alfvén instabilities driven by energetic particles in toroidal...

52
UCLA 1/12/08 Alfvén Instabilities Driven by Energetic Particles in Toroidal Magnetic Fusion Configurations W.W. (Bill) Heidbrink Calculated Alfvén Eigenmode structure in ITER Satellite measurements of Alfvén waves that propagate from the magnetosphere to the ionosphere VanZeeland Sigsbee, Geophys. Res. Lett. 25 (1998) 2077 Magnetosphere Freq Freq Ampl. Ampl. Ionosphere Time Time

Upload: others

Post on 13-Feb-2021

0 views

Category:

Documents


0 download

TRANSCRIPT

  • UCLA 1/12/08

    Alfvén Instabilities Driven by EnergeticParticles in Toroidal Magnetic Fusion

    Configurations

    W.W. (Bill) Heidbrink

    Calculated AlfvénEigenmode structure in ITER

    Satellite measurements of Alfvénwaves that propagate from themagnetosphere to the ionosphere

    VanZeeland

    Sigsbee, Geophys. Res. Lett. 25 (1998) 2077

    Magnetosphere

    Fre

    qFr

    eq

    Am

    pl.

    Am

    pl. Ionosphere

    Time

    Time

  • UCLA 1/12/08

    Instabilities Driven by Energetic Particles areof both scientific and practical interest

    V2

    R0V2

    Damage

    •Carbon coats DIII-D mirrors whenescaping fast ions ablate thegraphite wall1

    •Transport of fast ions by Alfvénwaves onto unconfined orbits causea vacuum leak in TFTR2

    Losses of charged fusionproducts must becontrolled in a reactor!1Duong, Nucl. Fusion 33 (1993) 749.

    2White, Phys. Pl. 2 (1995) 2871.

    Beam injection into the DIII-D tokamak

  • UCLA 1/12/08

    Source Angle Energy Distribution

    Neutral beams Beam angle Slowing down

    Fusion products Isotropic Slowing down

    Ion cyclotron Perpendicular Boltzmann

    Electron cycl. Perpendicular Boltzmann

    Runaway Parallel Flat

    Sources of Energetic Particles

  • UCLA 1/12/08

    Representative Fast-ion Diagnostics

    •Fusion Products Neutrons,gammas, escaping chargedparticles

    •Fast-ion Loss Escaping ions hitdetector

    •Neutrals Escaping neutrals

    •Charge-exchangespectroscopy Photons emittedby ions that gain an electronfrom a neutral beam

    •Equilibrium Spatial profile,velocity-space average

  • UCLA 1/12/08

    •Signal is the integral over

    velocity space of W*F

    •Averages are best for

    general assessment (e.g.,

    “Are fast ions lost?”)

    •Local diagnostics address

    detailed physics mechanism

    (e.g., “Which class of

    particles resonate with the

    wave?”)

    Fast-ion diagnostics are sensitive todifferent parts of velocity space

    Heidbrink, PPCF 49 (2007) 1457.

  • UCLA 1/12/08

    Representative Instability Diagnostics

    •Density

    -- Beam Emission Spectroscopy (BES)

    -- Far Infrared Scattering

    -- CO2 Interferometers

    -- Microwave Reflectometers

    •Electron TemperatureElectron Cyclotron Emission(ECE)

    •Magnetic field probes

  • UCLA 1/12/08

    Mode Identification -- frequency & parametric dependencies -- eigenfunction -- polarizationLinear Stability -- Energetic particle drive (resonant particles, velocity-space anisotropy or spatial gradient) -- Plasma dampingNonlinear Dynamics -- Amplitude, eigenfunction, & frequency evolution -- Fast-ion transportControl -- Alter linear stability -- Exploit instability to improve performance -- Affect nonlinear dynamics

    Key Issues for any Energetic

    Particle Instability

    Bandwidth and sensitivity

    Antenna studies of normal modes

    Spatial resolutionWhat fluctuates?

