Home
Full Text
Contents
1. ET O0 05 64 a aa PN 0 1 0 01 3 e J 2 0 001 bet Q 4 oO 0 0001 E 10 w cas aa za Aiabe 3 i b OD 1 F N RO J 10 bs lai 200 0 200 400 600 800 1000 1200 distance along beam mm y 1 0376 e 0 0020187x R 0 99905 decay length 520 mm y 0 073199 e 0 010061x R 0 9959 decay length 98 mm y 0 076357 e 0 028048x R 0 99786 decay length 37 mm JPN53881 attenuation 70 keV He beam HEBEAM Tey 0985 3 T E ae Gian S eee 0 01 ea ce ae ee engen c p ri 1 2 i w 0 001 R Aa a a ee E 5 A Ik i H i 3 Q a fo i i a f i f i 0 0001 iets a Se SO OOS Oucae cs ae pens 10 4 10 fees e i 200 0 200 400 600 800 1000 1200 distance along beam mm y 1 0065 e 0 001444x R 0 99981 decay length 690 mm y 0 048648 e 0 018066x R 0 99898 decay length 55 mm y 0 0435 e 0 056393x R 0 995 decay lenght 18 mm Fig 2 55 Modelled populations of the ground state and the two metastable states The left graph shows results of the scotty_fwd modelling and the right graph results from HEBEAM A clear difference in the dec
2. 2 eff gt 610 Cc S a 2 E 410 wo oO Ko 210 200 0 200 400 600 800 dist along beam mm Fig 2 44 Singlet beam emission profiles of two discharges with and without CD puff Take not of the significantly narrower peak which forms in the pulse with CD puff Two different Hel triplet emission profiles have been measured with KS5a using a 134 5 kV beam Both profiles overlaid in Fig 2 45 have essentially the same shape but differ in intensity At the radial position z 1 60 mm we note a deviation between the two profiles which is limited to one viewing line 4 It is likely that this deviation is caused by an impurity line contributing to the measured intensity of the Doppler shifted beam emission line which appears at the same spectral location for this particular viewing line Furthermore comparing the profile of the 2 P 3 S transition 706 5 nm measured at 135 keV beam energy with KS5a with that of the same transition measured with KS7 at a beam energy of 73 keV we also find identical shaped profiles Fig 2 46 indicating that their sensitivity to He beam energy is weak The fact that the same profile is obtained with two independent systems gives additional confidence in our measuring and evaluation procedures Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 56 Hel triplet emission KS5a 2 4 10 l i
3. The measured emission profiles have been compared with calculations from a numerical collisional radiative model This model has been extended and optimised in performance as part of this thesis The main goal of this work was to derive the electron density and temperature profiles from the measured Hel emission profiles The developed reversion code is based on a variational method which calls the model calculation code many times and therefore required a subroutine which was optimised with respect to speed For the first tests with the reversion code synthetic data were generated in order to overcome possible errors in the look up tables generated from the atomic data The code shows good convergence and both electron density and temperature could be derived from sets of two different Hel emission profiles In summary the results show convincingly that fast He beam emission spectroscopy can be used as an electron density and temperature diagnostics particularly for the plasma edge Furthermore there is also the prospect for other diagnostic applications Contents 1 Contents 1 Introduction 3 1 1 Motivation ansehen 3 1 2 Beam emission Spectroscopy 444444444nn nn nnnnnnnennnnnnnnennnnnnnnnnnnnnnnennnnnnnnannn 5 1 3 Collisional radiative model for calculation of beam emission profiles 6 2 Hel beam emission measurements on medium size and large fusion experiments 8 2 1 Helium doping system for D heati
4. and A Unterreiter Plasma Phys Contr Fusion 34 1173 83 1992 Appendix 118 Appendix B List of Abbreviations ADAS Atomic Data and Analysis Structure ASDEX Assymetric Divertor Experiment AUG ASDEX Upgrade BES Beam Emission Spectroscopy CCD Charge Coupled Device optical sensor array CER Charge Exchange Recombination Spectroscopy CXRS Charge Exchange Recombination Spectroscopy ECE Electron Cyclotron Emission EFIT Equilibrium Fitting code for calculating shape and position of the flux surfaces ELM s Edge Localised Mode s FWHM Full Width at Half Maximum GRC s Generalised Collisional Radiative Coefficient s H mode High confinement mode improved edge confinement IDL Interactive Data Language by Research Systems Inc ITB Internal Transport Barrier JET Joint European Torus JPF JET pulse file JET database for raw data JPN JET puls number LCFS Last Closed Flux Surface LIB Lithium Beam L mode Low confinement mode MHD Magneto Hydro Dynamic Appendix 119 List of Abbreviations cont MSE Motional Stark Effect Diagnostic NBI Neutral Beam Injection NIB Neutral Injection Box NTM Neo classical Tearing Mode PINI Positive Ion Neutral Injector Plug In Neutral Injector PPF processed pulse files JET database for derived signals SNR Signal to Noise ratio TOKAMAK Toroidalnaya Kamera sz Magnitnimi Katuschkami toroidal chamber with magnetic coil a particular magnet configuration for f
5. 2 00 2 00 Rs Cc S 1 50 1 50 2 3 1 00 1 00 Q Ko gt 0 50 I L 0 50 gt oO 0 00 0 00 track 2 3 4 5 6 7 8 9 10 11 Fig 2 19 Overlap calibration factors for four successive pulses with the spectrometer set to 667 8 nm The accuracy and reliability of the method is demonstrated in Fig 2 20 showing profiles from two repetitive plasma pulses which were taken with different spectrometers looking at different beams The profiles match very well after having been shifted against each other by 20 mm This could Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 25 either be due to a mapping error or a consequence of the sweeping mapping based on different time slices The model calculation confirms that the attenuation of the 2S state is essentially the same for a 75 keV and a 135 keV beam see Fig 2 21 For details of the modelling see chapter 4 2 The good match between the two profiles which are taken with different spectroscopic systems using two different He beams is therefore a confirmation of the accuracy of the measurement 587 6 nm 2 P 3 D beam emission arb units kota eggy spun Que uos sua weag 100 0 100 200 300 400 Rmaj mm Fig 2 20 Profile of the emission of a Hel triplet line measured with the spectrometer KS7 looking at a 75 kV beam and with KS5a looking at a 135 kV beam for two repetitive pulses The profile measured with KS7 was shifted radially by 20 mm
6. 210 FE a SE aaa JPN 53882 706 5 nm g ps gg J B z Beer ee JPN 53881 587 6 7 5 H Ce Ra i onm o E 1 9210 A 1 6107 3 o E f o c p Z pa ge 2 1 4410 1 2 10 Z 2 fee o P gt 15 14 3 zZ Z 9 6 10 810 3 on Cc S 3 a E 4810 410 nn re i 5 a A REN 2 o meet lee ee N 0 100 0 100 200 300 400 dist along beam mm Fig 2 45 Comparison of the beam emission profile for the 2 P 3 D transition at 587 6 nm JPN 53881 with the one for the 2 P 3 S transition at 706 6 nm JPN 53882 Hel 23P 33S 706 5 nm 210 aq nn 6 10 g 2 2 Q 1 610 4 8 10 3 a 3 3 ae n o 2 3 B 1 210 3 610 R n Z X a a k z wo 2 810 2 410 9 a E 8 2 T E o 410 1 210 _ E 3 2 5 a 0 0 100 0 100 200 300 400 dist along beam mm Fig 2 46 Comparison of Hel 2 P 3 S 706 5 nm beam emission profiles taken from comparable plasma pulses with different He beam energy KS5a observed a 135 keV doped He beam and KS7 a 73 keV pure He beam Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 57 Profiles obtained with KS7 72 6 keV pure He beam These measurements are different from the above ones in two respects 1 The pure helium beam is expected to have a lower initial metastable fraction than the doped He D beam as the fas
7. D 53880 D gt 10 0 1 9 468 6 HellCX 501 5 2 s 3 p 53881 D gt 10 0 1 9 706 5 2 gt p 3 S 667 8 2 p 3 D 53882 D gt 10 0 1 9 587 6 2 P 3 D 706 5 2 gt p 3 S Table XI Listing of pulses from the shift on 2 March 2001 with 134 5 keV doped He beam from octant 8 KS5a and a pure 72 6 keV He beam from octant 4 KS7 The first four pulses where used to set up the flow rate of the Dz and CD gas puff in order to get the same plasma density with either puff Profiles taken from KS5 134 5 keV doped He beam Two profiles from plasmas with CD puff JPN53872 and JPN53874 were more peaked than those from plasmas with D2 puff JPN53873 and JPN53875 This can be seen in Fig 2 44 where the first two pulses are overlaid using the same scales for both pulses Apart from the peak near z 1 00 mm both pulses are quite similar in profile shape and intensity as if the higher density in one pulse would be compensated by the higher Zef in the other pulse In the case of 70 keV beam energy for the triplet profile in Fig 2 2 43 the pulse with the higher impurity level has a Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 55 reduced peak intensity while in Fig 2 44 the 135 keV singlet profile shows a higher peak intensity for the pulse with the higher impurity level 667 8 nm Hel 2 P 3 D 110 JPN53873 Z 1 7 lid 11 9 10 m 8 10 JPN53872 Z 2 1 lid 8 9 10 m
8. The different behaviour of cross sections for spin conserving and spin changing collisions leads to a rather different temperature dependence of the excitation rate coefficients even for electron Chapter 1 Introduction 4 energies of some 10 keV which are typical for the plasma core Fig 1 1 shows the temperature dependence of some rate coefficients for impact excitation from the ground state 2 S and from the triplet metastable state 23S 10 b 4 D 10 b zl v E E ge 3 g E So 107 5 a g E 3 S 10 amp z 10 p l eel raul ril raal 10 10 10 10 10 electron temperature eV 107 E q ae 3 0 8 10 Fr E 10 b A 2 E J 17 F 7 o 10 z oO E oO 18 s 0 F 3 10 E T ai E rail risl toil toil gt 10 10 10 10 10 electron temperature eV Fig 1 1 Electron temperature dependence of rate coefficients for electron impact excitation out of the He ground state 1 S into several excited levels Chapter 1 Introduction 5 1 2 Beam emission spectroscopy The passive visible spectroscopy of the plasma is limited to its edge as essentially all atoms are fully stripped inside a hot fusion plasma By injecting fast neutral particles a wide range of parameters becomes accessible to active spectroscopy Most of the major fusion devices have to make use of neutral beam heating systems for active spectr
9. The fans of lines of sight are shown in Fig 2 8 together with the beam trajectories The optical head with the adjustable mirror is shown in Fig 2 9 The fibres form the two periscopes are distributed to three spectrometers KS5a b c making simultaneous measurement for different wave lengths possible Fig 2 10 shows a sketch of the KS4 5 system All three spectrometers are of Czerny Turner design and use a subset of up to 12 fibres and a single entrance slit The detectors are PC controlled back illuminated CCD sensors Due to coating of the plasma facing optical components and ageing of the fibres the light throughput has decreased by different amount from fibre to fibre and the signal Fig 2 8 Plan view of the JET vessel showing the neutral beam trajectories and the fans of viewing lines hatched areas Position A oct 7 and C oct 1 are the locations of the optical heads of the two periscopes B marks the vertical port of the single viewing line of KS4 Fig 2 9 Sketch of the optical head of the periscopes consisting of a stacked fibre bundle F lens system L vacuum window W and fold back mirror M Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 18 intensities were quite different for the various instruments Because KS5a provided the highest signal intensities the measurements have been performed mainly with this spectrometer Carbon and Neon Sin
10. as expected from its neutralisation in He The small metastable 2 S fraction is also obvious from the model calculations shown in Fig 2 48 Here the measured emission profile is compared with calculations assuming 0 and 0 5 of initial 2 S population In the profile calculated for 0 5 metastable fraction a characteristic peak just appears which can not be seen in the measurements indicating that the initial 2 S population of the pure He beam is well below 0 5 Hel 2 S 3 D beam emission 72 6 keV pure He beam 310 910 g 3 D 2 3 240 610 3 2 2 5 9 8 S o S E 3 110 310 gt E 5 E a E B 2 ne init pop 0 5 2 S 5 2 S init pop 0 metastable 0 0 100 0 100 200 300 400 500 dist along beam mm Fig 2 48 Comparison of the measured 2 P 3 D transition 667 8 nm with model calculations The two lines are calculations assuming either 0 solid line or 0 5 dashed line initial 2 S population of the He beam Fig 2 49 compares singlet Hel beam emission profiles from the 2 S 3 P and 2 P 3 D transition measured with a pure He beam The intensities were normalised to the model calculations In the outer region up to z 150 mm the profile shapes are noticeably different This is also reproduced by the model calculations Further inside the plasma the profiles are similar in shape and the intensity of the most intense singlet line 2 P 3 D 667 8 nm is by
11. hrenden Fusionsexperimente ASDEX Upgrade in Garching DE und JET in Culham UK durchgef hrt wurden Zur Erzeugung der schnellen He Strahlen wurden dabei Injektoren der Neutralteilchenheizung verwendet um reine He Strahlen oder He dotierte D Strahlen zu erzeugen Die Strahlemission der schnellen He Atome wurde mit Hilfe der Ladungsaustausch Spektrometer des jeweiligen Experiments gemessen Mehrere Hel Singulett und Triplett Linien ausreichender Intensit t konnten im sichtbaren Bereich gemessen werden wobei die Empfindlichkeit der verwendeten Spektrometer gegen beide Enden des sichtbaren Bereichs hin diskriminiert war Die Doppler verschobene Strahlemission ist als reine Linienstrahlung in der Regel ungest rt von Verunreinigungsemissionen gut me und auswertbar Die Emission von Triplettlinien ist auf die u eren 200 mm des Plasmas beschr nkt jene von Singulettlinien hingegen ber den gesamten beobachteten Bereich me bar jedoch am Maximum um eine Gr enordnung schw cher als die h chste Intensit t der st rksten Triplettlinie Aus der Form der Singulett Emissionsprofile konnte auch der anf ngliche Anteil an metastabilen 2 S Atomen im He Strahl bestimmt werden Emissionprofile der intensivsten Hel Singulettlinie 2 P 3 D bei 667 8 nm und der intensivsten Triplettlinie 2 P 3 D bei 587 6nm wurden an AUG f r unterschiedliche Plasmaentladungen gemessen Au erdem konnte die Streuung der Messergebnisse anhand einer Messreihe an nahezu
12. name yttocs was chosen for the reversion code IDL interactive data analysis language is used as development platform for various reasons Compatibility IDL is used for the data analysis at JET and AUG and also as the graphical user interface GUI in ADAS The graphics output which can be easily integrated into the programmes is ideal for testing the code during all development stages GUI windows so called widgets for a comfortable and interactive user interface are easy to create 4 1 Basic algorithm The basic algorithm of the conversion code is a variation method with two nested iteration loops The code calculates the Hel beam emission profile from an initial guess of the electron density electron temperature and Zer profiles forward calculation The input profiles are then adapted iteratively with the aim to minimise the quadratic deviation between calculated and measured beam emission profiles As profiles parametrised functions are being used This limits the number of parameters of the optimisation procedure and avoids unphysical profile shapes The flow diagram Fig 4 1 shows the two levels of iteration Within the two inner loops the parameters from either the electron density ne or the electron temperature Te profile are varied and the beam emission profile is calculated This iteration continues until the quadratic deviation between the calculated and measured emission profiles has converged i e
13. on Fusion Enfineering Monterey CA IEEE Cat No 87CH 1 2507 2 1987 H B Gilbody K F Dunn R Browing and C J Latimer Formation of metastable helium atoms by electron capture during the passage of fast He ions through gases J Phys B Atom Molec Phys 4 800 813 1971 H B Gilbody K F Dunn R Browing and C J Latimer Electron loss from fast metastable and ground state helium atoms in passage through gaseous targets J Phys B Atom Molec Phys 3 1105 1112 1970 http www ipp mpg de spek home html C B Markwardt mpfit http cow physics wisc edu craigm idl fitting html J Mor B Garbow and K Hillstrom minpack 1 Argonne National Laboratory http www netlib org minpack Institut f r allgemeine Physik TU Wien Wiedner Hauptstr 8 10 A 1050 Wien Austria Appendix 117 43 M Proschek S Mendart H D Falter H Anderson H P Summers A Stabler P Franzen H Meister J Schweinzer T T C Jones S Cox N Hawkes F Aumayr and HP Winter Proc 18th IAEA Fusion Energy conference Sorrento Italy 2000 44 M Brix Messung von Elektronentemperatur und dichte mittels Heliumstrahl diagnostik im Randschichtplasma eines Tokamaks PhD Thesis Ruhr Universitat Bochum 1998 45 M Brix Abschlu bericht zum Forschungsauftrag Nr 021 41362607 930 an die Forschungszentrum J lich GmbH 1999 46 J Schweinzer E Wolfrum F Aumayr M P ckl HP Winter R P Schorn E Hintz
14. 0 01 F 4 E iteration 15 0200 fs Ah hs TS gi ig gh a 14 09 2001 12 01 yttocs V0 0 0 500 1000 1500 number of fwd_calc calls Fig 4 5 Output sheet from yt tocs for a undamped variation sequence A description of the output sheet is given on the preceding page Chapter 4 Reversion code yt tocs 104 ne_profile Te_profile a a ee 3000 Tg 6x10 9b re 2000 v 3 8 2 4x10 9 b z D 2 5 5 2 8 1000F o u v B ax10 S olus olna 200 6 200 400 600 800 200 0 200 400 600 800 dist along beam mm dist along beam mm Hel emission 1 measured sim 3 1 D Hel emission 2 measured sim 3 3 P Fu u ea e ya mu va an ann ya mu men sor 5F E E 40F T F S E 3o 5 E 5 m s F 2 F Be SE ef Ss 20r 5 E T E 2 E 2 E Ff E F v H v L E J E 10F E J 2 Ir J 2 E E amp 1 E E J OF oF g E E E ss E S eR eh E Ce EE ON Eee ri E a Oy PT 200 6 200 400 600 800 200 0 200 400 600 800 dist along beam mm dist along beam mm deviation data fwd calculation result of fit 0 06 Oo F start situation 0 05P 1 artificial data F H 1 initial population 2 singet S 1 0 0 04 4 Be initial population 2 triplet S 10 a E J 2 003F J mean value Zeff 2 0 5 F 5 F core value Zeff 2 0 3 F 1 gt 0 02 4 E AN TEE J resolution of measurement 5 0 O 0 01 4 E iteration 13 006 rc Cr Cr ar We Pa or Oe Cs BER EL C
15. 2a u BEE RER ERTE N D e e J aoe a 1 F a Hebeam Bin ee gt ee ee 4 T A a 5 510 EE EE SE A ek POSEE i DER yas eet Fk o en 5 410 un ee nce treet 4 O HN J SI ee en 4 E C i a lS 2 JE i 24 F Fur x 4 2 72 107 ee ee an x RS EL Gun I Fy J 24 7 i i x 4 i ns een 2 ne q 0 E 4000 3800 3600 3400 3200 3000 radial position Rmaj mm Fig 2 22 Emission profile for the Hel singlet line 667 8 nm 2 P 3 D The symbols are data spaced 2 mm resulting from the fit of single frames 50 ms exposure time The solid line gives an average over 10 data points corresponding a distance of 20 mm Chapter 2 Hel beam emission measurements on large and medium size fusion experiments 27 2 4 Measurements at AUG 2 4 1 AUG July August 1999 30 keV pure He beam Our first He beam emission measurements were parasitic As described in chapter 2 1 a pure He beam at half the nominal energy was injected at the end of the heating beam injection for 200 ms program with low NBI power needs This was the case for the investigation of neo classical The conversion of one PINI to He operation was possible for an experimental tearing modes NTMs where pulses with only 3 MW NBI heating power were used The CER diagnostic was needed for the ion temperature measurements However not all of its fibres viewing lines are used by the spectrometer therefore it was possible to connect four CER
16. In the case of the AUG injectors the beam alignment is adjusted so that the power loading on scrapers above and below the beam are equal Chapter 3 Analysis of the spectroscopic data 93 Using the standard C CX line emission at 5290 5 A for which the CER spectrometer is calibrated and normally used for we find that the raw data in Fig 3 11 are well smoothed if we assume a beam divergence of 0 9 degree and no misalignment of the beam It is noteworthy that the quoted divergence of 0 9 is practically identical with the measured divergence of a D beam The dissociation of D2 molecules in the ion source leads to an additional energy spread of the beam particles and consequently to a higher beam divergence then in the case of the atomic He beam Therefore a divergence of 0 75 is assumed for the He beam or He fraction of a doped beam Fig 3 12 shows the Hel beam emission profile 667 8 nm 2 P 3 D of a doped He D beam The zigzag behaviour of the straight data solid dots is not compensated over the whole radial range when applying a correction with a beam divergence of 0 75 and no misalignment of the beam dashed line When assuming the same beam divergence but a misalignment of 0 15 the correction seems to work for a different radial range solid line A different alignment of the beam could be explained by stray or earth magnetic fields before the neutralisation occurs which influences the D and He fractions differently
17. Plasma temperature and density profiles of the swept L mode plasma measured with the standard diagnostics of JET are shown in Fig 2 13 The abbreviations of the used diagnostics are explained in Tab 1 ECE Michelson Interferometer Te profile ECE Heterodyne Radiometer Te fe LIDR LIDAR Thomson scattering KE9D edge LIDAR sea wepe e Table I Abbreviations for the standard electron density and temperature diagnostics at JET The measured electron densities and temperatures are mapped onto the beam axis and a subsequent fit gives smooth profiles used for our later model calculations for details see chapter 3 2 Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 20 JET shot 53881 at t 57 8000 sec rer aai r 4 ke9d 57 8770 19E A ky63 57 8499 eee 5 r N N x lidr 57 8800 z 40x109 E t E amp E tt 3 0x10 19E t 1 o i E E 2o10 E i E N 1 0x1019 E 2000 ke9d 57 8770 kk1 57 7857 OMA IN IN N kk3 57 8002 1500 Bee lidr 57 8800 S D 1000 v 500 3 0 3 2 3 4 6 3 8 4 0 3 Rmaj m Fig 2 13 Temperature and density profiles of a L mode plasma used for the sweep experiments as measured with the standard diagnostics i beam power 6 1 10x10 Oct 8 PINI 7 o 3 a amp 2 He fraction radial position of outer plasma edge LCFS at mid plane rbo cm w amp
18. ae ae p Sopot ill See re 350 400 450 500 550 600 650 700 750 Hel wavelength nm Fig 2 31 Intensity of the Doppler shifted Hel beam emission peaks solid dots and the corresponding unshifted passive emission open symbols Wavelengths where no beam emission signal could be identified are indicated on the abscissa signal to noise ratio JPN 49029 48 100 oO 2 o c 10 T c D Kr c O O 2 E oO E 1 oO oO oO oO a i Eo i 0 1 i L L L i i L 1 1 L 1 N i N 1 ji 1 ji 1 i 1 1 L 1 L fh 1 L 1 3 5 3 55 3 6 3 65 3 7 3 75 3 8 3 85 3 9 Rmaj m Fig 2 32 Signal to noise ratio SNR of the 6 most intensive Hel lines The low SNR of the 388 9 nm and 447 1 nm lines compared to their relative high absolute signal intensities is due to the low sensitivity of the spectroscopic system at this wavelength The lines are for guidance only Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 40 In Fig 2 32 the signal to noise ratio SNR of the Doppler shifted HeI beam emission signals are plotted for the 6 most intensive Hel lines The sensitivity of the KS7 system dropped sharply at 500 nm This can be seen in Fig 2 33 where the quantum efficiency of the KS7 CCD sensor and the transmission of the optical system are plotted vs the wavelength range of interest This drop explains why the two lines with the lowest wavelengths
19. against the other profile Another effect of using plasma sweeping is a large noise level of the beam emission data which manifests itself mainly in sections with small gradients Fig 2 22 The integration time of the spectrometer was set to 50 ms or 100 ms to keep the plasma movement during the exposure as small as possible 2 or 4 mm resp The large scatter was considerably reduced by smoothing over 10 frames which corresponds to a plasma range of 20 mm and is still small compared to the spatial resolution of the spectroscopic system Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 26 beam attenuation JPN 53881 sr ee a Se a nn a a a i 2 s population 70kV 7 0 08 i RIESE ENGER EE AE teste ns population 130kV 4 g L i J amp L a be a op Zu 006 Besen ee Ik D I 5 H S L J 0 04 F arene u nn 4 z L a Q L 2 L Bu 0 02 Het Agnes 4 em 20 0 20 40 60 80 dist along beam O LCFS cm Fig 2 21 Modelled beam attenuation of the 2 S level for two different beam energies 810 p 1 i i JPN 53874 KS5a 1
20. and saves the combined sensitivities int calibr pro A call of this function returns the intensity calibrated data It uses the sensitivity files generated with build_sens_file pro and merge_sens_files pro for the respective spectrometer At the moment sensitivity files for KS5a and KS7a are generated with the data from the latest available calibration measurements It is planned to extend the code with a history of sensitivity files to be able to apply the appropriate calibration for each pulse number Chapter 3 Analysis of the spectroscopic data 90 3 6 Alignment correction for AUG data During the HeI beam emission measurements at AUG the alignment of the CER diagnostic to the beam was not perfect making a correction of the measured beam emission intensities necessary beams 1 amp 4 south Geometry of the AUG SO injector A beams 2 amp 3 vacuum port east Fig 3 10 Geometry of the AUG 60 kV injector He beam emission measurements were performed with beam 3 Chapter 3 Analysis of the spectroscopic data 91 3 6 1 Geometry of the beam axis The co ordinate system of the SO injector is shifted and inclined against that of the AUG tokamak A point u v w of the injector frame converts into the AUG x y z frame through the transformation cf Fig 3 10 x ucosg vsing Rcos y y using vcosg Rsiny z w 0 0 10 with R 2 842m y 33 75 18 75 Using two points o
21. be performed by monitoring the emission of a standard white light source cavity with aperture with uniform emission in 27 with a known emission placed at the location of the emitting plasma In case of JET this is not possible Instead the optical head was taken out of the machine and placed in front of the light source leaving the vacuum window excluded In case of our measurements with the KS7a system during the campaign the situation was worse It was only possible to disconnect the fibres from the optical head and place them in front of the light source aperture leaving the optical head and the vacuum window excluded However it should be sufficient for a relative calibration from wavelength to wavelength because the wavelength dependence of the transmission of the optical head is expected to be small Comparing the measured intensity counts pixel after the background subtraction at a given wavelength with the known emission of the lamp gives the sensitivity of the system at this wavelength The spectrometer has been calibrated over the full range of wavelengths in steps of 10 nm In order to get a statistical information of the calibration measurement ten successive exposures with an exposure time just below signal saturation have been made for each measurement For the background subtraction dark count rate it is important to use the same exposure time for the background measurements Therefore the respective set of background mea
