Melting of Tungsten by ELM Heat Loads in

Melting of Tungsten by ELM Heat Loads in the JET Divertor Guy Matthews, Gilles Arnoux, Boris Basylev, Jan Coenen, …. …and JET Contributors 1 EX/4 -1, Fusion Energy Conference 2014, St. Petersburg

Outline 1. 2. 3. 4. Background The JET melt experiment Simulation of the results Conclusions and future plans

Decision on material for ITER first divertor 3 Risks for tungsten melting in early ITER operation had to be reviewed by IO

Bulk W melting well studied in medium sized tokamaks Bulk melting can cause disruptions = dangerous for ITER Bulk melting risk was considered low for ITER divertor due to tile shaping & protection systems ASDEX-Upgrade divertor manipulator

Looking down into the tungsten divertor - after The JET tungsten divertor St ac St k A ac k B Tungsten lamella shaping and divertor protection systems work in JET at ITER relevant inter-ELM heat fluxes 5 2011 As installed 2012 After ~3500 pulses ~20 Hrs plasma

W melting by ELMs QSPA ELM simulations looked worrying for ITER N. Klimov et al. , JNM 390 -391 (2009) Plasma flow direction QSPA-T 721 …but plasma pressure lower and B W target higher in ITER (and JET) stabilises surface waves q > 2. 2 MJm-2, t ~0. 5 ms 3. 8 MJm-2 for a 1. 5 ms pulse (ITER minimum t. TQ) Extrapolation to ITER requires a benchmark for the MEMOS code JET is capable of achieving a similar transient heat-flux at normal incidence 6

Typical temperature rise due to ELMs in JET Temperature rise ( C) Geometric factor for a vertical edge ~ × 20 Surface T depends mainly on pedestal pressure not ELM size! Existing data – normal lamellas T during ELMs So increase heating power or Ip to raise T [T. Eich] ITER ~105 Pa Pedestal pressure (Pa) 7 7

Outline 1. 2. 3. 4. Background The JET melt experiment Simulation of the results Conclusions and future plans

Exposing a tungsten edge in JET A JET pulse 84779 q|| AB 9 B

Typical JET W melt pulse 3 MA/2. 9 T 23 MW 300 k. J ELMs q|| ~3 GWm-2 ~0. 5 GWm-2

Limitations of the top IR viewing

Temperature from IR ( C) Melt pulses are reproducible and no disruptions Special lamella Standard (Ref. ) lamella Time (s)

84686 – Before melting q|| Special lamella 5. 5 mm LFS 13 HFS

After 84724 Special lamella LFS 14 HFS

After 84778 Special lamella LFS 15 HFS

After 84779 Special lamella LFS 16 HFS

After 84781 Special lamella LFS 17 HFS

After 84782 Special lamella LFS 18 HFS

After 84783 B 5. 5 mm LFS Jx. B Jthermionic HFS Erosion: 150 -300 m per pulse, 5 -10 m per ELM (frequency 30 Hz) Total volume moved: ~6 mm 3 19

Erosion centres on the ELM footprint Tref( C) 20

Indirect evidence for W droplet expulsion W droplet event during melt pulse Laser blow off with W target A few droplets reach the main plasma with diameters ~ 100 m - Small perturbations only and no disruptions 21

Outline 22 1. Background 2. The JET melt experiment 3. Simulation of the results MEMOS = key tool used for ITER predictions Stefan problem in 3 D geometry accounting surface evaporation, melting and re-solidification solved by implicit method - Surface power density vs time from IR is the input - Vapour shielding - Temperature dependent thermophysical data applied - Moving boundaries are attached to melt layers - All forces: Gradient of plasma pressure, gradient of surface tension, Jx. B, tangential friction force [Bazylev TH/P 3 -40]

Power density q|| from reference lamellas IR camera T(t) MEMOS 3 D B 23 q|| = qn / sin Theodor 2 D inverse code qn (MWm-2) Reference lamellas

Power density on the special lamella PIC model says: MEMOS input = qn from IR (Theodor) fs >0. 6 during ELMs fs iterated to match: evaporation rate, fs~1 inter-ELM and L-mode synthetic IR image and Planck radiation q|| from reference lamellas qn= q|| sin qs qqss == fss qq|||| cos 24 Special lamella

fs chosen to fit evaporation rate and Planck radiation Temperature #84779 – IR (unresolved) and peak (MEMOS) Melting IR MEMOS fs=1 W evaporation rate from WI (400. 88 nm) Best fit to all data with fs=0. 4 25 MEMOS fs=0. 4

W melt evolution #84779 – MEMOS [Bazylev TH/P 3 -40] fs=0. 4 26

Hierarchy of forces – MEMOS JET pulse #84779 - MEMOS Surface: -200 to +400 m m B [Bazylev TH/P 3 -40] Surface: -40 to +10 m m Shadow Jx. B HFS Melt depth and motion match JET data J 60 mm LFS Plasma pressure (6 k. Pa) + surface tension gradients + thermionic emission (Jx. B) 27 2. 5 mm Plasma pressure (6 k. Pa) + surface tension gradients

Conclusions W melting by ELMs in JET provided important inputs to ITER: • Shallow melts with a few small droplets ejected but no disruptions • JET melt results are consistent with MEMOS assuming J×B dominant • Unexpectedly large power mitigation factor found for exposed W edge Future JET plans New lamella at 15 : • Fully resolved IR temperature • Grazing field angle / more ITER-like • Simpler geometry 28

Thermionic emission during ELM - MEMOS JET pulse #84779 - MEMOS [Bazylev TH/P 3 -40] Unlike JET experiment, suppression of thermo electron emission is predicted in ITER due to grazing field angles 29

Larmor radius smoothing - PIC code [Dejarnac NF] Equivalent to fs~0. 8 - Calculated for ELMs only - insufficient to explain fs=0. 4 in H-mode 30 - No effect expected in L-mode where we find fs=0. 2

MEMOS suggests significant vapour shielding …. . but we are not able to prove it experimentally due to lack of a consistent physics model for fs=0. 4 in H-mode and fs=0. 2 in L-mode #84779 31