  • UCLA 1/12/08

    Outline

    1. Alfvén Gap Modes

    2. Energetic Particles(EP)

    3. Energetic ParticleModes (EPM)

    4. NonlinearDynamics

    5. Prospects forControl

    Pinches,Ph.D. Thesis

  • UCLA 1/12/08

    Shear Alfvén Waves are transverse electromagneticwaves that propagate along the magnetic field

    •Dispersionless: = kll vA

    •Alfvén speed: vA = B/(μ0nimi)1/2

    •Ell tiny for

  • UCLA 1/12/08

    Periodic variation of the magnetic field producesperiodic variations in N for shear Alfvén waves

    Zhang, Phys. Plasmas 14 (2007)

    Periodic Mirror Field in the LAPD

    Periodic variation in B Periodic variation in vA Periodic variation inindex of refraction N

    Frequency gap that isproportional to N

  • UCLA 1/12/08

    Periodic index of refraction a frequency gap

    Lord Rayleigh, Phil. Mag. (1887)

    “Perhaps the simplest case … is that of astretched string, periodically loaded,and propagating transverse vibrations.…If, then, the wavelength of a train ofprogressive waves be approximatelyequal to the double interval betweenthe beads, the partial reflexions from thevarious loads will all concur in phase,and the result must be a powerfulaggregate reflexion, even though theeffect of an individual load may beinsignificant.”

  • UCLA 1/12/08

    The propagation gap occurs at the Braggfrequency & its width is proportional to N

    Wikipedia, “Fiber Bragg grating”

    •Destructive interferencebetween counterpropagating waves

    •Bragg frequency: f=v/2

    • f/f ~ N/N

    frequency gap for shear Alfvén waves

    • f = vA/ 2 , where is thedistance between fieldmaxima along the field line

    • f ~ B/B

    n1 n2

    n3

    n3n2

  • UCLA 1/12/08

    Frequency gaps are unavoidable in atoroidal confinement device

    Field lines in a tokamak

    BmaxBmin

    R

    a

    •For single-particleconfinement, field lines rotate.

    •Definition: One poloidal transitoccurs for every q toroidaltransits (q is the “safety factor”)

    •B ~ 1/R

    • B ~ a/R

    •Distance between maxima is = q (2 R) so fgap = vA/4 qR

    Periodicity constraint on the wavevector: ~ ei (n m )

    •n “toroidal mode number”

    •m “poloidal mode number”

    • and toroidal and poloidal angles

  • UCLA 1/12/08

    Frequency Gaps and the Alfvén Continuumdepend on position

    Gap caused by toroidicity1

    1based on Fu & VanDam, Phys. Fl. B1 (1989) 1949

    •Centered at Bragg frequency vA/qR

    •Function of position through vA & q

    •Gap proportional to r/R

    •If no toroidicity, continuumwaves would satisfy = kll vA withkll ~ |n - m/q|

    •Counter-propagating wavescause frequency gap

    •Coupling avoids frequencycrossing (waves mix)

    •Crossings occur at manypositions

  • UCLA 1/12/08

    All periodic variations introduce frequency gaps

    Shear Alfvén wave continuuain an actual stellarator

    Spong, Phys. Plasmas 10 (2003) 3217

    BAE “beta” compression

    TAE “toroidicity” m & m+1

    EAE “ellipticity” m & m+2

    NAE “noncircular” m & m+3

    MAE “mirror” n & n+1

    HAE “helicity” both n’s & m’s

    T

    E

    N

    H

  • UCLA 1/12/08

    Rapid dispersion strongly damps waves inthe continuum

    ~ d (kll vA) / dr

    Radially extended modes inthe continuum gaps are moreeasily excited

  • UCLA 1/12/08

    Radially extended Alfvén eigenmodes aremore easily excited

    Pinches, Ph.D. Thesis

    Continuum Mode Structure

    •Imagine exciting a wave withan antenna--how does thesystem respond?

    •In continuum, get singularmode structure that is highlydamped (small amplitude)

    •Where gap modes exist, theeigenfunction is regular &spatially extended

  • UCLA 1/12/08

    Magnetic shear is the “defect” that createsa potential well for Alfvén gap modes

    •In photonic crystals,defects localize gapmodes

    •The defect createsan extremum in theindex of refraction

    Defect

    GapMode

    Extremumtype

    Couplingtype

    Alfvén continuum

    Magnetic shear(dq/dr) createsextrema

  • UCLA 1/12/08

    An extremum in the continuum can be the“defect”

    VanZeeland, PRL 97 (2006) 135001; Phys.Plasmas 14 (2007) 156102.

