ebook img

Direct observation of current in type I ELM filaments on Asdex Upgrade PDF

0.99 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Direct observation of current in type I ELM filaments on Asdex Upgrade

Directobservation ofcurrent intype I ELMfilaments onASDEX Upgrade N. Vianello,1 V. Naulin,2 R. Schrittwieser,3 H.W. Mu¨ller,4 M.Zuin,1 C. Ionita,3 J.J. Rasmussen,2 F. Mehlmann,3 V. Rohde,4 R. Cavazzana,1 M. Maraschek,4 and the ASDEX Upgradeteam4 1Consorzio RFX, Associazione Euratom-ENEA sulla Fusione, Padova, Italy 2Association EURATOM-RisøDTU, OPL-128 Risø, DK-4000 Roskilde, Denmark 3AssociationEURATOM/O¨AW,InstituteforIonPhysicsandAppliedPhysics,UniversityofInnsbruck,Austria 4Max-Planck-Institut fu¨r Plasmaphysik, EURATOMAssociation, Garching, Germany (Dated:January17,2011) 1 Magnetically confined plasmas are often subject to relaxation oscillations accompanied by large transport 1 events.Thisisparticularlythecaseforthehighconfinementregimeoftokamakswheretheseeventsaretermed 0 2 edgelocalizedmodes(ELMs).Theyresultinthetemporarybreakdownofthehighconfinementandleadtohigh powerloadsonplasmafacingcomponents. PresenttheoriesofELMgenerationrelyonacombinedeffectof n edgecurrentandtheedgepressuregradientswhichresultinintermediatemodenumber(n=10 15)structures a ∼ − (filaments)localizedintheperpendicularplaneandextendedalongthefieldline. Itisshownherebydetailed J localizedmeasurementsofthemagneticfieldperturbationassociatedtoanindividualtypeIELMfilamentthat 4 thesefilamentscarryasubstantialcurrent. 1 PACSnumbers: 52.35.PY52.55.Fa52.55.RK52.70.Ds ] h p Edge localised modes (ELMs) are short (ms) breakdowns DOP - m of the high confinement regime (H-mode), which is envis- 0.4 0.6 0.7 0.8 0.9 50 s aged to be used in power producing tokamaks. Due to the 40 a highpowerfluxesassociated,thepresenceofELMsposesde- z] 30 1.0 pl mands on the design of plasma facing components, that are f [kH 20 ce 0.8 EInltmer Elm . hard to meet. Thus understanding, with the aim of control- 10 n 0.6 s e sic lriensge,arEcLh.MMeovreenotvseri,saopnaertofrfotmhethfeorienmteoressttpfroirofruitsieiosninorfieunsitoedn 0.4 coher 00..24 hy pcilnasamtinags,atnhaeliongtieerseswtiitnhEsLpoMrapdhicyseixcpsliosseivnehaenvceendtsbayssoobmseerfvaesd- I [A]sat 00..23 0.0 5 10 15 20 25 30 p 0.1 f [kHz] forexampleinsolarflares[1]orinmagneticsubstorms[2]. 0.0 [ 4.184.204.224.244.26 ELMsareatpresentthoughttooriginatefroma combina- t [s] 4 tionofcurrentandpressuregradientdrivenMHDmodes[3], v 2 and are known to result in a medium number (n 10 15) FIG. 1. Top: Degree of Polarization (DOP) analysis as a function ≈ − 6 of structures [4–7], localised in the perpendicular plane and oftimeandfrequency. Bottom: Ionsaturationcurrent(Isat). Right: 3 extended along the magnetic field. These filaments travel CoherencebetweenIsatandbpcomputedintheELMandinter-ELM 2 throughtheScrapeOffLayer(SOL)andhavebeenmeasured phases. Thetimeintervalsareshowninthebottompanelwithcol- 0. usingLangmuirprobesonvariousmachines,seee.g.[8],and oredboxes. 