1. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Observation and analysis of pellet material  B drift on MAST L. Garzotti 1, K. B. Axon 1, L. Baylor 2, J. Dowling 1, C. Gurl 1, F. Köchl 3, G. P. Maddison 1, H. Nehme 4, A. Patel 1, B. Pégourié 4, M. Price 1, R. Scannell 1, M. Valovič 1, M. Walsh 1 1 Euratom/UKAEA Fusion Association, Culham Science Centre, Abingdon, Oxon, UK. 2 Association EURATOM-Österreichische Akademie der Wissenschaften, Austria. 3 Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA. 4 Association EURATOM-CEA, CEA Cadarache, Saint Paul-lez-Durance, France.

2. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Overview Experimental set-up Macroscopic features Visual analysis Quantitative interpretive analysis First principle simulations Conclusions

3. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 MAST pellet injection system On MAST deuterium pellets are injected vertically from the top of the machine into the high field side of the plasma. –Typical pellet speeds are between 250 and 400 m/s. –Nominal pellet masses are 0.6, 1.2 and 2.4 10 20 atoms. Typical MAST target plasmas: –I p =0.66 ‑ 0.76 MA, –B=0.47 ‑ 0.50 T, – =1.6 ‑ 7.5·10 19 m -3, –T e0 =0.7 ‑ 1.2 keV, –H-mode plasmas NBI heated (P NBI =1.1 ‑ 3.0 MW with neutral beams with energy 65 ‑ 67 keV). top pellet entry outboard pellet entry (not used in this study)

4. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 MAST pellet diagnostics Unfiltered visible images of the complete pellet trajectory inside the plasma taken with a fast camera: –frame rate 5 kfps, exposure time 7  s, –core region of the cloud saturated, –information limited to the edge of the cloud. Narrow spectrum (centre wavelength 457 nm and bandpass 2.4 nm) radiation (mainly brehmsstrahlung) emitted by the pellet cloud recorded by a second CCD camera: –frame rate 30 fps, exposure time 31 ms, –limited field of view including only the final part of the pellet trajectory, –images saturated on a smaller region of the pellet cloud, –more detailed information about the structure of the cloud. Density and temperature profile measured: –every 5 ms with a multiple-pulse, 34 radial points Thomson scattering system, –immediately after the end of pellet ablation with a single-pulse, 300 radial points Thomson scattering system.

5. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Deposition: the inner zone Adiabatic deposition creates a distinct zone:  n e > 0, doubled  lnT e Simulation indicates favourable increase of transport Overtaking the pedestal’s role

6. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Pellet retention time: measurement Encapsulates complex post-pellet losses: depends on fraction of gas/beam fuelling, non-exponential in time and inhomogeneous

7. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Pellet retention time Correlates with status of edge transport barrier Diffusive:  pe l  ( a – r pel )  CUTIE simulation in good agreement

8. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 The ratio  pe l /  E decreases for r pel  a For ITER-like pellets:  pe l /  E ~ 0.2 Further improvement: normalise to  E,pel =  E ( r pel ) (analogue to  E,ped ) Pellet retention time normalised to energy confinement time

9. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Illustration for ITER Assume density controlled only by pellets and  pe l /  E ~ 0.2 Then:  pel ~ 70 Pa m 3 /s ~ 70% of design steady-state value For 5mm pellets, f pel = 4  /  pel, faster than in today plasmas

10. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 EXB drift Pellet material deposited in a tokamak plasma experiences a drift towards the low field side of the torus induced by the magnetic field gradient. E R0R0 BB E  B drift B B

11. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Characteristics of the drift Potentially beneficial effects on the fuelling efficiency, since increases the deposition depth of the pellet material for pellets injected from the high field side of the plasma. Very difficult to observe, because of the fast time scale on which it occurs (~100  s) and the presence of other transport mechanisms in the plasma. Detected in the past on different machines (ASDEX-U, JET, DIII-D, Tore- Supra, FTU and MAST). Since the fuelling of ITER plasma will rely significantly on the beneficial effect of this  B drift to increase the pellet material deposition depth, it is crucial to analyse this phenomenon in detail: –develop codes to predict it, –compare the predictions with experimental results in present machines.

12. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Camera images t=0.2226 st=0.2236 st=0.2244 s Snapshots of the pellet cloud taken during pellet ablation. MAST shot 16335

13. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Timing Relative timing of the camera frames and the high space resolution Thomson scattering profiles. Camera frames Low resolution TS High resolution TS

14. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Image composition Superimpose all the frames taken during the pellet ablation at intervals of 200  s. Superimpose the image of the equilibrium map Superimpose grid at the toroidal location of the pellet injection plane to measure distances. LFSHFS

15. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Visual analysis Flux surfaces spaced by intervals of  N =0.1. The surface highlighted in red corresponds to  N =0.4 (innermost surface affected by the pellet perturbation according to Thomson scattering). Pellet ablates completely outside  N =0.5 ‑ 0.6. To affect the surface  N =0.4 the pellet material should drift by ~20 cm towards the low field side (LFS) of the plasma. End of the pellet trajectory is 45 cm above the equatorial plane. Clouds equally spaced vertically along the pellet path and pellet path follows an almost straight line. LFSHFS

16. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Brehmsstrahlung imaging Asymmetric structure of the pellet cloud extending towards the LFS is visible on the images of the final part of the pellet trajectory taken with the filtered camera. Suggests that a drift is taking place towards the LFS of the plasma. LFSHFS 45 cm above the equatorial plane

17. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Interpretive analysis (I) Interpretive analysis of the observations performed with the code PELDEP2D (Pégourié & Garzotti EPS Bertchesgaden 1997). Pellet advances along the trajectory in the cross section of the plasma. Ablation calculated at each point (NGPS). Material distributed along the magnetic field gradient with typical drift length Λ. Resulting 2-dimensional density distribution averaged over the magnetic surfaces to give a poloidally symmetric deposition profile. Adiabatic plasma cooling caused by pellet material drifting in front of the pellet taken into account. t1 t2 t3 Λ

18. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Interpretive analysis (II) The post-pellet ablation profile (no drift) falls outside the experimental data. Drifted (Λ~25 cm) profile fits well the experimental measurements. Drift along the magnetic field gradient ~35-40% of the plasma minor radius Displacement between ablation and deposition profile of 10-20% in terms of flux radial co-ordinate. Without pre-cooling pellet penetrates to 60 cm above the plasma equatorial plane (shorter than the observed penetration). With pre-cooling penetration reaches 50 cm above the equatorial plane (closer to the experimental observations).

19. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 First principle simulations (I) Simulations performed with a first principle code: –NGPS-type ablation, –four fluid Lagrangian drift model (plasmoid expansion). Details of the code: –B. Pégourié et al., Nucl. Fusion 47 44 (equations), –F. Köchl, this conference, today’s poster session, P4.099 (benchmarking). Good agreement with the experiment. Pre-cooling has to be taken into account.

20. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 First principle simulation (II) Simulations of the MAST experiments have been attempted also with another similar first principle code described in P.B. Parks and L.R. Baylor, Phys. Rev. Lett. 94 125002. The code underestimates the displacement of the deposition profile by ~50%. The reason for this is that the main mechanism driving the plasmoid drift is the reheating of the pellet cloudlet. In this model background plasma temperatures over 1 keV are required to build enough pressure in the cloudlet to accelerate it along the major radius. Therefore this mechanisms is predicted to be weak in MAST plasma simulations because of the relatively low background plasma temperature.

21. L. Garzotti – Workshop on Fuelling in Magnetic Confinement Machines – Hersonissos, Crete 16 th -17 th June 2008 Conclusions Fast visible imaging and high space and time resolution Thomson scattering have revealed the details of the pellet trajectory, ablation and deposition profile on MAST. The presence of a  B-induced drift, leading to a 10 cm displacement between ablation and deposition profiles, has been identified. Interpretive analysis shows that this displacement is compatible with a 20- 25 cm drift of the pellet material in the direction of the magnetic field gradient. There is evidence of the drift induced plasma pre-cooling in front of the pellet playing a role in increasing the pellet penetration depth. These results are predicted by one of the first principle ablation/deposition codes presently available, whereas a second code tends to underestimate the drift because the driving mechanism is predicted to be weak on MAST.