22. by statistical noise of the sensor Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 53 In the pair of pulses shown in Fig 2 43 JPN52798 and JPN52804 one pulse has been performed with a D gt puff Ze 1 7 the other with a CD puff Zete 1 9 The line average density was marginally 7 higher in the pulse with the CD puff The beam emission profile from the pulse with CD puff shows a significantly lower intensity by a factor 2 38 and a slightly reduced penetration compared with the profile from the pulse with D3 puff From this we conclude that the metastable triplet population is strongly influenced by the impurity level in the plasma 2 P 3 D Hel emission 587 6 nm 10 T T ji T T T T T T T T T c JPN 52798 Z 1 7 eff T P T w ll 3 N 10 3 o o N S LO Z 49 I a oO i N 5 D 2 z E N x g E R E PEPEE EE ETE AE EEE pees EA NIE FURHRENER 1 5 en Meee ee EEE SE ee oe hate es Be sacha ine FERNE HERE 10 of io ae HENSELAE 50 0 50 100 150 200 250 dist along beam mm Fig 2 43 Hel beam emission profiles of the 2 P 3 D triplet line 587 6 nm in plasma discharges with and without CD puff The CD puff causes a significantly reduced intensity right scale and a small reduction in penetration Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 54 2 5 3 JET March 20
23. data where generated for an initial metastable population of 1 singlet 2 S and 10 triplet 2 S this is marked with the solid black dot in Fig 4 8 Yttocs gives the possibility to include the initial singlet population in the variation procedure i e the initial singlet population is derived by the code Running the code with a wide range of initial triplet populations produces the initial singlet populations plotted in Fig 4 8 The correct initial singlet population is reproduced within a 20 error margin when the Chapter 4 Reversion code yt tocs 108 initial metastable triplet fraction is varied between 0 20 This means that it is possible to get a good estimate of the initial metastable singlet fraction for the full range of initial metastable triplet fractions which can be expected under the experimental conditions used resulting initial 2 S population 4 5 initial 2 S population fit result 0 10 20 30 40 50 60 assumed initial 2 S population Fig 4 8 Initial metastable singlet population 2 S derived by the yt tocs code for synthetic data generated with initial metastable populations of 1 singlet 2 S and 10 triplet 23S states solid black dot The open points represent the derived initial singlet population for a given initial triplet population For up to 20 2 S population not much difference of the 2 S population is found 4 5 3 Finite resolution of the observati
24. een _ eee see ie beam emission arb units 0 0002 Hf sasas a bec Keen Beeren 100 0 100 200 300 400 500 distance along beam mm Fig 2 27 Beam emission measurement and modelling of the Hel transition 2 P 3 D 587 6 nm for the AUG discharge 13665 The obvious discrepancy in the curve width is discussed in chapter 2 6 3 Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 34 The two points labelled in Fig 2 26 with their fibre numbers 7 and 8 show a clear tendency of too low values for different discharges and wavelengths See also Fig 3 12 in chapter 3 6 This could be due to calibration errors for these channels In Fig 2 27 an obvious discrepancy between the curve width of the measured and the calculated profile can be seen This problem is discussed in chapter 2 6 3 AUG standard H mode 2 P 3 D 587 6 nm low high density phase beam emission arb units 150 100 50 0 50 100 150 dist along beam mm Fig 2 28 Comparison of the triplet Hel beam emission 2 P 3 D for low and high density phase The profiles measured during the high density phase are shifted relative to the profiles measured during the low density phase by about 40 to 50 mm In Fig 2 28 the triplet emission profiles from the low and the high density phase are plotted in the same graph The abscissa is defined as the distance along the beam where the position of the LCFS defi
25. emission profiles of both lines Initial guess T 9 N profiles forward calculation N change of Ne Ne profile lt limit gt limit 1 2 3 Tes Nee 4 forward calculation i I br 8 T profile e lt limit gt limit lt limit gt limit final n T profiles Fig 4 1 Flow diagram of the basic algorithm of yt t ocs The wavelengths and the order of variations within the inner loop are only examples Chapter 4 Reversion code yt tocs 96 The code was developed in four stages 1 The forward calculation of the emission profiles i e the calculation of the Hel beam emission from given ne Te and Zerr profiles This code is based on the C code Scotty written for the He beam modelling 29 The code has been rewritten in IDL and is now called scotty_fwd The extension _f wd indicates that the actual calculation of the beam emission is performed by a subroutine called forward_calc which is a streamlined version of Scotty suitable for the variation method of the conversion code see chapter 4 2 2 Derivation of ne or T profile with all other parameters fixed These so called inner loops have been developed and tested separately The beam emission profile is fitted by varying the parameters of the n or T profile making use of the mpfit package 41 Part of this development step was to define a suitable parametrisation for
26. fibres to a spectrometer from the Li beam LIB diagnostics which remains unused for normal operation For about 20 pulses He was injected and the LIB spectrometer was set to the wavelength of one of the n 3 gt 2 or n 4 gt 22 Hel transitions for each pulse In that way the 11 most intensive visible Hel lines 29 were covered but no Hel beam emission signal could be identified Unfortunately it was discovered later that the image amplifier has not been working and the sensitivity of the CCD sensor on its own was not high enough to resolve the beam emission signal For some repetitive pulses it was not necessary to measure the ion temperature for each pulse therefore the CER diagnostic could be used to measure the HeI beam emission In that way the beam emission profile of three Hel lines see Tab V could be measured with the CER spectrometer These results were presented in 29 AUG pulse wavelength Hel transition beam energy equi beam maximum intensity nm keV current A of beam emission 12635 502 nm 2 S 3 P 27 keV 14 2 A 1 7 e19 ph m s 12643 587 6 nm 2P 3 D 27 keV 14 2 A 50 0 e19 ph m3s 12644 667 8nm 2 P 3 D 27 keV 14 2 A 4 0 e19 ph m3s Tab V Observed Hel transitions beam parameters and maximum intensity of the beam emission profiles for three successful pulses of the first campaign at AUG The intensity of the Doppler shifted beam emission was weak and the resultin
27. in fusion devices because it is frequently used for glow discharge cleaning of the vacuum vessel However significant equilibrium fraction of atomic He involves temperatures below 2 eV Hel emission from the plasma is therefore limited to the cold plasma edge This simplicity of the Hel spectrum is a big advantage as the spectral analysis can easily be automated and evaluated with high accuracy Doppler shift of beam emission The beam energies used for our beam emission measurements are in the range from 30 keV to 140 keV corresponding to particle speeds in the range of 1 2 2 6 10 m s The resulting Doppler shift of the emitted light is large enough for separating the beam emission peak from passive Hel plasma emission provided the angle between beam axis and viewing line is not close to 90 E g the beam emission of the 587 6 nm line 2 P 3 D cannot be separated from passive emission for observation angles in the range of 9045 2 2 1 Spectroscopic system at AUG CER diagnostics The CER diagnostics collects light from the plasma via a mirror mounted in a port of vessel segment 13 A lens system focuses this light onto a plate holding optical fibres These fibres define the viewing lines fanning out horizontally from the optical head to the beam from the ion source Q3 and Q4 of the SE injector Fig 2 4 shows a plan view of the set up The collected light is transmitted via optical fibres across the biological shield to a Cz
28. is measured wave length pixel by pixel The last calibration Fig 3 8 Example of a spectrum with a dead pixel within for the KS4 5 system was performed the Doppler shifted peak in 1999 In the data taken in 2001 the position of the pixels with the low sensitivity appears to have moved by one or two pixel The obvious drop in the number of counts for these pixels is independent of the wavelength setting of the spectrometer and was observed throughout the measurement campaign Fig 3 8 gives an example where a dead pixel is within the peak of the Doppler shifted beam emission The data shown were already calibrated using the 1999 pixel by pixel sensitivity At the position where the dead pixel was observed during the calibration measurements the intensity is being overcompensated On the other hand a too high sensitivity is being applied to the neighbouring pixels where the sensitivity has decreased since the last calibration measurements It is obvious that these inconsistencies in calibration can affect the peak height and width of the curvefit This problem could be overcome by a new calibration of the instrument which requires long torus hall access and was therefore not possible at short notice Chapter 3 Analysis of the spectroscopic data 86 3 5 Absolute Calibration of the JET spectrometer Ideally the absolute calibration of the whole optical system spectrometer fibres optical head and vacuum window would
29. lines with sufficient intensity could be identified in the visible range However on either end of the visible range the sensitivity of the spectrometer was too low The Hel beam emission appears in the spectrum as a clean Doppler shifted peak largely undisturbed by impurity emission Emission from the triplet levels is limited to the outer 200 mm of the plasma The singlet emission could be detected over the full observation range but its maximum is about one order of magnitude lower then the maximum of the most intense triplet emission The initial metastable 2 S fraction of the beam could be derived from the shape of the singlet Hel beam emission profile Emission profiles of the most intense Hel singlet line 2 P 3 D at 667 8 nm and the most intense triplet line 2 P 3 D at 587 6 nm could be measured at AUG for different plasma discharges Furthermore repetitive measurements of nominally identical pulses allowed to estimate the scatter in the data At JET measurements of emission profiles with good spatial resolution could be achieved by sweeping the plasma across the viewing lines during the measurement Besides the higher resolution this experimental technique also made it possible to cross calibrate neighbouring channels which yielded significantly reduced measurement errors Another interesting observation is the influence of the plasma impurity distribution on the triplet beam emission profile giving the prospect for a new Ze diagnostics
30. of 72 6 kV and for pulse 52797 a doped helium beam of 70 3 keV was used The pure helium beam passed through a plasma with a line density of 10 m compared with the line density of 0 75 10 m for the doped beam The intensity of the beam emission is a factor of 2 37 higher for the pure beam than for the doped beam From the known neutral beam current of the pure beam 16 A we can estimate the neutral He beam current in the doped beam as I doped N doped Sdoped line emission inensity doped He beam doped pure 4 ae pire Spure line emission inensity pure He beam With the data above we can derive a doped beam current dopeq Of 7 5 1 I 16 Sopen 10 2 37 5 06A 5 The respective beam emission profiles are shown in Fig 2 51 together with the modelled beam emission The profile from the doped beam shows a characteristic peak near the LCFS which is typical for the emission originated from metastable ions This peak is not present in the profile from the pure He beam indicating that the pure He beam has a much lower metastable 2 S population This is also reproduced by the modelling where a 1 initial 2 S population was assumed for the doped beam and no initial metastable 2 S fraction for the pure beam The scale for the model calculations is correct for the pulse with the pure He beam In order to match the measurement of the doped beam with the corresponding calculated intensity the lat
31. of the energy confinement time T from 0 06 to 0 13 s and the increase of the stored energy up to Wmm 0 36 MJ The plasma density electron density n increases as well although the gas flow decreases at the same time indicating an increased particle confinement For a short time after the L H transition the energy confinement time goes up to T 0 2 s and is lost again shortly thereafter with the first ELM event The ELMs edge localised modes can be identified from the Hea signal Balmer alpha line at 656nm which increases during an ELM event when the unstable plasma edge collapses During the first flat top phase t 2 5 4 5s the average plasma density settles at ne 5 10 m This means that during this phase the density is above the set point and only defined by the gas flux from the wall From t 3 to 3 5 s the beam power is ramped up to 7MW After the second power ramp up the first He injection into the ion source was triggered After t 4 4 s the control parameter for the plasma density is doubled ne 1 10 m This can also be seen in the gas flow At about t 5 5 s the density of the second flat top phase is reached and kept constant till t 7 s Other discharges at AUG Emission from the strongest He lines could also be measured in two other discharges that are characterised in table VII pulse amp number time s stored energy MJ heating power MW density 1 m Killer pellet 1364
32. plasma without reaching an equilibrium level The entire triplet beam emission is therefore dominated by the initial triplet population The actual initial metatstable triplet fraction can therefore only be estimated within the accuracy of the absolute calibration JPN53881 attenuation 70 keV He beam SCOTTY E 86 nord Sr 0 1 E i MANS ITT L Eq OOI A a E E 3a ll 5 one 5 F 2 s J 2 OOO Tee eer ee ne RNE 3 0 0001 gear E T i i i i 200 0 200 400 600 800 1000 distance along beam mm y 0 99776 e 0 0019298x R 0 9986 decay length 520mm y 0 074409 e 0 010161x R 0 99938 decay length 98mm y 0 064964 e 0 026783x R 0 99856 decay length 37mm Fig 4 7 He beam attenuation of initially populated levels modelled for the JET discharge 53881 Accuracy for estimating the initial metastable singlet population In order to assess the possibility of determining the initial 2 S population from the shape of the singlet emission profile e g 2 P 3 D 667 8 nm the yttocs code was tested with synthetic data see chapter 4 2 Emission
33. the JET test bed a He doping system with pre set timing was installed at the octant 4 injector 80 kV and first parasitic measurements of Hel beam emission could be made in November 1999 He doping system at AUG 60 kV Following the successful test at JET a similar He doping system has been installed at AUG at the so called SE injector This allowed us to use the so called diagnostic PINT Q3 with a more favourable viewing geometry instead of Q4 which has been used previously for the pure helium beam Successful measurements with a He doped D beam at the full beam energy of 60 kV were made in June 2000 new upper chords ee mi Be afl yt SUNOS Ae dopeg IT T a u Zi h eatin I 9 beam f N lower chords I 1 JET Fig 2 2 Cross section through the JET plasma showing the beam trajectories and the viewing lines used at the octant 4 injector projected to the cross section Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 11 Improved timing system at JET During the JET experimental C3 campaign November 2000 the pre set pneumatic timing system of the doping system was replaced by a more versatile electronic one During three dedicated sessions in December 2000 and March 2001 Hel beam emission measurements with doped He D and pure He beams were performed using either of the two beam injectors installed at octant 4 80 kV and octant 8 140 kV He beam inj
34. the ne and Te profiles see chapter 4 3 3 Derivation of n and T profile using Zer given and fix 4 Including Ze in the optimisation 4 2 Forward calculation scott y_f wd For the forward calculation i e model calculation of the HeI beam emission with given n Te and Zerr profiles both programms scotty _fwd and yttocs are calling the subroutine forward_calc Their algorithm is based on the original scotty code 32 However some substantial improvements as described in the following have been implemented The f orward_calc routine solves a collisional radiative model for a given He beam into a given D Plasma A short summary of the model is given in chapter 1 3 for more details see 29 The atomic data used by the code are accessed from lookup tables generated by the ADAS 311 code 31 In these ASCII files tables of so called generalized radiative coefficient matrices GRCs are stored for a certain number of plasma densities and temperatures The matrix element M gives the effective change in population of level j out of level i For i the matrix element M gives the loss rate due to ionisation The number of Hel levels to be included can be set for the generation of the lookup tables The other levels are assumed to be in equilibrium with the included non equilibrium levels The original scotty code used GRCs of the size 3 x 3 ground and the 2 metastable states and assumed all other levels to be
35. the neutral beam injector at octant 4 rated for 80 kV was operated with a doped deuterium helium beam at 70 3 kV The beam emission was measured with the KS7 spectrometer in its 6 fibre setup The individual pulses are listed in table X The column Gas puff indicates which gas was puffed into the plasma during the sweep phase The different gases were used in order to change Zef of the plasma The most intensive singlet and triplet Hel beam emission profiles are shown in Fig 2 41 The data are plotted in the coordinate system used for the model calculations which is the distance z along the beam axis starting from the LCFS Wavelength Hel line integrated Se nm 3 Transition asp density oa Zeit 52796 667 8 2P 3 D D 7 0 1 5 52797 4 a CD 7 5 1 9 52798 587 6 2 P 3 D D 7 0 1 7 52799 706 5 2P 3S D 7 2 1 7 52800 388 9 2 s 4 p ui 7 0 1 7 52801 728 1 2P 3 S 7 1 1 6 52802 501 5 2 S 3 P e 7 0 1 6 52803 656 1 Du a 7 0 1 6 52804 587 6 2 P 3 D CD 7 5 1 9 52805 587 6 2 P 3 D CD 8 5 2 0 Tab X List of plasma and spectrometer parameters used for the sweep experiments in Dec 2000 at JET Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 51 JPN 52797 667 8 nm 2 P 3 D 1 4 10 j Pe ee De ip ae apt 0 0001 ies SARE A dl calc scotty_fwd
36. the same trajectories The measurements of the HeI beam emission at the end of the discharges were parasitic However the conversion of one ion source to He reduced the available heating power which could be tolerated at AUG where enough heating power was available Unfortunately these first measurements suffered from problems with the spectrometer and an unfavourable beam geometry which made it difficult to separate the Doppler shifted Hel emission from the unshifted one However a first set of HeI beam emission results could be obtained at 30 kV beam energy Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 10 He doping system at JET octant 4 80 kV For most of the experimental program at JET all the available heating power is needed This led to the development of a so called doped beam in which temporarily a small amount of He gas was added to the deuterium gas fed into the ion source thus producing a deuterium beam with a helium minority fraction 28 This procedure has the advantage of producing a fast helium beam fraction without actually reducing the available heating power and with a negligible He influx to the plasma The pressure increase in the beam box caused by the unused helium gas was tolerable and the additional power load onto the fractional energy dump produced by the residual He atoms in the neutralised beam was within the specification of this dump After tests at
37. to temperature Due to the different collisional life times of singlet and triplet metastable states this feature is lost 2 At higher energies gt 2keV the impact of proton collisions cannot be neglected anymore 45 3 The light collected from a thermal He beam gives a local information directly from the collection volume also called active volume In contrast to this for the fast He beam the emitted light originating from a given position is the sum from processes along the beam trajectory preceding this position To extract the desired density and temperature profiles a non local deconvolution technique must be applied similar to that used for fast Li beam diagnostics 46 Chapter 5 Summary Conclusions and Outlook 110 5 Summary Conclusions and Outlook In order to assess if the emission from fast He beams can be used for electron density and temperature diagnostics of fusion plasmas Hel beam emission experiments have been performed at two relevant TOKAMAK Experiments namely JET Joint European Torus largest TOKAMAK world wide and AUG ASDEX Upgrade Germany by making use of the neutral beam heating systems to generate either a pure He beam or a doped He D beam as neither experiment has a dedicated fast diagnostic beam For the spectroscopic measurements of the Hel beam emission the CX diagnostic systems were used on both machines The diameter of the heating beam defines the spatial resolution of the beam emissio
38. w 1 1210 810 o 2 8 3 3 2 oO S 3 F 8 410 610 Oo O G n a aa a a Fi cai N Be ee en g o o n er ee 5 8 5 610 410 gt u Ren S o h E So Pee Q model parameters 2 o a EEE FR a Zeff 1 9 Carbon onl s 2810 N 210 UNSERER e IEEEBBRR SER SIR HIEEREEREE REE init 2 S pop 1 ai ae enee eaS pop 10 7 0 0 100 0 100 200 300 400 500 dist along beam mm JPN 52804 587 6nm 2 P 3 D 1 4 10 T T T T T T T T T T T i T j T 0 002 T 1 12 10 g JPN 52804 0 0016 z 5 ee oe ee es ee 3 2 2 C ee nr pee ene ee en 3 T 8 410 0 0012 amp oO oO a ze er tee Riesen it ee g Sm Qo MT Nr etf 21 9 Carbon only L 8 5 610 0 0008 gt E S 5 a a 15 2 810 0 0004 0 0 100 0 100 200 300 400 500 dist along beam mm Fig 2 41 He beam emission profiles from a doped 70 3 keV He beam observed with KS7 compared with model calculations scotty_fwd Upper graph JPN52797 Hel singlet line 2 P 3 D at 667 8 nm Lower graph JPN52804 Hel triplet line 2 P 3 D at 587 6 nm Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 52 For these experiments the fibre viewing the range around z 300 mm turned out to be inoperative As a consequence the fibre viewing the range around z 4 0 mm could not be cross calibrated for details on the cross calibration see chapter 2 3 2 The triplet emission was quite strong and peaked
39. will overlap with the unshifted background emission For this reason most of the viewing lines of the lower chords were not suitable for directly measuring a beam emission signal Using the 6 fibre setup had some advantages and disadvantages The switching between the standard setup for CX and the 6 fibre setup is very quick and only requires the turning of a mirror by hand and the changing of the wavelength setting remote controlled to switch over from CX observation to Hel observation and vice versa The spectral range is about 10 times higher compared to the standard setup and thus large enough to measure the Doppler shifted intensity and the unshifted intensity with similar sensitivity 6 fibres are not sufficient for a profile measurement over the whole plasma range Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 38 G 06 0 0 O Opper chords uncalibrated iii moO CCO ower chords calibrated Rmaj m 31 32 33 34 35 360 37 38 39 Fig 2 30 Radial position of the active volumes for the viewing lines of the KS7 diagnostic The solid dots indicate the selected fibres for described measurements Using the KS7 diagnostic for the Hel beam emission measurements implied that the ion temperature Tj of the plasma edge could not be measured For ten pulses dedicated to pellet experiments performed between 18 and 19 10 1999 the measurement of the edge T was n
40. 01 sweep experiment with 135 keV doped and 73 keV pure He beam During the shutdown preceeding this experimental phase the He doping system has been transferred from octant 4 to the octant 8 injector Therefore the doped He beam was operated from the octant 8 injector rated for 140 kV The octant 4 injector had been converted to operate with a pure helium beam Both spectrometers have been used KS5 looking at the doped He D beam from the octant 8 injector and KS7 looking at the pure He beam from the octant 4 injector The plasma discharges were of the same pulse type as in the Dec 2000 campaign a high clearance diverted L mode discharge characterised in table II Compared to the Dec 2000 campaign the sweep range was moved in radial direction in order to keep the strike point on the vertical target which was beneficial for keeping the plasma density constant The pulse list of this shift is shown in table XI pulse gas ae a KS5a doped L KS7 pure ay ies aii a en a Left ee transition ee transition 53872 CD 8 9 2 1 667 8 2 p 3 D 501 5 2 s5 3 p 52873 D gt 11 8 17 667 8 2 p 3 D 501 5 215 31P 53874 CD 9 3 2 1 667 8 2 p 3 D 447 1 2 P 4 D 53875 D gt 10 5 1 7 667 8 2 p 3 D 447 1 2 P 4 D 53876 D gt 10 0 1 8 656 1 Da 501 5 2 s5 3 p 53877 CD 10 0 2 1 653 6 Da 492 2 2 p 4 D 53878 CD 10 0 2 1 468 6 HelICX 587 6 2 P 3 D 53879 D 10 0 1 9 653 6 Da 587 6 2 P 3
41. 2 m al p S K al o E pS to a r D a e r L start end time 55 7650 57 5650 2x1026 change add paran start f it wo E 7 525 A 1 mask break downs fitqual trip chisq p 3000 o ri wl offset 2a ii sese Seso wave length A ero u m set to 1 save sf save ascii ERROR LOG unzoon REPLOT OL_cal CIDL SAVE RESTORE EXIT L 13 410 79 8 S S L Fig 3 3 Screen shot of the main widget of the IDL program he_wi d Loading raw data The framed region in the upper left corner contains the buttons and fields for loading the raw data The appropriate spectrometer can be chosen by operating the upper pull down list button The content of the second pull down list useable beams PINI numbers is automatically adjusted to the selected spectrometer After selecting the spectrometer PINI and pulse number the data access can be started by pressing ENTER pulse field needs to be the active field The following sequence is started loading and displaying of the time trace from the ppf data for plasma position XLOC RBO and beam power of the selected PINI loading raw data of the selected spectrometer call of spec_j pfreader by A Meigs JET applying background subtraction from the raw data applying absolute calibration to the data converts counts into photons Str m s pre setting the start and end time frames to the times of the extreme positions of the plasma h
42. 2001 Fig 3 9 Sensitivity of track7 KS5a displayed with the s hows pec software time wavelength dead pixel and etaloning can be clearly seen The lower two graphs show horizontal and vertical sections through the upper graph taken at the location of the cross The ratio between the measured standard deviation and the value for counts gain yields a wavelength dependent value If the noise would solely originate from photon statistics this ratio should be one and independent of the wavelength One reason why a wavelength dependent factor is being observed could be the continuous exposure of the CCD during the readout shifting of the image across the sensor Normally during the read out of the CCD a shutter is closed in front of the sensor At JET the shutters have been deactivated in order to reduce the minimum repetition time Especially with only a small number of tracks large binning the readout time is very short compared to the exposure time However if the intensities between different tracks are very different the influence of the exposure during readout of the weak track is not negligible Chapter 3 Analysis of the spectroscopic data 89 merge sens files pro The viewing lines of some diagnostic systems are grouped in two different periscopes and have to be calibrated separately This is the case for KS7a and for KS5c The program merge sens files pro helps to merge two sets of calibration files
