9
1 Modeling and Control for Fusion Plasma Stabilization by means of a Mechanical ECRH Launcher at TEXTOR B.A. Hennen, 1,2 P.W.J.M. Nuij, 2 J.W. Oosterbeek, 3 M. Steinbuch, 2 E. Westerhof, 1 B.C.E. Vaessen 1 and the TEXTOR-team 1 FOM-Institute for Plasma Physics Rijnhuizen*, Association EURATOM-FOM, PO Box 1207, 3430 BE Nieuwegein, the Netherlands, www.rijnhuizen.nl ; 2 Eindhoven University of Technology, Control Systems Technology Group, PO Box 513, 5600 MB Eindhoven, the Netherlands, www.tue.nl ; 3 Forschungszentrum Jülich GmbH, Institut für Plasmaphysik*, Association EURATOM-FZJ, 52425 Jülich, Germany, www.fz-juelich.de ; * Partners in the Trilateral Euregio Cluster e-mail address of main author: [email protected] Abstract. This paper discusses several relevant control aspects, regarding the specific task of feedback- controlled stabilization of Neoclassical Tearing Modes (NTMs), for the Electron Cyclotron Resonance Heating (ECRH) installation operated at TEXTOR. The work reported primarily focuses on the mechanics and control of an ECRH launcher system with a 2 rotational degrees of freedom (DOF) steer-able mirror. The dynamics of this instrument are measured and modeled in terms of the equations of motion. Nonlinear system behavior (mainly induced by friction) is characterized. The impact of external disturbances is considered. This system identification procedure allows design and implementation of a cascaded control strategy for improved actuation of the mechanical launcher and also provides information on the physical limitations of the system. The paper additionally presents a proposal, from a control engineering perspective, for development of NTM control scenarios, dedicated to the TEXTOR ECRH installation. 1. Introduction A tokamak is an open system, from which a wealth of control problems arise [1]. Dedicated control solutions differ from standard feedback control concepts, applied similarly in other disciplines, up to tasks constituting completely new control challenges, demanding innovative routes. Enhanced efficiency of a tokamak, requires higher plasma beta, plasma shaping and larger devices. Higher beta operation leads to the manifestation of Neoclassical Tearing Modes (NTMs). These instabilities are generated through reconnection of adjacent flux surfaces of the nested plasma topology, due to a lack of bootstrap current, which is caused by perturbations of the applied toroidal and poloidal fields associated with local distortions of their current density profile. NTMs hamper operation and can be the cause of disruptions. Hence, NTM suppression or controlled stabilization is of utmost importance. Localized injection of electron cyclotron (EC) waves [2]-[4] has demonstrated its feasibility to restore the distribution of the current profile by replacing the missing bootstrap current, thereby stabilizing NTMs. The missing bootstrap current can either be replaced by auxiliary heating in the island, i.e. the resistance across the island is reduced, which increases the current (ECRH), or by driving an additional (helical) current perturbation into the plasma parallel to the plasma current (ECCD), both of which can be achieved using EC waves.

Modeling and Control for Fusion Plasma Stabilization by ......1 Modeling and Control for Fusion Plasma Stabilization by means of a Mechanical ECRH Launcher at TEXTOR B.A. Hennen,1,2

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

  • 1

    Modeling and Control for Fusion Plasma Stabilization by means of a

    Mechanical ECRH Launcher at TEXTOR

    B.A. Hennen,

    1,2 P.W.J.M. Nuij,

    2 J.W. Oosterbeek,

    3 M. Steinbuch,

    2 E. Westerhof,

    1

    B.C.E. Vaessen1 and the TEXTOR-team

    1

    FOM-Institute for Plasma Physics Rijnhuizen*, Association EURATOM-FOM,

    PO Box 1207, 3430 BE Nieuwegein, the Netherlands, www.rijnhuizen.nl;

    2

    Eindhoven University of Technology, Control Systems Technology Group, PO Box 513,

    5600 MB Eindhoven, the Netherlands, www.tue.nl;

    3

    Forschungszentrum Jülich GmbH, Institut für Plasmaphysik*, Association EURATOM-FZJ,

