077-05/rs

R&D Areas for US Reference Blankets

US TBM

Test Modules

Predicative Capability Integration

Tritium Systems

1. Thermofluid MHD 2. SiC FCI Fabrication and Properties 3. SiC/FS/PbLi Compatibility &

Chemistry 4. RAFS Fabrication & Materials Prop.5. Be Joining to FS for First Wall6. Helium System Subcomponents

Tests 7. PbLi/H2O Hydrogen Production

Rates8. TBM Diagnostics 9.Ceramic Breeder Thermomechanics10. Breeder/Multiplier Pebble

Specifications11. Partially Integrated Mockups

Testing

1.Transport Model Development 2.Tritium Extraction from PbLi3.Tritium Extraction from He4.Purge gas flow advection

R&D tasks vary considerably in cost and scope

System Integration

077-05/rs

LM-MHD and FCI R&D for the DCLL

Contributors S. Smolentsev, K. Messadek, S. Sharafat, M.

Dagher, N.B. Morley, M. Abdou, M. Narula, A. Medina, N. Vetcha, N. Bhatt, A. Jousse – UCLA

R. Munipalli, P. Huang – HyPerComp

M.J. Ni, Chinese Academy of Science

Yutai Katoh, ORNL

Gerry Youngblood, PNL

Bob Shinavski, Hypertherm

Brian Williams and Matt Wright, Ultramet

2

077-05/rs

Physics based analysis and numerical modeling of key phenomena affecting transport

– Buoyancy-driven mixed MHD convection

– 2D-MHD turbulence

– Impact of wall slip on Hartmann/Shercliff layer formation

– DCLL flow design analysis

– Thermoelectric-MHD self-driven liquid metal cooled FW and Divertor Experimental Investigation of key issues with little existing data

– Flow distribution, pressure and velocity measurements in 3 channel manifold

– Velocity measurements using ultrasonic techniques in LM flows, including PbLi 3D simulations of complex geometry flow elements using HIMAG

– Flow distribution and flow balancing in 3 channel manifold

– Impact of FCI overlap gap electrical conductance on pressure drop and local velocity profile

– Fast flowing liquid metal films in divertors

– Thermocapillary-driven flow in heated LM pools

LM-MHD research strategy overview

077-05/rs

Several tasks on Flow Channel Insert (FCI) characterization

FCI fabrication SBIRs – Composite based (phase II beginning)– Foam based (phase II ending, new phase I)

FCI thermomechanical modeling SiC sample property control and

measurements SiC property modeling SiC irradiation, particularly swelling and

creep effects on stresses

4

Rapid solutions to MHD flow problems – Status

Ramakanth Munipalli, Peter HuangHyPerComp Inc., 2629 Townsgate Rd., Suite 105, Thousand Oaks, CA 91361

Collaborators in HIMAG development:

Mingjiu Ni, now at Graduate University Chinese Academy of Sciences

Neil Morley, Sergey Smolentsev, Mohamed Abdou, UCLA

Meeting at UCLA, March 14, 2008

HyPerComp Incompressible MHD solver for Arbitrary Geometry

HIMAG acceleration in new SBIR

(a) Novel mathematical convergence acceleration techniques:HIMAG has been extended to include multigrid methods

(sequence of meshes is generated automatically for complex geometries)Variety of implicit schemes included in HIMAG

Implicit procedures and sensitivity of numerical parameters has beenperformed, for optimal code performance

Jacobian-Free Newton-Krylov Method

(b) Rapid prototyping capabilities in accelerating problem setup, Template based DCLL mesh generation is now possible

(c) Variable fidelity modeling for developing better initial conditions and MHD-specific convergence acceleration procedures.

(d) Routine use of hybrid meshes is now possible

(e) Rapid interpolation across meshes and CPUs can be performed

Uniform grids

Un-nested, stretched grids

Non-uniform unnested grids

HIMAG is testing MHD flows on a sequenceof automatically generated non-uniform, un-nested computational meshes using a multigrid algorithm.

