Simulations of Free surface liquid metal layers and other topics…

1. Simulations of Free surface liquid metal layers and other topics… Neil B. Morley, A. Ying, M. Narula, R. Hunt, M. Abdou UCLA Fusion Science and Technology, UCLA R. Munipalli, P. Huang – HyPerComp M. Jaworski, D. Ruzic – UIUC

2. Outline • 3D MHD simulations of thermocapillary, thermoelectric, and buoyancy effects in liquid metal layers heated from the top • Initial tests of SiC/Ga compatibility at 750C • Initial studies of Be/F82H bonding for ITER TBM armor attachment • Separate Presentation - Simulations of active water cooling of the ITER FW/Shield components • 3D Simulations and experiments for fast flowing liquid metal layers (work completed this year, but not presented here)

3. 3D, MHD, multiple material, coupled transport (heat/mass) code for liquid metal free surface and closed channel (blanket) problems • Main development activities: Acceleration and validation exercises (see Munipalli poster) HyPerComp Incompressible MHD solver for Arbitrary Geometry

4. Thermocapillary Force Modeling • Surface heating can cause a temperature gradient at the free surface • Surface tension changes as a function of temperature (lithium factor is about -3.75%/100K) • Liquid moves along the surface from areas of lower surface tension to higher surface tension, thermocapillary (Marangoni) convectionFsurf= n – s where  = surface curvature, n = surface unit normal, s surface gradient • Rigid-Lid Boundary -- “Free” surface approximation with traction n u =sT where  = o + T = temp dependent surface tension coefficient, • Deformable free surface with improved surface tension model in HIMAG to include variable surface tension Fsurf= n – s  n – (T – nT) Gas/Plasma/Vacuum Tangential surface tension forces Free surface contours Liquid metal

5. SLIDE and Thick Liquid Layer Simulations with rigid-lid BC Surface heat flux • 1 x 10 x 10 cm cavity full of Lithium simulated with 3D-MHD HIMAG code • Surface Conditions • 10 mm wide shaped surface heat flux strip • LHF = 104 W/m2 peak • HHF = 106 W/m2 peak • Rigid-Lid “Free” surface approximation with traction •  n u =sT • Wall conditions • No-slip • Electrically insulated or perfectly conducting • Thermally insulated, sides • Isothermal, bottom Lithium cavity •  - dynamic viscosity •  - surface tension gradient wrt T B 

6. 1D Analytic Model – (Slightly modified from Jaworski) • (2u/y2) - B2u/ - g h/x = 0 • 1D Model including: • Surface height change • Conservation of flow • Gives solution: • dh/dx = • usurf= • u(y)/usurf= Constant surface temperature gradient T/x = b, Defines surface traction b = (/) u/y y B x

7. Surface Temperature Comparison of 3D HIMAG simulation with rigid-lid BC to 1D analytical model B T gradient ~ 104 K/m

8. Study of impact of field direction on Marangoni recirculation velocity B = 0.13 T • Surface temperature profiles not dramatically changed with field direction changes • Heat flux into the bottom wall was measurably changed B = 0.5 T B = 0.5 T

9. New rigidlid results – no symmetry and 3 component field relative to the walls / heat flux 15 B = 0.5 T 15

10. Looking from the top, the heated spot appears to be pulled slightly off center line by strong lateral velocity

11. Parameter b/gh (~Bond No.) provides estimate of the change in surface height Bond No. for 1 cm thick lithium film • dh/dx varies from 1.5x to 1x the dimensionless parameter b/ghwith increasing field • Surface deformation becomes important for SLIDE (~1 cm) when q’’ approaches 10 MW/m2 • Surface deformation becomes important for 1 mm when q’’ approaches 100 kW/m2 Bond No. for 1 mm thick lithium film

12. Test case with temperature dependent surface tension and deformable surface model Heat flux • 5 mm thick Li film in a 10 x 10 cm reservoir • Strip heat flux with 1 MW/m2 peak 1 cm width • Fluid ratio 100 • Small field (500 g) • Observed height change ~0.6 mm • Peak surface velocity 0.22 m/s in liquid phase ~gas lithium Slices from 3D simulation showing temperature and velocity contours streamlines