    52425 Jülich, Germany, www.fz-juelich.de;

    * Partners in the Trilateral Euregio Cluster

    e-mail address of main author: [email protected]

    Abstract. This paper discusses several relevant control aspects, regarding the specific task of feedback-controlled stabilization of Neoclassical Tearing Modes (NTMs), for the Electron Cyclotron Resonance Heating

    (ECRH) installation operated at TEXTOR. The work reported primarily focuses on the mechanics and control of

    an ECRH launcher system with a 2 rotational degrees of freedom (DOF) steer-able mirror. The dynamics of this

    instrument are measured and modeled in terms of the equations of motion. Nonlinear system behavior (mainly

    induced by friction) is characterized. The impact of external disturbances is considered. This system

    identification procedure allows design and implementation of a cascaded control strategy for improved actuation

    of the mechanical launcher and also provides information on the physical limitations of the system. The paper

    additionally presents a proposal, from a control engineering perspective, for development of NTM control

    scenarios, dedicated to the TEXTOR ECRH installation.

    1. Introduction

    A tokamak is an open system, from which a wealth of control problems arise [1]. Dedicated

    control solutions differ from standard feedback control concepts, applied similarly in other

    disciplines, up to tasks constituting completely new control challenges, demanding innovative

    routes. Enhanced efficiency of a tokamak, requires higher plasma beta, plasma shaping and

    larger devices. Higher beta operation leads to the manifestation of Neoclassical Tearing

    Modes (NTMs). These instabilities are generated through reconnection of adjacent flux

    surfaces of the nested plasma topology, due to a lack of bootstrap current, which is caused by

    perturbations of the applied toroidal and poloidal fields associated with local distortions of

    their current density profile. NTMs hamper operation and can be the cause of disruptions.

    Hence, NTM suppression or controlled stabilization is of utmost importance.

    Localized injection of electron cyclotron (EC) waves [2]-[4] has demonstrated its feasibility

    to restore the distribution of the current profile by replacing the missing bootstrap current,

    thereby stabilizing NTMs. The missing bootstrap current can either be replaced by auxiliary

    heating in the island, i.e. the resistance across the island is reduced, which increases the

    current (ECRH), or by driving an additional (helical) current perturbation into the plasma

    parallel to the plasma current (ECCD), both of which can be achieved using EC waves.

  • 2

    Mirror suspension

    V

    Mirror

    Servo actuators

    {

    acuum window

    An important problem for the proposed injection of high power microwaves into the plasma is

    control of the alignment between the island location and the current deposition location [5].

    Efficient, flexible and well-localized heating and current drive can only be achieved adopting

    accurate and precise feedback control, which should additionally sustain alignment whenever

    the island stabilizes or disappears. A steer-able launcher system is considered to be able to

    achieve these goals and will therefore be applied in future tokamaks. The work reported here

    primarily focuses on mechanics and control of such an ECRH launcher system.

    This paper is organized as follows. Section 2 will give a brief description of relevant

    components of the ECRH installation [6],[7] operated at TEXTOR, including a mechanical

    launcher system with a 2 rotational degrees of freedom (DOF) steer-able mirror. Section 3

    reports on the dynamical analysis and Frequency Response Function (FRF) characterization

    conducted for this instrument. The equations of motion are derived. Experiments demonstrate

    friction induced non-linear system behavior hence an additional friction term is incorporated.

    Section 4 considers the impact of external disturbances. The system identification procedure

    conducted, allows design of a cascaded control strategy for improved actuation of the

    mechanical launcher, which is the topic of Section 5, and also provides information on the

    physical limitations of the system. Finally, Section 6 initiates a proposal for development of

    NTM control scenarios dedicated to the TEXTOR ECRH installation.

    2. TEXTOR ECRH installation

    FIG. 1. Mechanical ECRH launcher

    The TEXTOR ECRH installation for island suppression [5],[7] consists of a 140 [GHz]

    gyrotron, able to generate an output level of 800 [kW] with a maximum pulse length of 10 [s].

    A quasi-optical line transmits the gyrotron power via several confocal mirrors directly into the

    tokamak, where the beam enters through a CVD diamond window. Inside the tokamak vessel,

    the beam is projected onto the stainless steel mirror of the mechanical ECRH launcher system.

