Transcript Mass transfer modeling for LM blankets

Mass transfer modeling for
LM blankets
Presented by Sergey Smolentsev (UCLA)
with contribution from:
B. Pint (ORNL)
R. Munipalli, M. Pattison, P. Huang (HyPerComp)
M. Abdou, S. Saedi, H. Zhang A. Ying, N. Morley, K. Messadek (UCLA)
S. Malang (Consultant, Germany)
R. Moreau (SIMAP, France)
A. Shishko (Institute of Physics, Latvia)
Fusion Nuclear Science and Technology Annual Meeting
August 2-4, 2010
UCLA
In this presentation:
• Status of R&D on development of MHD/Heat & Mass
Transfer models and computational tools for liquid metal
blanket applications
• Examples: corrosion & T transport
OTHER RELATED PRESENTATIONS at THIS MEETING
TITLE
Presenter
Oral/Poster
Tritium Transport Simulations in LM
Blankets
H. Zhang
UCLA
oral
Modeling Liquid Metal Corrosion
S. Saedi
UCLA
poster
R. Munipalli
HyPerComp
poster
Integrated Modeling of Mass Transport
Phenomena in Fusion Relevant Flows
Mass transfer in the LM flows is one of the key
phenomena affecting blanket performance and safety
Traditionally, major considerations
associated with the LM flows are the
MHD effects. But there are more….
Tritium permeation is an issue –
no solution has ever been proven
Corrosion/deposition severely limits
the interfacial temperature and thus
represents an obstacle to developing
attractive blankets at high temperature
operation
Blanket: “Hot” leg. Mass transfer coupled with MHD. Corrosion. T production. T
leakage into cooling He. Formation of He bubbles in PbLi and trapping T.
Ancillary system: “Cold” leg. Turbulent flows. Wall deposition and bulk precipitation.
T leakage into environment. T extraction. Cleaning up.
Main objectives of mass transfer modeling
Blanket:
• Revisit maximum PbLi/Fe t (470 ?)
and wall thinning (20 m/year ?)
• Estimate T leakage into cooling He
streams in the blanket
•
•
•
•
Ancillary system:
Estimate T leakage into environment
Model T extraction processes
Model clogging/deposition
Model clean up processes
Phenomena, design:
• Address “new” phenomena (i.e. He
bubble formation in PbLi and
trapping T by the bubbles)
• Find new design
solutions/modifications
Challenge!
The whole PbLi loop, including
the blanket itself and the ancillary
equipment, must be modeled as
one integrated system
What do we need?
• New phenomenological models for:
- interfacial phenomena
- nucleation/crystallization
- particle-particle/wall interaction
- MHD effects on mass transfer
He bubble transport and trapping T
- T transport physics
by the bubbles is not well understood
• New material databases (He-T-PbLi)
• New mass transfer solvers and their coupling
with existing MHD/Heat Transfer codes
What tools do we use?
• HIMAG as a basic
MHD/Heat Transfer
solver
• Many UCLA research
MHD, Heat & Mass
transfer codes
• CATRIS (in progress)
as a basic mass
transfer solver
• Many thermohydraulic /
mass transport codes
The R&D on the development of new
phenomenological models and their
integration into numerical codes is underway
CATRIS: MATHEMATICAL MODELS
1. Dilution approximation, Ci<Ci0
Ci
 (V)Ci  ( Di Ci )  qi
t
2. Lagrangian particle tracking, Ci>Ci0
 pV
dVp
dt
K
  Fk
k 1
3. Multi-fluid model, Ci>>Ci0
N
i
  i Vi   J ij
t
j 1
N
i Vi
k
k
k k
  i ViVi   σi  i g   Pij
t
j 1
MODELING EXAMPLES
Example
Description
Modeling status
#1
Riga experiment
Modeling of “corrosion” experiment in
Riga, Latvia on corrosion of
EUROFER samples in the flowing
PbLi at 550 in a strong magnetic field
Good match with experimental
data on mass loss. Addressing
groove patterns needs more
sophisticated modeling.
#2
Tritium transport
Numerical analysis of tritium transport
in the poloidal flows of the DCLL
blanket with SiC FCI under DEMO
blanket conditions
Analysis for the front duct of the
DCLL DEMO OB blanket has
been done using a fully
developed flow model.
#3
Magnetic trap
Modeling of extraction of ferrous
material suspended in the flowing
liquid in a magnetic trap
First “demo” results have been
obtained using Lagrangian
particle tracking model under
some assumptions for B~ 0.1 T.
#4
Sannier equation
Modeling of corrosion of
ferritic/martensitic steels in turbulent
PbLi flows to reproduce existing
experimental data and to address the
effect of a magnetic field
In progress. Computations are
performed using the UCLA
corrosion code (Smolentsev).
Turbulence in a magnetic field is
modeled via “k-eps” model.
Riga experiment 1/11: setup
Simulation of “CORROSION” EXPERIMENT in Riga
PbLi loop
EUROFER samples
B=0, B=1.7 T
T=550C
U=2.5 cm/s, U=5 cm/s
Time=2000 hours
Rectangular duct, 2.7x1 cm2
Courtesy of Dr. Andrej Shishko, Institute of Physics, Latvia
Two 12-cm sections of 10
samples in a row, one
section at B=0 and one at
B=1.7 T
Riga experiment 2/11: results
Macrostructure of the washed samples
on the Hartmann wall in 3000 hrs at 550
Mass loss, mg
Uo=2.5 cm/s
Uo=5 cm/s
B=0
#
B=0,T
B=1.7,T
B=0,T
B=1.7,T
1
376
593
437
743
2
245
564
338
757
3
303
481
330
623
4
193
486
283
605
5
223
456
251
506
6
257
440
-
-
7
163
483
248
482
8
198
484
310
512
9
214
566
321
463
10
205
502
314
474
Mass loss is almost doubled
in the presence of B-field
B=1.7 T
PbLi flow
Courtesy of Dr. Andrej Shishko, Institute of Physics, Latvia
Riga experiment 3/11: results
FLOW
~40m
Magnetic field
~ 500m
•Wall thinning: 1.5->1.4 mm
•Grooves: 40 m deep
Courtesy of Prof. Rene Moreau (SIMAP, France)
• In addition to wall thinning,
periodic grooves aligned with
the flow direction have been
observed on the Hartmann
wall
• Mechanism of groove
formation is still not well
understood
• A. Shishko (Latvia): higher
velocity in the surface cavities
causes higher corrosion rate.
The effect may be related to
specimen machining
• R. Moreau (France):
the grooves are due to
instability mechanism
associated with induced
electric currents crossing the
interface
Riga experiment 4/11:
Basic assumptions
•
•
•
•
•
Fully developed, laminar flow
Only Fe is considered
Purely dissolution mechanism
No oxygen passivation layer
Mass transfer controlled
corrosion
• Zero Fe concentration at x=0
mathematical model
 2U  2U B0 B 1 dP



