1 / 74

Real Time Control of advanced scenarios for steady-state tokamak operation

Real Time Control of advanced scenarios for steady-state tokamak operation. Xavier LITAUDON CEA-IRFM, Institute for magnetic fusion research (IRFM) CEA Cadarache F-13108 St Paul Lez Durance. France e-mail: xavier.litaudon@cea.fr.

elke
Télécharger la présentation

Real Time Control of advanced scenarios for steady-state tokamak operation

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Real Time Control of advanced scenarios for steady-state tokamak operation Xavier LITAUDON CEA-IRFM, Institute for magnetic fusion research (IRFM) CEA Cadarache F-13108 St Paul Lez Durance. France e-mail: xavier.litaudon@cea.fr Courtesy: J.F Artaud, A. Bécoulet, S. Brémond, D. Campbell, J. Ferron, G. Giruzzi, C. Gormezano, E. Joffrin, S. H Kim, D. Mazon, D. Moreau, P. Politzer,T. Suzuki, T. Tala

  2. OUTLINE TOWARDS REAL TIME PROFILE CONTROL ? • Challenges for continuous operation • continuous tokamak reactor operation • real time control requirement • Real time control of kinetic & magnetic energy • optimal profile for steady-state & MHD stable profiles • approaches to profiles control • Real time fusion D-T burn control • burn control with dominant bootstrap and a-heating ? • Control of core performance with the plasma facing components constrains • wall scenario compatibility issues • simultaneous control of core & edge

  3. Optimisation of tokamak concept BT IP IC BP IP = Iinductive+ INon-Inductive • Long-Pulse Operation → IP = INon-Inductive • Non-Inductive Current Drive • Externally driven, e.g. waves injection • To drive 15MA on ITERrequires 150MW • 150MW coupled power requires ~ 1GW fusion • Internally driven  Pression: bootstrap effect • Efficient reactor at high Q =Pfus/Padd relies on the optimisation of bootstrap current [e.g. Kikuchi M Nucl. Fusion 1990, Gormezano C ITER physics basis Nuc Fus 2007]

  4. Challenges of continuous tokamak operation • Fully non-inductive regime • High confinement & bootstrap current • Real time control of kinetic & magnetic configuration close to operational limits with a large fraction self a-heating & bootstrap • Technology of Long Pulse Operation • Coils, Plasma Facing Components, Structure Materials, Heating &Current Drive systems, Diagnostics, data acquisition, fuel cycle… Worldwide research activity: physics, modelling, technology

  5. Towards a continuous tokamak reactor DEMO ITER Tore Supra DD ~400s 0% 20% Q ~ 30 Continuous 80 to 90% 60-80% Q ~ 10 400-3600s 70% 10-50% A scientific and technical challenge JT60-SA JET Pfusion/Padd Q ~ 1 2s 10% 20% DD ~100s 0% >60% duration self-heating bootstrap Existence and control of a self-organised plasma state for continuous tokamak operation ?

  6. Bpol Btor • Toroidal • bt = P/pBtor • pBtor=B2tor/2m0 • Poloidal • bp= P/pBpol • pBpol=B2pol/2m0 Basic standard tokamak parameters • Safety factor q • number of toroidal turns for one poloidal turn • Confinement • H=t/tscaling H~1 • tscaling = I R2 P-2/3 • Stability • q >1 • bN =bt/(Ip/aB) ≤ 3

  7. stability limit low q95 limit bt [%] SS q95~4-5 Fusion gain optimisation Inductive q95~3 bp STEADY-STATE REACTOR : Optimisation of QDT & Bootstrap current QDT  btt B2 • fusion power+ bootstrap •  high bN , t , B • since bt bp  (1+k2) b2N • Optimise shaping • Stability • -q & pressure • -wall stabilisation • Confinement Bootstrap optimisation Iboot/Ipe½ bp=e-½q bN

  8. H-mode + internal transport barrier Plasma pressure L-mode H-mode Pedestal radius 0 1 Enhanced performance for non-inductive regimes

  9. Towards Long Pulse Operation on ITER Inon-inductive/Ip Ibootstrap/Ip • Inductive operation • Q  10 Ip~15MA 400s • ‘Intermediate’ • Q ~5-10 Ip~12MA 1000s • fully non-inductive • Q~5 Ip~9MA 3000s • Active research activity • Integration of physics & technology 20% 7% 50% 20% 100% 50%