13. Conclusions of Thermocapillary simulations • Thermocapillary surface velocity on the order of tens of cm/s possible with 1 MW/m2 heat flux on 1 cm deep lithium, even in 0.5 T magnetic fields • Thin film motion more strongly influenced by small surface normal field than the larger toroidal field • SLIDE with surface normal field only should give similar flow and thermal response to divertors • 1D model with surface normal field appears to give good estimate of base flow at the surface • Impact of convection flow should provide measurable variation in heat flux at the bottom surface • Flows for different field alignment cases are very different • Surface deformation should become appreciable at heat fluxes > 1 MW/m2 or for thinner films • Cases with no symmetry assumption/ 3D fields, some asymmetric effects seen – more interpretation necessary

14. Thermoelectric Currents and Forces • Surface heating can cause a temperature gradient at the solid/liquid interface • This induces an electric current via the Seebeck effect (lithium/Iron couple is about 20 V/K) • Electric current interacts with the magnetic field via JxB Lorentz force • Include S T into Ohm’s law calculation j = (– + v x B - ST) Slices from 3D simulation showing thermoelectric current emerging from steel wall at the hot interface and returning at cooler interface

15. Li has unusually high Seebeck constant • Li is very thermo-electrically active • Li/Iron hasP = SLi – SFe ~ 20 • Impact of this effect is suspected in SLIDE results

16. Preliminary test of TE current effect on SLIDE • Test problem with electric potential at bottom wall imposed to approximate S.grad(T) effect from SLIDE data • 1 MW/m2 heat flux stip, 500 g vertical field, 15 K temp difference at bottom wall, rigid-lid “free” surface • Result clearly shows cyclonetype vortical motion ~0.3 m/s on surfaceand strong smearingof the heat flux at the “free” surface • Will be strongdifference betweenvertical and horizontal fields 3D simulation showing thermoelectric induced vortical motion

17. Next steps • Finish coupling / testing of HIMAG changes • Higher resolution tests, higher field tests • simulation of representative SLIDE cases • simulations of effect for thin heated films with coplanar fields (NSTX, melt layers, DiMES revisted) • Combined deformable surface, thermocapillary, thermo-electric, buoyancy simulations • ISFNT paper • Consideration of future experiments in BOB or QTORmagnets (using Ga alloy, Hg, Pb alloy) MTOR Thermofluid/MHD facility

18. Magnetic Intervention for Inertial Fusion:Cusp magnetic field keeps ions off the wall(in Plasma Physics terms: Conservation of P = rA = 0) Polar cusp (2) Axis Equatorial cusp Plasma expansion initially spherical Ion cloud deforms as it encounters cusp Ions, at reduced power, leak into external dumps • Physics demonstrated in 1979 NRL experiment: • R. E. Pechacek, et al., Phys. Rev. Lett. 45, 256 (1980). • NRL experiment modeled by D. Rose at Voss Scientific (2006)

19. 7 7 6 6 8 8 9 9 1 1 3 3 2 2 4 5 5 4 An example of a Magnetic Intervention Chamber Ions deflected downward by magnetic fields Ion energy absorbed in Gallium Rain Ion Dissipaters Chamber radius: 5 m Point cusps: 16 T Main coils: 0.75 T Energy absorption in Ga: 85% in first 10 mg/cm2 15% in next 100 mg/cm2 Only first layer evaporates Gallium inventory enough so mean temp rise < 300C coils beam tubes chamber ion orbits Gallium Droplets NB Vapor P of Ga = 10-6T at 720 C A.E. Robson, NRL (ret)

20. Preliminary Compatibility Tests between CVD SiC and Pure Ga Quartz crucible containing SiC disk and Ga, in quartz vacuum tube – before test After test – Bands of white (located inside furnace) and black (located outside furnace) deposits following 25 hours