Preliminary demonstrations were made in Phase-Iwith encouraging results. Full implementation during recently Awarded Phase II

Figures here show residual convergence withconventional and a two-level multigrid methodin uniform and non-uniform meshes.

Sample un-nested mesh sequence is shown on left.

Multigrid version of HIMAG

Interpolation techniques:Point-element relations for standard interpolationElement-element relations based on intersectionPoint-point relations for matching grids/nearest neighbor

E

P

E1

E2

P

Mesh sequencing and the role of interpolation techniques

Octree search procedures

CAD based surface data interpolationis being developed

Sample hybrid mesh used in modeling nozzle flow under prior applications (note the enforcement of orthogonality near the solid walls, and a gradual conversion to an unstructured framework).

Meshes must reflect gradients in solution. For instance, the gap regions in MHD channel flows can be resolved by 3-4 cells or 10-15 cells, depending on the aspect ratio of the channel. As seen above, low aspect ratios present more oscillatory velocity profiles.

Routine use of hybrid meshes

Hybrid meshes: Design of automatic optimal meshes for MHD

Sample meshes for channel flow with expansion

Current lines (above) and potential contours(below) computed on a hybrid mesh, Ha = 300

HIMAG translators created for use with CUBIT mesh generation program from SNL

2 CPU restart2 to 4 CPU restart

Prism mesh used for this calculation

Robust and accurate restart across multiple CPUs

077-05/rs

Magnification, 1 x 2.5 x 2.5Length Units, mmVelocity Units, m/s

Velocity Profile Simulation in 3 channel manifold

077-05/rs

Manifoldentrance

expansion

parallel

inlet

077-05/rsexpansion

outlet

Manifoldoutlet

077-05/rs

15

Simulations match preliminary experimental results on flow distribution in high N region, but not for moderate N

3D simulations including expansion and contraction

(a) Mean velocity versus mean inlet velocity(b) Normalized mean velocity versus Ha2/Re

Courtesy of MessadekWhy?•Himag is just wrong? Numerics or grid?•Jet unsteadiness affects momentum transport?•Parallel channel pressure is higher than simulated (electrodes or 3D field?)•Other 3D field effect?

077-05/rs

U

Comparison against 2D perfectly conducting expansionPressure Contours and Velocity Streamlines

X

Y

-2 -1 0 1 2 3 4 5-1

-0.5

0

0.5

1

1.5

2

p: 1E+06 3E+06 5E+06 7E+06 9E+06 1.1E+07Pressure is dimensional

B

Axial Velocity at Various Cross-sections

Centerline Velocity

077-05/rs

FCI overlap gap electrical conductance impact on PbLi flow profiles in the DCLL

Courtesy Ultramet Courtesy DMS

Machined-Lip overlap between FCIs

SiC Brackets to restrain FCI ends

077-05/rs

Impact of 3-D MHD effects in FCI overlap region

Reformation of near wall M-shaped velocity jets can sweep hot FW regions

– Jets are potentially unstable and can decay into fluctuations

– Jets will overlay profiles/fluctuations resulting from mixed convection effects

Additional pressure drop that can affect flow distribution between parallel channels Divided FCI

Continuous FCI

FCI overlap gaps impact the current closure, and hence disturb velocity and pressure drop (Ha=1000; Re=1000; =5 S/m, cross-sectional dimension expanded 10x

Courtesy Hypercomp

077-05/rs

X Y

Z

-10

-5

0

5

10-1 0 1

Z

-101

X Y

ZNew study to look at overlap region gap conductance effect

3D simulations with HIMAG with resolved FCI gap region

Gap Conductance0.01-1

N = 1000 (Ha = 500, Re = 250)

