R Felton 1 (22) Session Leader Training November 2010 Plasma Density Control R Felton Session Leader Training November 2010 Torus radius 3.1 m Vacuum vessel 3.9m x 2.4m Plasma volume 80 m3 Plasma current < 5 MA Toroidal field < 4 Tesla Unique technical capabilities : Tritium Beryllium

R Felton 2 (22) Session Leader Training November 2010 Plasma Density Control Objectives Inject Gas for breakdown and initial current rise Inject Gas to achieve required density for (i) Experiment (ii) Neutral Beam Injection Inject special gasses for experiment e.g. Tritium, Neon, Argon Inject special gasses to assist RF / LH coupling e.g. Hydrogen

R Felton 3 (22) Session Leader Training November 2010 Ways and Means Gas Bottles, Lines and Matrices Gas Introduction Modules Plasma Density Measurements Plasma Density Controller (PDLM) Gas and Density Pulse Schedule Programme Co-ordination JET Operating Instructions

R Felton 4 (22) Session Leader Training November 2010 Gas Bottles, Lines, Matrices 9 gasses in Gas Store 4 lines from Gas Store to Vessel 11 Injection Modules 2 Valve Matrices Control:- GIS on VC

R Felton 5 (22) Session Leader Training November 2010 Gas Introduction Modules Mid-Plane –1,2,4 Top & LH –5,6,7,8 Divertor –9,10,11,12 Tritium –15

R Felton 6 (22) Session Leader Training November 2010 GIM Control & Calibration Piezo moves valve against spring Valve opening depends on voltage across piezo (programmable psu) –Not linear –Stable enough PDLM controls –0.. 10V == % open –Needs calibration curve Calibration Pulses –In Restart

R Felton 7 (22) Session Leader Training November 2010 Density Measurements Line-Integrated Density (LID), usually the Vertical Line-Of-Sight 3 –KG1 Interferometer (main) –KS3 Bremsstrahlung (back-up) Range e 18 /m 2 Occasional problems with –Fringe-jumps –Calibration of Bremsstrahlung New signal processing –KG1 real-time –KG1 fringe counting –KG4 Polarimetry KG1 KS3

R Felton 8 (22) Session Leader Training November 2010 Plasma Density Validation Previously –Main density signal for Density Control selected from KG1 LIDs, blended in KVS –Backup from KS3 Bremsstrahlung, scaled to the main density Future, –Density signal from several LIDs KG1v+KVS, KG1rt, KG1c (fringe counter), KG4 polarimetry (Cotton-Mouton) KS3 Bremsstrahlung –Density signals for Density Control in PDLM NBI Enable in PEWS Real-Time Experiment control in RTCC –New PDV pulse schedule

R Felton 9 (22) Session Leader Training November 2010 Feedback Theory U = Line-Integrated Density (e 18 /m 2 ) Uref = Reference waveform (e 18 /m 2 ) E = Control Error Yfb = Feedback Yff = Feedforward waveform (0..10V) Y = Control output 0..10V == % opening Reference Waveform - PID Control + Feedforward Waveform UrefEYfbYff GIM PSU and Valve Plasma Measurement & Validation YU Control Error E = Uref – U Feedback Yfb = K p E + K i ∫ Edt + K d dE/dt Kp = G G gain Ki = G / TiTi integral time constant Kd = G * TdTd differential time const. Control Output Y = Yfb + Yff Feedback as Laplace transform

R Felton 10 (22) Session Leader Training November 2010 Feedback Practice Sensor: KG1/LID3 PID: G=1, Ti=1000s, Td=0.7s (  =1) Actuators: GIM8: FB + 20% FF P only Density exceeds reference Sensor: KG1/LID3 PID: G=1, Ti=2s, Td=0.7s (  =1) Actuators: GIM8: FB + 20% FF PI Density on target Error Density GIM Electron Flow

R Felton 11 (22) Session Leader Training November 2010 Feedback Practice Sensor: KG1/LID3 PID: G=1, T i =3s, T d =0.7s (  =1) Actuators: GIM8: FB + 20% FF X-point formation increases pumping e(t) ≠ k v(t) Gas flow e(t) and v(t) n(t) and n ref (t) t=61s t=62s