    The launcher is able to steer the mirror in 2 rotational degrees of freedom (DOF), by means of

    two translational servo actuators, which transfer the motion of their driving shafts via a

    suspension of rods and hinges towards the mirror, see Figure 1. The servo actuators consist of

    a rotational AC permanent magnet synchronous motor and a spindle to convert motor axis

    rotation into linear translational motion. The motors encompass three phase sinusoidally

    distributed stator windings and a rotor with permanent magnets.

    For the remainder of this paper the first DOF (in the horizontal plane) will be denoted as

    rotation, while the second DOF (in the vertical plane) is the elevation. The steering range of

    the mirror is limited between -45° and 45° in rotational direction and from -30° to 30° in

    Vacuüm break

    Servo actuators

    Mirror suspension

    Mirror

  • 3

    elevational direction, where the mirror in its center position is oriented at 0° elevation and

    rotation. The elevational DOF corresponds to movement of the microwave beam in poloidal

    direction through the plasma, while the rotational degree corresponds to the toroidal direction.

    Requirements for sufficient speed of response of the launcher are defined in terms of a 10°

    rotation in 100 [ms], with a positioning accuracy of 1°, which is based on a typical island

    growth rate of 10 [ms] or more. Note that the ECRH microwave beam can be steered at equal

    speed in both DOFs, which adds special flexibility and efficiency in the application of heating

    and current drive on every possible location within a plasma cross-section.

    NTMs feature a flat electron temperature profile. Detection of islands is therefore

    accomplished through measurement of the perturbation the islands cause on the electron

    temperature profile. If an island is located between two ECE channels, the fluctuations on

    these two channels will be 180° out of phase, which can be exploited to determine the exact

    location and width of the NTM [8]. Cross-correlation among different ECE channels can be

    used to increase resolution and limit the number of channels required. Furthermore, Mirnov

    coils, soft X-ray and equilibrium codes are employed for island detection. An accurate

    positioning and alignment feedback control loop can either make use of the gyrotron power

    deposition- and island location or on the island width as a control variable to steer the

    launcher system in real-time.

    An alternative ECE diagnostic, envisaged for installation on TEXTOR, is accomplished

    through an additional optical system, or line of sight scheme [7], which is integrated halfway

    in the transmission line to monitor island perturbations on the electron temperature profile.

    Within the system, the reflected ECE emission, 'gathered' from the plasma, is separated from

    the transmitted gyrotron power by means of a frequency selective dielectric plate, which has

    periodic minima and maxima in transmission and reflection following the selected ECE

    observation frequencies. An optical construction with mirrors, an additional dielectric plate

    and a 6 channel radiometer, with channels spaced in a range of frequencies separated 3

    [GHz] and starting at a frequency of 132.5 [GHz], measures the EC temperature profile.

    3. Dynamical characterization

    The dynamics of the ECRH launcher are analyzed experimentally using frequency response

    measurement techniques. Open loop measurements of in- and output signals are used to

    estimate the Frequency Response Function (FRF) [9]. For a linear system with excitation (or

    input) )(tx and response (or output) )(ty one can write in the time domain: ).()()( txthty ⊗=

    In the frequency domain this convolution integral (with impulse response )(th ) reads

    ),(ˆ)()( fHfSfS xxxy ⋅= where the auto and cross power spectral density )( fS xx and )( fS xy

    are estimated using fast Fourier transforms, which yields )(ˆ fH as an estimate for the FRF.

    As a measure for the linearity of this approximation, the coherence function is defined by

    ,)()(

    |)(|)(

    2

    2

    fSfS

    fSf

    yyxx

    xy

    xy⋅

    =γ with .1)(0 2

  • 4

    100

    101

    102

    −160

    −140

    −120

    −100

    −80

    −60

    −40

    Am

    plit

    ude

    [dB

    ]

    Rotational DOF

    100

    101

    102

    −200

    −100

    0

    100

    200

    Pha

    se [

    deg.

