using the FRTC code without SOL PDI also observed in these discharges (Cesario, Nature Comm, 2010).

Development of full-wave field solvers in the LHRF has been a major advance in modeling (Wright, PoP, 2009; Meneghini, PoP, 2009, Shiraiwa, PoP, 2010) Important tool for understanding multi-pass damping regimes and wave behavior at higher density: Approach includes full-wave effects such as diffraction, scattering at a cut-off, and poloidal mode coupling. Wave equation is solved using either a semi-spectral ansatz for E (TORLH) or a pure finite element method (FEM) approach (LHEAF): 2 c 2 4 i E SE id( b E ) PE J ANT Field solvers have also been coupled to 3D Fokker Planck codes: TORLH+CQL3D (Wright, PoP, 2009) LHEAF+VERD (Shiraiwa, PoP, 2011)

Full-wave LH simulations reveal spectral broadening seen in ray tracing that is needed to fill the spectral gap in weak damping regimes (Wright, PoP, 2009) TORLH simulation for C-Mod at n e ~ 5-7 10 19 m -3, Te(0) = 2.3 kev, n // (0)=-1.55. Damping is based on quasilinear distribution from coupling to CQL3D. Poloidal power spectrum on several flux surfaces for TORLH-CQL3D simulation exhibits spectral broadening due to poloidal mode coupling and diffraction.

Full-wave/ Fokker Planck and ray tracing / Fokker Planck simulations have shown that toroidicity can close the spectral gap in EAST (C. Yang, ASIPP, to be published, 2013). GENRAY/CQL3D and TORLH/CQL3D simulations reveal that toroidal effect is sufficient to close a large spectral gap despite high aspect ratio of device [R 0 /a = 1.85m / 0.45m]: n e (0) = 1.0 10 19 m -3, T e (0) = 1.5 kev, and I p = 480 ka, f 0 = 2.45 GHz, and n // (0) = 2.1 Spectrum of Re(E ) versus m Seld (MW / m3) TORLH+CQL3D GENRAY+CQL3D FFT (Re(E )) 10 0 ~ r/a 0.10 0.30 0.50 0.70 0.90 10 2 1.00 10 4 10 6 rho 10 8 1000 500 0 500 1000 Poloidal mode number - m

Full-wave FEM solver LHEAF (Meneghini, PoP, 2009) models SOL and vessel geometry accurately simulations at high density reveal spectral broadening at edge that can possibly contribute to HXR decline (Meneghini, PhD Thesis, 2012) Power abs in r/a bin [kw] 35 30 25 20 15 10 5 Forward Power Only LHEAF/VERD GENRAY/CQL3D 0 0 0.5 1 1.5 r/a LHEAF-VERD simulation for C-Mod at n e ~ 1.3 10 20 m -3 shows high Landau absorption inside the LCFS compared to the collisional damping outside the LCFS predicted by GENRAY-CQL3D.

Edge losses in C-Mod due to collisional damping in the SOL and full-wave spectral effects in the edge cannot fully explain the decline in HXR at high density Points to possible role of nonlinear effects in HXR decline (Baek, PPCF, 2013; also Baek, Invited Talk I2.7).

In LSN C-Mod plasmas, ion cyclotron PDI is excited near the inner plasma edge above e 1x10 20 m -3 (Baek, PPCF, 2013; RF Conf. 2013, Invited Talk I2.7) At n e 1.2 x10 20 m -3 no ion cyclotron PDI occurs at the outer edge within the detector sensitivity Inner wall probe detects PDI occurring at the inner edge (decay of LH pump wave into an ion cyclotron quasi-mode). Underlying cause for onset of PDI on HFS at lower density than expected (Porkolab, PF, 1977) thought to be due to weaker penetration of LH pump wave as the density increases.

HXR decline in C-Mod may be mitigated by insuring single pass damping of LH waves (Shiraiwa, RF Conf., 2013) Proposed experiment Double LHCD power to 2MW level 2 nd launcher is located off-midplane in order to realize high single pass absorption Goal is to demonstrate recovery of LHCD efficiency at high density by enhanced single pass absorption

Nonlinear PDI has been investigated extensively in FTU as the cause of the HXR decline seen at high density in Standard Regime plasmas (n e ~ 1.5 10 20 m -3 and low outer T e ) Ion sound quasi-mode is found to be excited in FTU Standard Regime plasmas. Mode is reduced in New Regime plasmas created with pellet injection and characterized by high outer T e. Standard Regime Process has been simulated using the LHSTAR code (Cesario, PRL, 2004). New Regime Cesario, Nature Comm, 2010

Extensive work has also been done to interpret the FTU results in terms of LH wave scattering from density fluctuations in the SOL (Pericol-Ridolfini, NF, 2011) Calculated an optical thickness ( OPT ) of the plasma SOL due to LH wave scattering from density fluctuations (following Andrews and Perkins, PF, 1983). Spectral broadening of LH pump wave simulated by scattering found to be consistent with experimental measurement. Found that fluctuation amplitude and OPT are reduced in New Regime discharges.