    Gap structure in a DIII-D plasmawith a minimum in the q profile

    •Gap modes reside in effectivewaveguide caused by minimum in qprofile

    •These gap modes called“Reversed Shear AlfvénEigenmodes” (RSAE)

    •Many RSAEs with different n’s

    •All near minimum ofmeasured q

    •Structure agreesquantitatively with MHDcalculation

    Measured Te mode structure in DIII-D

  • UCLA 1/12/08

    fmeas

    dampB

    fTAE

    time

    In the toroidal Alfvén Eigenmode (TAE), modecoupling is the “defect” that localizes the mode

    Fasoli, Phys. Plasmas 7 (2000) 1816

    JET #49167

    Using an external antenna to excite an=1 TAE on the JET tokamak •The frequency of the

    measured TAE followsthe frequency gap asthe discharge evolves

    •Can infer the wavedamping from the widthof the resonance

    •Width is larger whenthe eigenfunction“touches” thecontinuum

  • UCLA 1/12/08

    GAE

    dominated by m = 3 (n=1)TAE

    generated by coupling of m = 5,6 (n = 2)

    MHD

    Code

    Soft X-ray

    Tomogr.

    Predicted spatial structure is observedexperimentally for both types of gap mode

    (coupling)(extremum)Data from W7-AS stellarator

    Weller, Phys. Plasmas 8 (2001); PRL 72 (1994)

  • UCLA 1/12/08

    Shear Alfven wave polarization confirmed

    Magnetics data from NSTX spherical tokamak

    Fredrickson

  • UCLA 1/12/08

    Part 2: Energetic Particles

    Pinches, Ph.D. Thesis

    1. Alfvén Gap Modes

    2. Energetic Particles(EP)

    3. Energetic ParticleModes (EPM)

    4. NonlinearDynamics

    5. Prospects forControl

  • UCLA 1/12/08

    Fast-ion orbits have large excursions frommagnetic field lines

    Plan view

    Elevation (80 keVD+ ion in DIII-D) •Perp. velocity

    gyromotion

    •Parallel velocity follows flux surface

    •Curvature & Grad Bdrifts excursionfrom flux surface

    Parallel ~ v

    Drift ~ (vll2 + v 2/2)

    Large excursions for large velocities

  • UCLA 1/12/08

    Complex EP orbits are most simply describedusing constants of motion

    Projection of 80 keV D+

    orbits in the DIII-D tokamakConstants of motion on orbital timescale:energy (W), magnetic moment (μμ),toroidal angular momentum (P )

    Roscoe White, Theory of toroidally confined plasmas

    Distribution function: f(W,μ,P )

  • UCLA 1/12/08

    Orbit topology is well understood

    Zweben, Nucl. Fusion 40 (2000) 91

    Prompt losses of D-T alpha particlesto a scintillator at the bottom of TFTR

    Edge loss detector onthe TFTR tokamak

  • UCLA 1/12/08

    The drift motion must resonate with a waveharmonic to exchange net energy

    Time tocompletepoloidal orbit

    Time tocompletetoroidal orbit

    Tct

    e tomplete

    vllEll 0 (transverse polarization)

    Parallel resonance condition: = n + p

    Write vd as a Fourier expansionin terms of poloidal angle :

    Energy exchange resonance condition: n + (m+l) =0

    (main energyexchange)

    Drift harmonicWave mode #s

  • UCLA 1/12/08

    • Resonance condition, np = n – p – = 0

    n=4, p = 1

    n=6, p = 2

    n=3, p = 1

    n=5, p = 2

    n=6, p = 3n=7, p = 3

    Prompt losses

    E [MeV]4.5 5.0 5.5 6.0 6.5 7.0 7.5

    Calculated resonances with observed TAEs during RF ion heating in JET0

    -50

    -100

    -200

    -150

    -250

    Log

    (f E

    /n

    p)