  10. 5 2 1 0 1 ITER STEADY-STATE OPERATION Steady-State operation at Q ~ 5 (Pa~Padd) with full non-inductive current drive + optimized current & pressure profiles q95~5 (9MA) at high k, d • Iboot/Ip50% • bN~3, H98(y,2) ~1.5 • nl~7x1019m-3 • Ti/Te ~ 1 • tD~3000s qo-qmin~0.5 qmin ~ 1.5-2.5 [Gormezano Nuc Fus 2007, Campell Pop (2001), Green et al PPCF 2003 & ITPA steady-state group]

  11. nHe/ne=6% t*He/ tE =5 Gyro-Bohm scaling bN=3.5 bN=2 CONTROL OF CORE CONFINEMENT IN STEADY-STATE ITER OPERATION ITER-AT: Ptot=73MW, Vloop=0 CRONOS-0D • Control of high confinement • for steady state: • DH/H~20% DQ/Q~50% (at Q~5) • favourable effect of density peaking while nHe/ne< 6% b0 ITER SS target b-0.9 [Litaudon PPCF 2006]

  12. The decisive role of real time control • Bringing Fusion to its “Reactor Era” requires an innovative programme of“discharge mastering”, combining: • real time control of the magnetic/kinetic configuration (non-linear and time effects) • real time control of component integrity • high-level algorithms and control schemes • a consistent set of simulation tools: • first principles (“PFlops”) • integrated modelling (“CPU hours”) • fast simulators (“~ 10 ms”) [A. Becoulet & G.T. Hoang PPCF 2008 and Joffrin et al PPCF 2003 ]

  13. Pol. Field Magnetization Burn termination Current Ramp Down Current Ramp Up Current Flat Top Breakdown Heating Control requirement: plasma scenario PF Reset And Recool Fusion Power Time(s) Plasma Current Inductive Flux D-T Fuelling Plasma Density a-particle Fraction Additional Power

  14. Example of scenario: JET plasma JET #67687

  15. Real Time plasma profile reconstruction(EQUINOX code) #57336 [Joffrin et al PPCF 2003, Joffrin et al PPCF 2007 ]

  16. Plasma scenario : 15MA H-mode ITER scenario Tokamak simulation: free-boundary equilibrium (DINA-CH) & transport evolution (CRONOS) ITER S.H. Kim et al, PPCF 2009

  17. [R. Buttery, EFPW 05] MHD Stability issues & real time control • Edge Localized Modes • Damage to Plasma Facing Components • Neo-classical tearing modes • Limiting pressure, risk of disruption • Resistive wall modes • Limiting pressure • Disruptions • Device safety • Fast particle modes • Limiting -heating, CD Necessity for Real Time feedback control & localized CD

  18. PROFILE CONTROL REQUIREMENTS FOR STEADY-STATE OPERATION ITER • ITER SS operation above the no-wall limit • at ITER wall position, the marginal b is sensitive to details in q and pressure profiles • qmin~2.4,bN <3.85 • qmin~2.1,bN<2.6 P q r [Shimada et al NF 2004, Polevoi et al IAEA 2002]

  19. ITER Heating & Current Drive actuators: FLEXIBILITY for profile Control 170 GHz 5 GHz 40-55 MHz 1MeV/D- ITER 33MW/CW 20MW/CW 20MW/CW 20MWCW

  20. Control of a self-organised state ? • two time scale: fast (blue) & Slow (red) • a-heating dominant in burning reactor Actuators Plasma [Politzer et al NF 05]

  21. Magnetic/Kinetic configuration RT control

  22. PROFILE CONTROL REQUIREMENTS • Oscillation in confinement observed in steady-state tokamak plasmas  limiting the fusion perfomance • Tore Supra* & DIII-D** (i) non-linear coupling between j & TjLH(j,T),jboot (j,T),c(j,T) (ii) non-linear interplay of heating, CD & MHDs<0, double tearing, ideal MHD limits … • ITER SS  extra coupling via a-heating (i) non-linear coupling between j & Tjboot(j,T), c(j,T), Pa(T) (ii) non-linear interplay of heating, momentum, CD & MHD Pa(T),b limits, TAE (a-particles), … *Giruzzi et al PRL 03 **Politzer et al NF 05

  23. LH Power (MW) Transformer flux (Wb) Te(R,t) Te(0) = 4.8 keV density (x1019m-2) Ti(0) =1.6 keV TORE SUPRA Neutron (x1010/s) Zeff ~2 Non-linear behaviour in non-inductive regime • oscillations of core electron temperature • non-linear interplay between q-profile, transport and heat sources (and MHD) Vloop=0 for 6 min, 1 GJ 2003 [Giruzzi PRL 03; Imbeaux PRL 06; Maget Nuc Fus 06]