22

B

a = b = 1tFCI = tgap =

0.05wgap= 0.016

077-05/rs

Velocity Profiles at Gap conductance = 0.3

23

u

00.5

11.5

Y

-10

1

Z

-1

-0.5

0

0.5

1

u

00.5

11.5

Y

-10

1

Z

-1

-0.5

0

0.5

1

u

00.5

11.5

Y

-10

1

Z

-1

-0.5

0

0.5

1

u

00.5

11.5

Y

-10

1

Z

-1

-0.5

0

0.5

1

u

00.5

11.5

Y

-10

1

Z-1

-0.5

0

0.5

1

u

00.5

11.5

Y

-10

1

Z

-1

-0.5

0

0.5

1

x = -1

x = -0.5 x = 0

x = -0.5

x = 0 (gap)

x = 2x = 1x = 0.5

077-05/rs

Current and potential at Gap conductance = 0.3

24

X

Z

-6 -4 -2 0 2 4 6

-1

0

1

2

x B

Y

Z

-1 -0.5 0 0.5 1-1

-0.5

0

0.5

1

077-05/rs

Pressure Profiles for 3 gap conductances

25X

P

-5 0 5 10

100

200

300

400

500

600

700

GC=0.32, CenterGC=0.32, SidewallGC=0.064, CenterGC=0.064, SidewallGC=0.64, CenterGC=0.64, Sidewall

FD regions match analytic solutions

Very sharp pressure gradient near the wall ~ Ludford layer thickness

3D pressure drops much larger FD regions

077-05/rs

Velocity Jet Maximum and 3D Pressure Drop

Similar looking dependence on Gap Conductance

Velocity peak varies from 3x to 6x base flow over reactor relevant range

K3D varies from .1 to .45 over reactor relevant range

26

Approx. Reactor Range

K3DN = P3D / (U2/2)

(b/(f a2)) / (tgap/(gapwgapa))

077-05/rs

Velocity Jet Maximum and 3D Pressure DropLog Plot dependence on overlap gap conductance

27

Approx. Reactor Region

K3DN = P3D / (U2/2)

SQRT

LINEAR

Peak Velocity varies ~ with sqrt of gap conductance

3D pressure drop varies closer to linear with gap conductance

077-05/rs

Still in progress on overlap gap conductance study / to do…

More variations of N, Re, Ha Impact of higher SiC conductivity (right now it is

very nearly perfectly insulated) Better resolution and unsteady effects Full DCLL parameter simulation (Ha=15k, Re =

60k)

Initial conclusions: Even one, extremely snug fitting overlap gap will – change overall pressure drop of the single channel

significantly– Disturb velocity in local region ~2a around the gap

location28

077-05/rs

Bulk and contact resistance measurements of CVD SiC surface coatings for FCIs

Work of Albert Medina, UCLA CONCEPT: measure voltage across SiC

specimens of different thickness in order to determine bulk and contact resistanceVs = Is *(Rbulk + Rcontact)

– Rbulk= t/(bulkA)

– Rcontact= rcontact/A Samples: High Resistivity

(RH SC-001) CVD disks– 1, 2, and 3 mm thick

x 25 mm dia– 0.5 um surface roughness

29

Is

Vs, Ts Vr

077-05/rs

SiC disk sample mounted in MACOR insulator

MACOR PbLi Reservoirs

Measurement leads in atmosphere controlled oven during assembly

077-05/rs

Proceedure Preclean disk with HCL/ethanol solution, rinse with

ethanol Cover disk faces with masking tape Cement disk into holder ring Remove masking tape and clean disk surfaces

again with ethanol Inspect cemented joint for flaws Assemble sample holder Bake sample and holder for 2 hours in Argon Pour in molten PbLi into both reservoirs Stir PbLi for 30 sec each side Introduce measurement leads into PbLi melts in

each reservoir and put on reservoir lids Seal oven again and reflush with Argon Begin measurement run / temperature ramp

31

077-05/rs

Reference EC data on high purity, high resistivity CVD SiC (Rohm-Haas)

2 different measurements on the same grade high purity SiC

– Vendor datasheet for HR-Grade

– PNL-HP with Au electrodes

32

077-05/rs

UCLA measurement with PbLi electrodes – Sample 1 (3mm thick)