43. 388 9 and 447 1 nm show a high intensity in Fig 2 31 but a low signal to noise ratio in Fig 2 32 The transmission of the optical system including the optical head 5m Superguide fibre and 70m QSF 600 fibre was calculated from manufacturers data 34 wavelegth dependence of the KS7 system 80 a i zu Te T TTY 0 8 ia transmission peas re va ee k of KS7 optics ee f gt a 5 Z 60 a a A 3 O oO o 7 S z 3 3 5 40 T hr oO z E E 3 I Quantum Efficiency 3 20 6j H a 0 0 200 300 400 500 600 700 800 wavelength nm Fig 2 33 Measured quantum efficiency of the KS7 CCD sensor and calculated transmission of the optical system Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 41 Influence of the He doping on the global Zer of the plasma These first measurements using a He doped beam with 500 ms of He puffing into the beam source showed that the pressure in the beam duct stayed well below the limiting pressure and that the reliability of the doped PINI 6 was unaffected by the addition of helium The doped He D beam produced an increase of the passive Hel line emission from the plasma edge Fig 2 34 The red peak is the Doppler shifted beam emission which appears during the He doping The passive emission shows a sharp rise during the He doping to twice the value and a slow decay over a number of seconds after the end
44. 7 8 2 5 0 476 4 99 5 09 10 ITB 13742 3 1 6 0 64 5 38 4 18 10 Table VII Main plasma parameters of two further AUG pulses with He beam emission data The killer pellet pulse is a pulse driven towards a high density disruption and then terminated by a killer pellet The increase of the density starts at 2 9 s termination with the pellet occurs shortly before 6 seconds Helium for the doped beam was injected into the ion source before the density rise at 2 5 s At his time the discharge is a flat top H mode discharge without additional gas fuelling and consequently very similar in density to the first phase of the standard H mode pulse The second pulse type was a so called ITB discharge A characteristics for this type of discharge is that additional heating is already supplied during the current rise phase At the time of the He injection 1 6 s there is no additional gas fuelling The density is lower than in the case of the standard H mode pulse while the stored energy is higher Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 31 Density profiles Density profiles can be obtained from different diagnostics The signal DCN L1 from the ddf database is a combination of the profile obtained from several laser channels of the Interferometry DCN and the profile delivered by the Li beam diagnostics limited to the scrape off layer The profile is smoothed
45. 7 for the T parameter Comparing the convergence plots in Fig 4 5 without damping and Fig 4 6 with damping clearly shows the improved convergence achieved by the damping Apart from a faster convergence the damping also results in better fits as can be seen from the smaller total deviation in the case of damping or from the better profile match between fitted result and actual input profile In the case of a calculation with damping the temperature profile is correct in the range where the beam emission data are sufficiently large On either side of the beam emission profile of the triplet line the data converges towards zero and the fit returns very high temperature values This unrealistic high temperature at 800 mm pushes the returned density up beyond its correct value Description of the yt t oc s output sheet Fig 4 5 4 6 The output sheet stored by ytt ocs contains five plots and a text section The two plots of the first row show the input profiles n and T The dotted lines show the start profiles at the beginning of the optimisation the dashed lines the artificial input profiles used to generate the data and the solid lines the result of the optimisation The knot points defining the parametrised profiles are shown as diamond symbols in the same plots The two plots of the second row show the two emission profiles both the assumed measured data diamonds and the results of the fit solid lines The
46. 9 central el temperature 1 5 2 keV 53873 D 1 7 11 8 toroidal field 2 4 Tesla 53874 CD 21 93 Plasma current 2 5 MA 53875 D gt 17 105 add heating 3 MW on 2 n Table II Main plasma parameters of the ar ae me JET discharge used for the He beam ll a a 109 emission experiments 53879 D 1 9 10 0 53880 D 1 9 10 0 53881 D 1 9 10 0 53882 D 1 9 10 0 Table III Line average density in 10 m and average impurity level Ze for the pulses of our second measuring campaign Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 22 One problem with the first plasma sweep pulses was caused by dropping of the plasma density when the strike point passed over the gap between horizontal and vertical target in the divertor see Fig 2 16 This drop in density can be explained by a accordingly stronger pumping The plasma density variation during the sweep could be kept below 5 by limiting the sweep to a range where the outer strike point stayed on the vertical target tiles The measurements were performed with a series of nominally identical pulses However the Zerr Was varied by injecting either deuterium or CD into the plasma Table II shows the main plasma parameters of the discharge and table III shows the reproducibility of the plasma density when Zef was varied Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 23 2 3 2 Cross calibration of t
47. DISSERTATION Towards fast He beam edge plasma diagnostics ausgefiihrt zum Zwecke der Erlangung des akademischen Grades eines Doktors der technischen Wissenschaften unter der Leitung von o Univ Prof HP Winter E134 Institut fiir Allgemeine Physik eingereicht an der Technischen Universitat Wien Technisch Naturwissenschaftliche Fakult t von Dipl Ing Michael PROSCHEK Matr Nr 9025445 Br uhausgasse 11 15 A 1050 WIEN Wien am 21 Dezember 2001 Kurzfassung Die genaue Kenntnis der Randschichteigenschaften ist f r die Entwicklung von Hochtemperaturplasmen mit reaktorrelevanten Parametern von entscheidender Bedeutung Dabei werden hohe Anforderungen bez glich r umlicher und zeitlicher Aufl sung gestellt um steile Dichteprofile von H mode Plasmen und sogenannten Advanced Scenarios r umlich aufl sen zu k nnen Eine gute Zeitaufl sung ist notwendig um Instabilit ten in der Randschicht ELMs und Fluktuationen untersuchen zu K nnen Aktive Spektroskopie mit injizierten Atomstrahlen hat sich in vielen Bereichen der Plasmadiagnostik als fundamentale Untersuchungsmethode f r mehrere Plasmaparameter etabliert In der vorliegenden Dissertation wird die Entwicklung einer aktiven Strahlemissionsdiagnostik mit einem schnellen neutralen He Strahl beschrieben und deren Eignung als Dichte und Temperaturdiagnostik untersucht Der zentrale Teil der Arbeit besch ftigt sich mit orientierenden Messungen die an zwei der f
48. Furthermore the smaller divergence of the He beam leads to an even higher sensitivity to misalignment AUG 13647 killer pellet pulse 2 P 3 D 667 8 nm OFT corr 0 15 0 75 0 35 oP 4 00 0 0750 Ppp E F straight data J BD 03 ee ee been ge pop T J gt u i 4 Et a 5025 een a ak pa ara Var cm 4 c C al fa i i p L E 02 een ge A Ye Ee ae i E oO I i g N eo A i A 50 15 po en nn Sa a Ma 7 H v oOo Ari yw e 0 1 E ee eee ete penne Zee St ere SE a a 0 05 C 1 L L L 1 3 L 1 L L l L 1 N 1 A N 1 4 L L 1 1 6 1 7 1 8 1 9 2 2 1 2 2 2 3 R maj m Fig 3 12 HeI beam emission profile 667 8 nm 2 P 3 D of a doped He D beam The straight data and two corrections with different parameters are shown Raw data are the output data of the cer wi d program These are already processed data but without the correction for the beam alignment Chapter 4 Reversion code yt tocs 94 4 Reversion of Hel beam emission profiles into density and temperature profiles code yttocs In this chapter the algorithm of the reversion code yttocs and some first results are presented The code extracts desired electron density and temperature profiles from a set of Hel beam emission profiles The inverse problem modelling of the beam emission from given density and temperature profiles is calculated by the code scotty therefore the
49. I peak can be easily recognised and separated from other emission in the close spectral vicinity In both machines it has been observed that the HeI beam emission appears in the spectra as clean Doppler shifted peak and is thus largely undisturbed by impurity peaks for most of the wavelengths For the doped He D beams only one singlet 2 P 3 D and two triplet lines 2 P 3 D and 2 P 3 S could be measured with sufficient signal to noise ratio On either end of the visible range the sensitivity of the CCD sensor was low yielding a weak signal in spite of a relatively intensive beam emission e g 2 S 3 P at 388 9 nm With the pure He beam also the emission profile of the 501 5 nm line 2 S 3 P could be measured At AUG Hel beam emission profiles for the two strongest lines transition 2 P 3 D at 667 8nm and 2 P 3 D at 587 6 nm were obtained for four Chapter 5 Summary Conclusions and Outlook 111 different discharges Repetitive measurements allowed additionally to estimate the scatter in the data to 10 A systematic deviation of some channels indicates a non perfect calibration which was partly due to an unfavourable alignment of the CER viewing lines to the beam axis At JET singlet and triplet HeI beam emission profiles with good spatial resolutions could be obtained as a result of dedicated experiments using radially swept high clearance L mode plasmas The problem of inaccurate calibration of the spectroscopic system
50. The parameters are displayed along the beam taking the plasma movement into account For the time being the mapping is approximate in the sense that the shape of the flux surfaces is regarded as fixed This is the case for the slow sweeping experiments The radial positions of the different active volumes fixed in space are supplied by other programs and stored in tables The relative plasma movement over time is represented in an imaginary movement of the active volumes of the different tracks the plasma is regarded as fixed in space The relative movement is given by the relative change of the major radius of the plasma boundary at the mid plane of the machine XLOC RBO The active time is taken for the time point where the mapped values equals the actual positions of the active volumes F Doppler shift lf rel hight Doppler peak I FIUHM Doppler peak FF Intensity Doppler peak FF Intensity PRIM peak F Int Doppler peak integr 0 K Fig 3 4 Widget for choosing parameters to display Chapter 3 Analysis of the spectroscopic data 79 Choosing parameters The selection of the displayed parameter is done via a separate widget which can be called by pressing the par_Ist button i e parameter list button Fig 3 4 shows the widget with the list of parameters to choose from All parameters but the last are fit parameters The last one intensity Doppler shifted peak integrated is given by integration su
51. a factor of 2 8 higher than for the 2 S 3 P line 501 5 nm Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 59 singlet Hel beam emission 726 keV pure He beam 8 10 Tit mrp aa ririt Fi a a Fri 0 0001 z 710 g D 5 5 3 s 610 2 5 o 8 510 s S ao S E 410 a co 3 5 310 D g S 2 2 T g 210 5 5 c E 1 1 3 S 105 667 8 nm 2 P 3 D clac 7 g E 2 JPN53881 norm 0 L 1 i 1 i ib L L i 1 i i 1 i 1 1 i i i ii 1 i 1 i 1 ii 1 i L 0 0 50 100 150 200 250 300 350 400 dist along beam mm Fig 2 49 Comparison of the 2 S 3 P singlet line profile at 501 5 nm with the profile from the 2 p 3 D transition 667 8 nm The plasma discharges are comparable and for both discharges a pure He beam was used The triplet Hel beam emission 2 P 3 D at 587 6 nm of a pure He beam is plotted on a logarithmic scale in Fig 2 50 As this was the strongest obse
52. am injector Situation at ASDEX Upgrade AUG At AUG two injectors with 4 ion sources each operating at 60 kV SE injector or 100 kV NW injector extraction voltage are installed Pure He beam at AUG 30kV At AUG the beam box is equipped with ion getter pumps therefore He is not being pumped during the pulse and therefore only short He beam pulses are possible We did our first Hel beam emission measurements in August 1999 at AUG with ion source Q4 from the SE injector converted to pure He operation The duration of the He beam pulse was limited to 300 ms and placed at the end of the heating phase to keep the duct load to a minimum After neutralisation the remaining ions are swept out of the beam by a bending magnets The configuration of the bending magnets in the AUG injector provides the same bending power in all four magnets In case of one He 4 amu beam and 3 deuterium beams 2 amu the extraction voltage of the He beam had to be set to 30 kV halve the value of the D beams in order to divert the He beam properly into the ion dump U gt Mye ZU p Mp When operating the beam system with D2 a small fraction of singly charged molecular ions can be left in the beam after the neutraliser These ions are deflected with 2 the bending radius compared to D ions ending up on the so called fractional energy dump designed to take up the power load of these molecular ions He ions with the same energy as the D ions follow
53. and also hinder the convergence A good description of the electron density appears to be a spline curve plasma core joined to an exponential decay plasma edge For Te and Zr continuous spline curves or stepwise linear functions see Fig 4 2 are suitable The code is written in a way to make the addition of new profile parametrisation quite simple When choosing the parametrisation one has to be aware of the fact that it defines the possible profile shapes but also may influence the convergence behaviour Chapter 4 Reversion code yt tocs 99 Te_profile 2000 p start profile target profile 3 fitted profile v 5 STEPWISE LINEAR AP 5 g monotone e 1000 v 5 Piot D v v 0 L L 4 J lL L 1 it L L 1 ji L 4 4 gt 200 0 200 400 600 800 Xo X X2 X3 nak Rye x dist along beam mm Fig 4 3 Illustration of the convergence problem due to parameter restriction Left Parametrisation for monotonously increasing function stepwise linear Right An illustration of the resulting convergence problem The dashed line shows the target profile the dotted line the start profile of the fit procedure and the upper solid line the converged result A typical convergence problem due to parameter restriction is shown in Fig 4 3 On the left side the parametrisation of a stepwise linear function is sketched The positions Xo Xp 2 are given and the parameters Po Py are restricted to positive values allowing only monoton
54. and regarded as most reliable Thomson scattering and reflectometry are other diagnostics yielding density profiles Not all diagnostics are available for every pulse and the range of some diagnostics can be limited For some pulses the Thomson scattering system was moved to view the divertor and was therefore not available for evaluation of the core plasma The density profile can normally be well presented by 1 for P lt p an p normalised flux radius n p n x 1 ap x A Joroze 3 P radial position of step in profile as Fig 2 24 shows by using interferometric data AUG 13646 standard H mode density profiles DCN Li 1 4 10 m Ss 1 210 H n 1 m 3 e 19 DCN Li 5 98 lt t lt 6 2 s fit 0 0 2 0 4 0 6 0 8 1 1 2 Fig 2 24 AUG standard H mode Density profiles at 3 6 and 6 s taken from DCN and Li beam diagnostics at time intervals with He doped beam Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 32 Temperature profiles AUG 13646 standard H mode electron temperature at 3 6 s At AUG the main diagnostics 3500 FT Ts for electron temperature 3000 profiles are Thomson scattering and the ECE 2500 radiometer From Thomson 2000 scattering also a fitted profile Te eV called Thomson fit is stored in the AUG ddf L database When avai
55. around z 60 mm Ring 3 65 m For z gt 300 mm the signal was below the noise level The singlet emission was by one order of magnitude lower in intensity than the triplet emission even for the most intensive line but could nevertheless be detected over the whole range covered by the spectrometer triplet emission profiles 3 10 l l 2 8 10 2 25 10 2 110 1 5 10 1 4 10 7 510 beam emission JPN52798 587 6 nm wu 9 907 66LZSNdf uoissiwa wesq 50 0 50 100 150 200 250 dist along beam mm Fig 2 42 Smoothed HeI beam emission profiles from two different triplet lines measured in successive basically identical discharges The shape of the profiles is identical but the intensity differs by one order of magnitude The pulses JPN52798 and JPN52799 were almost identical with only a small difference in the line average density 7 0 and 7 2 10 m The measured He beam emission profiles of the 2 P 3 S and the 2 P 3 D lines at 587 6 nm and 706 5 nm are basically identical in shape From this we conclude that no additional information is to be gained from measuring the second triplet line profile The intensity of the two profiles differs by a factor 10 7 Fig 2 42 nevertheless the amount of scatter is the same for both lines in spite of the large difference in intensities This indicates that at least in this case the scatter is caused by the plasma turbulence rather than
56. axis Neutral Chapter 3 Analysis of the spectroscopic data 92 beams at a sufficiently large distance from the beam source are normally well described by a Gaussian curve and the correction factor is therefore 2 _4 min c e withA 1 tano 14 v l is the distance between beam source and cross over point and is the divergence angle of the beam The correction is required as the individual lines of sight are staggered in z direction and thus have an alternating distance from the beam axis leading to a zigzag profile The exact value of the correction is critically dependent on the quality of the alignment The beam alignment is normally different from the optical alignment of the ion source due to deformation of the vacuum housing sagging of the accelerator grid support bending and thermal expansion of the grids and deflection by earth magnetic or stray magnetic fields Therefore the accurate alignment of a powerful neutral beam can only be achieved by using the beam itself and it is limited by the quality of the beam diagnostics AUG 13727 c CX emission 529 05 nm 6 5 A 4 2 T c 3 2 wn 2 E x 2 O 4 0 1 6 1 7 1 8 1 9 2 2 1 2 2 R maj m Fig 3 11 The zigzag of the data retrieved with the cer wid evaluation software disappears if the intensities are corrected by the beam intensity at the crossover between beam and viewing line Assumed beam divergence 0 9
57. ay of the initial metastable population can be seen One possible reason for the slower decay in the initial metastable fraction could be an error for the calculation of the distance Scotty_fwd uses an Chapter 2 Discussion of the Results 68 optimisation of its step width and an error in the distance calculation could arise from this complex procedure For testing the code beam emission profiles for given density and temperature profiles with periodic squares were calculated see Fig 2 56 The resulting populations of the different levels show the expected behaviour Furthermore the numerical stability of the code has been tested Therefore it is likely that the deviation is due to a problem in the atomic data base 10 o ax oO ne 1 m zei oO SI 2500 2000 1500 Te eV 1000 oO level population uonejndod 343 0 001 dist mm Fig 2 56 ne and T Testprofiles periodic square profile and the modelled population of some Hel levels The resulting emission profiles show the expected spatial correlation indicating correct calculation of the output distance Scotty_fwd uses look up tables generated by a modified ADAS 311 code whereas the first version of Scotty used look up tables generated by the 311 code of the official ADAS release Scotty_fwd uses the same atomic data but the
58. be introduced for other configurations then the ones used for the calibrated systems This was the case for the He beam emission measurements A clear deviation was observed for two lines of sight The corresponding beam emission signal was too low for all measured wavelengths This can be seen in the triplet emission profile of Fig 2 38 where the two tracks at the radial position Rmaj 3 78 m and 3 82 m appear as clearly too low JPN 49503 587 6 nm 2 P 3 D 0 0016 HH 110 hi oe eee e JPN 49503 e ISBN E E EE URAN 2 E _ 0 00128 s10 8 2 DR io eet eee SE CO merce 3 SD PETER Le E EEEE Seen Aenean ERSERDER Renee meee 5 s oO 2 i 0 00096 610 lt BT rs a E E E a E E a 8 B E o oo P a iam aa a 8 0 00064 410 S 5 3 nn i OBERE Oe es es es res Q oO 7 a 15 0 00032 240 0 fe se E i iii een ame i N o fis i iii 0 385 38 3 75 37 4365 36 355 3 5 R m maj Fig 2 38 Beam emission profile of the 587 6 nm He I line 2 P 3 D measured during the pulse JPN 49503 The intensities of the two tracks at the radial position Rmaj 3 81 m and 3 82 m are too low indicating an incorrect calibration 3 cross calibration of new upper chords In order to get signals from further inside the plasma and to bridge the Doppler shift gap of the lower chords also fibres from the new upper chords have been used In Fig 2 30 one can see tha
59. beam emission measurements on large and medium size fusion experiments 43 Hel beam emission profile measurements using 16 fibre setup After this first successful test of the Hel doping system on JET and the assessment of the different Hel line intensities emission profiles for the three most intensive lines have been measured In order to obtain a better spatial coverage for these measurements KS7 was used in its standard setup with 16 fibres Tab IX shows a list of these pulses with doped He beam injection parasitic beam emission measurements JET pulse Info Hel wavelength Pyer Prr lt n dl gt transition nm MW MW 10 1 m2 49225 2 P 3 D 587 6 3 2 z 18 6 49226 2 p 3 D 667 8 3 2 F 29 49227 2 P 3 S 706 5 3 2 24 49503 H mode 2 P 3 D 587 6 15 1 2 73 49504 H mode 2 P 3 D 667 8 14 7 0 97 10 49555 modulation 2 P 3 D 587 6 12 1 1 0 8 9 Tab IX Pulses with KS7 in standard setup 1 19 11 1999 The analysis of this data shows that the reduced bandwidth of KS7 in its standard 16 fibre setup severely limits the quality of the measurement as further detailed below Limitations of the KS7 spectrometer in standard mode 1 Non uniform magnification A detailed description of the KS7 spectrometer is given in chapter 2 2 2 and 33 34 It basically consists of two Czerny Turner spectrometers in series with a 4x4 array of entrance s
60. calibration IDL save Will save actual program memory not yet working RESTORE Will restore saved program memory not yet working EXIT Kills the main widget and ends the main program Chapter 3 Analysis of the spectroscopic data 81 Start Parameter For the fit procedure an initial guess set of start parameters is required The start values for the peak positions are given by the unshifted wavelength of the transition in question wavelength setting of spectrometer and the Doppler shift which is calculated from the geometry of the observation system and the He beam energy The peak height is estimated by finding a local maximum in the vicinity of the estimated peak position The peak width is given by a constant value based on experience Parameter widget The start FIT parameter entry BEE parameter for additional ire Te disturbing peaks are entered oaf interactively By clicking the a change add param button in the main widget a new widget appears showing the spectrum of Dae the first active track at the active 0 7050 7060 7070 7080 7090 time see Fig 3 5 By clicking wave length into the plot on the position of the S hai disturbing peak dragging the F poso 1 ao cursor to the flank of the peak at vr Ss 2 p 200 halfe maximum and releasing the 4 Bm 3 5 00 mouse button an additional B c000 0000 Gaussian distribution is added and a on s isco the position and FWHM are de
61. cted The main advantage of the global approach is that it is easier to incorporate global properties as the initial metastable population in the beam or the Chapter 4 Reversion code yt t ocs 98 impurity profile in the plasma That was the main reason for using a global approach for the development of yt tocs The function prof_funct calculates the profile defined by the fit parameters varied and matched to the data by the fit code and some fixed parameters The latter are passed via common block variables The fixed parameters are the name of the defining function e g spline exp and the positions of spatial fixed points e g spline knots A I I Ax SPLINE STEPWISE LINEAR Fig 4 2 Examples for the parametrisation of the profiles P are the parameters changed by the fitting procedure In both examples they are restricted to positive values The x are the fixed spatial positions of the knot points Left SPLINE curve plasma core joined to an exponential decay plasma edge is used to describe the electron density profile Right Stepwise linear profile with positive values only but no further restrictions is used to describe the electron temperature profile Part of the development was to define a suitable parametrisation compromising between excessive freedom and too many restriction Excessive freedom can lead to unphysical solutions too many restrictions can suppress features of the profile
62. ctions of all transitions from levels with n gt n which means that these levels are treated as being in equilibrium with the levels up to ns So far this modified ADAS 311 code constructed by H Anderson 31 can not be run at TU Vienna due to incompatibilities between the sun UNIX used at Strathclyde and the LINUX system used at TU Vienna The lookup tables ASCI files were generated by S Loch in Strathclyde and subsequently transferred to the LINUX computer at TU Vienna For a certain beam energy Ep the GRCs are stored in lookup tables for a list of electron temperatures 7 and densities ne covering the parameter range of the plasma that is going to be analysed The stepwise calculation of the populations by using the GRCs matrix M is given in Equation 2 Chapter 1 Introduction 7 N x dx N x M E n T N x nel 4 2 V N and N denote the population densities of levels i and j respectively and ne the electron density at the position x The step width dx necessary to obtain a numerical stable solution is typically 0 1 mm for n 3 11x11 matrices For increasing n and higher velocities v of the beam the step width has to be reduced From each location along the beam axis the matrices with the next higher and lower density and temperature are located and the relevant GRC is created by logarithmic interpolation in the case of density and by linear interpolation in the case of temperature The typical siz