1 observedthroughhighspeedcameras[7]orGasPuffImaging 9 diagnostic[9]. Inthisletterwepresentresultsofdirectmea- 0 outatASDEXUpgrade(AUG)tokamakbymeansofanewly surementsof all three componentsof the magneticfield per- : constructed probe head described in detail elsewhere [12]. v turbationsassociated to ELM filamentstructures in the SOL i Theprobeheadconsistsofacylindricalgraphitecase((cid:31)=60 X togetherwithanestimateofthecurrentcarriedbyfilaments. mm)whichholdssixgraphitepins. Oneofthetipsmeasured r Magnetic fluctuations associated with ELMs are usually theionsaturationcurrent,onetipwasswept, whereasallthe a believed to originate from MHD activity. Measurements of othershadbeenkeptfloating. Insidethecase,20mmbehind the magnetic activity have previously been performed using the front side, a magnetic sensor measuring the time deriva- magnetic pickup coils close to the vessel wall [7, 10] or on tiveofthethreecomponentsofthemagneticfieldismounted. insertable probes [11]. These measurements generally take The sensor has a measured bandwidth of 1 MHz. A similar placefarfromthefilamentsincomparisontotheirradialex- probe with combined electrostatic and magnetic signals has tent. This makes it difficult to observe the magnetic pertur- already been used in AUG [11], but with a wider magnetic bation goingalong with individualfilamentsand to examine coil and a larger radial separation between electrostatic and the magnetic fine structure of the ELMs. Both would result magneticsensors.Theprobehead,mountedonthefastrecip- in important information such as the excursion of magnetic rocatingmidplanemanipulator,isinsertedfromthelowfield fieldlinesfromtheirequilibriumpositionandiftheELMsare side of the torus for 100 ms 12 mm inside the limiter posi- associatedwithreconnectionevents[1]. tion.Thedatahavebeenobtainedintype-IELMyplasmadis- Here we report on measurements of type I ELMs carried charges(# 23158,23159,23160,23161,23163)with a toroidal TypesetbyREVTEX 2 A] 0.4 (a) (a) (b) 5.1 I [0sat.0001..020 0.010 T] 1.7 (c) m 0.005 br bp (b) 0.005 [bλ2−1.7 [T] 0.000 b [T]p 0.000 −5.1 −6.3 −2.1 2.1 6.3 −0.005 −0.005 bλ1 [mT] −0.010 −0.010 4.234 4.236 4.238 4.240 −0.0100−0.00330.0033 0.0100 FIG.3. (a)TrajectoryofELMfilamentassociatedmagneticfieldex- t [s] b [T] r cursions in all three spatial directions. The direction of minimum varianceisshown(blackline)togetherwiththedirectionoftheequi- libriumfield(blueline). (b)Hodogramofthemagneticperturbation FIG.2. (a)Ionsaturationcurrent(b)Poloidalandradialcomponent associatedwithELMcurrentfilamentreconstructedinthemaximum ofthemagneticfield. Theredboxhighlightsthetimeintervalwhen varianceplane.Theredlineshowstheellipticalfit. the two components change their phase relation. (c) Hodogram of theradial/poloidalmagneticfieldcomponent.Theclosedloopinred referstothetimeintervalshighlightedinpanel(a,b). tudeincreases,theradialandpoloidalcomponents(b andb r p )changetheirphaserelationashighlightedbythecolorbox. magneticfieldof-2.5T,0.8MAofplasmacurrent,6.5 1019 Wenowconsiderthehodogramoftheperpendicularmagnetic m 3centralelectrondensityandaq valueof5.2. × perturbation(Fig. 2 (c)), i.e. the magneticfield perturbation − 95 trajectoryintheb b plane,duringthepassingoftheELM. During the analysis we have used the ion saturation cur- r− p Hereoneclosedorbitcorrespondstothetimeintervalmarked renttoinferthepassingofatype-IELMfilamentstructurein in Figure 2 b. Closed loops are compatiblewith the passing frontoftheprobe,asdoneforexamplein[13]. Inanalyzing of currentfilamentsin frontof the probe. Outside the ELM, the ELM events we employ the idea that the magnetic sig- the magnetic field perturbation exhibits an almost linear po- nalduringtheELMcanbeseparatedintodifferentfrequency larizationintheperpendicularplane. Thisshowsclearlythat components.Higherfrequencies,abovefewhundredkHz,we magneticactivityinbetweenELMs(wavelike)differsqualita- presumeto be generatedmostly by