Simulations by Bertelli (PPCF, 2013) applied Monte Carlo wave scattering technique in GENRAY-CQL3D to study role of fluctuation scattering in density limit on C-Mod Effect of scattering is sensitive to matching between k -LH and k -Fluct : Found that although scattering could broaden RF power density and HXR profiles, it could not explain decline in HXR emission at high density. n e = 0.53 10 20 m -3 n e = 1.47 10 20 m -3

Linear and nonlinear coupling studies of LH waves has been an active area experimentally and theoretically Large gap (~15 cm) needed to couple LH waves in ITER has been tested experimentally on JT60-U (Ide, NF, 2000): In plasmas with H-mode edge. Gap was up to 14 cm. Discharges were diverted, but relatively quiescent.

LH waves have been coupled over large distances (~ 11cm) in JET using a gas puff technique (Ekedahl, RF Conf., 1997; NF, 2005; PPCF, 2012) First experiments were done with D 2 gas puffing into L-mode plasmas (Ekedahl, RF Conf, 1997). Later experiments used CD 4 in H-mode plasmas with edge localized mode activity (Ekedahl, NF, 2005). More recently found that gas injection from outboard midplane was more effective than top gas injection system (Ekedahl, PPCF, 2012)

Recent measurements from C-Mod have confirmed nonlinear modifications to density profile during LHRF injection Profiles measured with an X-mode SOL reflectometer (C. Lau, MIT PhD Thesis, 2103) Profiles have been simulated using the POND code (Meneghini, MIT PhD Thesis, 2012): Accounts for nonlinear ponderomotive force modifications of the density profile by the LHRF power. LH off dashed lines LH on solid lines Simulations solid lines with circles Measured millimeter-like vacuum gap in density profiles: Consistent with what is needed to reproduce experimental values of RC in linear coupling codes ( Grill code, SWAN code, ALOHA).

New experimental results may point way to unexpected advantages of LHRF injection Co- and counter- current ion toroidal rotation is observed on C-Mod after injecting LH waves (A. Cushman, PRL, 2009) and has been studied theoretically (J. P. Lee, MIT PhD Thesis, 2013): Initial change in rotation is found to be consistent with simulated torque density from LH injection. But need to account for intrinsic rotation in order to explain longer time behavior of rotation (J.P. Lee, RF Conf., 2013).

Co-current rotation also observed in EAST with LHRF injection (Shi, PRL, 2011) Rotation change is ~ 40 km/s in the L-mode plasma core and ~20 km/s in the edge. Simple model based on turbulent equipartition and thermoelectric pinch can explain this rotation.

LHRF injection into edge plasma has dramatically modified edge plasma in EAST Strong mitigation of edge-localized modes has been observed on EAST, when LH power is applied to H-mode plasmas with ICRF heating. ELM peak particle flux ( ion ) decreases by ~ factor 2 and stored energy increases during LH pulse. I P = 0.5 MA, n e-av ~ 4.7 10 19 m-3, B 0 = 1.8T, f 0 = 2.45 GHz, P LH = 2 MW. LH waves appear to modify magnetic topology by driving helical current filaments along field lines in the SOL. Leads to a splitting of the divertor strike-point with stabilization effects similar to what is associated with RMP coils.

Pedestal characteristics actively modified with LHRF injection in C-Mod (Hughes, APS, 2010) Apply 1MW of LH power to an EDA H-mode produced with ~1.5 MW of ICRF heating. Improved pedestal characteristics with lower density and higher T e at the edge. LH waves not accessible to the plasma core.

Innovative concepts may be needed for LHRF applications to ITER and DEMO Integrated modeling now includes most advanced LH actuator models available: Makes it possible to optimize and define the ITER Baseline Heating and Current Drive System and to study the feasibility of new actuator concept (see F. Poli, Invited Talk I3.1, RF Conf., 2013). Several integrated modeling frameworks are now available (ITM, SWIM, TASK, CRONOS, TRANSP/TSC, PTRANSP). HFS launch of LH wave may prove to be useful: Lower n // is accessible to same density so that it s possible to burn through edge pedestal.

