1 Plasma formation in MAST by using the double null merging technique P. Micozzi 1, F. Alladio 1, P. Costa 1, A. Mancuso 1, A. Sykes 2, G. Cunningham 2, M. Gryaznevich 2, J. Hicks 2, M. Hood 2, G. McArdle 2, F. Volpe 2, Y. Dnestrovskij 3 1 Associazione Euratom-ENEA sulla Fusione, C.P. 65 Frascati, Roma, 00044 Italy 2 Euratom/UKAEA Fusion Association, Culham Science Centre, Abingdon, OX14 3DB UK 3 Kurchatov Institute, Institute of Nuclear Fusion, Moscow, Russia Outline Start-Up Techniques for Spherical Tori Equilibrium Simulation of the Double Null Merging (DNM) in MAST Experimental Results of the DNM in MAST Magnetic Reconstruction during the DNM Future Perspectives
2 In a Spherical Tokamak A=R/a~1, so very few space is left for the central solenoid (wound around the central rod) MAST Only a small inductive flux can be stored
3 In a ST based CTF (or Power Plant), the central solenoid would be bombarded by neutrons (no space for internal protection): needs of different start-up techniques Steady State ST Reactors must rely upon non-inductive CD (e.g. NBCD) and high Bootstrap + Diamagnetic fraction to sustain the toroidal current
4 A possible solution is to use the flux of the poloidal field coils in order to obtain start-up & initial build-up of I p without central solenoid MAST The Merging/Compression (M/C) scheme (developed in START and successfully used in MAST) inductively forms plasma toroids around a coil internal to the vacuum vessel (P3) and then merges them
5 Up to 400 ka of I p without cental solenoid (decay time ~ 200 ms) Up to 500 ka of I p with solenoid Hot final plasma, due to reconnection Problems in CTF design with M/C : in-vessel coils increase radial build, may generate impurities, need protection from neutrons
6 These problems can be removed by Double Null Merging (DNM) scheme * : Break-down is obtained in a low-order null between two coils external to the vacuum vessel * Y. Ono 20th IAEA 2004 IC/P6-44 No in-vessel coils in CTF with DNM
Experimental set-up in MAST for DNM 7 DNM Simulation Disconnected Central Solenoid Solenoid Feeder on P2 P2 P3 P4 P5 Capacitor Bank on P3 Eddy Currents in passive components has been extimeted by ANSYS code Currents in the PF feeders computed by McArdle code
8 Simulation performed with a free boundary Equilibrium Code with multiple contact points Δψ ind + Δψ res =0.02+ ( 1+ C E ) μ 0 R 0 I p [ L ext + l i 2] Flux balance of formation and merging with 20 mwb lost flux & Ejima coefficient C E =0.7 P2 P3 I p iteratively computed
Equilibrium modelling gives more current than in experiments Without central solenoid I p ~340 ka, lasting ~0.3 s 9 Equilibrium modelling of P4 current Plasma is hot T e (0)~0.5 kev and dense n e (0)~9 10 19 m -3 In experiment P4 current needs earlier rise Good NBI target
10 I p I P4 +I P5 R ext BV ramp-up effect is clearly seen: in red shot BV increases, plasma is bigger, I p current increases (~1.2 MW of NBI added) Difference between M/C and DNM: in M/C I p x R proportional to I P3 in DNM I pl R does not depend on I P3 in M/C W tot proportional to I P3 2 in DNM W tot does not depend on I P3 (30% variation on I P3 )
11 P2 High speed CC: some evidence of plasma ring P3 between P3 and P2 (e.g. #13206) P3 support Fast camera images: visible light suggests merging faster than in equilibrium modelling
12 Magnetic Reconstruction of the DNM initial phase (plasma and passive currents effects must be distinguished) #13201 (ZERO) 4.4 ms Passive effects induced by PF currents and plasma Zero shot (no plasma) passive effects of PF currents: measured PF coil currents are subtracted (casings included) least-square fit on all magnetic probes (~100) determines: Currents in passive elements (7 couples) (cross-check with ANSYS in progress) Spherical external multipoles M ne (r ext ), n=1,3,5,7 describing eddy currents upon vacuum vessel Effect of plasma on passive currents is then similarly analyzed in ( Plasma shots - Zero shot ) Input data for equilibrium solution
#13198 EQUILIBRIUM 4.4 ms Iterative (spherical geometry) Grad-Shafranov solver: (over)determine amplitude of 2 or 3 (given) functional dependences of p(ψ) and I dia2 (ψ) 13 Hollow pressure profile: p(ψ)~(ψ-ψ edge ) -0.5 -(exponent < 0 to get p(ψ) > 0) I dia2 (ψ)~(ψ-ψ edge ) 0.5 Boundary: largest ψ among contact points (none on P3) After equilibrium convergence Bpol fit Flux fit r in r ext
3.6ms 4.4ms 5.6ms merging 7.0ms 9.0ms 14 #13198 Equilibrium reconstruction #13212 - no equatorial plane plasma is produced 4.4ms 6.4ms 8.0ms
#13198 Evolution of discharge after formation 15 9.0ms 14.0ms 16.0ms 20.0ms 28.0ms Pink contours and shade Frascati ODINsph code Flux & Bpol measurements used Only up/down symmetric plasma and eddy currents Blue & black contours EFIT Bpol and outer edge from H α
Comments upon Magnetic Reconstruction Results are preliminar (only few shots analysed, no comparison with M/C), but: 16 Plasma seems to form around P3, and not at the X-point between P2 & P3 It is not clear if a secondary break-down happens on the midplane (this could explain the I p dependence with BV and not with I P3 ) The ratio between I P2-P3 and I P4 is critical (for wrong values plasmoids do not merge, e.g. #13212) One can guess that the presence of coils inside the vacuum chamber does not allow for proper DNM Vloop close to P3 (~100 V) much higher than one at the X-point, moreover the quality of the null is poor
17 What to do to improve DNM experiments on MAST? Try to form plasmoids on X-point between P3 & P5, then push with P4 Insert a toroidal limiter surrounding P3, and/or add a further coil to improve null multipolarity and push plasmoids Risk to still form plasmoids around P3 Is it still possible to obtain break-down?
18 Conclusions DNM experiments on MAST produce hot/dense final plasma that is a good targed for NBCD in view of fully non-inductive start-up & sustainement of the discharge (no central solenoid) There are remarkable differencies between M/C and DNM: in particular, final I p does not depend upon I P3 in DNM, but only on I P4. Moreover clear BV-ramp effects are observed but Preliminar results of magnetic reconstruction seem to suggest that - in present MAST configuration - break-down still occours around P3: the differencies with the standard M/C could be due to the effect of P2 current ramp-down Probably only major modification of the MAST hardware could allow for proper DNM start-up, but the price to pay is to loose the possibility of using M/C
13198 MAGNETOSTATIC 4.4 ms PLASMA SHOT: Zero shot is subtracted, Currents in passive elements (7 couples) due to plasma, added to same currents of zero shot Spherical external multipoles M ne (r ext ), n=1,3,5,7 describing eddy currents due to plasma upon vacuum vessel, added to same of zero shot Spherical internal multipoles M ni (r ext ), n=1,3,5,7 Spherical external multipoles M ne (r in ), n=1,3,5,7 describing plasma current within [r in, r ext ] Magnetostatic assumption: j φ =constant within plasma range [r in, r ext ] Res1 Bpol fit Flux fit r in r ext