Alfve´nicactivity or high tivelyfromthemagneticperturbationduringanELM,which frequencyturbulence.Thesearenotconsideredinthepresent seemstobeduetothemotionofcurrentfilaments. analysis,alsobecauseofthefrequencycut-offbythegraphite shielding[12]. AdditionalMHDactivitywillstillbe present Themeasurementsofallthreecomponentsofthemagnetic inthesignalatfrequenciesbelow20kHz,butduringELMfil- perturbationallowstocheckthealignmentofthecurrentfila- amentsmostofthesignalispresumedtooriginatefromslowly mentsdirectly. InFigure3(a)the trajectoryofthemagnetic varyingcurrentsconvectedwiththefilaments.Thisisjustified fieldexcursionduringanELMeventinallthreemagneticfield bythe so-calledDegreeof Polarization(DOP) analysis[14], componentsisshown. Thetimeintervalselectedcorresponds whichisatestforaplanewaveansatz,quantifyinghowwell totheonehighlightedinFigure2. Aclosedellipticallooply- the relation k B=0 is satisfied. It is based on the evalu- inginaplaneslightlytiltedwithrespecttothelocalframeof · ation and diagonalization of the spectral matrix S= B B , referenceisobserved,asexpectedforfilamentarystructures. h ∗i ji calculatedinFourierspace. TheDOPrepresentsameasureto The directionnormalto this plane can be determinedby us- determinewhetherS representsa purestate quantifyinghow ingtheminimumvarianceanalysis[15]. Themethodisbased much one single eigenvectorapproximatesthe state. A high onthesolutionoftheeigenvalueproblem(cid:229) 3n =1MµBn nn =l nµ valueofDOPimpliesthatthefluctuationsconsideredareco- where MµBn = hBµBn i−hBµihBn i is the Magnetic Variance herentoverseveralwavelengths,andthusthemethodcanbe Matrix,thebracketsindicatingthemeanvaluesaveragedover usedtodistinguishbetweenpropagatingmodesandcoherent the time the structure spends traveling in front of the probe, localized fluctuations. Figure 1, panel (a), shows the results µ,n =1,2,3denotingthecartesiancomponentsoftheX,Y and of DOP analysis as a function of frequency and time. The Z ofthechosensystem,l andnbeingrespectivelytheeigen- temporal evolution of the ion saturation current is depicted valuesandeigenvectorsofthesystem. Theeigenvectorscor- in the lower panel. A sudden drop in the DOP is observed respondingtotheeigenvaluesl ,l andl ,representthedi- 1 2 3 in correspondance with a steep increase of I signals: this rectionofmaximum,intermediateandminimumvarianceof sat implies that the magnetic field fluctuations during an ELM the magneticfield respectively. Theeigenvectorcorrespond- canbebetterrepresentedbycoherentstructuresthanbyplane ing to the minimum eigenvalue corresponds to the direction wavepackets. A coherence analysis between the ion satura- of minimumvarianceand is normalto the plane spannedby tioncurrentandthepoloidalcomponentofthemagneticfield the magnetic field perturbations. This direction is observed (Fig. 1(c))revealsanincreaseofcoherencebetweenthetwo tobeparalleltothatoftheequilibriummagneticfield,shown signals during the ELM activity. A closer look on the be- inthesameplotbyablueline. Thisconfirmsthehypothesis haviorofmagneticfluctuationsassociatedtoanELMeventis that the ELM filament is aligned with the equilibrium mag- showninFigure2(b). Whenthemagneticfluctuationampli- netic field. From the sense of the polarization the direction 3 10 B 0.006 (a) 0.006 (b) 8 M 0.004 bλ2 0246 [T] −000...000000022 bλ2(λ1=a)Δ t1 Δ t2 bλ2(λ1=a) 2 1/22 ) [T] + b(bλλ2100..000024 χχ22 MBiopnoolapro =la 3r .=3 81 .