  24. Oscillation in bootstrap-dominated regime DIII-D Vl=0 (open transformer circuit) q95~10 0.7 Ip (MA) #119767 #119770 0.6 0.5 5 1 4 2 3 Time (s) • DW/W~50% • Iboot/Ip~85% • bN~bp~3.3 r r [Politzer et al NF 2005]

  25. Non-optimal : narrow profiles and steep gradients Optimal ITB : broad profiles with moderate gradients MHD stability broad Jboot and Jtot broad ne for reduced impurity accumulation control of ITB radius, strength & width ACCESS TO HIGH bNOPTIMAL PROFILES ? ITB radius p=cst ITB width

  26. Hybrid Reversed shear QDB 4 3 bN 2 1 2 4 6 p0/<p> High bN requires broad pressure profiles ITPA database • Stability limit improves with ITB radius and width* control of confinement & q-profile ? *Lao et al APS 99 [Sips IAEA 2004 , Litaudon et al PPCF 2004]

  27. REAL TIME CONTROL OF KINETIC ENERGY JET • Operation close to the no-wall stability limit while avoiding disruptions • Real time control of neutral beam heating to match a neutron yield production • Similar results obtained on DIII-D, JT-60U t= [G. Huysmans et al., Nucl. Fusion, 1999, C. Gormezano et al FST 2008 ]

  28. Control of electron temperature gradient • PLHCD to slow down q(r,t) • PNBI RT controlled by neutron • PICRHRT controlled by rs/LTe where LT= T/T • proportional-integral [Mazon, Litaudon, Moreau et al PPCF 02]

  29. TORE SUPRA RT control of magnetic energy • feedback control for non-inductive operation: • Primary voltage  Vloop-Vloop, ref • PLHCD Ip ref - Ip • n//-LHCD Liref- Li with Li b2q / b2q (a) • More recently* n//-LHCDHard X Ray width representative of LHCD absorbed & J profile [Wijnands Nuc Fus 1997, Litaudon PPCF 1998, *Joffrin Nuc Fus 2007]

  30. RT control of minimum q, qmin DIII-D • Feedback control of qO or qmin during the plasma current ramp-up phase • Change of plasma conductivity through electron heating • ECRH or NBI • RT q-profile using MSE data [J. Ferron et al Nuc Fus 2006]

  31. 1.31.7 RT control of minimum q, qmin JT-60U • High-bp ELMy H-mode • Ip=0.8MA, Bt=2.5T, q95=5.8, ne=1.8x1019m-3, bp=1.2-1.5. • Off-axis LHCD control: • dPLHCD/dt=a(qmin,ref - qmin) • a=2MW/s • Without control • qmin down to 1.3 • Interaction j & Te • Requirement for T & J control interlocks by arcing N//=1.7 [Suzuki et al NF 2008]

  32. Multi-Input-Multi-Output (MIMO) profile control All transfer functions are in matrix form Controller disturbance K(s) noise G(s) sensors actuators q(s) qtarget P(s) sensors plasma actuator dq(s)=K(s) dP(s) Controller dP(s)= G(s) dq(s) G(s)= gc[1 + 1/(tis)] K(0)inv First approach: control based on pseudo-inverse of the steady-state gain matrix [D. Moreau et al Nucl Fus 2003, D. Moreau et al Nucl Fus 2008]

  33. RT q-profile control with off-axis LHCD in low b-phase • control in the prelude phase • "Model Based" control on 5 q-values • PLHCD is controlled to minimise (q-qtarget) in the least square sense • Access to various q-profiles [ D. Mazon et al PPCF 2003, D. Moreau et al Nucl Fus 2003]

  34. setpoints start of control end of control undershoot RT q-profile control in high b-phase Multi-Input, multi-output control 3 actuators: LHCD, ICRH, NBI Model based SVD control: steady-state gain matrixdeduced from open loop experiments [D. Moreau et al Nucl Fus 2003]

  35. q-profile Jplasma pressure pressure  Internal Transport Barrier Jbootstrap bootstrap   pressure Jext. Control of kinetic & magnetic profiles • non-linearly coupled profiles • two time scales • tres ~10s-100s • tE ~1-5s

  36. ρ*T 1 0 x Control of kinetic & magnetic profiles ICRH q or i=1/q LHCD 1 0 x NBI modulations around a reference steady-state: Galerkin’s projection of 1/q and ρ*T [D. Moreau et al Nuc Fus 2008, T. Tala et al Nuc. Fus 2005]

  37. Control of kinetic & magnetic profiles q profile and r*Te control The controller minimizes quadratic error: [Laborde et al., PPCF (2005), Moreau et al. Nuc Fus 2008]