63. d by the KS7 system is small for most lines of sight the beam emission peak was overlapping with the passive Hel emission peak For these tracks either the fit procedure failed or the uncertainties of the fit results were large The data was therefore additionally analysed with a different procedure A so called time trace procedure integrates the spectrum over the spectral range of the Doppler shifted peak and analyses its variation with time In this procedure it is important to apply the correct time dependent background subtraction An interpolation between the spectrum recorded just before and the one just after the He doping could be used for the background subtraction because the He doping was relatively short 500 ms and took place during the flat top phase of the plasma discharge However a linear interpolation over time is only meaningful as long as the unshifted peak is outside the integration range As seen in Fig 2 34 the intensity of the unshifted peak increases rapidly at the beginning of the He injection and shows only a slow decay after the He injection Assuming that the increase of the passive He emission is caused by the He gas influx from the neutral injection box NIB the time dependence of the unshifted background emission can be modelled using a constant flow rate while the valve admitting the helium into the beam source is open and an exponential decay in the helium flow when the trapped volume discharges into the beam s
64. dx Z 2 OV ee Parga N j z L OV p aepo Parga N j L A N u L AN 1 j i j i j j gt i J J lt i N and N denote the population densities of level i and j respectively v the velocity of beam particles lt ov gt the rate coefficient of the collisional processes Marget the density of the target in this case plasma electrons or protons and Aj the spontaneous emission coefficients for the emission from levels j to levels i All excitation de excitation and loss processes can be included into one matrix for a given set of parameters ne Te Ep Zett The coefficients of this collisional radiative matrix Mj are called generalised collisional radiative coefficients GRCs ADAS 311 calculates this GRCs up to high level numbers including cross sections from the ADAS data base and numerical approximations 30 The ADAS code subsequently merges this big matrix into a more convenient 3x3 matrix for the transitions between the He ground state and the two metastable levels 2s S and 2s S in Fig 1 2 which covers the influence from the excited levels This code provides the matrices in the ne vp plane which is not suitable for beam emission spectroscopy with a fixed beam velocity v but variable plasma temperatures Te We used a different version of the ADAS 311 code not released which is able to merge the GRCs into an arbitrary matrix size n Xn and presents the matrices in the n T plane The condensed GRCs contain proje
65. e y 1 1515e418 eX 0 025567x R 0 99448 14 L L ji ji L L L ji L L L 1 L 1 L L 10 50 100 150 200 250 300 350 dist along beam mm Fig 2 50 Triplet Hel beam emission 2 P 3 D at 587 6 nm for injection of a pure He beam JPN 53879 plotted on a logarithmic scale The solid line is an exponential fit to the decay inside the plasma Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 61 2 5 4 Conclusions from the Experiments at JET As result of the radially swept high clearance L mode plasma experiments singlet and triplet profiles could be measured with good spatial resolution The results were also suitable for the development of a reversion code The problem from inaccurate calibration of the spectroscopic systems could be overcome by in situ cross calibration of neighbouring channels The Hel beam emission appears in the spectra as s clean Doppler shifted peak undisturbed by impurity peaks for most of the wavelength region For the doped He beams only one singlet 2 P 3 D and two triplet lines 2 P 3 D and 2 P 3 S could be measured with sufficient SNR The SNR for lines on both ends of the visible range were low due to a low sensitivity of the detector Especially emission at the 388 9 nm line 2 S 3 P is high but it shows only a weak signal With the pure He beam more suitable line transitions were accessible and an emission profile for the 501 5 nm line 2 S 3 P cou
66. e beginning of October 1999 for details see chapter 2 1 The doped He D beam was operated close to the nominal extraction voltage of the octant 4 injector of 80 kV and He was injected into the ion source for 500 ms Wavelength scan of Hel lines 18 19 10 1999 The aim of the first series of measurements was to asses the signal intensities of the modelled n 3 gt 2 and n 42 Hel lines and to investigate whether the He peak can be easily recognised and separated from other emission in the close spectral vicinity For observation of the Hel beam emission from the He doped beam at octant 4 the KS7 spectrometer CXRS was used Its observation system provides more fibres then the 16 used for the CX measurements Six spare fibres with suitable viewing lines were connected to the single entrance slit of the 6 fibre setup of the spectrometer details see chapter 2 2 2 The radial positions of the active volumes cross section of the beam and respective lines of sight of the viewing lines from the upper and the lower chords are given in Fig 2 30 The positions corresponding to the six selected fibres are marked by solid dots The expected Doppler shift for the different viewing lines is indicated by the grey triangles No Doppler shift occurs at a major radius Rmaj 3 4 m viewed by the upper and at Rmaj 3 75 m viewed by the lower cord Within a radial range of about 100 mm of these positions the Doppler shifts are too small and the Doppler shifted peak
67. e next step a triplet emission profile e g 587 6 nm would lead to a n profile which is a weighted combination of both results In some cases the convergence of the whole problem is improved in that way However it is important to choose the right weights A too small weight can lead to a false convergence too close to the start values and a too high weight leads to oscillations The code can deal with all possible combinations and permutations To make the idea behind it clearer three examples are shown below 1 Derivation of the electron density by fitting one emission profile can be realised by defining the following sequence loop_seq ne level seq 1 weight _seq 1 The electron temperature is given and remains unchanged 2 The electron density can be also derived by including the information from two emission profiles with different weights and leaving the given temperature profile unchanged This would be realised with the sequence loop_seq ne ne level seq 1 2 weight _seq 1 0 4 3 The standard sequence used for the analysis of the measurements loop_seq ne te level_seq 1 2 weight _seq 0 8 0 7 Chapter 4 Reversion code yt t ocs 106 4 5 Factors influencing the accuracy of the result We have to distinguish between uncertainties of the atomic data set and errors of the measurements This chapter focuses on the latter 4 5 1 Absolute Calibration The yttocs code pro
68. e of the ASCI files generated by ADAS 311 is depending on the matrix size 1 to 10 Mbytes It contains approximately 500 matrices linearly spaced on a logarithmic scale singlet triplet 4 P 4 D 45P 4 D oe 2 4 1 aal 13 Fig 1 2 Level diagram of Hel states with quantum numbers n lt 5 and related photon emission in the visible range The original scotty code 32 used for the modelling of the beam emission in the work of S Menhart 29 involved a model treating the ground and metastable states as mutually independent This simplified model becomes less well applicable with increasing beam energy The code solving equ 2 has therefore been rewritten in IDL scotty_i d to extend the number of independent states to a principal quantum number ns where n can be freely selected Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 8 2 Hel beam emission measurements on medium size and large fusion experiments 2 1 Helium doping system for heating beams Due to the lack of dedicated fast diagnostic beams at AUG and JET the D heating beam injectors were utilised in both cases to generate fast He beams Operating a neutral beam injector with He gas leads to an increase of the pressure in the beam box because He is not pumped by the ion getter or standard cryo pumps used in the beam injectors in order to pump the gas outflow from the neutralisers The pressure
69. e sensor is still at the right location In these cases it is not possible to find a spectrometer setting suitable for all entrance slits The situation is even worse for the Doppler shifted line of the beam emission E g the grey shaded peak in Fig 2 37 is situated in a position at the wing where the transmission is less then half the flat top value Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 45 800 600 400 200 o o 20 40 60 B8 1 8 1200 1000 800 600 400 200 o 20 40 60 8 1 8 800 600 400 200 o o o 20 40 60 8 1 8 3000 2500 2000 1500 1000 500 ra o O 20 40 60 80 100 JPN 49227 o N o gt o S o 8 800 600 400 200 o 20 40 60 8 8 3000 2500 2000 1500 1000 500 o 20 40 60 8 8 KS7 Spectrometer raw data A 706 5 nm 600 1000 800 so 600 200 400 200 0 e O 20 40 60 80 100 0 20 40 60 80 100 1200 1200 1000 1000 800 800 600 600 400 400 200 200 o o 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 O 20 40 60 80 100 500 800 o9 600 300 a 5 400 200 8 so 200 o o 0 20 40 60 80 100 0 20 40 60 80 100 pixel CCD design wavelength 529 1 nm Fig 2 37 The 4x4 spectra solid lines and respective transmission functions dotted lines The centre wavelength is 706 5 nm The position of the unshifted peak equal to the centre wavelength varies with
70. ection has been performed from one beam source only either PINI 6 in octant 4 or PINI7 in octant 8 During these He beam emission experiments the He PINI doped or pure was the only beam source used in the respective injector By restricting the number of beam sources it was possible to operate the doped deuterium helium beam for up to 6 seconds without exceeding the limiting pressure in the beam duct The long pulse duration was required for experiments with plasma sweeping cf chapter 2 3 A projection of the beam axes of the octant 4 injector onto the vertical plasma cross section is shown in Fig 2 2 together with the viewing lines of the KS7 diagnostic Experiments with pure He beams could be performed during a dedicated helium campaign providing measurements with higher He beam currents and therefore higher intensity of the Hel beam emission Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 12 2 2 Spectroscopy At both experiments for the HeI beam emission measurements the spectroscopic system of the respective charge exchange diagnostic CXRS was used The CXRS diagnostic monitors visible charge exchange lines from impurity ions Derivation of the ion temperature impurity concentration and collective velocity of the ions is achieved from the width height and Doppler shift of the emitted spectral line respectively In most cases the 529 1 nm line emitted after electron capture from the fast beam pa
71. emission profile 2 P 3 D can be characterised by a sharp intensity increase near the plasma edge and a much slower decay towards the centre Only in the range with the large gradient we see a small drift of the Doppler shift Sufficiently deep inside the plasma the emission profile is almost constant and the Doppler shift stays unchanged during the sweep For a beam energy of 134 keV and an observation angle 30 a change in the observation angle of 0 5 would lead to a change in the Doppler shift of 0 05 nm for the A 667 8 nm line The observed drift of 0 1 nm is therefore consistent with a very steep increase of the emission profile Chapter 3 Analysis of the spectroscopic data 84 2 Disturbing peaks Some of the observed drifts of the peak position are too high to be explained by the influence of the beam divergence In some of these cases an overlapping disturbing peak of low intensity could be identified If the fit procedure only interprets the two overlapping peaks as a single one the spectral position of the compound peak will depend on the relative intensities of the two peaks During the sweep of the plasma the disturbing peak changes the intensity differently to the beam emission peak leading to a drift of the compound peak This can be overcome by adding the disturbing peak to the fit function used in the he wid software However if the peaks are too close together this fitting proced
72. emission profiles which were forward calculated from the start profiles are plotted as dotted lines The bottom plot shows the convergence behaviour by giving the total deviation between fit and measurement for the iterations of the outer loop The abscissa gives the number of f wd_ cal c calls Chapter 4 Reversion code yt tocs 103 ne_profile Te_profile Ago a 3000 Po J 6x10 9b E 2000 4 v usa 5 D 9 d 2 axi0 9b z D 2 5 5 3 8 1000F 4 o r v E 4 2x1019 OL oly eb pb 200 0 200 400 600 800 200 0 200 400 600 800 dist along beam mm dist along beam mm Hel emission 1 measured sim 3 1 D Hel emission 2 measured sim 3 3 P Tg ET ar u Fa man rem Du an aa aa zer Dana 5F E E 40r AF S E 3o 5 E 5 m s of gs E Be SE ef 5 20F 5 E T E 2 E 2 E 25 E E v H o L E J E 10F E J ze Ir 4 Oo E E g 1 E E J op OF amp E F F gt E Sn eh E S E E EE E10 Bar ep a E ae ee re i 200 0 200 400 600 800 200 0 200 400 600 800 dist along beam mm dist along beam mm deviation data fwd calculation result of fit CS ee F start situation 1 artificial data F 1 initial population 2 singet S 1 0 _ 0 04 H 4 Be initial population 2 triplet S 10 a f J 2 ooz3b J mean value Zeff 2 0 5 F 5 F core value Zeff 2 0 3 F 1 gt 0 02F 4 E AN u J resolution of measurement 5 0 O
73. emperature at the plasma edge via C charge exchange 33 Two so called optical heads are situated in vertical ports above and below the beams of octant 4 see Fig 2 6 The original design of the optical head 34 consisting of a lens system and a fibre plate was designed to provide a fan of viewing lines covering the plasma edge with good spatial resolution gt 20 mm The optical head of the upper port has been redesigned to cover the full range of the plasma from the edge to the core with a higher light throughput but lower spatial resolution gt 60 mm These viewing lines are used for measuring the collective velocity of the plasma plasma rotation from the Doppler shift of the CX signal fibers Octant 4 upper chords fiber panel Fig 2 6 A sketch of the KS7 diagnostic with its main components optical head fibre panel and the two spectrometers The light collected by the optical head is transmitted through optical fibres to the spectrometer situated outside the biological shield A special feature of the KS7 system is a fibre panel where a subset of fibres can be easily selected and plugged into the panel Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 16 In normal operation of the KS7 diagnostics two Cerny Turner spectrometers are used in series in the following called standard mode see Fig 2 7 The first spectrometer acts as a band bass filter and the sec
74. energy into account The actual value of the Doppler shift converged value of the curvefit parameter can change during the sweep of the plasma There are two reasons for the drift of the Doppler shift 1 Beam divergence The finite beam divergence typically 0 7 leads to a weighted active volume sum of different Doppler shifts In Fig 3 6 this situation is illustrated The angle between line of sight and the trajectory of the beam particle varies along the viewing line passing through the beam In case the beam emission is almost constant within the active volume the resulting beam emission profile Doppler shifted peak is only broadened but not shifted In case Fig 3 6 Sketch of the influence of the finite beam the beam emission has a steep divergence on the net Doppler shift gradient at the position of the active volume one side of the beam has a higher weighting due to higher emission intensity In Fig 3 6 the beam emission is symbolised by the red shading In the displayed case the emission is dominant in the upper region smaller angle and the observed Doppler shift would be bigger then the expected beam axis value In Fig 3 7 the measured beam emission profile upper plot and the respective Doppler shifts lower plot from a sweep experiment are displayed In the Doppler shift plot the different tracks corresponding to different viewing lines can easily be identified by their Doppler shifts The shown singlet
75. entry of the beam into the plasma is called the initial metastable population The initial beam composition is of large influence on the intensity as well as on the shape of the beam emission profiles This is due to the fact that the metastable population of a fast neutral He beam decays by orders of magnitude from the initial population to the equilibrium value inside the plasma The decay length is typically of the order of 100 mm Fig 4 7 shows the beam attenuation modelled for a JET discharge and its decay lengths are approximated through exponential fits One can see that an equilibrium for the 2 S state is only reached at about 200 mm inside the plasma For the correct ne and T deduction an accurate knowledge of the initial metastable population is important However at the moment there are no independent measurements available under the conditions of a neutraliser of the heating beam cf chapter 2 6 2 The initial metastable singlet 2 S population leads to a characteristic hump in the singlet beam emission profile This hump can be used to obtain a reasonably accurate estimate of the initial metastable singlet population in the beam Chapter 4 Reversion code yt tocs 107 A change in the initial metastable triplet population 2 S on the other hand does not change the shape of the triplet beam emission profile This can be explained by the fact that the metatstable triplet population shows an exponential decay through the
76. equation 16 the parameter value p after the iteration i outer loop is calculated The weight w must stay within the range 0 the old parameter p stays unchanged maximum damping 1 the new parameter p is used unchanged no damping A weight w 0 5 leads therefore to a parameter value which is the mean value of the preceding and the current value Because the calculation of the beam emission forward calculation is non linear in ne and T the combination of both parameters is not necessarily a solution However once converged old and new parameters are the same and therefore their combination is also a solution of the forward calculation Examples for a yttocs calculation with and without damping of the parameters are shown in Figs 4 5 and 4 6 The output sheet for each run consists of the density and temperature profiles at the top the beam emission profiles of a singlet and a triplet line middle and the convergence plot bottom The input profiles emission data and profile parametrisations have been the same for both runs Measurements shown Chapter 4 Reversion code yt tocs 102 as symbols in the beam emission plot are synthetic data The density and temperature profiles which the reversion code should return are therefore known see chapter 4 2 and shown as dashed lines in the density and temperature profile plots For the damped case the weights are 0 8 for the n parameter and 0
77. er tubes with interference filters For routine application of the fast He beam as a plasma edge diagnostic a dedicated diagnostic beam is essential as the diameter of the heating beam limits the spatial resolution Hel beam emission spectroscopy based on He doped heating beams gives valuable additional information to the existing CX and BE spectroscopy As all these measurements are made with the same observation system they originate at the same location This reduces uncertainties otherwise introduced by mapping errors Furthermore the Hel beam emission spectrum is much simpler than the complex D spectrum This has already been exploited for the cross calibration and could also be beneficial for measuring plasma fluctuations In summary the results of this thesis show convincingly that fast He beam emission spectroscopy can be used as an electron density and temperature diagnostics particularly for the plasma edge Furthermore there is some prospect for other diagnostic applications impurity profiles plasma fluctuations Appendix 114 Appendix A Bibliography 1 2 3 4 5 6 7 8 9 10 11 12 13 14 M G von Hellermann and H P Summers JET report JET P 93 34 1993 R C Isler L E Murray S Kasai J L Dunlap S C Bates P H Edmonds E A Lazarus C H Ma M Murakami Phys Rev A 24 2701 1981 R J Groebner F H Brooks K H Burrell L Rottler Appl Phys Le
78. erny Turner spectrometer type BM 100 The fibres are arranged in a single vertical line array in front of an adjustable entrance slit A programmable CCD Camera is operated with typical exposure times of 20 50 ms and the data read out time is about 15 ms The used Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 14 2400 lines mm grating leads to a dispersion of 0 09 A pixel and can operate in the range 300 680 nm Wavelength settings above 680 nm are prohibited by the control software although the spectrometer could be operated at higher wavelengths Therefore the 728 nm 2 P 3 S and the 707 nm 2 P 3 S Hel emission lines could not be measured at AUG Doppler shift For the measurements at AUG we used the ion sources Q3 and Q4 When using the source Q4 the observation angle was too close to 90 for many viewing lines see above The situation is better when using the ion source Q3 see Fig 2 4 Doppler shift for AUG 12664 0 8 rir yp eg er a ee IT ep ee N N 0 6 Q gt Dikbler shift nm nm go Lia ee ee pise araj rrii 74 76 78 80 82 84 86 88 observation angle 4 in 7 f Fig 2 5 Expected solid line and measured dots Doppler shift of 667 8 nm 2 P 3 D Hel beam emission observed during AUG pulse 12644 The open dots are used for tracks where the fit procedure failed Fig 2 4 Plan view of a segment of AUG One can see the fan of viewin
79. exity As a consequence the code is not being updated as regularly as in other experiments A considerable number of magnetic sensors have failed The field ripple is only negligible near the magnetic axis but becomes noticeable near the coils Reciprocation probe measurements of the plasma edge location show discrepancies of up to 100 mm at the position of the probe compared to the plasma edge calculated with EFIT Chapter 3 Analysis of the spectroscopic data 75 3 3 Analysis code for the sweep experiments he_wi d For the analysis of the raw data from experiments at JET a code called he_wid has been developed It is written in IDL and makes use of IDL widgets graphical user interfaces Although specifically designed for the sweep experiments the code has been kept versatile and is able to deal with data from different spectrometers However the code includes a lot of JET specific commands Nevertheless it should be possible to adapt the software to other raw data sources file or AUG with only minor changes Especially the curve fittig routine is independent of the origin of the data The code incorporates the following features e Access to JET database JPF PPF raw data e Absolute calibration of the spectra software module i nt_cali br e Versatile spectral fit making use of mpf i t 2 Estimation of the Doppler shift automatic setting of the start parameters Option of limiting parameter ranges Additional Gaussian pea
80. fined These values are listed J7 no text output I display fits J display D peak 1 flat weight _remove line OK below the spectrum and can be edited there The checkbox left of the fields indicates whether the fields are active or not Fig 3 5 Widget for the entry of the parameter for additional disturbing peaks If there are no additional disturbing peaks selected only the first two rows of fields are active unshifted and shifted Hel peak The initial guess of the fit function is automatically displayed in colour on top of the spectrum to ease trouble shooting The displayed spectrum can be changed by browsing through the active tracks with the back and forward buttons However the start values for position and FWHM of the disturbing peaks are the same for all tracks Chapter 3 Analysis of the spectroscopic data 82 The bottom row of checkboxes in the widget have the following effect on to output of the fit program No text output If active the extensive text output of mpf it will be suppressed Display fits The result of the fit for all spectra of the active tracks for all frames of the selected time window is being plotted instantly Display D peak Instant display of the Doppler shifted peaks i e the measured spectrum subtracted from the resulting fit leaving out the Doppler shifted peak Thus only the Doppler shifted peak remains in the spectrum provided that the fitting procedure was successful Fla
81. finition of Ze reflects that the impurities mostly originate from the plasma wall or from impurity puffing The actual impurity profiles in fusion plasmas are not well known yet and the present definition has to be regarded as a first artificial step Chapter 4 Reversion code yt tocs 101 4 4 Weighting and mixing The independent variation of electron temperature and density profiles can cause the following minor problems 1 For some test data the convergence of the whole calculation not of the single fits showed an oscillation between the two independent optimisation loops fit of singlet and triplet emission profile 2 Although the used Hel transitions are mainly sensitive to either electron temperature Te or density ne in particular for the triplet lines there is a certain sensitivity to the respective other parameter as well Features from the T profile can for example be seen in the beam emission profile used to derive the n density profile However the fitting algorithm tries to reproduce this T feature in the n profile These problems can be overcome by introducing some kind of damping or by modifying the sequence of fitting beam emission profiles Weighting In order to damp possible oscillations an underrelaxation parameter w is introduced The weight w determines to which extent the old parameters p are changed towards the values pi 1 returned by the fit Pin Pit Din Pi Ww 16 In
82. fit by Craig Markwardt 41 It is based on Minpack 1 Lmdif f by Mor and collaborators and performs a Levenberg Marquardt least squares minimisation This technique is a particular strategy for iteratively searching for the best fit This particular implementation is drawn from MI NPACK 1 42 and appears to be more robust than routines provided by IDL curvefit In comparison to the standard curvefit another advantage is the option of assigning limits to the individual parameters This turned out to be essential especially for spectra with impurity lines Chapter 3 Analysis of the spectroscopic data 13 3 2 Mapping of plasma parameters onto beam axis In a magnetically confined toroidal plasma so called nested flux surfaces are formed by the magnetic field lines twisting around the magnetic axis Along these flux surfaces transport is fast and many plasma parameters are therefore kept constant Consequently the flux surfaces provide the obvious coordinate system for the plasma The various diagnostics and their lines of sight are geometrically related to the fixed vacuum vessel and intersect different parts of the plasma Plasma parameters e g electron density or temperature measured by means of the various diagnostics must therefore be mapped from the origin of the measurement to the system of nested flux surfaces Both AUG and JET have implemented an equilibrium code called EFIT which calculates the location and shape of t