21. Preliminary Compatibility Tests between CVD SiC and Pure Ga SiC Disk after Ga Exposure SiC Disk after Exposure and Cleaning 10 mm Ga can be removed by scrubbing with a soft cloth and HCL/ethanol mixture – slight surface discoloration remains Ga partially wets the SiC, roughly have the sample surface area

22. Preliminary Compatibility Tests between CVD SiC and Pure Ga • Initial Conclusions • White material deposition has gallium oxide and both C and Si present • Partial wetting of the SiC disk was observed • Sample weighing did indicate a mass loss was 6 mg after only 25 hours

23. Preliminary Compatibility Tests between CVD SiC and Pure Ga Test Description • CVD SiC disk sample (initially 0.806 g) • SiC cleaned with ethanol and acetone, then blown with dry air • Pure Ga metal (~3 ml) from Atlantic Metals • In quartz (Si02) crucible, with quartz insert to keep SiC submerged • Crucible placed on steel support in quartz tube, pumped on with mechanical and turbo pump (but no vacuum gauge available) • Aluminum witness plate placed in cold region of the quartz tube • Quartz tube inserted into tube furnace • Temperature brought to 100C for 24 hours to fully pumping/baking on tube and sample before beginning high temperature exposure • Temperature brought to 400C for 2 hours to continue pumping/baking, some minor deposition of dark material on cold quartz tube wall immediately above heater (see photos) • Temperature brought to 700C for 24 hours, significant deposition of dark material seen on cold quartz tube wall immediately above heater (see photos) • Test terminated due to concerns an unexpected reaction was taking place • White material deposition seen on the quartz tube wall, and steel support inside the hot region (see photo) • Partial wetting of the SiC disk was observed • Sample cleaned with HCL solution (good for dissolving Ga oxides), Fantastic (good for dissolving Ga oxides), water, acetone and blown with dry air. Mass loss was 6 mg.

24. Perform direct bonding of Be to Reduced Activation Ferritic/Martensitic Steel (Hunt, Ying (ULCA), Goods (SNLL)) Application: ITER requires a 2mm coating/armor of Beryllium on plasma facing surfaces of TBMs Be used as armor layer F82H (or EUROFER97) used as structural material Task: create a robust diffusion bond between two dissimilar metals Beryllium & RAFM steel (F82H) Metallize Be (w/ Ti, Cu) to enhance HIP bonding characteristics. Status of Experiments • Phase 1  HIP Cu to RAFM steel • 650*, 700, 750, 800, 850* C *completed • Tensile test, shear test, microstructure analysis all underway • Phase 2  Not yet underway. Proceed with Be bonding studies

25. Cu to RAFS HIP @ 650 & 850 C (Tensile Test) 650 C Failure at material interface  650 C creates insufficient bond 850 C Failure in Cu bulk material  850 C diffuses enough to create strong bond

26. Cu to F82H HIP at 650 & 850 C (Shear Test) 650 C Failure at material interface  650 C creates insufficient bond 850 C Failure in Cu bulk material  850 C diffuses enough to create strong bond

27. HIP Test Findings/Progress • AES in progress to characterize elemental composition of fracture surfaces and HIP bond interfaces • Initial AES (auger electron spectroscopy) shows very little Cu across interface in 650 C HIP  Insufficient metallurgical bond • 850 C (HIP for 2hrs @ 103MPa) high enough temp for sufficient diffusion of Cu/RAFM • Proceed with HIP at 800, 750, 700 C to find lower suitable temp than 850 C

28. Metallization • Proceeding with a series of diffusion studies to optimize Ti and Cu film thicknesses via EMP/SEM characterization • Anneal substrates of Cu with PVD Ti, and electroplated Cu • Microprobe Analysis • 650 C – Significant percent of the 20 mm Ti metallization film is reacted • 850C – TBD • Proceed with 700, 750, 800 to match HIP test temperatures • Oxidation levels found higher than expected. PVD chamber potentially problematic. Figure showing Cu bulk (light color on top, Ti deposition layer (dark in middle), and Cu electroplate layer (light color on bottom)