    ]

    100

    101

    102

    0

    0.2

    0.4

    0.6

    0.8

    1

    Coh

    eren

    ce [

    −]

    Frequency [Hz]

    0.1 [V]

    0.2 [V]

    0.4 [V]

    0.6 [V]

    0.8 [V]

    1.0 [V]

    1.2 [V]

    100

    101

    102

    −160

    −140

    −120

    −100

    −80

    −60

    −40

    Am

    plit

    ude

    [dB

    ]

    Elevational DOF

    100

    101

    102

    −200

    −100

    0

    100

    200

    Pha

    se [

    deg.

    ]

    100

    101

    102

    0

    0.2

    0.4

    0.6

    0.8

    1

    Coh

    eren

    ce [

    −]

    Frequency [Hz]

    0.1 [V]

    0.2 [V]

    0.3 [V]

    0.4 [V]

    0.5 [V]

    0.6 [V]

    0.7 [V]

    FIG. 2. Frequency Response Function measurements for both degrees of freedom

    A proper FRF measurement [9] usually adopts uniform random noise as input signal )(tx in

    order to guarantee excitation of the system in a broad range of frequencies. Figure 2 shows the

    FRF estimates obtained for both DOFs of the ECRH launcher, where noise with different

    amplitude levels in a frequency band of 0 – 200 [Hz] is exploited to excite the system. This

    input signal is provided directly as a torque to both servo actuators. The output signal is a

    linear acceleration measurement from piezoelectric accelerometers mounted outside the

    vacuum vessel on both driving shafts of the mirror suspension. The obtained FRFs are

    integrated twice in the Laplace domain to obtain the Bode plots in Figure 2.

    The Bode diagrams reflect dynamic behavior of the ECRH launcher, which is dominated by

    the actuators. Both axes of motion display comparable dynamics. Note that resonances are

    encountered near 22 [Hz] and 60 [Hz], which magnitudes are excitation amplitude dependent.

    They originate from the mirror suspension. One can furthermore observe phase transitions

    occurring near these frequencies, which also indicates system dynamics. Note that, starting at

    -90°, the phase decreases smoothly over all measured frequencies, which suggests the

    presence of a time-delay.

    Whenever the root mean square (RMS) value of the excitation signal is increased, a transition

    towards higher frequency response amplitude levels occurs as observed in the plots (following

    the arrow directions). This transition demonstrates a nonlinear dependence with respect to the

    linearly increasing RMS values of the noise excitation and appears to be dominant in the

    frequency envelope 0-20 [Hz] and for low excitation levels. This justifies the hypothesis that

    this behavior might be caused by static friction, i.e. for low amplitudes of excitation, the

    system operates in the stiction zone, whereas, after a transition zone with stick-slip behavior,

    the system operates in full slip if a certain friction force threshold is exceeded.

    Based on kinematical and dynamical modeling techniques the equations of motion for the

    mirror suspension system of the ECRH launcher can be derived [9]:

    ,)cos()cos()sin()sin(

    )cos()cos()sin()sin()sin()cos(

    33

    2

    3

    2

    2

    211

    fricelevelevelevelev FxkxbaFMJJ

    JJJJ

    −−−⋅==++

    −−−

    &&&&

    &&&&&&&

    θθϕθθϕϕ

    θθϕθθϕϕθθϕθϕ (2)

  • 5

    100

    101

    102

    103

    −90

    −80

    −70

    −60

    −50

    −40

    −30

    −20

    −10

    0

    Frequency [Hz]

    Po

    wer

    Sp

    ectr

    al D

    ensi

    ty [

    dB

    /Hz]

    # 103079

    Measurement outside plasma present phase

    Measurement within plasma present phase

    −15 −10 −5 0 5 10 15 20 25−0.5

    0

    0.5

    1

    Time [s]

    Acc

    eler

    om

    eter

    Sig

    nal

    [V

    ]

    # 103079

    −15 −10 −5 0 5 10 15 20 25−200

    0

    200

    400

    Pla

    sma

    curr

    ent

    [kA

    ]

    Toroidal magnetic field x 2 [Tesla]

    # 103079

    Fre

    qu

    ency

    [H

    z]

    −10 −5 0 5 10 150

    500

    1000

    1500

    2000

    2500

    3000

    −120

    −100

    −80

    −60

    −40

    −20

    0

    20

    −10 −5 0 5 10 15−0.5

    0

    0.5

    Am

    pli

    tud

    e [V

    ]