83. g emission profile had a large scatter As for the viewing lines close to the plasma edge the Doppler shift was too small the beam emission peak could not be separated from the unshifted passive emission from the edge Therefore the plasma edge was outside the accessible radial region and it was also not possible to identify the beam emission as an increase of the total emission shifted plus unshifted Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 28 2 4 2 AUG Mai 2000 60 keV doped He D beam After the first campaign at AUG chapter 2 4 1 successful measurements with a He D doping system have been made at JET chapter 2 5 1 Consequently a similar system has been installed at AUG in April 2000 Between 25 4 2000 and 8 6 2000 the new system was tested and Hel beam emission measurements have been carried out The He D doping system can be used for parasitic measurements without restrictions to the NBI heating power By installing the doping system at the ion source Q3 the accessible radial range could be extended details see chapter 2 1 For the beam emission measurements the CER spectrometer was needed therefore only pulses where the ion temperature measurements have not been needed could be used for our parasitic experiments Furthermore we were interested in a series or repetitive pulses in order to obtain a set of comparable beam emission profiles The ideal candidate meeting these require
84. g lines can easily be excluded provided they are distinguishable from the Doppler shifted lines The software has a zoom history storing a limited number of zoom ranges By pressing the UNZOOM button the displayed spectral range goes back one step in the zoom history Wavelength offset The expected position of the unshifted peak is marked with a solid triangle at the bottom of each spectrum plot When the Doppler shift values are loaded the expected position of the respective peaks are marked by open triangles Due to inaccuracies in the wavelength calibration the whole spectrum can be slightly shifted This shift can be incorporated by typing the wavelength shift in the field wl offset and acknowledging with ENTER The estimated peak position must be sufficiently close to the actual peak position for the fit procedure to succeed 3 Time trace mapped radius mode This mode is for displaying the parameter of the fit and is therefore l background Order only available when the fit procedure has been started at least once for the current data I background 1st Order In this mode another toggle button becomes available which allows to i background 2st Order switch between time trace plot display and mapped plot wavelength PRIN peak Time trace plot The selected parameters are plotted over the rel hight PRIM peak selected time range start end time 1 FH PRIM peak mapped plot
85. g lines crossing the beams from ion source Q3 and Q4 The beam from Q3 can be operated as a doped He D beam marked red As mentioned above for our first measurements at AUG one ion source was converted to He operation During the CX diagnostics measurements it was necessary to operate Q3 in D Therefore Q4 had to be used for the generation of the pure He beam In Fig 2 5 the expected solid line and measured dots Doppler shift of the 667 8 nm 2 P 3 D Hel beam emission observed during AUG pulse 12644 is shown Because the Doppler shift is too small for observation angles above 85 the beam emission could not be distinguished Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 15 from the passive plasma emission and the curve fit failed open dots The beam emission peak measured with an observation angle of 77 was disturbed by a Si impurity line For the He doping system the ion source Q3 could be used yielding better peak separation due to higher Doppler shift 2 2 2 Spectroscopic system at JET KS4 5 7 diagnostics For the experiments at JET two different CX diagnostics have been used depending on the injector used for generating the He beam For the 80 kV injector at octant 4 the so called KS7 diagnostics and for the 140 kV injector at octant 8 the so called KS4 5 diagnostics was used for observing the beam emission KS7 system The KS7 diagnostics is mainly used for measuring ion t
86. g of the Hel line 2 P 3 D The two abscissas meas and calc were shifted against each other by 40 mm Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 36 2 4 3 Conclusions from Experiments at AUG The generation of a fast He beam at AUG was achieved for both pure and doped D He beams The doped D He beam had the advantage that the diagnostic PINI could be used for it which provided a better viewing geometry giving access to a wider radial range Another advantage was that the timing of the He doping is freely selectable and that the available NBI power stays unchanged This gives access to relevant phases of a plasma discharge whereas experiments using the pure He beam had to be scheduled at the end of the heating period Such beam emission experiments could be carried out parasitically with the only restriction that the CER diagnostic is not available for the ion standard temperature measurement The Hel lines 2 p 3 s 2 P 3 S and 2 S 3 P were not accessible due to technical limitations of the spectrometer With regard to the other n 3 gt 2 and n 4 gt 22 Hel transitions the sensitivity of the CER spectrometer was only high enough for the two strongest lines transition 2 P 3 D at 667 8 nm and 2 P 3 D at 587 6 nm For these two accessible lines we obtained beam emission profile for 4 different discharges and made repetitive measurements which allowed to estimate the error in the data to 10 Th
87. generated data of the look up tables are represented in a ne T plane rather than the Ne Ep plane used by the official version of ADAS 311 ne plasma density Te electron temperature E beam energy In a comparison of results from Scotty with those from the new scotty_fwd it turned out that the decay length problem only occurs when using the new look up tables It is therefore likely that the n T projections from the ADAS data are responsible for the too wide triplet beam emission profiles Chapter 2 Discussion of the Results 69 Due to manpower restrictions on the ADAS side and some incompatibility problems between the ADAS and the TU Wien computer systems it was not possible to solve these problems in the time frame of this paper However to avoid this problem for the development and testing of the reversion code suitable test data were generated using the same look up tables used later for the reversion see chapter 4 Chapter 3 Analysis of the spectroscopic data 70 3 Analysis of the spectroscopic data In this chapter the analysis of the Hel beam emission spectra is described Both at AUG and JET the existing analysis codes where not entirely suitable for our purposes and therefore additional software had to be written AUG At AUG we used the existing IDL code cer_wid to analyse the He beam emission data 40 The code written for the analysis of charge exchange CX data is not fully suitable for a
88. getic lithium beams 25 has been successfully used as a diagnostic of plasma density Due to the low ionisation energy of only 5 4eV the excitation rate coefficient of lithium quickly looses its electron energy dependence with increasing electron energy This being beneficial for a pure density diagnostics makes lithium unsuitable as a temperature diagnostics Helium with its much higher ionisation energy of 24 6 eV has some potential for simultaneous density and temperature diagnostics for the plasma edge This has been verified with thermal helium beams 26 Both diagnostics fast lithium beams and thermal helium beams are limited in range by the penetration depth of the neutral particles to the outer regions of the plasma Energetic helium atoms penetrate much deeper into the plasma than either lithium atoms of similar energy or thermal helium atoms and therefore offer the prospect for measuring TOKAMAK plasma parameters inside of the H mode and even internal transport barriers ITB Helium as an atom with two electrons has two different spin systems the singlet system i e the spins of the two electrons are antiparallel S s 0 and the triplet system i e the spins of the two electrons are parallel S X s 1 The triplet system can only be populated by spin changing processes from the ground state i e electron collisions whereas the singlet levels are mainly populated by spin conserving processes from the ground state
89. gle fibre link eco cameras splitter Heliurn and Beryllium or Nitrogen Deuterium or R Carbon and Argen Haliurn and Berylliurn A o Nitrogen Heated multiple fibre links idagical shield OCD Cameras Fig 2 10 Viewing lines of the KS5 spectrometer used for the 140 kV beam from the octant 8 injector estimated Doppler shift for KS5a The viewing lines from KS5a are Saree Pr rp err rg crossing the beams from both ion sources 7 Yi YY 3 PINI 6 and 7 In Fig 2 11 the estimated en a 1 Doppler shifts for the 587 6nm dl dea aa ren er o q 2 P 3 D beam emission fora 140keY Ale rast ee He beam PINI 6 and 7 are plotted for as ddd R da VIA the KS5a viewing lines When installing ole ae i 1 the He doping system at PINI 7 the i ale alan ee ee ee 7 Doppler shift was sufficient for all 25 i ie er tracks The shaded areas indicate 3 enrii ei A S eer unfavourable values of the Doppler shift 28 3 3 2 3 4 3 6 38 4 an ae R maj overlapping the beam emission in the ae 587 6 nm spectrum with either passive Fig 2 11 Estimated Doppler Hel emission or an impurity line shifts for the KS5a viewing lines for the 587 6 nm 2 P 3 D beam emission coming from a 140keV He beam PINI 6 and PINI 7 Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 19 2 3 Plasma sweep experiments During the first HeI beam emi
90. he KS7 system have been performed in either the standard or in the 6 fibre mode It turned out that in the standard setup with 16 fibres the bandwidth was too narrow for some of the fibres and could thus not cover the range of the unshifted and the Doppler shifted peak Although it was in principle possible to correct the data subsequently such a correction introduced additional intolerable errors Therefore during the campaigns in 2000 and 2001 KS7 was only used in the 6 fibres setup To compensate for the smaller number of fibres a sweep experiment was introduced see chapter 2 3 KS7 provides a large number of lines of sights which can be changed easily on a shot by shot bases However an absolute calibration of the system has so far only been performed for one set of fibres Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 17 These fibres where chosen according to the following requirements e Doppler shift sufficiently large to separate beam emission from the passive emission e Avoiding viewing lines for which the Doppler shifted He emission is masked by emission from impurity lines e Distance between neighbouring active volumes intersection of line of sights with the beam small enough for overlapping calibration lt 100 mm KS5 system The KS5 charge exchange system uses fibres coming from two periscopes octant 1 and 7 looking at the beams of the octant 8 injector
91. he flux surfaces from magnetic flux loop measurements coil currents and the magnetic properties of the major machine parts e g influence of the iron core For each pulse the flux surfaces for a certain number of time slices at JET typically every 200 ms are calculated and stored in a database Both AUG and JET provide code libraries called flush routines at JET and kk li brary at AUG respectively for mapping from one coordinate system into another poloidal flux radius For the modelling of the He beam emission it is necessary to know the electron density ce set eRe ting bea gr and temperature along the beam path Data taken by the EEN FERE RETI various diagnostics are stored 02 Fins aoe dee eo kee T ee ee together with the location of 0 the measurement Either the mapped T profile actual position e g R o y or 4000 position of the CER viewing lines the flux surface coordinates at this position are given In 3000 some cases the position is mapped to a standard plane 2 e g the mid plane of the P00 a ES a E AAT SUAS es f machine and the respective radial position major radius 1000 Rmaj is given al 0 i Jt e Loo Loi oo 200 0 200 400 600 800 1000 1200 1400 distance along beam mm Fig 3 2 Poloidal flux radius along the beam upper graph and mapped and interpolated electron density along the beam lower graph Chapter 3 Analysis
92. he spectrometer channels using plasma sweep data JET53874_KS5a_7_mean_2 110 KH rn i2 i7 i3 is 810 E ee NS eee Pe ee oe i4o i9 H 5 i10 l i6 0 1 59 90s beam emission arb units ee fige ae 57 57 5 58 58 5 59 59 5 60 time s Fig 2 17 Photon flux of the HeI beam emission 667 8 nm measured with the KS5a diagnostic during a horizontal plasma sweep JPN 53874 Different colours for different lines of sight JET53874 Hel beam emission 667 8 nm 2 P 3 D 25 110 ai a a a a F i i i track 2 J H l j track 3 4 810 u Doreen u track 4 4 H track 5 4 H track 6 4 H i f N track 7 4 2 6407 bi ae s wies ze ee 1 m tacks 4 L track 9 4 A L track 10 J a fa r i a 24 E410 gt L T L o 3 L 210 H 710 C 0 70 J 610 E 116 I n J B F 7 s510 F I tee eee a 102 4 Sal z F amp 410 P N os 4 5 F T F i 0 80 7 es a E 310 F J ao H z E u J g 210 H 4 C a 110 P e aaa 4 4000 3800 3600 3400 3200 3000 radial position Rmaj mm Fig 2 18 Upper graph Data shown in Fig 2 17 mapped to respective major radius at t 58 s Lower graph Same data after overlap calibration As the plasma sweep was wide enough to overlap neighbouring
93. identischen Plasmaentladungen bestimmt werden Bei JET wurden Emissionsprofile mit guter r umlicher Aufl sung bestimmt Diese hohe Aufl sung konnte dadurch erreicht werden dass das Plasma w hrend der Messung quer zu den Sichtlinien verschoben wurde Neben dieser erh hten Aufl sung erlaubte diese neue Experimentiertechnik auch benachbarte Kan le des Spektrometers relativ zueinander zu kalibrieren was zu einer deutlichen Reduzierung der Messfehler f hrte Eine weitere interessante Beobachtung ist dass die Form der Triplett Emissionsprofile auch von der Verteilung der Plasmaverunreinigungen beeinflusst wird Dies ffnet die M glichkeit zur Entwicklung einer Zer Diagnostik Die Emissionsprofile werden mit einem numerischen Sto Strahlungsmodell simuliert und mit den Messungen verglichen Das Modell wurde im Rahmen dieser Arbeit erweitert und bez glich Rechenzeit optimiert Das eigentliche Ziel der vorliegenden Arbeit war es Dichte und Temperaturprofile aus den gemessenen Emissionsprofilen ableiten zu k nnen Dazu wurde der Umkehrcode yttocs entwickelt Dieser beruht auf einer Variationsmethode in der die Modellrechnungen oftmals aufgerufen werden was u a die Entwicklung einer geschwindigkeits optimierten Subroutine notwendig machte Die ersten Tests dieses Umkehrcodes wurden mit synthetischen Daten vorgenommen um etwaige Fehler in der Aufbereitung der atomaren Daten zu vermeiden Der Code zeigt gutes Konvergenzverhalten sowohl Dichte al
94. ii A B Izvozchikov V I Marasev A I Kislyakov E A Mikhailov M P Petrov and G V Roslyakov JETP Lett 29 1 1979 A A E van Blokland E P Barbian A J H Donn A F van der Grift T W M Grimbergen Th Oyevaar F C Sch ller H F Tammen H W van der Ven T F Vijverberg and F D A de Winter Rev Sci Instrum 63 3359 68 1991 H Takeuchi K Tobita Y Kusama M Nemeto T Itoh Y Tsukahara and JT 60 team Rev Sci Instrum 59 1652 7 1988 G Schilling T A Kozub S S Medley and K M Young Rev Sci Instrum 57 2060 2 1986 A H Sarkissian Rev Sci Instrum 69 923 5 1998 E S Marmar J L Terry W L Rowan and A J R Wootton Rev Sci Instrum 68 265 8 1997 T Itoh S Matsuda M Matsuoka Y Ohara M Shitomi and K Watanabe Proc 4th Int Symp on Heating in Toroidal Plasma Int School of Plasma Phys Varenna Italy June 1984 2 1081 6 1984 Y J Kim R A Breun D A Brouchous and R J Fonck Rev Sci Instrum 61 3046 8 1990 A S Schlachter J W Stearns and W S Cooper Rev Sci Instrum 59 1729 31 1988 Y Kusama K Tobita T Itoh M Nemeto Y Tsukahara H Kimura and H Takeuchi Rev Sci Instrum 61 3220 2 1990 E Wolfrum F Aumayr E Hintz D Rusbiilt R P Schorn D Wutte and HP Winter Rev Sci Instrum 64 2285 92 1993 B Schweer M Brix and M Lehnen J Nucl Mater 266 269 673 8 1999 W Mandl Development of Active Balmer Alpha Spectr
95. in equilibrium with them The new subroutine f or ward_calc was designed for variable matrix sizes however it was only tested for 11x11 matrices all levels with n lt 4 For the concept of an iterative conversion code it is essential to keep the computation time of the forward calculation sufficiently low Chapter 4 Reversion code yt t ocs 97 The new scotty_fwd code needs less then 5 seconds which is acceptable for the reversion code This performance improvement could be achieved by the following measures 1 The new code loads the whole lookup table into the memory eliminating the slow file access for each calculation step several 1000 steps per profile 2 The step width is dynamically changed keeping the total number of steps to a minimum 3 The size of the matrix number of non equilibrium levels included in the calculation can be changed The gain of calculation speed with smaller matrices is twofold All the matrix operations are faster and the numerical stability of the code is better leading to bigger step widths and consequently a smaller total number of steps Synthetic emission data For the development of the reversion code it is adventageous to generate synthetic measurement data The small code art_data performs a forward calculation call of scotty _fwd extracting beam emission profiles from density and temperature profiles This emission profiles are then modified in order to get realistic measure
96. is relatively high scatter is due to the low intensity of the signal A systematic deviation of some channels indicates their non perfect calibration The alignment of the beam to the CER viewing lines was unfavourable and made an intensity correction of the different channels necessary In particular the outer lines of sight are very sensitive to misalignment Furthermore the intensity of the observed signal is lower The triplet emission was only measurable over the outer 200 mm of the plasma This range is only roughly twice the range of the 50 kV lithium beam Since our last measurements the alignment and calibration of the CER diagnostic has been improved which should lead to a higher beam emission signal and a lower scatter of the derived emission profile The settings of the spectrometer slit width and exposure time can be optimised in view to the signal intensity At JET a cross calibration of neighbouring channels could be achieved by sweeping the plasma across the viewing lines see chapter 2 3 2 The application of this procedure to AUG is being investigated and would improve the achievable profile quality Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 37 2 5 Measurements at JET 2 5 1 JET October November 1999 80 keV doped He D beam The He doping system has been developed and tested at the JET test bed 28 and was then first installed at PINI 6 of the octant 4 injector at JET in th
97. itional helium pumping capacity is required Such doped beams can therefore be made available without adverse effects on the availability of the neutral beam heating system Using the heating beams for beam emission spectroscopy limits the spatial resolution of the measurement as these heating beams have a diameter of typically 150 300 mm This is a severe restriction for diagnosing the narrow edge region Another potential disadvantage of using the powerful heating beam is that the latter might strongly influence the local plasma parameters Both these problems could be overcome with a dedicated diagnostic beam However for the proof of principle experiments described in this paper only the heating beams were available Chapter 1 Introduction 6 1 3 Collisional radiative model for calculation of beam emission profiles The model used in this paper has been described in detail in the Ph D thesis of S Menhart 29 and can be summarised as follows The level population of fast He beam particles interacting with a plasma of given density and temperature distribution is calculated stepwise starting with a given initial population In order to calculate the change in the population of the Hel levels within a step dx along the beam we have to solve the statistical balance equations which represent the rates at which the excited levels of an atom are populated and depopulated In general these equations have the shape dN dN dt Vp
98. ivity and standard deviation are saved as an IDL save file This file is afterwards accessed by the calibration routine int_calibr It is important to keep the information of the pixel because the sensitivity can change substantially from pixel to pixel In Fig 3 9 the calculated sensitivity is displayed as a contour plot together with one horizontal and one vertical cut Because the standard display program was used the wavelength axis is marked with time instead of wave length A On can clearly see that some pixel have a significantly lower sensitivity over the full wavelength range dead pixel or dust particle In the near infra red NIR region A gt 730 nm the CCD sensor shows etaloning an interference effect of back illuminated CCD s They consist of thin films typically 10 20 microns thick which become semitransparent in the NIR Reflections between the front and the back surface of these devices cause them to act as etalons leading to unwanted fringes These fringes can be seen in Fig 3 9 if the bluish interference pattern around 750 nm Chapter 3 Analysis of the spectroscopic data 88 22 cal shot O pixel 315 track 7 time 7300 00 intens 4 0079e 24 107 10725 pixel 10724 4500 5000 5500 6000 6500 7000 7500 time 3 5x10725 3 0x10725 2 5x10725 2 0x10725 1 5x10725 1 0x10725 5 0x10724 4500 5000 5500 6000 6500 7000 7500 50 100 150 200 250 300 350 time pixel Thu Jul 5 09 33 26
99. ks can be added for disturbing impurity lines Interactive input of the parameters for these additional Gaussian functions e Different display options Spectra including resulting fits free choice of tracks Interactive zooming range definition for fit Time trace of beam data Interactive input of start end and current time frames Plot of the resulting fitting parameters free choice of parameters Toggle between time trace and distance along beam mapping e Possibility of background subtraction for modulated beams interpolated background e Automatic identification of frames with NBI break downs e overlap calibration for sweep experiments MAIN WIDGET When starting he_wi d the main widget appears see screen shot in Fig 3 3 see chapter 3 1 Chapter 3 Analysis of the spectroscopic data 76 FIT program for Hel beam emission KS7 Iof x x108 pee Dr m mwee reae load D_shift KS5a pulse saben me E 1 1550 87 3 0000 4 h x102 pn srl Io 2 tes nom 4 3 0000 rw time frame KI gt oxi 028 5 7650 B wv time tr spectr w beam D ims yy ntti par_ist gt saa 8 3 a on ow or oy Er mM N N N x x e y o gt D i 8 S Ss L D M ue oO o a g Ss S L S640 58650 SB60 5870 wave length A modulated beam EN ya 8 8 L S S PS PS Tb 8 8 2 8 8 S Ss t S E e o Erz mye Er oy Er Ft r
100. lable 1009 the Thomson fit data is being used for the 1500 500 temperature profile For 0 pulses without Thomson fit data the ECE radiometer data Thomson fit 13646 looks most promising Gaussfit 3000 0 08 Hi res Thomson scattering t 3 52 3 67 although its scatter can be Thomson scattering t 3 5 3 7 considerable Fig 2 25 shows the electron Fig 2 25 Electron temperature profile for the standard temperature profile for the H mode discharge 13646 at t 3 6s Symbols are straight standard H mode pulse data from Thomson scattering standard and high resolution 13665 as derived from Pe fit is from the standard Thomson scattering Data points and Thomson fit show a centrally peaked profile However also a Gau fit is consistent with the data due to its large scatter Thomson scattering data Hel beam emission for the standard H mode discharge The straight Hel beam emission profiles show a high systematic scatter As described in chapter 3 6 this can be explained by the alignment of the CER diagnostic to the beam and is being corrected Even for the most intensive singlet line the signal intensity was close to the noise level leading to a considerable scatter in the beam emission profile For evaluation of the Hel spectra we used the standard CER fitting routine cer _wi d which was not optimised for our requirements Figs 2 26 and 2 27 show the corrected beam emission
101. ld be measured additionally The triplet emission line intensity is dominated by the initial fraction of the 2 S level and its attenuation inside the plasma The lower intensity of the equilibrium emission could not yet be resolved by the available spectroscopic systems A significant peak in the singlet profiles measured with a doped He beam indicates an initial metastable 2 S population of about 1 although the metastable population should be predominantly in the 2 S state 35 For the most intense lines the noise was dominated by the plasma even for an integration time of only 50 ms and therefore also the prospect of measuring plasma fluctuations has to be mentioned This would however require a faster detector with higher sensitivity e g photomultiplier tubes with interference filters Variation of the impurity content showed that Zerr is of influence on the shape of both the singlet and the triplet lines The profiles of the triplet emission becomes narrower with increasing Ze The potential of using this sensitivity to impurity levels for developing a Zerr profile diagnostics should be examined further Chapter 2 Discussion of the Results 62 2 6 Discussion of the Results 2 6 1 Estimate of He flux in the doped beam The two beam emission profiles for pulses JPN 52797 and JPN 53881 were measured with the spectrometer KS7 for plasma pulses with a line averaged impurity Zerr of 1 9 Pulse 53881 was observed with a pure He beam
102. line integrated density KG1 LID7 a o lid 10 m 8 5 54 56 58 60 62 64 time s Fig 2 14 Injected beam power top location of the LCFS in the mid plane middle and plasma density The duration of He doping is visible as increase in the injected power and marked by shading Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 21 zzp rL r rrr yp rrr pr rrr rr ry z NV t 56 8928 s IR J t 60 1220 s PER RT DR TB RT tye tee La RZ RR 2 5 r 4 5 210 220 230 240 250 260 270 280 290 300 310 Major radius m 53873 7 7 170 If sweep m _ Ex beam E FE USP a aaa tan i H 180 JPN 52799 58 6s Fig 2 15 Flux surfaces from EFIT for the extreme Fig 2 16 Strike point location for plasma positions of the radial sweep increased pumping The plasma was swept radially by 120 mm with a ramp rate of 40 mm s The location of the last closed flux surface LCFS in the mid plane is shown in Fig 2 14 together with the injected beam power and the line integrated density The plasma shape was made very slim in order to avoid interaction between vessel wall and plasma at the extremes of the sweeps The two extreme positions of the flux surfaces are shown in Fig 2 15 pulse D CD 4 Lett lt n dl gt line averaged density 7 10 10 ni 53872 CD 2 1 8