×5 11 0×− 110.0−10. -2 −0.004 −0.006 0.000 -4 4.236800 4.236867 4.236933 4.237000 4.23680 4.23687 4.23695 4.23702 -6 t [s] t [s] -6 -4 -2 0 2 4 6 8 10 12 bλ1 FIG.5.(a)Timetracesofthetwoperpendicularcomponentsofmag- FIG.4. Hodogramscalculatedforamonopolar currentdistribution neticfieldduringanELMfilament.(b)Totalperpendicularmagnetic (M )andbipolarone(B). fieldasafunctionoftime. Superimposedtheresultsofafitforan expectedmonopolar(red)andbipolar(blue)currentdistributionboth modulatedbyanexponentialdecay. ofthecurrentisfoundtobecolinearwiththeplasmacurrent. The method allows the determinationof the rotation matrix, sothatthehodogramintheplaneperpendiculartothecurrent betweenthemaximumofbl andthemaximum/minimumof 1 filaments may be reconstructed. This hodogramis shown in bl ,wherebl andbl aretheprojectionsofthemagneticfield 2 1 2 Figure 3 (b) together with a fit to an ellipse. The shape of in the maximum-intermediate variance plane. Thus within thehodogramcalculatedintherotatedframeofreferencecan theseapproximationsthecurrentmaybeestimatednotingthat be used to determine the type of current distribution associ- |bl 2(l 1 =a)|= 21µ20pIa0 = 4pDµ0tIv0l 1 Applied to the experimental atedtoanELM.TheorieshavebothproposedELMfilaments data,thiscomputationisequivalenttocalculatethequantities tobeassociatedtomonopolar[4,16]andbipolarcurrentdis- depictedinFigure5,whereD t hasbeencalculatedbothatthe tributions [17]. Up to now no clear evidence of one mecha- maximum and at the minimum of bl . The two values D t1 2 nismpredominantwithrespecttotheotherhasbeenreported. and D t are equal to 38 µs and 42 µs respectively. The esti- 2 Figure4showstheanticipatedshapeofthe hodogramin the mateof thecurrentreliesonthe knowledgeoflocalvelocity bipolar and monopolar cases. The hodogram in the bipolar in the l direction. This propagation has been reported for 1 case exhibits a cardioid-likeshape with a cusp at the origin, ASDEXUpgrade,e.g. [11,20]. Theexperimentalsetupdoes which represents a distinct signature and which can not be not allow a reliable local measurements of v or v , neither r p recognized in the experimental data (see Fig 3 (b)). To re- can we rely in the presented shots on correlation analysis of inforce the statement we note that a bipolar-like hodogram wallmountedprobes(whichare 160degreesseparatedfrom with the presence of a cusp has been previously recognized the manipulator in the toroidal direction). Most likely these in[18](cfr. fig3(b))whereithasbeenassociatedtoadirect signals are indeed dominated by MHD mode activity in the measurementof a bipolarcurrentstructures. We emphasize, confinedplasmaratherthancurrentfilamentsintheSOL.As however, that the filaments observed in [18] are of a com- thebestapproximationwethusassumeasradialpropagation pletelydifferentoriginasinducedbyDrift-Alfve´nturbulence themostprobablevaluev =1.2km/sas determinedin [20] r [19]. A further corroboration of the hypothesis of monopo- andcalculatevl =vr/cos(∠(l 1 r)),where∠(l 1 r)isthe lar current filaments results from considering the quadrants naglebetweenl 1 andtheradialdi−rectionwhichforth−epresent 1 occupied by the magnetic field trajectory. Indeed while the caseisoftheorderof7 . From[20]wealsoestimateastan- ◦ monopolar current hodogram regularly occupies two of the darddeviationofv ass =700m/s,whichensurethat71% r vr quadrants, the bipolar one always spans three of the quad- ofthe eventsare observedwithin(v s ). With these val- r± vr rants: thislatterbehaviorisnotobservedintheexperimental uestheaveragedistanceofclosestapproachisapproximately