    -5

    -6

    -7

    -8

    -9

    -10

    Pc

    i [M

    eV

    ]

    A typical distribution function has many resonances

    Pinches, Nucl. Fusion 46 (2006) S904

  • UCLA 1/12/08

    Alphas, TFTR tokamak4

    TAEs

    & R

    SA

    Es

    Beam injection, CHS stellarator1

    Tremendous variety of resonances are observed

    Electron tail, HSX stellarator3

    RF tail ions, C-Mod tokamak2

    EA

    Es, TA

    Es, &

    RSA

    Es

    GA

    EH

    AE &

    TAEs

    1Yamamoto, PRL 91 (2003) 245001; 2Snipes PoP 12(2005) 056102; 3Brower; 4Nazikian, PRL 78 (1997) 2976

  • UCLA 1/12/08

    The spatial gradient of the distributionusually drives instability

    •Slope of distribution function atresonances determines whetherparticles damp or drive instability

    •If the wave damps0v/

  • UCLA 1/12/08

    TAEs in TFTR: avoid energy damping by beamions, use spatial gradient drive by alphas

    •Strong beam-iondamping stabilized AEs duringbeam pulse

    •Theory1 suggested strategy toobserve alpha-driven TAEs

    •Beam damping decreasedfaster than alpha spatial gradientdrive after beam pulse

    •TAEs observed2 whentheoretically predicted

    Wf /

    AEs observed after beaminjection in TFTR D-T plasmas

    2Nazikian, PRL 78 (1997) 29761Fu, Phys. Plasma 3 (1996) 4036;Spong, Nucl. Fusion 35 (1995) 1687

  • UCLA 1/12/08

    EP drive is maximized for large-n modesthat are spatially extended

    Theory1•EP drive increases with n (strongertoroidal asymmetry)

    •But mode size shrinks with n

    •Weak wave-particle interactionwhen orbit is much larger than themode

    Drive maximized when orbitwidth ~ mode size (k EP~ 1)

    Large n anticipated in reactors

    Most unstable mode numbervs. theory (many tokamaks)2

    1Fu, Phys. Fluids B 4 (1992) 3722; 2Heidbrink,Pl. Phys. Cont. Fusion 45 (2003) 983

  • UCLA 1/12/08

    Part 3: Energetic Particle Modes (EPM)

    1. Alfvén Gap Modes

    2. Energetic Particles(EP)

    3. Energetic ParticleModes (EPM)

    4. NonlinearDynamics

    5. Prospects forControl

    Pinches, Ph.D. Thesis

  • UCLA 1/12/08

    EPMs are a type of “beam mode”

    Normal Mode (gap mode)

    nEP

  • UCLA 1/12/08

    EPMs often sweep rapidly in frequency asdistribution function changes

    Simulation withkinetic fast ionsand MHD plasma

    Modes observed during intense negativeneutral beam injection into JT-60U

    Briguglio, Phys. Pl. 14 (2007) 055904

    Shinohara, Nucl. Fusion 41 (2001) 603

    Radius

    Fre

    qu

    en

    cy

    EPM

    TAE

    μμ

  • UCLA 1/12/08

    Part 4: Nonlinear Dynamics

    Pinches, Ph.D. Thesis

    1. Alfvén Gap Modes

    2. Energetic Particles(EP)

    3. Energetic ParticleModes (EPM)

    4. NonlinearDynamics

    5. Prospects forControl

  • UCLA 1/12/08

    •Analogy between “bump-on-tail”and fast-ion modes:

    velocity-space gradient

    configuration-space gradient

    •Resonant ions get trapped inwave: they bounce in wavepotential ( b) & scatter out ofresonance ( eff)

    •Behavior depends on drive, damp,

    b, eff

    •Wide variety of scenarios possible

    1D analogy to electrostatic wave-particle

    trapping describes many phenomena

    Berk, Phys. Pl. 6 (1999) 3102

  • UCLA 1/12/08

    Striking Success of Berk-Breizman Model

    Pinches, Plasma Phys. Cont.Fusion 46 (2004) S47

    Fasoli, PRL 81 (1998) 5564

    Chirping TAEs during beam injectioninto the MAST spherical tokamak

    Simulation of first chirp

    Nonlinear splitting of TAEsdriven by RF tail ions in JET

    Fre

    qu

    en

    cy

    Fre

    qu

    en

    cy

    Time

    Time (ms)