R Felton 12 (22) Session Leader Training November 2010 Gas Pulse Schedule Gas species Reservoir pressure Control mode –Puff –Dose –Feedback –RTC Control Times Density Reference for Feedback

R Felton 13 (22) Session Leader Training November 2010 GIM Set-up Puff: 100% opening of GIM for a time FBack: Feed-back in time window with feed-forward (Dosing) waveform Dosing: time window in waveform V2 valve open increases volume

R Felton 14 (22) Session Leader Training November 2010 PDLM (aka PDF) Schedule Requested Density waveform Control Mode = PID Gain G Integral Time Constant Ti Differential Time Constant Td and strength α

R Felton 15 (22) Session Leader Training November 2010 Plasma Density Feedback Input reference : target density waveform units = “Fringes” (10 18 m -2 ) starts at GAS=39.5, allow for pump-out at X-point formation allow for heating phase Input signal : DENS1 (usually) Output signal : Vopen ( V) to any GIM Vopen = VopenFF + VopenFB Proportional Gain:  Vopen/  LID A simple flat unit gain works well, add extra GIMs to increase effective gain at appropriate time (e.g. Xpoint). Select GIMs for appropriate speed and location.

R Felton 16 (22) Session Leader Training November 2010 Plasma Termination Normal –PDLM follows requested density, usually a ramp down from experiment level to zero Soft Stop –When Plasma Termination Network (PTN) triggers a Soft Stop, PDLM ramps down from the last-good density to zero, proportional to plasma current. GIM Stop –When Plasma Termination Network (PTN) triggers a GIM Stop, PDLM switches off the GIMs designated to be switched off by the GIM Stop.

R Felton 17 (22) Session Leader Training November nd Feedback Loop Experimental Control a Signal e.g. Frad using Gas –Better to use RTCC Target waveform –Normalised range -1 to +1 –Starts at GAS = 39.5s Signal –Normalised range -1 to +1 –Selected and scaled in RTSS pages PID Feedback –P I D tuning needs several pulses –P I D can be estimated from a step response Step Gas, observe change in Density and Signal G ~∆Signal / ∆Density, Ti ~ rise-time

R Felton 18 (22) Session Leader Training November 2010 Real-Time Control RTCC output PDF1 –Sets the reference waveform for the Main Loop (Density) –Modify target density dynamically, according to events RTCC output PDF2 –Sets the reference waveform for the Auxiliary Loop (Norm. Sig.) –Modify target signal dynamically, according to events RTCC output GIM groups A,B,C –Sets the GIM opening directly (0..100%) –Trigger puffs/doses on events, e.g. NTM, Pellet Centrifuge –Experimental Feedback Control, e.g. 3 He, Frad

R Felton 19 (22) Session Leader Training November 2010 Real-Time Experiments 3 He Control for Mode Conversion Y = 100Φ He / (100Φ He + 18Φ Da ) U = 3 He gim Detached Plasma Y = Φ Da / Φ Br U = D 2 gim Guards for Tritium puff Detached Plasma serendipitidous gain

R Felton 20 (22) Session Leader Training November 2010 Gas Inventory Gas injected per pulse Gas loaded on cryo pumps

R Felton 21 (22) Session Leader Training November 2010 Programme Coordination The Gas Form –Programme Intranet SL requests –Gas species –Reservoir pressure Approved at Weekly C/O Coord Printed for –Shift Techs –Vac Group –EiC D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D 2 +10% H 2 H2H2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 CD 4 D 2 +10% H 2 H2H2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 CD 4 D 2 +10% H 2 H2H2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 D2D2 CD 4 D 2 +10% H 2 D2D2 D2D2 D2D2 D2D2 D2D

R Felton 22 (22) Session Leader Training November 2010 JET Operating Instructions JOI 4.2 –Pews Density Limits for Neutral Beam Injection Valid in any Plasma Configuration JOI 4.7 –Alternative Pews Density Limits for Neutral Beam Injection, Valid in Fat or Superfat Plasma Configuration Only JOI 7.2 –Vacuum Conditions for Tokamak Operation JOI 7.4 –Operation of Gas Introduction Matrix JOI 7.5 –Avoiding Potentially Explosive Mixtures