103. lits The first spectrometer acts as a band bass filter the second one as the spectral analyser The purpose of the band pass spectrometer is to mask the spectra allowing to fit four of then side by side onto one row of the CCD sensor and to prevent cross talk from adjacent spectra The CCD is partitioned into 4 tracks corresponding to the 4 rows of entrance slits The four spectra from each track are extracted during the data analysis and the respective wavelengths are assigned The transmission function for each entrance slit of the 4x4 array is determined by measuring the signal produced by a white light source Fig 2 36 shows the transmission measurement for entrance slit 2 of 4 from the first row over the whole width of the CCD sensor all 385 pixel The pixel position on the CCD sensor corresponds to a certain wavelength One can see that the flat top region of the transmission function is typically 25 pixel wide which corresponds to a wavelength range of about Inm Without the band pass spectrometer the intensity would be constant over the whole sensor width Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 44 4 Due to the anamorphic 10 Ta 25 pixel flat top i g 4 magnification of the Cerny Turner 3 3 10 spectrometer the image of the entrance slit array on the CCD 3 Fel 04 sensor is distorted In order to get S an undistorted image the entrance fad 04 slits are shaped to c
104. ly of deuterium in case of a doped He D beam and helium in case of a pure He beam It is expected that the metastable population is predominantly in the 2 S state The exact composition of the metastable fraction could not be derived as this would require an absolute measurement of the beam emission signal The impurity content of the plasma has been varied by puffing alternatively CD or D2 into the discharge for some pulses It could be observed that Zerf has an influence on the shape of both singlet and triplet emission profiles In particular the peak width of the triplet emission profiles becomes smaller with increasing Zerr The code for modelling of the HeI beam emission scotty has been rewritten in order to meet the requirements of the reversion code reduced run time and to increase the number of independent states involved The new code scott y_fwd was designed as a subroutine and optimised in performance The atomic data are accessed via look up tables generated by the ADAS 311 code In case of the new scotty_fwd code a modified ADAS 311 code Chapter 5 Summary Conclusions and Outlook 112 not part of the official ADAS release was used which represents the atomic data in projections that are more suitable for beam modelling Good agreement between modelled and measured beam emission profiles has been obtained for the singlet emission A significant mismatch is however observed between modelled and measured triplet e
105. m the interference with the unshifted passive Hel line can be eliminated but at the same time the total beam emission signal is halved For the measured triplet line 2 P 3 D the signal intensity was nevertheless strong enough in spite of the beam modulation However the limited accuracy of the calibration of the individual viewing lines still degraded the quality of the resulting beam emission profiles Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 50 2 5 2 JET December 2000 sweep experiment with 70keV doped He D beam For this campaign the pre set pneumatic timing system of the He doping system has been replaced by a more versatile electronic one From the experience of the previous measurements in 1999 it was clear that 1 KS7 can only be used in its 6 fibres setup and consequently the loss of spatial coverage must be compensated 2 a cross calibration of the different channels viewing lines is crucial for an improved quality of the measured profiles As a full experimental shift of approximately 15 plasma pulses was dedicated to our HeI beam emission experiments a suitable pulse type could be chosen The outcome was the sweep experiment described in chapter 2 3 which met all the above criteria The experiments were all performed using the same pulse type a high clearance diverted L mode discharge characterised in table II During the experimental campaign in December 2000 PINI 6 from
106. ment data The modifications include the averaging of the emission data over a given radial range typical 50 mm to simulate the limited resolution of the BES and the addition of a realistic level of noise typical 5 The advantage of using synthetic data is twofold First the original input data are known and therefore the correct result of the reversion and secondly the influence of the uncertainties in the atomic data set cancel out 4 3 Profile parametrisation There are two different ways of defining the input profiles for the fitting procedure a global and a local approach 1 global The entire input profiles are represented by parametrised functions some 5 parameters and the profile as a whole is matched 2 local The profiles are defined step by step starting from outside of the plasma The word local is under quotation because the measured beam emission is not only a function of the local density and temperature but also dependent on the local population of the atomic levels in the beam produced in the layers which the beam has already passed In the stepwise attempt errors would sum up quickly and propagate with progressing forward calculation in beam direction The method would therefore be very sensitive to noise in the data and local imperfections of the calibration The global approach on the other hand can introduce unintended limitations in the profile shape if unsuitable fit functions are sele
107. ments was the standard H mode pulse Standard H mode discharge The AUG standard H mode pulse is routinely performed during the start up sequence of AUG It is well characterised and the measurement of the ion temperature is not necessary Fig 2 23 shows the time trace of some important signals of the standard H mode discharge 13665 AUG pulse number 13646 13665 13667 13782 time s stored energy MJ 0 589 0 594 0 625 0 562 3 5 heating power MW 4 88 4 87 4 88 4 94 3 5 density 10 1 m 6 29 5 13 5 27 5 12 3 5 stored energy MJ 0 567 0 537 0 56 0 558 6 3 heating power MW 4 76 4 86 4 84 4 85 6 3 density 10 1 m 12 5 10 26 10 32 10 38 6 3 Table VI Reproducibility of the AUG standard H mode pulse at low density The two flat top phases one with a line average density of 5 10 m between 2 5 and 4 5 s and another one with a line average density of 10 m between 5 5 and 7 s had been used for the HeI beam emission measurements The stored energy of the AUG plasma is roughly the same in both periods which means that the electron temperature is correspondingly lower in the high density phase of the discharge Helium gas for producing a doped He D beam was injected into the ion source between 3 5 and 3 7s and between 6 1 and 6 3 s The reproducibility of the discharges was very good with only one pulse 13646 showing a 20 higher density in table VI The
108. mission profiles It is likely that the generation of the look up tables by the modified ADAS 311 code is responsible for the discrepancy Due to manpower restrictions on the ADAS side it was not possible to solve the problem in the time frame of this work Therefore it was only possible to test the reversion code with synthetic data generated with the same look up tables The newly developed reversion code yt tocs for the determination of electron density and temperature is based on the variation of the respective profiles Starting from an initial guess for electron density and temperature the code adapts these profiles iteratively until calculated and measured emission profiles are well matched The convergence of the code with synthetic data and realistic noise levels is satisfactory The singlet beam emission e g 2 P 3 D is used to determine the electron density and the triplet emission to determine the electron temperature profile As the singlet emission was visible over the whole plasma range but the triplet emission is limited to the outer part the density can be determined over the whole range while the temperature determination is limited to that range where the triplet emission is actually visible outer 300 mm The quality of the results depends largely on the accuracy of the atomic data therefore additional effort in bringing the data base up to date would be beneficial The influence of the impurity content Zer on the beam emi
109. mmation over the Doppler shifted peak after background subtraction For the background subtraction the resulting fit is used leaving out the Doppler shifted peak Thus only the peak of interest remains in the spectrum provided that the fit procedure was successful 4 Masking of frames time points Breakdowns During the operation of the heating beams high voltage breakdowns can occur from time to time In the case of a breakdown the beam is turned off for typically 30 ms This affects the intensity of the beam emission and must be taken into account The software looks for breakdowns in the beam data files and identifies the respective frames With the toggle button mask break downs the masking of frames with a breakdown during the exposure can be turned on and off Fit qualtity The intention of implementing this feature was to make it possible to mask fits with bad quality The first attempt was to mask all frames where the chi square value of the fit exceeds a certain trip level field right to the toggle button This option does not appear to be useful but it is still implemented Activating the fitqual toggle involves another masking of the data If at the end of the fit procedure a parameter still tends to overcome its parameter limit it will be indicated by the fit code If the fit quality masking is turned on these frames are masked as well 5 Error bars Error bars are calculated for each frame and
110. n aus der Arbeitsgruppe in Wien und vor allem deren Leiter Friedrich Aumayr bin ich f r zahlreiche kritische Diskussionen und f r das angenehme Arbeitsklima dankbar das w hrend der gesamten Zeit herrschte Und zu guter letzt verdanke ich dem Zufall dass ich von meinem letzten Aufenthalt in Oxford meine nunmehrige Lebensgef hrtin Rebecca mit nach Hause bringen konnte Diese Arbeit wurde im Rahmen der Assoziation EURATOM AW durchgef hrt und von der Friedrich Schiedel Stiftung f r Energietechnik gef rdert Further acknowledgements I am very grateful to the JET team and in particular to Dragoslav Ciric Carin Giroud N Hawkes Tim Jones Andy Meigs and Klaus Dieter Zastrow for all their support and advice Out of the colleagues from the European associations I met at JET in particular Mathias Brix has been very supportive in numerous discussions The excellent teamwork and the professional and friendly attitude of this team is greatly apreciated I want to thank specifically Mathias Brix but also many other colleagues from European associations from the good teamwork I found at JET From the AUG team I am particularly thankful for the help from Hans Meister Joe Schweinzer Albrecht Stabler and Lisl Wolfrum Furthermore I would like to thank Prof Hugh Summers from Strathclyde University Glasgow for his support and advice
111. n profile Using a comparatively large heating beam results in a low resolution of the resulting beam emission profile which is therefore only of limited use for plasma edge diagnostics However it is sufficient for the presented proof of principle experiments At AUG beam emission experiments were performed using a pure 30 keV He beam and a 60 keV doped beam respectively At JET the octant 4 injector was used to generate both pure and doped He beams with about 75 keV beam energy and the octant 8 injector to generate a 135 keV doped He D beam The doping of D heating beams with He worked so successfully and reliably at both machines that the so called diagnostic beam source could be used for the experiments with doped helium beams As the spectroscopic systems were aligned to this diagnostic beam it provided the most favourable viewing geometry The doping of the diagnostic beam does not reduce the available NBI power and since the timing of the He doping is freely selectable it gives access to interesting phases of plasma discharges The first beam emission experiments in 1999 were carried out parasitically during steady state phases of the plasma discharge and plasma operation was only affected by the restriction that for the respective pulse the CX diagnostic was not available for ion temperature measurement The signal intensities of n 32 and n 4 gt 2 Hel lines have been assessed and it has been investigated whether the Doppler shifted He
112. n the beam axis namely the centre of the grid u v w 6 5 0 47 0 6 and the cross over of the beam axes from source Q3 and Q4 with the co ordinates u v w 0 0 0 0429 we get the equation of the beam axis in the form x7 X eh 272 Pe YOY _222 11 XTX TM 225744 a b c 3 6 2 Geometry of the optical lines of sight From the optical lines of sight the locations of the mirror and of a calibration target are known The calibration target was installed inside the AUG vacuum chamber near the axis By converting into Cartesian co ordinates we get the equations for the individual lines of sight in the same form as for the beam axis using b and J Z for the respective constants The minimum distance between beam and line of sight axis is then given by equation 12 and can found in Bronstein x 7 Yy I 2 dimin 12 The angle a between the line of sight and the and the beam axis is given by cosa o 13 lef where x and x are vectors parallel to He beam axis and the line of sight Finally the radius of the point of minimum distance between beam axis and line of sight can be estimated from the projection of the two axes into the z 0 plane This is justified as the cross over is actually very close to the mid plane From the minimum distance between beam axis and line of sight we can calculate a correction factor which compensates for the reduced beam current when moving away from the beam
113. nalysing the He beam emission spectra The Doppler shift of the beam emission peak is among other things defined by the viewing angle and varies with viewing line Within one profile the values vary over a wide range more than Inm The Doppler shifts of the different tracks are so different that the code fails to find the Doppler shifted peak for many tracks Therefore one has to set the estimated position for the peak of interest start value for the fit by hand for every single track An automatic calculation of the expected Doppler shift would therefore be very helpful but has not been implemented yet In some cases impurity lines with wavelength and intensity near that of the Doppler shifted He line are observed An automatic calculation of the expected Doppler shift would also help to distinguish between Doppler shifted He beam emission and such disturbing impurity lines JET At JET a complex spectroscopic fitting software called KS4fit was developed by M von Hellermann for the IBM mainframe computer Due to the complexity of this code it would have been difficult however possible to adapt it to our needs Furthermore it was clear from the start of our experiments that the then available IBM platform will be phased out As the new computer platform LINUX cluster was already available we decided to develop our own analysis code there using IDL for LINUX Both at AUG and JET the data acquired during a pulse are stored in big da
114. nes zero For the high density case the position of the maximum emission appears to be moved outwards beyond the LCFS and only the decaying part of the emission profile is within the range of the CER lines of sight In the low density case Fig 2 26 and 2 27 the position of the steep increase of the emission profile is in agreement with the model calculations For the high density case the modelled emission profiles need to be shifted outwards to match the measurement In the plot of Fig 2 29 good agreement between the measured data and the model calculation could be achieved with a shift of 40 mm This shift could be explained by a mapping error EFIT or by an error in the density profile used for the modelling Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 35 2 P 3 D 667 8 nm AUG standard H mode high density phase dist along beam calc mm 60 40 140 240 340 440 1 6 I I 1 8 10 L average 1 333 i AUG 13667 2 j AUG 13880 o E F i D F F i i 3 2 i aisea a AE gs 1087 N g gt a al F o 3 F i i 5 as occa en baa c L i v o g Z j i i 5 05333 H nn ia ey eee 3 5 2 D 0 2667 Hp E Dapena en AGENS J 310 j calc 1 init 2 S population J 0 footer Be eo 940 100 0 100 200 300 400 dist along beam meas mm Fig 2 29 Beam emission measurements AUG 13667 13880 and average and modellin
115. ng beams unnnuuuunnnunnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 8 2 2 SPOCITOSCODY una en 12 2 2 1 Spectroscopic system at AUG CER Diagnostics rsnersnnnnnnnennnnnnnnnennnnnnnnnn 13 2 2 2 Spectroscopic system at JET KS4 5 7 Diagnostics nersnsennseennnnnnnnnennnnnnnnenn 15 2 3 Plasma SWEEP experiments uuuuuuuuununnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 19 2 3 1 Radially swept high clearance X point L mode plasma nenenenn 19 2 3 2 Cross calibration of the spectrometer channels using plasma sweep data 23 2 4 Measurements at AUG unssssssnnnnnnnnnunnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnen 27 2 4 1 AUG July August 1999 30 keV pure He beam 2 u 2400nnnnnnnnnnnnnnnnnnnnnnnnnnn nenn 27 2 4 2 AUG Mai 2000 60 keV doped He D beam ccceceeeeeceeeeeeeeeeeeeeeeeeeeeeeseaeeesaeeneneeenaes 28 2 4 3 Conclusions from Experiments at AUG 0usnssssnsnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn nn 36 2 8 Meas rements at JET 2 2 a anderer 37 2 5 1 JET October November 1999 80 keV doped He D beam nernnnnnsnennnnenn 37 2 5 2 JET December 2000 sweep experiment with 70 keV doped He D beam 50 2 5 3 JET March 2001 sweep experiment with 135 keV doped and 73 keV pure He beam 54 2 5 4 Conclusions from Experiments at JET uunneensnsensnnnnnnnnnnnnnenn
116. nnnnnnnnnnn nn 91 3 6 2 Geometry of the optical lines of sight srsnussneennnnnennnnnnnnnnnnnnnnnnnnnnnnnnnnnnn nen 91 4 Reversion of Hel beam emission profiles into density and temperature profiles code yt t ocs 94 AT Basic 3lg0rlih uu ae an 94 4 2 Forward calculation scotty_fwd sssssssscsesessecseeesseesseensseeesseesseeneseeeseenssensseessees 96 4 3 Proflie paramelrls llon 2 97 4 4 Weighting and MDG 2 22 22 ee 101 4 5 Factors influencing the accuracy of results 222uuuuuuuunnnnnannnnnnnnnnnnnnnnnnnnnnnn 106 4 5 1 Absolute calibration 2 422 822 ler Eee 106 4 5 2 Initial metastable population of the beam 240r422nn0nnennnnnnennnnnnnennnnnnnnnnnn nenn 106 4 5 3 Finite resolution of the observation system 24ussessnnnnennnnnnnnnnnnnnnnnnnnnnnnn nme nn 108 4 6 Comparison with evaluation procedures for other beam emission diagnostics 109 5 Summary Conclusions and Outlook 110 A Bibliography 114 B Abbreviations and Symbols 118 In this corner of the following pages you will find a flip book showing the flux surfaces of the JET discharge JPN 52799 printed version only Please pay attention to the plasma sweep at the end of the discharge Chapter 1 Introduction 3 1 Introduction 1 1 Motivation The development of novel diagnostic tools providing reliable quantitative values for plasma parameters has been a key factor in the progress of fusion research towa
117. nnnnnnnnnnnnnnnnnnnnnnnn nen 61 2 6 Discussion of the Experimental Results uuuuunnunsssnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 62 2 6 1 Estimate of He flux in the doped beam s4s44esennnnnnnennnnnnnennnnnnnennnnnennnnnnnennnnnnn 62 2 6 2 Initial metastable fractions in the He beam 00 0 eeeeeeeeeeneeeeeeaeeeeeeaaeeeeeaaeeeeeeaaeeeeeeaas 64 2 6 3 Discrepancy with the modelled Triplett profiles_ uu44444n eeeeeeeeeeeeaeeeeeeaaeeeeeeaas 66 Contents 2 3 Analysis of the spectroscopic data 70 3 1 Spectral fit procedure hespec_fit mpfit sssccccsssssseeeeeeeeeessesseeseeeeeeeeesseeseees 71 3 2 Mapping of plasma parameters onto beam AXIS 1sccccccccccceseeceeencnseseseneeeseseeees 73 3 3 Analysis code for the sweep experiments he_ Wid sssssssssssscssssseeesseesereessessens 75 3 4 Factors influencing accuracy Of fit results eunnnnnnannnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 83 3 4 1 Change of the Doppler shift during sweep experiments ceesceceeeeeeeeeeeeteeeeeeeetees 83 3 4 2 Dead pixel of the CCD sensor nurssssnssennnnnnnsnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn 85 3 5 Absolute Calibration of the JET spectromelter 22000000000000000000n nn nnnnnnnnn nn 86 3 6 Alignment Correction for AUG data uuuuu02000002000000000nnnnnnnnnnnnnnnnnnnnnnnnn nn 90 3 6 1 Geometry of the beam AXIS unsensnsnnssnnnnnnnnnnnnennnnnnnnnnnnnnnnnnnnnnnsnnnnnnn
118. of the He injection This is due to the influx of He gas from the beam box The passive Hel emission is only emitted from the cold outer part of the plasma edge and gives therefore no information about the core plasma 2000 gt s00 5 a 1000r An il Pre AN N IM M w HN Sey EL 4 DH Oo al nt DOM H N Fig 2 34 Time development of a Hel spectrum recorded for one viewing line The appearance of the Doppler shifted beam emission peak red indicates the time of He doping At the same time the unshifted peak from the passive edge emission increases In Fig 2 35 the time trace of the global Ze measurement is plotted together with the injected power of the doped He D PINI The time of the He doping can be identified by the increase of the injected power when He is being fed into the ion source between 55 and 55 5 s The subsequent decay of the helium partial pressure in the beam source manifests itself by the decay in the beam current No increase of Zerr can be seen during or after the He injection Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 42 P Ww 49029 Z i Pi 1 8 10 1 7 10 1 6 10 1 5 10 1 4 10 1 3 10 1 2 10 1 1 10 52 53 54 55 56 57 58 Fig 2 35 Time trace of the mean Zer together with the injected beam power from the He doped PINI During and after the He injection no increase in Zer can be seen Chapter 2 HeI
119. of the main widget The start and end times define the time range displayed in the time trace mode and the frames to which the fit procedure will be applied to 2 Spectrum mode The spectra of the selected tracks at the active time frame are displayed Selecting tracks On the right side of the main widget is a list of fields buttons Each row belongs to a track The fields are for Doppler shift and relative calibration of each track The checkbox on the right side of the fields is for activating or de activating the respective track Selecting active time frame A time can be typed in the time field By pressing ENTER the active frame is set to the closest frame and the respective time and frame number are displayed In case spectrum is the active mode the respective spectra are plotted Scrolling from frame to frame is possible with the two buttons to the right of the frame field The active frame can be set interactively in the beam plot Chapter 3 Analysis of the spectroscopic data 78 Zoom unzoom The spectral range can only be selected interactively The start and end value of the wavelength range are set by clicking within a plot spectrum dragging the cursor to a second position within the same plot and subsequently releasing the mouse button The spectral range is applied to all tracks For the input of the fit routine only the displayed part of the spectrum is used In this way disturbin
120. of the spectroscopic data 74 The poloidal flux radius p is a natural coordinate system for flux surfaces Rho poloidal the normalised poloidal flux radius is defined by Y Y Y P pol 9 Y is the poloidal flux the index s refers to the separatrix location and the index a to the magnetic axis p extends from the plasma axis P 0 at magnetic axis to the separatrix pol Pp 1 at the separatrix and beyond p gt 1 pol AUG mapping code A small code map_beam has been written which provides P for equidistant points along pol the beam path by mapping the cylindrical coordinates of the beam trajectory to the flux surfaces see upper graph in Fig 3 2 From the so called AUG ddf database plasma density and temperature profiles can be accessed in poloidal flux surface coordinates By interpolating this data for the p values of the beam trajectory the density and temperature profiles along the beam are derived Fig 3 2 shows the output of the mapping code upper graph and the interpolated plasma density along the beam path lower graph The dots mark the intersection of the lines of sight of the CER diagnostics with the beam trajectory At JET a similar code is being used The precision of the EFIT reconstruction at JET is limited by some peculiarities JET has an iron core transformer and the saturation of the iron must be incorporated into the code adding to its compl
121. ompensate the w non uniform magnification This E 0 shape has been calculated for the S design wavelength A 529 1 nm 1 1 04 by ray tracing through two 6 100 200 300 400 mr position on CCD pixel For wavelengths different from the Fig 2 36 The transmission function of one slit second design wavelength the centre of of the first row The numbers 1 4 indicate the the transmission function and the position of all four transmission function of the first row position of the centre wavelength of the slit array are shifted against each other The direction and amount of the shift depends on the position on the CCD sensor The shift is more pronounced for slits further away from the optical axis This can be seen in Fig 2 37 where all spectra solid lines and transmission functions dotted line of the 4x4 spectra are plotted for a measurement with the central wavelength of 706 5 nm The positions of the central wavelength are indicated by the black arrows and the positions of the centre of the transmission function is at the centre of each plot pixel 50 A problem arises if a spectral line of interest is situated in the wings of the transmission function For a spectrum near the edge of the CCD Image and a wavelength setting sufficiently far away gt 60 nm from the design wavelength the central wavelength is already situated in the wing of the transmission function while the central wavelength of a spectrum in the middle of th
122. on system Because of the finite size of the observation volumes the modelled emission profile needs to be smeared out somewhat to reproduce the measurement The active volume covers typically some 40 mm of the beam trajectory depending on the instrument and the line of sight and an additional factor for the finite diameter of the heating beam typically 150 mm Chapter 4 Reversion code yt t ocs 109 4 6 Comparison with evaluation procedures for other beam emission diagnostics For the thermal He beam diagnostics line intensity ratios are used to derive the local electron temperature and density The origin for line rations which are mainly sensitive to either electron temperature or density is the different behaviour of the rate coefficients for spin changing population of the triplet out of the singlet system and spin conserving electron collisions for details see 44 For the fast He beam line ratios which are mainly sensitive to one of these parameters do not exist anymore because of the following reasons 1 The generation of a fast neutral He beam from accelerated He ions via charge exchange leads to significant initial populations of metastable states 2 S and 2 S which are by orders of magnitude higher than their equilibrium levels which in the fast beam will be reached after passing through a sufficiently large section of the plasma In a pure ground state beam singlet to triplet line rations are mainly sensitive