dataascanbeverifiedbycomparingFigure3(b)andFigure 4cmandusingtheaveragevaluebetweentheminimumand 4. These results make us confidentthat the magnetic fluctu- maximum of bl we obtain an estimate of 1.9 kA. The to- 2 ationsobservedareindeedgeneratedbyamonopolarcurrent tal perpendicular field b2 +b2 is shown in figure 5 (b). distribution. Under this basic assumptionthe currentcarried q l 1 l 2 The magnetic field variation can be fitted with a function of bythisfilamentmaybeestimated. Intherotatedframeofref- einretnhceel f1odriraecctiirocnu,lathremtwonooppeorlpaernsdyimcumlaertrmicagfinlaemticenfitelddricfotimng- the form f =(cid:18)√(t−at0)2+b −g (cid:19)e−(t−t0)/t e, t0 being the time ponents may be written as bl 1 = −l r210+Ba02a and bl 2 =−rl021B+0al 21 iunlsatranfitelcdorarnesdptoendthineget-ofothldeinmgatxiimmeu,manodftahsesutmotianlgpethrpeemndaigc-- whereB = µ0I0 anda isthedistancebetweenthetrajectory netic field is the response from a passing monopolarcurrent 0 2p r0 ofthecenterofthefilamentsandtheprobe,representingalso filament. Thegoodqualityofthefitprovidesadditionalsup- the distance of closest approach, r is the radius of the fila- portforthemonopolarnatureofthefilament,alsocomparing 0 mentsandI itscurrent. Thedistanceacanbeapproximated, withthefitexpectedfromabipolarcurrentdistributionshown 0 assuming the filament to propagate with a constant velocity in blue line which exhibitsan higherc 2. The average decay alongthe l 1 direction: a=D tvl whereD t is thetime delay time is determined as t e 200 µs. In order to increase the 1 ≈ 4 PdF thatthesamecurrentdensitywasestimatedin[11]assuming Min Max 14 themagneticperturbationtobe inducedby arotatinghelical 25 (a) 12 (b) structurewithabi-directionalcurrentcloseto,butstillinside, 20 10 mT] 15 # 8 the separatrix. The value foundis higher than the measured b[δλ2105 246 kjsAat/mcu2)r.reCnotndcelnusdiitnygtointthheisiolentt-ebriawseedhativpe(parpopvriodxeidmeavtiedlyen5c0e 0 thatELMfilamentscarryconsiderablecurrentsforwhichwe 10 20 30 40 50 0 20 40 60 80 100 dt [μs] ∠(λ−r) [degree] havefoundareasonableestimate. Themagneticsignalsdur- 1 10 ing ELM filamentsdiffer substantially from wave activity in SOL (c) 8 kA] 6 between ELMs. We have shown that the current in the fila- [M 4 mentsis co-alignedto theplasma currentand approximately IEL 2 of a magnitude as expected for the edge. The current flows 0 along the unperturbed magnetic field lines and has a unidi- −2 0 2 4 6 8 r−r [cm] rectionalnature. This poses the question where the filament sep currents close in the SOL and why such high current densi- FIG.6. (a)JointPDFofD t andbl (l 1=a)(b))Histogramofthe tiesaresustainedintheELMfilaments. We hopethatfuture anglebetweenl 1 andtheradialdire2ction. (c)Filamentscurrentes- experimentswillcontributetoanswerthesequestions,which timatevsdistancefromtheseparatrixestimatedwithvr=1.2km/s. willthrownewlightontheinstabilitymechanismsforELMs. Errorbarsresultsfromthepropagationofvelocityuncertaintyinthe This work, supported by the European Communities under estimateofaandI. thecontractofAssociationsbetweenEURATOMandENEA, Risø/DTU, O¨AW Innsbruck, and IPP Garching, was carried statisticreliabilityofthepreviousestimatewehaveanalyzed out within the framework of the European Fusion Develop- eventsfrom5differentshotswithsimilarconditions. There- ment Agreement. This work was also supported by project sults are shown in Figure 6. In the panel (a) the joint PDF P19901oftheAustrianScienceFund(FWF). of