  38. Control of kinetic & magnetic profiles • Limitation of steady-state gain matrix approach: controller response slow during fast events • Development of an optimal control (time dependent model) q Measured Reference 1/q r*Te [Laborde et al., PPCF (2005), Tala et al Nuc. Fus 2005, Moreau et al Nuc Fus 2008]

  39. Slow model: resistive time scale Fast model: momentum / thermal confinement time scale, t = e t e<<1 Fastcontroller : Two time-scale modelsslow: magnetic profiles, fast: kinetic profiles [Moreau et al Nuc Fus 2008] Slow controller : opt. feedback opt. feedback

  40. x=0.1 x=0.2 x=0.3 x=0.4 x=0.5 x=0.6 x=0.7 x=0.8 x=0.9 x=1.0 Identification of a dynamical model Future: closed loop experiments 1/q(x) & V(x) • Generic approach: can be applied to any tokamak with any set of actuators and real-time measurements • Model identified on JET, JT-60U and DIII-D (on-going) Slow model for 1/q(x) x =0.5 x =0.4 x =0.6 x =0.5 x =0.7 x =0.8 x =0.6 x =0.9 x =0.7 N°70318 [Moreau et al Proc. 48th IEEE Conf. on Decision and Control China 2008]

  41. Modelling of real time control of Te and q • q control @ 300s • Te control @ 400s • 4 actuators • transport model based on global confinement scaling law • real-time update of the static models ITER model based profile control (CRONOS) : 12MA hybrid mode ITER S.H. Kim et al, EPFL thesis No 4500 sub to PPCF

  42. RT control in dominated bootstrap & a-heating regimes : open issues • existence of a stable and unique state with self-consistent pressure and current ? • control at high Iboot with PaPadd ? • rely mainly on q-profile control with minimum external CD? • pressure control requirements should be minimized • model based control? • strong requirements in terms of integrated transport modelling • ‘simulate’ in present day experiment a-heating with additional electron heating source • Experiments performed on JET & JT-60U to mimick a-heating in standard ELMy H mode regimes: how to extend to non-inductive operation ?

  43. SIMULATION OF ALPHA PARTICLE PLASMA SELF-HEATING USING ICRH UNDER REAL-TIME CONTROL • ICRH applied in response to real-time measured plasma parameters (e.g. neutron rate) simulating the self-heating effect • part of the external heating plays the role of auxiliary heating • Demonstrate stable control of the simulated burn? [T. Jones, EPS 2001]

  44. SIMULATION OF ALPHA PARTICLE PLASMA SELF-HEATING USING NBI UNDER REAL-TIME CONTROL [Takenaga et N Fus 2008]

  45. Mimick the self-alpha heating and self-driven current in present day non-inductive experiments • Integrated fusion burn control experiment to prepare Long Pulse Operation on ITER & DEMO • ICRH/ECRH ‘mimic’ the a-power→ Pa and Pfus • ECCD/LHCD ‘mimic’ bootstrap→ fBoot > 50% • Remaining powers for control→Pcontrol → Qeff = Pfus/Pcontrol~ 5-20 • Could be tested on long pulse tokamaks : Tore Supra, JT-60SA, EAST etc … • “Proof of principle” through modelling using a simplified version of CRONOS, METIS • Combination of H&CD powers & density actuators are required for burn control: • Powers : fast and precise control • Density : slow and coarse control

  46. Fusion burn simulation at high bootstrap fraction • high bootstrap burning plasma • a-heating mimicked using ICRH/ECRH • bootstrap mimicked using LHCD • Fuelling, Heating/CD feedback control • Equivalent fboot and Q Slow time scale

  47. METIS : A tool for (burn) control simulation METIS work-flow organisation • Mixed 0D and 1D equations • Coupled to “Simulink” for real time control design • Fast dynamic simulation • ~ 1minutes for 300 time slices • 2s per time slice when coupled to Simulink • Included in the CRONOS suite to prepare integrated modelling [Artaud, Litaudon et al EPS 2008]

  48. TORE SUPRA Modelling of Burn control with an L-mode confinement Control phase [Artaud, Litaudon et al EPS 2008]

  49. TORE SUPRA Modelling of Burn control with an L-mode confinement Control phase • Burn control@ 10s • control feasible for range of Pa (gfus): • Q = 15 obtained • target is missed • system runs away Target: Q = 15 Target = 80 % • Add nD control (slow): widens the operational space available for burn control while minimizing the external heating sources.

  50. TORE SUPRA Modelling of Burn control with ITB: Q~15 at fboot~70% at Vloop=0 Control phase

More Related