340 likes | 488 Views
Tearing modes control in RFX-mod: status and perspectives. P.Zanca , R.Cavazzana, L.Piron, A.Soppelsa Consorzio RFX, Associazione Euratom-ENEA sulla Fusione, Padova, Italy. Milestones. Intelligent Shell: egde radial field control (2005)
E N D
Tearing modes control in RFX-mod: status and perspectives P.Zanca, R.Cavazzana, L.Piron, A.Soppelsa Consorzio RFX, Associazione Euratom-ENEA sulla Fusione, Padova, Italy
Milestones • Intelligent Shell: egde radial field control (2005) • Clean Mode Control (de-aliasing of the measurement): TM wall-unlocking (2007) • Coils amplifiers improvements: maximum current and rensponse time (2008, 2010) • MHD model of the feedback (RFXlocking) (2007-2010)
Optimizations • Edge radial field reduction: get closer to the ideal-shell limit (determined by vessel/copper shell) • Increase the QSH duration: non-zero reference for the dominant mode • m=2, n=1 control in tokamak dicharges
Optimizations • Edge radial field reduction: get closer to the ideal-shell limit (determined by vessel/copper shell) • Increase the QSH duration: non-zero reference for the dominant mode • Control in tokamak dicharges
Latency reduction m=1,n=8-16 br(a) Empirical gains
Latency reduction m=1, n=8-16 br(a) Empirical gains
Latency reduction m=1, n=8-16 br(a) Empirical gains
Derivative control • db/dtis currentlyobtained numerically from b • Acquisition of the derivative signal db/dt is preferable • This allows a better PD gains optimization
m=1, n=-7 m=0, n=7 m=1, n=7 m=2, n=7 Brm,n (T) Icoilm,n (A) decoupler ON Dynamic decoupler • The dynamic decoupler reduces the side-harmonic components of the magnetic field produced • A “modal” decoupler could be designed considering a limited number of harmonics (i.e. only the poloidal sidebands)
bφ systematic errors correction • Unavoidable misalignment of the pick-up coils determines a spurious Ip contribution to bφ • Real-time subtraction of this term Similar for m=±1,2
M=0 control • Little affected by the feedback • High gains test on m=0, n<6 has not shown any improvement on F shallow discharges • m=0 control at deeper F still to be investigated • m=0 n≥7 spurious contribution should be removed by the dynamic decoupler
M=0, n<6 feedback with the toroidal circuit • Enhance the natural reaction of the 12 toroidal sectors to the m=0 low n TM • Present circuit too slow to follow the m=0 dynamic (2.5ms delay according to 2006 experiments) • Upgrade of the internal circuit control by reducing the latency
Independent feedback on br and bφ Iref = Kr br + Kφbφ • Suggested by J.Finn and co-workers • A more general control could allow finding a new optimum • Preliminary RFXlocking simulations are planned
Feedback on the plasma response bplasma = br – bcoils(vacuum) • bcoils from the cylindrical model used in the de-aliasing or from a state-space model which includes the shell frequency response • The hope is to reduce the TM amplitude at the resonant surface • According to RFXlocking edge br is comparable to the standard feedback case upon PD gains optimization
Synopsis • Control system upgrade • Latency reduction • db/dt acquisition • Improved toroidal circuit control • New algorithms • Dynamic decoupler • bφ sistematic errors removal • M=0 low n control with the toroidal circuit (partially developed) • Plasma response • Independent br bφ feedback • Other schemes • M=0 control at deep F • Non-zero reference control to sustain QSH } Gains optimization
RFXlocking • Semi-analitical approach in cylindrical geometry • Newcomb’s equation for global TMs profiles • Resonant surface amplitudes imposed from experiments estimates • Viscous and electromagnetic torques for phase evolution • Radial field diffusion across the shell(s) • Feedback equations for the coils current
Feedback limit Sensors Vessel Coils plasma
Feedback limit Sensors Vessel Coils plasma
Feedback limit Sensors Vessel Coils plasma br=0 everywhere: impossible
Single-shell: discrete feedback Δt = latency of the system
Toroidal circuit dynamic response Control system + internal control latency: 2.5 ms Power supply time constant: 3 ms
Toroidal circuit dynamic response Simulation of power supply behaviour with latency = 1.6 ms - Kp = 0.04 Simulation of power supply behaviour with latency = 0.1 ms - Kp = 0.7