Time-dependent simulations evolve plasma equilibrium and H&CD source profiles consistently Target plasma R=6.2, a=2.0, =1.8, ~0.45 n/n G >0.75 PF coil currents feedback system Transport model Coppi Tang, CDBM (GLF23, BgB, MMM) TSC (free boundary) Discharge scenario T, n, (R,z) eq H&CD profiles SWIM (fully consistent) TRANSP (analysis loop) Linear MHD stability JSOLVER (refines eq.) BALMSC (ballooning) PEST (kink) (NOVA K) NB : NUBEAM ICRH : TORIC ECRH: TORAY, GENRAY LH: GENRAY, (LSC) 2D Fokker Planck: CQL3D Francesca Poli 20 th RFPPC, Sorrento, Italy, June 25-28

Although HFS launch has a clear theoretical advantage, it is necessary to test the concept (Shiraiwa, IAEA-SSO, 2013) In reactors, combination of high density + high temperature makes LH wave penetration difficult. Higher magnetic field allows use of lower launched n, therefore reducing the likelihood that LH waves will be absorbed too close to the edge. Test case on C-Mod shows the possibility of realizing high single pass absorption with lower n // Lower-Inside-launch envisioned in the Vulcan reactor study (Podpaly, FED, 2012)

Summary-1 LH current profile control physics needed for ITER has been demonstrated in a number of devices: Core ITB control and sustainment in JT60-U and JET. Current generation at ITER relevant n e, B 0, and f 0 Demonstrations still needed at slightly higher density (n e ~ 1 10 20 m -3 ) and at longer pulse lengths > L/R Development and validation of a simulation capability for LH physics has seen great progress: Widespread use of diagnostics for hard x-ray emission (HXR), MSE, and synthetic diagnostic codes. Combined ray tracing / Fokker Planck solvers with 2D SOL for ray tracing and 3D (r, v, v // ) for Fokker Planck. Development of full-wave solvers and combined with Fokker Planck. Many plausible theories for closing spectral gap exist (toroidicity, scattering, full-wave effects, PDI): But no direct measurements have been made to support a specific theory. Understanding underlying cause is important even if ITER will be in the single pass damping regime. Different mechanisms may be operative in different experiments.

Summary-2 Anomalous decline in LH wave electron interaction at n e > 1 10 20 m -3 still not completely understood: Several plausible theories exist for the anomalous decline in HXR emissivity: Collisional damping, large spectral upshifts due to full-wave effects (diffraction at edge), nonlinear PDI, wave scattering. Experimental evidence does not point to one specific cause for HXR decline in all experiments. Hotter plasma (single pass damping) and hotter edge/sol would probably help in all cases. Experimental evidence, supported by ray tracing / Fokker Planck simulations supports notion that fast electrons used for LH current profile control in ITER should undergo minimal spatial diffusion. Long distance coupling needed for ITER (10-15 cm) has been demonstrated on a number of devices although further work is needed: Linear coupling codes work well once the density profile in the SOL is known. Most recent JET results did not find optimal coupling using gas injectors that were magnetically connected to LH launcher, although gas injection from injectors at the midplane is still quite effective. LH power causes nonlinear modifications to SOL density that are only now starting to be simulated with some confidence.

Summary-3 Un-anticipated benefits of LH power injection have been observed: Observed changes in toroidal ion rotation direction can be significant (up to 50 km/s): Mechanism is not well-established and may be different in each device. Modification of ELM behavior and pedestal characteristics with edge LHCD is quite desirable: Again mechanisms are not well-understood. Integrated modeling frameworks have been developed that should allow optimization of the ITER Heating and CD System: Will also be a useful tool for predicting performance at DEMO relevant parameters. Innovative concepts for overcoming reactor limitations should be pursued more aggressively: HFS launch of LH power may make it possible to burn through the high pedestal temperatures expected in DEMO.

Acknowledgements Thank you to colleagues for discussions and material MIT: John Wright, Syun ichi Shiraiwa, Greg Wallace, Seung Gyou Baek, Amanda Hubbard, Cornwall Lau, Orso Meneghini, Bob Mumgaard, Ron Parker, Miklos Porkolab PPPL Nicola, Bertelli, Randy Wilson Tore Supra: Yves Peysson, Joan Decker, Gerardo Giruzzo, Marc Goniche, Mélanie Preynas FTU: Angelo Tuccillo, Emilia Barbato, Roberto Cesario EAST: Bojiang Ding, Cheng Yang IPP Garching Marco Brambilla

1 ] Count Rate (Chords 9 24, 40 200 kev) [s 10 6 10 5 10 4 C-Mod: Measured HXR found to be consistent with ray tracing / Fokker Planck if 2D SOL is added to GENRAY ray tracing code 10 7 Line Integrated HXR Count Rate n =1.9, 5.4T, 800kA n =2.3, 5.4T, 800kA n =1.9, 7.0T, 800kA n =2.3, 7.0T, 800kA n =1.9, 5.4T, 1.1MA n =2.3, 5.4T, 1.1MA 10 3 0.4 0.6 0.8 1 1.2 1.4 1.6 x 10 20 Line Averaged n e [m 3 ] GENRAY-CQL3D w/o SOL GENRAY-CQL3D with SOL Wallace, PoP, 2010; NF, 2011 Evidence from MSE that J LH is moving outward as n e increases Mumgard, RF Conf., 2013 Rapid decline in HXR at n e > 1.1 10 19 m -3 consistent with collisional damping of LH waves in SOL (f 0 =4.6Ghz, B 0 =5.4T): HXR simulated using GENRAY-CQL3D and a synthetic code Caveat: PDI also observed in these discharges.