Initially follows Vendor data

– Coincidence? Then jumps to PNL data

– Wetting event Stays on PNL line during 2 hr cooling

33

077-05/rs

UCLA measurement with PbLi electrodes – Samples 1 (3mm thick) and 2 (2 mm thick)

Sample 2 has very low conductance initially– Significant contact

resistance? Sample 2 increases

up to Sample 1 and PNL data by 500C

Sample 2 is kept at 500C for two hours and tested with different current (20mA and 50 mA)– Current variation

shows no effect

34

077-05/rs

UCLA measurement with PbLi electrodes – Samples 1 (3mm thick) and 2 (2 mm thick)

35

Retesting sample 1 after sitting in contact with PbLi for a day showed contact resistance that did not go away when reheated

077-05/rs

Initial Conclusion on PbLi/SiC surface contact resistance Ground smooth, cleaned, but otherwise as-

received high purity CVD SiC showed no significant contact resistance with PbLi at 500C after 2 hours of exposure– 3 mm and 2 mm samples gave the same

measurement, overlapping with PNL measurements of similar grade material

Initially, it takes the sample some time before good electrical contact is achieved

Impurities in the system increase the contact resistance over time by a significant amount, dominating the total resistance– Nature of surface film still TBD

36

077-05/rs

PNL measurements on Composites

37

10-3

10-2

10-1

100

101

102

103

0 200 400 600 800 1000

Bar HyperTherm 2D-Nic S/CVI (multi PyC)Bar GE 2D-5HS Nic S/CVI (150 nm PyC)Bar HyperTherm 2D-Tyr SA/CVI (100 nm PyC)Disc HyperTherm 2D-Tyr SA/CVI (100 nm PyC)Disc HyperTherm 2D-Nic S/CVI (multi PyC)Disc, GE 2D-5HS Nic S/CVI (150 nm PyC)Bar CVD-SiC (Weber)Bar CVD-SiC (Hi-Purity)

Ele

ctr

ical C

on

du

cti

vit

y (

S/m

)

Temperature (oC)

20 S/m

50

0 oC

CV

D-S

iC R

ange

SiC Foam Measurements

077-05/rs

B

g = J x B

LM-MHD for Plasma Facing Components

077-05/rs

Rayleigh-Taylor Instabilities applied to melt layer motion during gas injection disruption mitigation events in ITER – Linear theory and 2D viscous simulations

3D simulations of MHD thermocapillary convection in a lithium tray using HIMAG– Comparison of field angle and low/high heat flux cases

2D simulations of Thermoelectric MHD driven flows for divertor and FW self-driven cooling– Initial calculations…Just a teaser

Fast flowing thin film flows– Narula thesis

LM-MHD for Plasma Facing Components

077-05/rs

Introduction to Thermoelectric MHD-driven divertor or FW flows

40

Li is very thermo-electrically active

Li/V (look at Nb in figure) has large thermo-electric power, P = SLi – SV > 20

Could self-pumped geometries for PFCs be envisioned?

– LTX, NSTX– DEMO FW

077-05/rs

2D TE-MHD Test CaseLong thin grooves filled with Li

400 m wide x 1 cm deep Li channel made from 100 m thick Molybdenum

1 MW/m2 uniform surface heat flux

Lithium flow driven by TEMHD currents generated from surface heat flux

Coupled fully developed TE-MHD flow and heat transfer calculated

New “thin” conducting wall BC with TE terms used to simulate conducting wall

Surface heat flux

B

T

u

Molybdenum

Lithi

um

077-05/rs

2D Velocity and Temperature Profile of TEMHD driven Li flow in Moly channels Peak surface velocity ~ 20

cm/s Surface temperature rise

(from bulk to surface), ~ 40 K

Bulk temperature rise in flow direction 10 K per cm (TE currents from this gradient not modeled)

q’’ = 1 MW/m2B= 0.5 T

T (K)

Red lines indicate TE current paths

1D Result Summary from Jaworski using Shercliff solutionVelocity scales like (S1-S2) T /B

(0.2 mm)