    Time [s]

    [dB]

    ,)sin()cos()cos(

    )sin()cos()cos()cos()sin()sin()cos()sin()sin(

    22,11,

    2

    2

    2

    13

    2

    333

    fricrotrotrotrotrot FxkxbaFaFMJ

    JJJJJ

    −−−⋅−⋅==+

    −+++

    &&

    &&&&&&&&

    θθϕϕ

    θθϕϕθϕθϕθϕϕθϕϕθ

    (3)

    where 3...1J is the mass moment of inertia, θ respectively ϕ the rotation and elevation DOF.

    elevrotM , is the angular moment forced upon the mirror by the mirror suspension, consisting of

    a force elevrotF , , delivered by the servo actuators in both DOFs and a moment arm 3...1a defined

    with respect to the center of mass of the mirror. elevrotk , and elevrotb , respectively represent the

    stiffness and damping coefficient of the mirror suspension. x is the translational motion of

    the servo actuator driving shafts. The equations of motion additionally incorporate a nonlinear

    friction component fricF , which is modeled in terms of the Leuven friction model [10]. Note

    that the friction characteristics encountered are of typical concern for tracking of NTMs, since

    tracking requires only small launcher movements, which justifies the need for a model.

    4. Disturbance characterization

    The launcher system is operated in an environment where external disturbances of various

    nature might affect dynamic performance and can harm smooth system operation in a

    feedback loop. Magnetic fields, diagnostics, electrical equipment and the plasma itself might

    all, to some extent, superimpose noise and other frequency components onto sensor signals.

    External forces or forces caused by induced currents interacting with the mechanical part of

    the launcher can likewise affect servo system operation. Disturbance characterization

    experiments have therefore been conducted during operation of the TEXTOR tokamak, to

    give an impression as to what extent sensor noise originating from such disturbances can

    influence control system behavior.

    FIG. 3. Time-domain plot, auto power spectra and time-frequency spectrogram of an acceleration

    measurement for the rotational DOF of the ECRH launcher during TEXTOR shot # 103097.

  • 6

    Figure 3 shows the analysis of a signal measured during a TEXTOR discharge from a linear

    accelerometer located on the actuator driving shaft of the rotational axis of motion. The

    launcher is not purposely excited using an input signal. Hence, all contents of the sensor

    signal can be attributed to disturbances. The plasma current pI and toroidal magnetic field

    TB are depicted as well. At 0 seconds, the plasma is initiated. The accelerometer signal

    clearly responds to the ramp up of the magnetic field and the presence of the plasma. A peak

    is observed in the accelerometer signal when the plasma disappears. The toroidal magnetic

    field is ramped down and the signal amplitude returns to its original level.

    Power spectral densities (PSDs) and a time-frequency mapping are depicted to obtain insights

    in the frequency components of the signal and their propagation in time. PSDs are derived for

    a time frame before plasma and during plasma operation. Note that all plots show a clear

    periodic electric grid component at 50 [Hz]. Whereas the signal’s frequency content remains

    at a low energy level before plasma, ramp up of the magnetic field lifts the signal to a higher

    energy level and adds additional frequency components, including a periodic component near

    600 [Hz]. After breakdown of the plasma, even more energy is added, including a strong

    periodic at 1300 [Hz], which probably originates from the vertical field control system. Note

    that the energy content of the disturbing phenomena is most likely not high enough to cause

    actual mirror motion or induce currents in the mechanical parts of the launcher.

    The observed disturbances might, however, affect control system behavior and therefore it is

    preferential to invest in the design of proper filtering, disturbance rejection and data

    acquisition techniques. One could, for example, employ low-pass filtering to reduce the

    operational frequency band of the control signals. In this context it is important to realize that

    the launcher is a motion system, which typically demonstrates a rather low bandwidth or

    speed of response as formulated in the requirement of a 10° rotation in 100 [ms] with a

    positioning accuracy of 1°. Filtering of its control signals with a rather low cut off frequency

    of, for example, 200 [Hz] will therefore not affect system functionality and performance.