the independent experimental values D t and bl used for 2 thecurrentevaluationisshown,highlightinghowthebulkof the filaments have values of approximately25 µs and 5 mT. In the panel (b) the histogram of the angle between l and 1 [1] W.Fundamenskietal.,PlasmaPhys.Control.Fusion49,R43 the radial direction is shown, showing that the l direction 1 (2007). isacomplexcombinationofradialandpoloidalpropagation. [2] A.Lui,IeeeTrans.PlasmaSci.28,1854(2000). Finally in the panel (c) the current estimated accordingly to [3] P.B.Snyder,H.R.Wilson, andX.Q.Xu,Phys.Plasmas12, the aforementioned formula is plotted vs the position of the 056115(2005). filaments with respect to the separatrix, taking into account [4] A.Kirketal.,PlasmaPhys.Control.Fusion48,B433(2006). thepositionofthemanipulatorandthedistanceofclosestap- [5] M.Fenstermacheretal.,Nucl.Fusion45,1493(2005). [6] A.Herrmannetal.,J.Nucl.Mat.363-365,528(2007). proach estimated. The errors shown represent the influence [7] A.Kirketal.,Phys.Rev.Lett.96,185001(2006). of the velocity uncertainty on the estimate of a and conse- [8] A.W.Leonardetal.,PlasmaPhys.Control.Fusion48,A149 quently on I. The bulk of the distribution of these filaments (2006). are thus observed in the SOL, even taking into account the [9] RMaquedaandRMaingiandNSTXteam,Phys.Plasmas16, uncertainty on v : the median of the distribution of the fila- 056117(2009). r mentsdetectedintheSOLwithintheerrorintheirpositionis [10] H.Takahashi,E.D.Fredrickson, andM.J.Schaffer,Phys.Rev. 1.4 kA, correspondingto a j 4.5MA/m2 for 1 cm radius Lett.100,205001(2008). filaments. Thesevaluesareckon≈sistentwithmeasurementsof [11] A.Herrmannetal.,ininProceedingsofthe22ndEnergyFu- sionConference,Geneva,IAEA-CN-165/EX/P6-1(2008). edgecurrentasseenforexamplein[21]. Ithasalreadybeen [12] C.Ionitaetal.,J.PlasmaFusionRes.Series8,413(2009). supposed in [22] that, in case ELM breakdown is driven by [13] M.Endleretal.,PlasmaPhys.Control.Fusion47,219(2005). peeling instabilities, this will lead to a flattening of the cur- [14] J.SamsonandJ.Olson,GeophysicalJournalInternational61, rent density profile over the separatrix and a subsequent in- 115(1980);O.Santol´ık,RadioSci.38,13(2003). crease of currentsflowing into the SOL. These currentswill [15] D.R.Weimer,J.Geophys.Res.108,12(2003). benearlyforce-freeandwillbeaccompaniedbypoloidalhalo [16] J.R.Myra,Phys.Plasmas14,102314(2007). [17] V. Rozhansky and A. Kirk, Plasma Phys. Control. Fusion 50, currentsclosingthroughthedivertortiles. Actuallythesecur- 025008(2008). rentshavebeenmeasured[12,23]withvaluesuptofewtens [18] M.Spolaoreetal.,Phys.Rev.Lett.102,165001(2009). of kA, supportingthe presenceof rapid flow of toroidalcur- [19] N.Vianelloetal.,Nucl.Fusion50,042002(2010). rents from the plasma into the SOL The resulting histogram [20] A.Kirketal.,tobepublished onPlasmaPhys.Contr.Fusion may also be compared with the estimate for ELM filaments (2010). in JET given in [24]: in this case the most probable current [21] D.Thomasetal.,Phys.Plasmas12,056123(2005). has been found of the order of 450 A, but with a lower ra- [22] E.Wolfrumetal.,inProceedingsofthe22ndIAEAEnergyFu- sionConference,Geneva(IAEA,2008)pp.EX/P3–7. dial velocity postulated. For completeness it must be noted 5 [23] P.McCarthyetal.,inInproceedings30thEPSConferenceon PlasmaPhysics,St.Petersburg,Vol.27A(2003)pp.P–1.064. [24] PMigliucciandVNaulin,Phys.Plasmas17,072507(2010).

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.