123. ond one as the spectral analyser The band pass spectrometer is used to fit spectra from 4 fibres onto a row of the CCD sensor without cross talk which makes it possible to image an array of 4x4 fibres simultaneously The band pass spectrometer is a McPherson model 207 of KS7 spectr II analyzer Cerny Turner design with a focal length of 0 67m and a spectr bandpass 600 lines mm grating This kind of spectrometer suffers from anamorphic magnification In order to get an undistorted image on the CCD sensor the entrance slit is shaped to compensate for this effect This shape is calculated for the design central wavelength mirror A 529 1 nm by ray tracing turnable 6 fibers 4x4 fibers through the spectrometer The analysing spectrometer a McPherson model 209 Cerny Turner spectrometer with a focal Fig 2 7 Schematic view of the KS7 spectrometer with two spectrometers in series standard mode of operation length of 1 33 m has been fitted with a 2400 lines mm ruled grating It is possible to bypass the band pass spectrometer by changing the mirror position see Fig 2 7 and using another entrance slit single slit In this mode the number of observable fibres has to be reduced from 16 to only 6 fibres This yields a 4 times bigger wavelength range and there is no distortion of the image in the following called 6 fibres mode During the 1999 campaign several parasitic measurements with t
124. or the selected wavelength was below the detection limit In Fig 2 31 the intensities of the Doppler shifted HeI beam emission peak solid dots and the corresponding unshifted passive emission open symbols are plotted for the pulses 49029 48 For wavelengths where no beam emission signal could be identified the respective symbol is placed on the abscissa The data are taken from the viewing line with the radial position Rmaj 3 8 m for all wavelengths because both singlet and triplet line profiles are close to the maximum at this radial position Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 39 Hel beam emission intensities track 4 JPN 49029 49048 1077 a RR RE ELL C i i E F E 7S 10 eye cls EN an ANIE TEER ETE ED a E EEEE POS oan re E SE E j o i E go E es 3 3 OoOo E z O NE 3 o G O E 15 io ao 10 Ee hia O Dan e en re Ct _o_ E Q 1O co f 3 7 E i po 10 Ne a P 3 H O ee E t P O 14 N eee en eae sie era E RE SA ne Dr hci tne ep as te En Se A E Shy _ a e gt E i J3 oe L i 4 D 5 5 ih unshifted passive emission wat BERN PER 4 A 22 beste ste een 10 E Doppler shifted beam emission 3 12 E E EIE 10 Sa eile ee a ai E E 9 n 9 90 E no beam rt Ir aw i J NO o Sh 0 0 4 emission 09 J signal gt ke ER o ee Hip nts ae ep
125. orizontal sweep displaying the spectra at the start time Absolute Calibration The absolute calibration is carried out by a subroutine called i nt_calibr more details see chapter 3 4 Chapter 3 Analysis of the spectroscopic data T11 Display modes The second framed region comprises the buttons and input fields for changing the plot display There are three display modes available row of toggle buttons 1 time trace mapped radius 2 spectrum 3 beam During the loading procedure the display is set to the beam mode Once all data are loaded the display changes to spectr mode and displays the spectra of the active frame 1 beam mode Display of the time traces for beam power and plasma position beam power The nominal injected power of the selected PINI ppf data NBIx NBLy where x is the number of octant of the beam 4 or 8 and y is the number of the PINI used plasma position The outer position of the Plasma boundary LCFS at the mid plane of the machine ppf data XLOC RBO Start end and active frame are display as vertical lines in the plot Interactive mouse actions A single mouse click left mouse button within a plot sets the active time and changes the display mode to spectrum By clicking within a plot dragging the cursor to a second position and subsequently releasing the mouse button the start and end frame are set Start and end time can also be typed into the respective fields
126. oscopy since dedicated diagnostic beams are only available in a few experiments The emission of the beam particles Do are analysed by the so called beam emission spectroscopy BES however due to the complexity of the spectrum it is not being used as a standard diagnostic for plasma density 27 Both JET and AUG use a fast Li beam for deduction of the edge density profile via impact excitation spectroscopy Li IXS No fast H D diagnostic beam is installed on either experiment for spectroscopic measurements However dedicated beam sources of the heating beam systems are routinely used for CXRS and MSE As shown in chapter 2 2 the HeI beam emission spectrum is much simpler than the beam emission spectrum of hydrogen and therefore offers the prospect for density and temperature diagnostics This has so far not been possible for routine operation as the production of helium beams requires a specialised helium pumping capacity which is not available during normal operation This problem has been overcome by using a so called doped beam 28 In this mode of beam operation a small quantity of helium gas is additionally injected into the beam source operating with hydrogen or deuterium The injection of helium is restricted to time periods during which the fast helium atoms are required This and the fact that the hydrogen gas in the neutraliser is used for neutralisation of the helium ions allows to reduce the helium gas flow to a level at which no add
127. oscopy at JET PhD Thesis JET report JET IR 92 05 H D Falter M Proschek S Menhart F Aumayr D Ciric S Cox A Dines D Godden N Hawkes T T C Jones and HP Winter Rev Sci Instrum 71 3723 7 2000 Appendix 116 29 30 31 32 33 34 35 36 37 38 39 40 41 42 S Menhart On the applicability of fast neutral He beams for fusion plasma diagnostics PhD Thesis TU Wien Austria HP Summers Atomic data analysis structure user manual JET internal report JET IR 94 06 1994 H Anderson PhD thesis Univ of Strathclyde Glasgow 1999 G Dragosits Modellierung von Heliumstrahlen in Randschichtplasmen project report TU Wien Austria N Hawkes and N Peacock Rev Sci Instrum 63 5164 1992 N Hawkes Experimental Studies of Ion Pressure Impurity Flows and their Influence on Transport in the JET Tokamak PhD thesis Imperial Collage of Science Technology and Medicine 1995 W G F Blair R W McCullough F R Simpson and H B Gilbody Excitation of helium in He H4 and He He collisions The influence of metastable helium atoms J Phys B Atom Molec Phys 6 1265 1276 1973 R S Hemsworth et al Nerutralisation mearurements for the JET injector Proc 13 Conf on Contr Fus and Plasma Heating Schliersee FDR 297 300 1986 R S Hemsworth et al Testing of the upgrade JET neutral injector proc 12th Symp
128. ot required and KS7 could thus be switched to HeI beam emission The Hel beam emission experiment were carried out with a doped deuterium helium beam Helium for the beam doping was injected during the flat top phase of the discharge just before the first pellet was injected The plasma parameters of these pulses as shown in Tab VIII varied considerably from pulse to pulse as these experiments included plasma density and heating power scans Consequently the electron density and temperature profiles varied from pulse to pulse and the resulting Hel beam emission profiles were not directly comparable However in this way the observability of the Hel lines could be assessed JET pulse Hel transition wavelength nm Pxs MW Prr MW n dl 10 1 m2 49029 2 P 3 D 667 8 6 z 11 49030 2 P 3 D 587 6 4 5 Z 15 49031 2 P 4 D 447 1 5 4 14 6 49032 2 P 3 S 706 5 4 6 14 4 49044 ay cae 2 1 16 7 0 19 6 2 P 4 S 505 49045 2 P 3 S 728 6 1 16 4 5 21 6 49046 2 P 3 D 388 9 1 16 4 5 22 2 49047 2 P 4 D 4920 1 16 4 5 23 7 49048 2 P 3 D 587 6 1 16 4 5 20 6 Tab VII Wavelength settings of the KS7 spectrometer and plasma parameters of JET pulses used for the HeI beam emission measurements 18 19 10 1999 All discharges involved a toroidal magnetic field of B 3 2 T and a plasma current of I 2 5 MA Grey font is used for pulses where the beam emission f
129. ource after valve closure 28 The background intensity from the last frame before and the first frame after the He doping is being used for normalisation By modulating the beam during the He doping the background incl the passive Hel line can be monitored and used for a more precise background subtraction For the pulse JPN 49555 the doped He D beam has been modulated with 100 ms beam on and off time The exposure time of the KS7 spectrometer has been set to 50 ms This guarantees that one entire frame is taken while the beam is either off or on Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 48 2 0x10 slit 12 1 5x10 1 0x10 intensity counts s i overlapping passive He line Bremsstrahlungs background 60 61 62 63 64 time s Fig 2 39 Time trace of the HeI beam emission integrated over the wavelength range of the Doppler shifted peak for the pulse JPN 49555 Besides the beam emission the integrated value includes the continuous Bremsstrahlung background dark grey and the fraction of the passive He line light grey in the spectral range schematically The arrows mark the frames used for the spectra in Fig 2 40 Fig 2 39 shows the time trace of the He beam emission integrated over the wavelength range of the Doppler shifted peak for viewing line 12 On top of the beam emission the integrated value includes also the continuous Bremsstrahlung background emission dark g
130. ously increasing profiles This restriction leads to a convergence problem illustrated in the right plot of Fig 4 3 The dashed line shows the target profile i e the original temperature profile used to generate synthetic emission data the dotted line shows the profile at the beginning of the fit procedure and the upper solid line shows the resulting converged profile Obviously the value of the knot defined by P3 is too high although the chosen profile type would allow to reproduce the target profile The reason behind this is that parameter P is defined from P through the increment AP34 Reducing the value of P only also changes the value of the nodes P4 and Ps in the way indicated by the lower solid line The profile segments where this parameter change would have a beneficial effect on the total deviation are shaded green and marked with a plus symbol and those with adverse effect are shaded red and marked with a minus symbol One can see that reducing the value P3 would increase the total deviation between emission data and forward calculation A combined variation of P3 increasing and P4 decreasing would lead to a better match However the number of permutations to cover all possible combinations is quite high already for a small number of parameters Equation 15 gives the total number of combinations N for a given number of parameter n is calculated taking positive and negative changes into account N 3 15 For
131. profiles for the most intensive singlet 2 P 3 D and triplet 2 P 3 D line measured during the low density phase of standard H mode discharges In order to obtain a smoother profile an average over 3 measurements from 3 equivalent discharges has been made as shown in Fig 2 26 The singlet beam emission measurement can be reproduced by the model calculations if an initial population of the metastable 2 S state of 1 is assumed For these model calculations scott y_fwd was used with ne and T from Fig 2 24 and 2 25 mapped onto the He beam axis Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 33 2 P 3 D 667 8 nm AUG average standard H mode low density phase 0 0001 calc 1 initial 2 S population calc 0 initial 2 S population 810 average 3 pulses 7 i T Te 5 5 610 2 wn 2 410 E oO a 310 0 100 0 100 200 300 400 500 distance along beam mm Fig 2 26 He beam emission measurements average over AUG 13646 13667 13880 and modelling of the Hel transition 2 P 3 D 667 8 nm The measurements can be reproduced by the model calculations if an initial population of the 2 S state of 1 is assumed 2 P 3 D 587 6 nm AUG 13665 standard H mode low density phase 0 0012 calc 9 initial 2 S population exp AUG 13665 0 001 teases eal aan A 0 0008 H fee eee een een m 0 0006 Hr oa aN mn een en 4 0 0004 Hr A
132. r was not perfect From the standard edge ion temperature measurements KS7 diagnostic emission profiles from the CY CX line 529 1 nm are available for very different sweep pulses L mode H mode different sweep ranges The overlap factors derived for these pulses support the assumption of an incorrect calibration Table IV shows the mean values of the derived relative calibration factors for 5 neighbouring tracks The small scatter in the overlap factors for these quite different plasmas shows that the difference in measured intensity is indeed caused by calibration errors for the different lines of sight fibre id chord number In the case of the calibration factors in table IV the procedure was started at the plasma core and proceeded stepwise towards the edge This implies an accumulation of the error from the centre towards the edge Rmaj_ fibre id lower chord ol factor Tab IV overlap factors ol factor for 5 lines of sight lower chords fibre id of the KS7 system derived from a least square fit of the overlapping C CX emission profile This in situ cross calibration was performed for each pulse In Fig 2 19 the introduced overlap calibration factors are shown for four successive pulses with the KS5a spectrometer at a fixed wavelength setting 667 8 nm 2 50 2 50 gt m53872 m53873 53874 153875 667 8 nm 2 P 3 D
133. rds break even conditions Of special importance in this area has been the use of neutral beams for active charge exchange spectroscopy which has led to a revolution of spectroscopic techniques 1 Charge exchange recombination spectroscopy CXRS makes use of the emission of radiation following electron capture by fully stripped plasma ions from neutral beam particles Low Z impurity density and temperature can be deduced from the intensity and width of the emitted line while the impurity flow velocity plasma rotation is obtained from the Doppler shift 2 3 4 5 6 Beam emission spectroscopy BES is used to measure plasma density fluctuations 7 8 9 10 the local pitch angle of the magnetic field is determined from the polarisation of the Stark or Zeemann emission and the total magnetic field strength from the wavelength splitting of multiplets 11 12 13 14 Another application based on the injection of fast neutral particles is the measurement of the plasma density through Rutherford scattering of these fast beam particles 15 16 17 Several experiments make use of dedicated diagnostic beam lines frequently operating with helium 18 19 20 21 22 Helium beams offer several advantages over hydrogen beams deeper penetration the option of diagnosing alpha particles via resonant charge exchange 23 24 and reduced intensity of background radiation from the scrape off layer which can mask the measurements Optical emission from ener
134. re 14 09 2001 13 23 yttocs VO O 0 200 400 600 800 1000 1200 number of fwd_calc calls Fig 4 6 Output sheet from yt tocs for a undamped variation sequence A description of the output sheet is given on the page 102 Chapter 4 Reversion code yt t ocs 105 Variation of the fitting sequence Instead of following the sequence shown in Fig 4 1 where the density profile is sequentially used to fit the beam emission profile of the singlet line and the temperature profile to fit the emission profile of the triplet line a predefined sequence can be used In the actual yt tocs code this sequence can be defined easily by editing the three defining arrays e loop_seq Array containing the code for the parameter set to be optimised e level_seq Array containing the index for the emission profile to be fitted to e weight_seq Array of the weights Example loop_seq ne te ne ne te te ne te kevell seg si ly 4 La ctl bald oil god foes pedis i weight _seq 0 8 0 7 0 8 0 4 0 4 0 7 0 8 0 7 In order to see the effects from both ne and T profile on the beam emission profile a sequence of combinations of emission profile and fit parameter set can be run through Different weighting for each step results in a defined combination of the results from consecutive steps E g two steps with both n profile variation the first for optimising a singlet emission profile e g 667 8 nm and th
135. rey and the fraction of the passive He line emission light grey within the spectral integration range The modulation of the beam can be seen in the beam emission part of the integrated signal Fig 2 40 shows the 16 Hel spectra 587 6nm 2 P 3 D measured during the pulse JPN 49555 for the viewing lines used In each graph the spectrum measured during the beam on phase at t 61 3 s the background spectrum and the difference between the two blue are plotted The background spectrum is the linear interpolation of the spectra measured during the beam off times at t 61 2 and 61 4 s Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 49 Fig 2 40 The Hel 587 6 nm 2 P 3 D spectra measured in the pulse JPN 49555 for 16 viewing lines In each graph the spectrum measured during a beam on phase t 61 3s the background spectrum linear interpolation of the spectra from t 61 2 and 61 4 s and the difference between the two blue are plotted The three arrows in Fig 2 39 indicate the timing of the spectra used for the background calculation grey and for the beam emission black Spectra with well separated beam and passive emission e g 2 and 3 plot in the first row of Fig 2 40 show that the correct background subtraction is being applied as the passive emission peak of these spectra disappears completely and only the Doppler shifted beam emission is left By modulating the bea
136. rise causes an accordingly higher power load of the beam duct and must therefore be limited At JET it was possible to convert a beam box to pure He beam operation This requires to cover the cryo pump with an argon layer before each pulse which is time consuming and also reduces the reliability of the system Furthermore under normal operating conditions the influx of He into the plasma is unwanted and operation of the beam injector with He was only performed very occasionally Full Energy Neutral Injector Box lon Dump Magnet Neutraliser Central Support Column Scrapers Ir y TY eee yA N Li Ett lt _ ieee L DENE ERREL m jee er Torus Rotary Port Valve N Calorimeter N I lon Source Fractional Energy Dumps ELEVATION OF JET NEUTRAL INJECTOR Fig 2 1 Side view onto the JET neutral beam injector Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 9 Situation at JET JET has two beam injectors with 8 beam sources each so called Positive Ion Neutral Injectors or PINIs The injector installed at octant 4 was operating with 80 kV beam sources rated for approximately 50 A of deuterium beams The injector at octant 8 was operating with 140 kV beam sources rated for 30 A of extracted deuterium current The 8 beam sources are arranged in two vertical banks of four sources each Fig 2 1 shows a side view onto the JET neutral be
137. rmalised 100 0 100 200 300 400 500 distance along beam mm Fig 2 54 Triplet beam emission 2 P 3 D at 587 6 nm of the pulse JPN 53881 Open symbols are the measurements the solid curves shows the model calculation with scotty_fwd and the dashed line results from the calculations with HEBEAM In Fig 2 55 the modelled populations of the ground state and the two metastable states are compared from the two codes A clear difference in the decay of the initial metastable population can be seen The longer decay lengths in case of the scotty _fwd modelling about twice the values from HEBEAM gives rise to the wider peak widths Well inside the plasma z gt 1000 mm the initial metastable population has decayed and the equilibrium population reached flatter part The decay length of the equilibrium population are quite similar from the two models Chapter 2 Discussion of the Results 67 JPN53881 attenuation 70 keV He beam scotty_fwd
138. rticles deuterium into the hydrogen like C state is used by the CXRS diagnostics The observation systems have about 15 viewing lines crossing the beam at different locations thus covering the plasma from the edge to the core Both the CX and the direct beam emission are located at the position of the beam neglecting plume and halo effects Therefore the measurements are local and the measured beam emission and CX signals are located at the intersection of the viewing lines with the beam so called active volumes The spectrometers used at both AUG and JET can be set to a wide range of wavelengths which covers the visible n 3 2 and n 4 2 transitions of atomic helium cf Fig 1 2 Hel spectrum A typical HeI beam emission 2 P 3 D 587 6 nm spectrum left plot is shown in Fig 2 3 together with a Dg beam emission spectrum right plot measured with the same diagnostic JET KS5 In order to be able to compare the peak widths the Hel beam emission peak red shaded is reproduced in the right hand plot not at its correct wavelength Hel 23P 3 D Spectrum active Da spectrum 1x1026 edge D18 edge y _8x1 02 emission 7 2 14 c Doppler shifted vi 3 25 12 g 6x10 beam 7 Ss signal 10 2 4x1025 g 0 8 S 06 S u 2x102 gt 04 F a 0 2 0 0 0 586 5865 587 5875 588 A nm 650 652 664 656 658 A nm Fig 2 3 Comparison of the 2 P 3 D 587 6 nm Hel beam emission spectrum lef
139. rved line the signal to noise ratio is excellent in a wide range and the decay can be observed over two decades From z 200 to z 300 the decay of the triplet line is very well represented by an exponential decay with a decay length of 39mm In that region the density of the plasma is almost constant at ne 4 10 1 m3 and the electron temperature increases from 900 to 1200 eV Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 60 Hel 2 P 3 D beam emission 72 6 keV pure He beam 17 10 SS a a re ee ee ee re A zZ 16 S10 2 S Cc 2 wn aa E f i 2 E 10 l l j D JPN 53879 587 6 nm si exponential fit ft nn n
140. s auch Temperaturprofile konnten aus den Emissionsprofilen zweier unterschiedlicher Heliumlinien extrahiert werden Damit konnte gezeigt werden dass schnelle He Strahlen f r Elektronendichte und temperaturdiagnostik geeignet sind und dar ber hinaus auch f r die Bestimmung anderer Plasmaparameter anwendbar sein k nnten Abstract A precise knowledge of the plasma edge parameters is essential for the development of reactor relevant plasmas High spatial and temporal resolution is required in order to resolve the steep profiles of H mode plasmas in so called advanced scenarios A good temporal resolution is advantageous for investigating edge instabilities ELMs and fluctuations Active spectroscopy of injected atomic beams is a well established diagnostics for a wide range of plasma parameters In this thesis the development of an active beam emission diagnostics with fast neutral He beams is described and its applicability as density and temperature diagnostic investigated This thesis mainly deals with proof of principle measurements performed at two of the leading fusion experiments ASDEX upgrade in Garching GE and JET in Culham UK For generation of fast He beams the injectors of the neutral particle heating system have been used to produce either pure He beams or He doped D beams For observation of the He beam emission the charge exchange spectroscopic systems have been used at both experiments Several Hel singlet and triplet
141. s could partially be overcome by the in situ cross calibration of neighbouring channels The shapes of the singlet and the triplet HeI beam emission profiles are significantly different The triplet emission profiles are dominated by the initial fraction of the 2 S population and its strong attenuation when passing through the plasma The triplet emission is therefore peaked and only measurable in the outer part of the plasma 200 mm in case of AUG and 300 mm in case of JET The intensity of the equilibrium emission further inside the plasma is much lower and could not yet be resolved by the spectroscopic systems The singlet beam emission profile could be observed over the full range from the plasma edge to its centre However even the most intensive singlet Hel beam emission 2 P 3 D was about by one order of magnitude lower in intensity than the maximum of the strongest triplet emission 2 P 3 D As the equilibrium emission can be measured emission originating from the initial metastable 2 S population can be seen as a significant peak near the plasma edge This emission is again dominated by the strong attenuation of the metastable singlet population passing through the plasma By comparing model calculations with the measured profiles an initial metastable 2 S fraction of about 1 for a doped He beam and less then 0 5 for a pure He beam could be determined The initial metastable fraction differs as the neutraliser target consists main
142. s not well known due to the interaction between the powerful heating beam and the target gas From the observed degree of neutralisation it could be deduced that the actual target thickness is only in the range of 1 3 to 1 2 of the target thickness in absence of the beam 36 37 The total metastable He fraction after neutralisation in a helium or hydrogen gas target and the associated cross sections have been measured 38 39 By stepwise solving differential equations for the destruction and production of ground and metastable atoms using the cross sections from the above papers the population from an initially pure ion beam which has passed through a neutraliser with given target thickness can be calculated Fig 2 52 shows such results for a He target and Fig 2 53 for a hydrogen target respectively The range of the probable target thickness of the JET neutraliser is indicated by dotted lines From our Hel beam emission measurements only the 2 S metastable fraction could be estimated By comparing model calculations with the measured 2 S 2 D profiles an initial metastable 2 S fraction of about 1 for the doped He beam D gt gas target with energies of 60 70 and 135 keV and of less than 0 5 for a 73 keV pure He beam He gas target could be determined This is in agreement with 35 where it is stated that the metastable population is expected to be predominantly in the 2 S state Chapter 2 Discussion of the Results 65 ne
143. ssion is already included in the reversion code but only as a global parameter The potential of developing this sensitivity to impurity levels into a Zerr profile diagnostic has to be examined further The profiles obtained from the JET sweep experiments are suitable for testing of the beam model and the development of the reversion code However emission profiles of this quality are only available for one pulse type L mode For benchmark purposes and further development of the code beam emission profiles from other pulse types with different electron temperature and density profiles are needed In situ cross calibration of neighbouring channels by sweeping the plasma across the viewing lines was very successful at JET The application of this method to AUG would improve the quality of the emission profile Since the last measurements the alignment and calibration of the CER diagnostic at AUG has been improved which should lead to a higher beam emission signal and a lower scatter of the profiles The settings of the spectrometer slit width and exposure time offers an additional handle for optimising the signal intensity For the strongest lines the noise are dominated by the plasma even at an integration time of only 50 ms and therefore also the prospect of measuring Chapter 5 Summary Conclusions and Outlook 113 plasma fluctuations is given This would require a faster detector with a higher sensitivity e g photomultipli