    5. Controller design

    The servo actuators of the ECRH launcher are operated in combination with servo amplifiers

    to provide the motor with appropriate input trajectories and control the servo system in closed

    loop. A cascaded control structure is implemented for accurate and precise control of the

    system. The cascaded control structure, as depicted in Figure 4, consists of three distinct

    feedback control loops, i.e. position-, velocity- and current (torque) control. The bandwidth or

    speed of response increases towards the innermost loop.

    The system measurements and modeling performed in previous sections allow derivation of

    transfer function models, which in turn can be applied to derive dedicated controllers for the

    cascaded control structure. The closed current (torque) control loop guarantees a constant

    motor torque. Since the current control loop is considered as part of the servo actuator

    dynamics )(sH in the FRF measurements, controller design solely involves the speed and

    position control loops. Controllers are obtained using a frequency domain tuning procedure,

    which exploits estimated transfer function models retrieved from a frequency response fit

    routine.

  • 7

    FIG. 4. Cascaded control loop

    The stabilizing controller for the speed control loop consists of a standard proportional

    integral (PI) controller, a lead/lag compensator and a low-pass filter as defined by

    ,

    13/2

    11

    32

    1

    13/2

    1

    2)( 21

    +⋅⋅

    +⋅⋅⋅

    +⋅⋅

    ⋅⋅+

    ⋅=

    sf

    k

    sf

    sf

    s

    fsksCspeed

    ππ

    ππ (4)

    where 1k and 2k are gains and f is the desired cross-over frequency, which determines the

    bandwidth of the control loop. Applying this controller, a closed loop transfer function model

    can be derived for the speed control loop, which is adopted subsequently for design of the

    position controller, which results in a classical PI controller, of the form

    .2

    )( 3s

    fsksCposition

    ⋅+⋅=

    π (5)

    Performance of the feedback control system is demonstrated in the open loop transfer function

    ( ))()( sHsC , sensitivity ( )( ))()(1/1 sHsC+ and Nyquist diagram depicted in Figure 5. The open loop plot reveals that the 0 [dB] point is crossed at 8 [Hz], i.e. the cross-over frequency

    or bandwidth of the controlled system is 8 [Hz]. The sensitivity and Nyquist graphs show that

    enough amplitude and phase margin is taken into account to guarantee system stability. The

    fourth subplot of Figure 5 shows control system response on a reference trajectory for pure

    cascade feedback as well as with additional torque and speed feed-forward, which reduces the

    response time and servo error considerably. As observed, the controlled system is able to

    rotate the launcher mirror by approximately 25° in 100 [ms] with a maximum positioning

    error of 0.4°, which corresponds to a servo actuator translation of 0.30 [m] in 100 [ms], with

    a maximum positioning error of 5·10¯³ [m]. Hence, performance is well within the earlier

    specified requirements. Similar control structures and controllers can be applied for both

    DOFs simultaneously.

    6. NTM control scenarios

    Development of NTM control scenarios, dedicated to the TEXTOR ECRH installation [5],[7],

    will require extensive simulation incorporating mechanical as well as relevant physics

    models. Since controller design usually requires only approximating dynamic system

    modeling, most physics models will require simplification while sustaining agreement with

    theoretical and experimental verifications. Controllers designed and validated throughout

    these procedures, should, for example, be capable of magnetic island stabilization on different

    flux surfaces and localization and tracking of fluctuating or perturbed islands.

  • 8

    100

    101

    102

    −80

    −60

    −40

    −20

    0

    20

    40

    Am

    pli

    tude

    [dB

    ]

    Open loop

    100

    101

    102

    −200

    −150

    −100

    −50

    0

    50

    100

    150

    200

    Frequency [Hz]

    Phas

    e [d

    eg.]