    Small eff

    Appreciable eff

  • UCLA 1/12/08

    Changes in canonical angular momentumcause radial transport

    •Magnetic momentconserved ( μ = 0)

    •Energy changes lessthan angularmomentum: W/W ~0.1( P /P )

    • P (radialtransport)

    •Leftward motion ongraph implies outwardradial motion

  • UCLA 1/12/08

    Four mechanisms of EP transport aredistinguished

    •Leftward motion ongraph implies outwardradial motion

    1) Convective lossboundary (~ Br)

    2) Convective phaselocked (~ Br)

    3) Diffusive transport(~ Br

    2)

    4) Avalanche(Br threshold)

    Convective loss boundary (~ Br, small %)Fluctuations in equilibrium push EPs across lossboundary

    2) Convective phase locked (~ Br, large %)EPs stay in phase with wave as they “walk” outof plasma

    3) Diffusive transport (~Br2)

    Random walk due to multiple resonances

    4) Avalanche (Br threshold)

    EP transport locally increases gradient,destabilizing new modes

  • UCLA 1/12/08

    Convective transport often observed

    García-Muñoz, PRL 99 (2007) submitted

    Edge scintillator on Asdex-U tokamak

    Image on scintillator screenduring TAEs

    Coherent fluctuations in loss signalof RF tail ions at TAE frequencies

    •Fast ions cross loss boundaryand hit the scintillator in phasewith the waves

  • UCLA 1/12/08

    Both convective and diffusive losses areobserved

    Nagaoka (2007)

    0 10 20 30 400

    10

    20

    30

    40

    i (

    arb

    . u

    nits)

    B (arb. units)

    fast responsefBurst

  • UCLA 1/12/08

    Avalanche phenomena observed

    Fredrickson, Nucl. Fusion 46 (2006) S926

    Magnetics data during beam injectioninto the NSTX spherical tokamak •When n=4 & n=6 TAE bursts

    exceed a certain amplitude, alarge burst with many toroidalmode numbers ensues

    •Fast-ion transport is muchlarger at avalanche events

  • UCLA 1/12/08

    Quantitative calculations of EP transport areunsuccessful

    Radial Te profile duringbeam injection into DIII-D

    Radial fast-ion profile

    Heidbrink, PRL 99 (2007) in pressVan Zeeland, PRL 97 (2006) 135001

    •Measured mode structure agrees well with MHD model

    •Input these wave fields into an orbit-following code

    •Calculate much less fast-ion transport than observed

    •What’s missing?

  • UCLA 1/12/08

    Part 5: The Frontier

    Pinches, Ph.D. Thesis

    1. Alfvén Gap Modes

    2. Energetic Particles(EP)

    3. Energetic ParticleModes (EPM)

    4. NonlinearDynamics

    5. Prospects forControl

  • UCLA 1/12/08

    Diagnose nonlinear interactions

    Crocker, PRL 97 (2006) 045002

    •This example shows that theTAEs (100-200 kHz) arenonlinearly modified by a low-frequency (~20 kHz) mode

    •Similar analysis of AE wave-wave interactions and wave-particle interactions areneededBicoherence analysis

    Filtered reflectometer ne signalduring beam injection into NSTX

  • UCLA 1/12/08

    Recent observations indicate kineticinteraction with the thermal plasma

    V2

    R0V2

    Calculated n=40 RSAE that agreeswith ne measurements on DIII-D

    •High-n modes are probably drivenby thermal ions.1

    •Alfvén modes driven by low-energybeams.2

    •New unstable gap modes fromcoupling of acoustic and Alfvénwaves.3

    •Wave damping measurements thatdisagree with fluid plasma models.4

    •New treatments of thermal plasmaare needed

    1Nazikian, PRL 96 (2006) 105006;Kramer, Phys. Pl. 13 (2006) 056104

    2Nazikian, JI1.01; 3Gorelenkov, Phys. Lett. A370/1 (2007) 70; 4Lauber, Phys. Pl. 12 (2005)122501

  • UCLA 1/12/08

    Use control tools to alter stability

    RSAEs

    RSAEs

    ECH deposition location is variedrelative to mode location ( qmin)

    Deposition near qmin stabilizesbeam-driven RSAEs in DIII-D

    Van Zeeland, Plasma Phys. Cont. Fusion 49(2007) submitted

    •Localized electron cyclotronheating (ECH) alters stability andconsequent fast-ion transport

    •Can we turn off deleteriousmodes in a reactor?