144. ssion measurements at JET by using KS7 it turned out that this diagnostics could only be used in its 6 fibres setup see chapter 2 2 2 with a corresponding loss in radial resolution introduced 2 3 1 Radially swept high clearance X point L mode plasma This type of plasma has been used previously for measuring the edge density with an interferometer channel which on its own can only measure the line integrated density By sweeping the plasma in radial direction the line integrated density varies with time and the local density can be extracted by means of an Abel inversion Fig 2 12 shows that the edge density profile measured in this way is well reproducible which is confirmed by edge LIDAR measurements derived from Thomson scattering 2 5x10 9 9 2 0x10 1 5x10 9 1 0x10 9 5 0x1018 Ne 1 m 360 Fig 2 12 Density profiles for a series of identical L mode sweep pulses derived by the Abel inversion of edge JPN 52796 57 13 s LCFS KG1v LID3 Abel inversion JPN 52796 core LIDAR 52798 52803 edge LIDAR 370 380 Rmaj cm interferometer data during sweeping To recover in radial resolution the sweep experiment was 400 For this plasma sweeping a so called L mode plasma has been used in order to avoid the noise introduced in H mode plasmas by ELMs One consequence of using a L mode plasma was that the heating power had to be kept below the H mode threshold
145. standard H mode pulse was used for measuring the beam emission of all visible He lines which were accessible the spectrometer could not be set to wavelengths above 7000 A Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 29 syun que i 3 3 S 3 3 S a i i 2 i je i 12 4 t He F T i 1 i H i i E i i i i ke i 3 1 He V x i wi i 7 gt i lt F i i i i I i i o gt ec I Q i T fl 1 z D A i i H i 1O i 8 Joo N i T i 1 ab E eee a aimee eer es eee ere Wy erie Fig 2 23 Plasma parameters of the AUG standard H mode discharge 13665 He doping marked by grey bars was performed at 3 5 and 6 1 s The transition into H mode at t 2 s is characterised by an increase in confinement stored energy W unn confinement time T and electron density ne Other abbreviations Py injected neutral beam power I plasma current H passive plasma H emission ELM activity Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 30 The ramp up of the plasma current up to Ip 1 MA occurs during the first second of the discharge ohmic phase Subsequently the NBI heating power is being ramped up between 1 4 and 2 1 s to about 3 75 MW on off modulation At t 2 2s the transition into the H mode occurs This can be seen from the significant increase
146. surements with no light coming through the entrance slits has been made The single measurements are saved by the control software of the spectrometers into a binary file All the files of the calibration measurements of one instrument are copied to a directory including a ASCI file containing a file list with the important information of the measurements Three different IDL programs have been developed for processing the calibration measurements 1 build_sens_file pro 2 merge sens files pro 3 int_calibr pro Chapter 3 Analysis of the spectroscopic data 87 build sens file pro This program reads in the ASCII file containing the file list all the binary files listed there and the calibration file of the lamp used It checks the consistency of the parameters stored in the file list with the information in the file headers of the binary files Sequence of processing the data from the calibration measurements Building mean value and standard deviation for the 10 exposures both measurements background and light Background subtraction dark count rate Calculate the wavelength for each pixel of the spectrum Interpolate calibration curve of the lamp for each pixel corresponding wavelength Calculate sensitivity for each pixel from mean value of the measurement and the interpolated lamp emission Store the calculated sensitivity in a three dimensional matrix pixel wavelength track The 3D matrices of sensit
147. t He ions are neutralised in pure helium rather than in a deuterium dominated He D gas mixture for the doped beam for details see chapter 2 6 2 2 The helium beam current is higher in the case of the pure helium beam and the beam divergence is likely to be lower Both effects are leading to a higher He flux density of the pure helium beam Fig 2 47 shows the Hel emission profile from the 2 S 3 P singlet line at 501 5 nm The transition is sufficiently intense to give a profile as obtained from the smoothed data 9 point average Hel 2 S 3 P beam emission m 72 6 keV pure He beam 1 10 T T T T T T T T T T T I T T T T T T T 8 10 A T 610 S i xe wn 2 5410 Eee ee ae OEE ase et ERLERNEN A oO t f 8 gt i i O pre Lae Haein en E a a ee 4 oO i 210 cz JPN 53880 data straight bo a EE JPN 53880 9 point average r i i 0 50 100 150 200 250 300 350 dist along beam mm Fig 2 47 Hel emission profile from the 2 S 3 P singlet line at 501 5 nm Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 58 The singlet profiles from the doped beam show their peak at about z 100 mm which is characteristic for emission originating from metastable atoms This peak is not present in singlet profiles from the pure He beam indicating that the pure He beam has a considerably lower metastable 2 S fraction than the doped beam
148. t graph with the D spectrum right graph The shaded peaks He red and D blue results from the respective excited beam particles For the Hel spectrum the spectrometer was used with a relatively wide opening of the entrance slit in order to get high signal intensities This leads to an instrumental function of the spectrometer with a typical width of Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 13 0 1 0 2 nm FWHM Therefore the real width of the peaks in Fig 2 3 are significantly smaller The highest peak of the spectrum is the unshifted passive plasma emission from the edge The Doppler shifted emission from the excited beam particles He red D blue is marked by shading Obviously the Dg beam emission spectrum is much more complex than that from the Hel emission due to four facts 1 Helium in contrast to deuterium is a minority species of the plasma and therefore the charge exchange halo of the beam is much weaker 2 The linear motional stark splitting of the Dg is much larger than the quadratic effect in the case of He which is below the resolution of the spectrometer 3 The deuterium beam in contrast to helium is composed of three components with 100 50 and 33 of the beam energy respectively 4 The passive Hel emission appears within a narrow peak and therefore well separated from the beam emission even for relatively small Doppler shifts Helium is a common impurity
149. t position with Rmaj lt 3 58 m cannot be accessed with the lower chords The optical head for the upper fan of chords had just been installed during the shutdown prior to the operating campaign and no absolute intensity calibration was available at the time of the measurement Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 47 Several attempts to cross calibrate these fibres against the old lower chords were either unsuccessful or of low quality It was also attempted to use the Bremsstrahlung background for a cross calibration This failed as well since the Bremsstrahlung signal for viewing lines close to the plasma edge most of the KS7 viewing lines is too weak The signals of the 667 8 nm Hel emission generated by a short D beam pulse into He gas were extremely weak SNR about unity Possible differences in the beam alignment with and without magnetic fields added additional uncertainties and this attempt had also to be discarded A comparison of carbon concentration profiles derived from CX measurements by KS4 5 and KS7 was not successful either This was due to the large error in the measurement of the C profile the different beam energies 75 and 135 keV used for the two systems and different wavelengths used for this calibration attempt and the Hel beam emission time trace procedure modulated beam JPN 49555 Since the Doppler shift of the HeI beam emission observe
150. t weight Overrules the standard weighting by a constant value for all pixels In the standard mode the weighting decreases linearly with the distance outside the range of interest defined by the two Hel peaks shifted and unshifted Overlap calibration In horizontal sweep experiments with a sufficiently large sweep range the mapped data for adjacent viewing lines are overlapping Due to inaccuracies in the absolute calibration the profiles of neighbouring viewing lines are shifted in intensity against each other By introducing an additional track dependant calibration factor the overlapping profile parts can be matched against each other leading to an increased quality of the profile shape The automatic procedure performs a least squares optimisation of the calibration factors During the run of the procedure all the overlapping profile sections are plotted The resulting factors are displayed in the second column of the track information table Below the table is a button set to 1 for resetting this additional calibration factors to 1 The size of the different active volumes are not taken into account which means they are included in the track dependent calibration factor Chapter 3 Analysis of the spectroscopic data 83 3 4 Factors influencing the accuracy of the fit result 3 4 1 Drift of the Doppler shift during sweep experiments The Doppler shift is estimated by the analysis software he_ wid taking the measured beam
151. tabases and can later be accessed through analysis software The amount of data was particularly large for the plasma sweep experiments at JET see chapter 2 3 in which spectroscopic data for many time frames and channels have been stored This large amount of data can only be handled by means of automatic access and evaluation software For this purpose the code he_wi d was developed In chapters 3 1 to 3 3 the algorithm and the application of this analysis code and its subroutines is being described IDL Version 5 3 from RSI research systems INC Chapter 3 Analysis of the spectroscopic data 71 3 1 Spectral fit procedure hespec_fit mpfit A typical Hel beam emission spectrum from a single viewing Hel 2 P 3 D Spectrum line is shown in Fig 3 1 The 1x1026 intensity of the beam emission edge corresponds to the area under the 8x102 gt emission Doppler shifted peak shaded g area In the algorithm used a 5 p g a a eam curve fit of the spectrum is 5 signal performed and the beam emission gt 25 ee z 4x10 intensity is calculated from the S resulting fit parameters peak 2x1025 width and height This procedure ensures that the 0 back d d tiall nn i e 586 5865 587 5875 588 A nm overlapping lines if present passive emission or impurity Fig 3 1 Hel beam emission spectrum of the lines are correctly taken into 2 P 3 D transition 587 6 nm account In case of he_wid software
152. ter had to be multiplied with the factor 0 32 This factor is the ratio of the He flux in the pure and the dopedHe beam and is in agreement with the upper estimation Chapter 2 Discussion of the Results 63 Hel 2 P 3 D 667 8 nm 340 22410 2 T gt n O g wn O oO w 3 c 3 O 15 n 1 2 10 A wn oO 2 5 2 E J D 3 2 610 3 22 100 0 100 200 300 400 500 dist along beam mm calc scotty_fwd init pop 0 metastable calc scotty_fwd init pop 1 2 S 10 2 S factor 0 32 Fig 2 51 He beam emission of the 2 P 3 D transition at 667 8 nm measured with pure and a doped helium beams The modelled beam emission was calculated for an initial metastable 2 S population of zero in the case of the pure beam and of 1 in the case of the doped beam Chapter 2 Discussion of the Results 64 2 6 2 Initial metastable fractions in the He beam The fast He beam consists of atoms produced by charge exchange from accelerated ions These charge exchanged atoms are either in the 1 s ground state or in one of the two metastable states 2 S and 2 S The fraction of metastable atoms depends on the gas used for neutralisation and its target thickness line integrated density in the neutraliser The neutraliser was operated with He in case of a pure He beam and predominantly with D in case of a doped He D beam The target thickness in the neutraliser of neutral beam heating systems i
153. the change in the quadratic deviation of consecutive steps is below a given limit The optimisation of these inner loops is performed by the fit procedure mpf i t which is also used in the spectral fit software he_wi d see chapter 3 1 The code of the outer loop is kept versatile in the sense that the sequence of variations e g Ne Te Ne Te and the choice of the respective emission profiles to be fitted ones can be changed easily More details on the sequence of variations can be found in chapter 4 4 The straightforward approach is as follows A sensitivity analysis 43 has shown that the 667 8 nm line 2 P 3 D is almost exclusively sensitive to electron density and several of the investigated triplet lines incl the intensive 2 P 3 D transition 587 6 nm are sensitive to both electron temperature and density Therefore within Chapter 4 Reversion code yt tocs 95 the n loop first inner loop the density profile is optimised based on the 667 8 nm emission profile Subsequently the density profile is kept fixed and the temperature profile is optimised based on the 587 6 nm emission profile second inner loop Both loops are being repeated outer loop until a truncation condition is reached This is the case when the change of the total deviations of consecutive steps from the outer loop is below a given limit The total deviation is defined by the sum of the quadratic deviations between measured and calculated
154. the case discussed above 6 parameters the number of forward calculations for one iteration of the fit procedure would be N 728 Taking the typical duration of one forward calculation 5 seconds a single iteration of the fit procedure taking all N combinations into account would last more then one hour and is therefore impracticable Chapter 4 Reversion code yt tocs 100 Additional global parameters Some additional parameters i e parameters which do not define the shape of ne and Te profile can be included in the list of parameters to be adjusted by the fit routine e The initial population of the He metastable singlet level 2 S e The initial population of the He metastable triplet level 2 S e The shape of the Ze profile Fig 4 4 The Ze profile is assumed to be parabolic The mean value is treated as input parameter while the value of Zet at a certain position e g the plasma core can be used to improve the fit This definition allows to define the parameter range for hydrogen plasmas to 1 lt Leff core lt Zeff mean Zeff core lt Zeff mean A flat profile is defined by Zeff core Zeff mean parameterised Zeff profile Zeff mean Zeff Zeff core 0 20 40 60 80 100 position in plasma O edge 100 core Fig 4 4 Parametrisation of the Zeff profile defined by only two values Zetrmean given and fixed and Zetrcore to be adjusted by the fit procedure This de
155. the curve fit is performed by a subroutine called hespec_ fit The model function f x for the fit is defined as follows 2 A background Gauss function at A4 polynomial unshifted passive emission 2 2 1 x A A 1 t Axa A exp frwmm A46 EXP lt frwam Sis 6 2 A 2 Ay shifted Gauss function at A4 Ag Gauss functions Doppler shifted beam emission impurity lines The A are the so called fit parameters which are adjusted by the fitting procedure to match the data of interest The background of the spectrum is represented by a polynomial 2 Order parameters Ao Az and each emission line by a single Gaussian distribution function Impurity lines can be added to the two Hel emission lines consisting of the passive emission from the plasma edge and the Doppler shifted beam emission Every Gaussian peak Chapter 3 Analysis of the spectroscopic data 72 is defined by its spectral position A4 Az Ag A3i s its peak height As Az A3i 7 and its peak width As A7 and A3i 7 Using the full width at half maximum FWHM for the peak width parameter is convenient but requires an additional factor frwum in equ 6 Using the definition of the FWHM and the Gaussian function FWHM 2 1 m gt 7 the resulting value for frwumis given by fewa 2 2 In 1 2 2 355 8 The subroutine hespec_fit itself calls the procedure mpf it fun which is part of the freeware mp
156. the position on the CCD sensor black arrow The signal intensities can be corrected for most of the slits by dividing the measured spectrum intensity by the normalised transmission function However the noise in the range of the wing is increased significantly leading to a big scatter for the derived beam emission profile For wavelength far from the design wavelength gt 60 nm the introduced error thus becomes intolerable large 2 Problem for the absolute calibration due to different sets of chord fiber slits The optical system of the KS7 diagnostic can be divided into the optical head mounted in the port of the torus the fibres leading from there to the diagnostics room and the spectrometer incl fibres from the fibre panel to the entrance slits The configuration of chords viewing lines fibres and entrance slits can be changed on the fibre panel This makes the system flexible but introduces problems with its calibration For the absolute calibration the transmission of the whole system for 5291A has been measured for a limited number of chord fibre slit configurations Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 46 Subsequently the total transmission of the system has been divided into the transmission of chord fibres and entrance slit by a least squares fit using the measured set of calibrated configurations Because this division is not exact errors are likely to
157. track by the fitting procedure The starting point for the calculation is the statistical error of the detector counts gain for each pixel of the spectrum Additional calculations of the error are done by the mpfit code some more information see http cow physics wisc edu craigm idl fitting html There appear to be some inconsistencies in the error bars coming from mpfit Therefore the error bars should not be used at the present stage of the software development For the integral intensity parameter the program calculates an error which is defined by the difference between the area under the fitted peak and the integral value This error is regarded as a meaningful empirical value Chapter 3 Analysis of the spectroscopic data 80 Bottom row of buttons fields save ps saves the currently displayed plot to a post scrip file The user is prompt to enter the name of the file in the terminal window save ascii saves the currently displayed plot to a ASCII file The user is prompt to enter the name of the file in the terminal window error A toggle button to turn the display of the error bar on and off log A toggle button for changing between logarithmic linear y axis display unzoom To go one step back in the zoom history replot Not all changes will automatically update the plot In these cases clicking the replot button will update the plot Ol_cal Calculates the relative calibration factors see overlap
158. tt 43 920 1983 R J Fonck D S Darrow K P Jaehnig Phys Rev A 29 3288 1984 A Boileau M von Hellermann L D Horton J Spence and H P Summers Plasma Physics Controll Fusion 31 779 804 1989 W L Rowan R D Bengtson R V Bravenec H He J Jagger D M Patterson D W Ross P M Valanju A J Wootton E S Marmar J H Irby J A Snipes and J L Terry Rev Sci Instrum 68 300 3 1997 R J Fonck P A Duperrex S F Paul Rev Sci Instrum 61 3487 1990 R D Durst E R Denhartog R J Fonck J S Kim and Y Karzhavin Rev Sci Instrum 66 842 4 1995 G Schilling S S Medley and S J Zweben Rev Sci Instrum 61 2940 2 1990 H Evensen D Brouchous D Diebold M Doczy R J Fonck and D Nolan Rev Sci Instrum 63 4928 30 1992 F M Levinton R J Fonck G M Gammel R Kaita H W Kugel E T Powell D W Roberts Phys Rev Lett 63 2060 1989 K Takiyama T Katsuta H Toyota M Watanabe K Mizuno T Ogawa T Oda J Nucl Mater 241 43 1222 7 1997 N Nishino H Kubo A Sakasai Y Koide N Akaoka T Sugie Y Tsukahara T Ito and H Takeuchi Rev Sci Instrum 62 2695 9 1991 R Gianella JET report JET P 93 47 1993 Publication officer JET Joint Undertaking Abingdon OX14 3EA United Kingdom Appendix 115 15 16 17 18 19 20 21 22 23 24 25 26 27 28 E V Aleksandrov V V Alfrosimov E L Berezovsk
159. ure will fail As a rule of thumb if the merged peak can be visually identified as two peaks the procedure has a good chance to succeed beam emission arb units JET53874 KS5a 667 8 nm 2 P 3 D 8 10 rreri aa ae ae ie re troie riro m oo gt oa oO N per oO oO oO oO Oo oO Tv to N N N N R R R R R R i 2 Fs a a a a a a a i oO to R I r Heibeam _ SP oa a ee ae a 1 5 2 Doppler shift nm N ol I i oo eae en a 3 5 EEE PO CRONE ae el UA at RO SAPO STMT at ere et 3900 3800 3700 3600 3500 3400 3300 3200 3100 radial position along beam mm Fig 3 7 Intensity and Doppler shift of the Hel 2 P 3 D beam emission measured during a sweep experiment At the location with high emission gradient a changing Doppler shift is noticeable Chapter 3 Analysis of the spectroscopic data 85 3 4 2 Dead pixel of the CCD sensor For calibrating the KS4 5 diagnostics a white light source of know RER i dead pixel position emissivity 1S observed see this measurement f calibration measurement chapter 3 5 Some pixels picture elements of the CCD sensors show for the whole wavelength range significant lower sensitivity than intensity arb units their neighbouring pixels These so called dead pixels can be identified in Fig 3 9 In order to take the pixel dependence into et oes 6645 6650 6655 6660 6665 account the sensitivity
160. usion plasmas List of Symbols Ne electron density 1 m Te electron temperature eV Lett effective nuclear charge Z charge of impurity bnz td Sa Laz j Rmaj Major radius of the TOKAMAK plasma m Z distance along beam axis mm z 0 for LCFS Computer codes and their names are written in this font Danksagung Ich m chte mich an dieser Stelle herzlich bei meinem Doktorvater Herrn o Univ Prof Hannspeter Winter f r die Anregung dieser Arbeit und deren stete F rderung bedanken Vor allem meine damit verbundenen Aufenthalte bei JET und AUG waren f r mich sehr lehrreich Durch das Vertrauen das er in mich gesetzt hat hatte ich auch w hrend l ngerer Auslandsaufenthalte weitgehende Freiheit Die zahlreichen internationalen Workshops und Konferenzen die ich dank Prof Winter besuchen konnte haben mir einen Einblick in die multidisziplin re Welt der Fusionsforschung gew hrt und auch sehr zu meiner Motivation beigetragen Besonders dankbar bin ich Herrn Dr Hans Falter von dessen Erfahrung ich w hrend unserer engen Zusammenarbeit sowohl fachlich als auch pers nlich sehr profitieren konnte Nicht zuletzt dank seiner Hilfe wurde ich bei JET so herzlich aufgenommen Mit seinen Anregungen Diskussionen und Hilfestellungen hat er wesentlichen Anteil am vorliegenden Ergebnis F r die zeitaufwendigen Korrekturen dieser Arbeit bin ich neben Hans Falter auch Prof Winter zu Dank verpflichtet Den Kollege
161. utralisation of He in He 10 Pra total metastable fraction target thickness cm neutralisation of He in H 60 keV 75 keV 135 keV total metastable fraction target thickness cm Fig 2 52 2 53 Total metastable fraction 2 S 23S of He beam obtained by neutralisation of He in a gas target for different beam energies The probable target thickness range of the JET neutraliser is indicated by dotted lines Left graph He gas target Right graph H gas target Chapter 2 Discussion of the Results 66 2 6 3 Discrepancy with the modelled triplet profiles For all beam energies the singlet beam emission could be well reproduced by the model calculations scott y_f wd see Fig 2 41 For the triplet emission profiles on the other hand a significant discrepancy occurred for the width of the peaked profiles at all beam energies In Fig 2 54 the triplet emission 2 P 3 D at 587 6 nm measured for the pulse JPN 53881 open symbols is compared with two different model calculations using the same input parameters The solid line shows the result of the scott y_fwd calculations and the dashed curve resuts from the HEBEAM code from M Brix The intensity of all profiles was normalised to 1 Hel beam emission 2 P 3 D 587 6 nm 3 D norm SCOTTY ADAS ID norm HEBEAM M Brix JPN 53881 KS7a emission intensity no
162. vides absolute values for the density and temperature profiles This output is only correct provided if the spectrometer is absolutely calibrated the atomic data are without errors and the current density profile of the beam is known The absolute calibration of the spectrometer is a difficult task and therefore its accuracy is limited and can change with time e g darkening of the window The atomic cross sections are normally regarded as more reliable in the energy dependence than in the absolute value and finally the current density profile in particular of the doped beam is only approximately known Therefore it is necessary to calibrate the output of the code against measurements from other diagnostics In general one calibration factor for each of the beam emission profiles used in the fitting routine is required in the code the array m factors Reliable measurements of the plasma core density are generally available and therefore an automatic determination of the calibration factor would be possible on a shot by shot basis However this calibration of the system is actually only required in case the calibration of the spectroscopic system has changed 4 5 2 Initial metastable population The fast neutral helium beam consists of atoms produced by charge exchange from accelerated ions these atoms are either in the 1 S ground state or in one of the two metastable states 2 S and 2 S In the following the population of the latter two at the
163. viewing lines it was possible to check the calibration of the viewing lines and to cross calibrate them where required Fig 2 17 shows the photon flux of the HeI beam emission 667 8 nm measured with the KS5a system at different lines of sight different colours plotted vs time Over the time span displayed 57 60s the plasma has been moved horizontally by 120 mm across the viewing lines The upper graph of Fig 2 18 shows the same data mapped to the respective major radius at t 58 s One can clearly see that the different profile sections match very well in shape but the absolute intensities differ from each other To overcome these inconsistencies an additional calibration factor overlap factor has been introduced for each track These factors were derived by matching overlapping areas of adjacent channels by means of a least squares procedure The individual calibration factors are restrained by the condition that the average over all tracks is unity This leads to unambiguous values and keeps the influence on the absolute calibration of the data small Chapter 2 HeI beam emission measurements on large and medium size fusion experiments 24 The result of the overlap calibration of the data from Fig 2 17 is shown in the lower graph of Fig 2 18 The observation that the individual viewing lines require individual calibration factors to yield a smooth profile indicates that the calibration of the spectromete
Download Pdf Manuals
Related Search
Related Contents
Intel Server 747116-011 User's Manual Fireball Hex 72™ Télécharger la fiche technique contrat de location vierge Lithonia Lighting FM54 ACLS LP M4 Installation Guide Chief FSB4203B mounting kit 『GLOIRE Reserve ドライバー』 2012年12月より発売(pdfファイル) Harmonic Measurement User Manual Kenmore 50 gal. Liquid Propane Water Heater Energy Guide Copyright © All rights reserved.
Failed to retrieve file