    FRF Measurement

    Transfer function estimate "frfit"

    100

    101

    102

    −35

    −30

    −25

    −20

    −15

    −10

    −5

    0

    5

    10

    Am

    pli

    tud

    e [d

    B]

    Sensitivity

    Frequency [Hz]

    FRF Measurement

    Transfer function estimate "frfit"

    −2 −1.5 −1 −0.5 0 0.5 1−2

    −1.5

    −1

    −0.5

    0

    0.5

    1

    Real part

    Imag

    inar

    y p

    art

    Nyquist diagram

    FRF Measurement

    Transfer function estimate "frfit"

    0 0.2 0.4 0.6 0.8 1 1.2 1.4−20

    −15

    −10

    −5

    0

    5

    10

    15

    20

    Time [s]

    Rota

    tion a

    ngle

    lau

    nch

    er m

    irro

    r [d

    eg.]

    reference trajectory

    feedback response

    feedforward response

    feedforward servo error

    FIG. 5. Open loop response, sensitivity, Nyquist diagram and time domain response on a reference

    trajectory for the cascaded control design

    To further enhance the potential of NTM control, one could also think in terms of alternative

    control schemes and derivation of more complex and robust controllers, which again requires

    extensive validation and testing in a simulation environment before application in an actual

    system. Repetitive control techniques might, for example, be applicable to the NTM

    suppression problem, since in a fixed coordinate frame the local distortions caused by the

    islands typically constitute periodic phenomena. In mechanical systems, repetitive control

    combines actual feedback control with the derivation of improved feed-forward signals to

    enhance control performance in subsequent trajectory repetitions. In the sense of NTM

    stabilization, repetitive control could potentially offer enhanced suppression due to control of

    the gyrotron power in phase with periodic oscillations of the island width.

    7. Conclusions

    As discussed, many elements of a dedicated NTM suppression and plasma stabilization

    method will possess a high level of complexity. But, as observed, specific control expertise,

    e.g. on modeling of process dynamics and system identification, is readily available in

    different engineering disciplines and well applicable for this task. Observations verify that

    mechanical system features induced by resonance events, friction and disturbances, limit

    performance and functionality of the mechanical ECRH launcher concept operated at

    TEXTOR. However, adopting modeling, experiments and simulation, dedicated controllers

    can be designed, which improve the mechanical properties of the launcher and will guarantee

    proper functioning of the system in real-time feedback controlled suppression and

    stabilization of magnetic islands.

  • 9

    Acknowledgements

    This work, supported by the European Communities under the Contract of Association

    between EURATOM/FOM, was carried out within the framework of the European Fusion

    Programme. The views and opinions expressed herein do not necessarily reflect those of the

    European Commission.

    References

    [1] PIRONTI, A., WALKER, M., et al., “Special sections on Control of tokamak plasmas”,

    IEEE Control Systems Magazine 25(5) (2005) 24-92 and 26(2) (2006) 30-91.

    [2] ISAYAMA, A., et al., “Complete stabilization of a tearing mode in steady-state high- pβ

    H-mode discharges by the first harmonic electron cyclotron heating/current drive on

    JT60-U”, Plasma Physics and Controlled Fusion 42 (2000) L37-L45.

    [3] GANTENBEIN, G., et al., “Complete Suppression of Neoclassical Tearing Modes with

    Current Drive at the Electron-Cyclotron-Resonance Frequency in ASDEX Upgrade

    Tokamak”, Physical Review Letters 85(6) (2000).

    [4] PETTY, C., et al., “Complete suppression of the m=2/n=1 neoclassical tearing mode

    using electron cyclotron current drive in DIII-D”, Nuclear Fusion 44 (2004) 243-251.

    [5] HUMPHREYS, D.A., et al., “Active control for stabilization of neoclassical tearing

    modes”, Physics of Plasmas 13 (2006) 056113.

    [6] WESTERHOF, E., et al., “Electron Cyclotron Resonance Heating on TEXTOR”,

    Fusion Science and Technology 47 (2005) 108-118.

    [7] OOSTERBEEK, J.W., et al., “Design of a feedback system to stabilise instabilities by

    ECRH using a combined ECW launcher and ECE receiver”, 24th SOFT (2006).

    [8] BERRINO, J., et al., “Automatic Real-Time Tracking and Stabilization of Magnetic

    Islands in a Tokamak Using Temperature Fluctuations and ECW Power”,

    IEEE Transactions on Nuclear Science 53(3) (2006) 1009-1014.

    [9] DE KRAKER, A., and VAN CAMPEN, D.H., Mechanical vibrations, Shaker

    Publishing, Maastricht (2001).

    [10] CANUDAS DE WIT, C., et al., “A New Model for Control of Systems with Friction”,

    IEEE Transactions on Automatic Control 40(3) (1995) 419-425.