  • UCLA 1/12/08

    Use control tools to alter nonlinear dynamics

    Maslovsky, Phys. Pl. 10 (2003) 1549

    Interchange instability driven by energeticelectrons in the Columbia Dipole •In this experiment, a small

    amount of power (50 W)scattered EPs out ofresonance, suppressingfrequency chirping &eliminating large bursts

    •Can we use analogoustechniques to eliminatedamaging bursts of lost alphasin a reactor?

  • UCLA 1/12/08

    Alfvén Eigenmodes can improve performance

    Similar discharges with differing levels of AEactivity during beam injection into DIII-D

    Weak AE

    ModerateAE

    Huge AE

    •Three discharges withdifferent levels of modeactivity•Fast-ion redistributionbroadens current profile•Optimal redistributiontriggers an internaltransport barrier muchbetter confinement•How can we exploit AEsin a reactor?

    Wong, Nucl. Fusion 45 (2005) 30.

    Ti=12 keV

    8 keV

    7 keV

    Huge

    ModerateWeak

  • UCLA 1/12/08

    Bmax

    Bmin

    Conclusions

    •Periodic variations of the index of refractioncause frequency gaps

    •Gap modes exist at extrema of Alfvén continuum

    •Use constants of motion to describe EP orbits

    •Wave-particle resonance occurs when: n + (m+l) =0

    •Instability driven by EP spatial gradient

    •EPMs are beam modes (not normal modes ofbackground plasma)

    •Berk-Breizman analogy to bump-on-tail problemoften describes nonlinear evolution

    •Fast-ion transport not quantitatively understood

    •Use thermal transport techniques to understandnonlinear dynamics

    •Develop tools to control Alfvén instabilities oreven improve performance

    Energy

    P

    8070 7264 6866

    Time [ms]

    14

    0

    12

    0

    10

    0

    qy

    []

    Fre

    qu

    en

    cy

    (kH

    z)

    RSAEs

    RSAEs

  • UCLA 1/12/08

    Acknowledgments* & additional resources

    Thanks to all who sent me slides:

    Brower (UCLA), Crocker (UCLA), Darrow (PPPL),Fasoli (CRPP), Fredrickson (PPPL), García-Muñoz(IPP), Kramer (PPPL), Mauel (Columbia),Nazikian (PPPL), Pinches (UKAEA), Sharapov(UKAEA), Shinohara (JAEA), Snipes (MIT), Spong(ORNL), Toi (NIFS), VanZeeland (GA), Vlad(Frascati), Weller (IPP), White (PPPL), Vann(York), Vincena (UCLA), Zhang (UCI), Zweben(PPPL)

    Special thanks for teaching me theory:

    Liu Chen, Boris Breizman, and Sergei Sharapov

    *Supported by the U.S. Department of Energy

    Clear explanation of basic theory: Firstchapters of Pinches’ Ph.D. thesis,http://www.rzg.mpg.de/~sip/thesis/node1.html

    Experimental review through 1999(especially TFTR results): King-Lap Wong,PPCF 41 (1999) R1.

    Experimental review of fast ions intokamaks (AE material dated): Heidbrink& Sadler, NF 34 (1994) 535.

    Lengthy theoretical review paper: Vlad,Zonca, and Briguglio,http://fusfis.frascati.enea.it/~Vlad/Papers/review_RNC_2.pdf

    Differences between burning plasmas &current experiments: Heidbrink, PoP 9(2002) 2113

    ITER review: Fasoli et al., NF 47 (2007) S264

    Recent theoretical review: Chen & Zonca,NF 47 (2007) S727