0
z 2 y 2 0 z  dx
2 B 2 B
U



B
0
0 0
2
2
z
y
z
z  b : U  0,
1 B 1 B

0
 z  w tw
y   a : U  0,
1 B 1 B

0
 y  w tw
C
 2 C  2 C  2C
U
 D( 2  2  2 )
x
x
y
z
x  0:
z  b :
Two BC types have been tested
(C0 is the saturation concentration at given t)
y  a :
C 0
C
 K (C0  C )  0 or C  C0
z
C
D
 K (C0  C )  0 or C  C0
y
D
Riga experiment 5/11:
* At 550C
** Based on equation of Sutherland-Einstein
***Recommended by Riga people (=0.676 wppm).
Solubility of Fe in PbLi
100
10
Solubility experiments
Barker et al., 1988
Borgstedt et al., 1991
Grjaznov et al., 1989
Riga group, 2006
1
C0, wppm
• Diffusion coefficient Fe-PbLi:
6.4E-09 m2/s**
• Saturation conc. C0: 6.26 g/m3 ***
• PbLi viscosity: 1.08E-07 m2/s
• PbLi density: 9300 kg/m3
• PbLi electrical conductivity:
0.7E+06 S/m
• Ha=0 and 227.3 (1.7 T); Cw=0.78;
Re=1157 and 2314
material properties*
0.1
0.01
0.001
0.0001
1E-005
600
650
700
750
800
850
T, K
Co: more than THREE order
of magnitude difference ???
Riga experiment 6/11: modeling results
B=1.7 T, Cw=0.78, U=2.5 cm/s
0.05
B=0, U=2.5 cm/s
B=0, U=2.5 cm/s
Velocity, m/s
0.04
0.03
B=1.7 T, U=2.5 cm/s
0.02
0.01
0
-0.005
-0.003
-0.001
0.001
Z, m
0.003
0.005
Riga experiment 7/11: modeling results
Mass loss, m/year
MASS LOSS: comparison with the experiment
430
215
BC :
C
D
 K (C0  C )  0
n
Riga group: C0=6.26 g/m3, K=4.27E-05 m/s
BC :
C  C0
Konys: 700 m/year
500C, 0.22 m/s, 0T
Grjaznov et al: C0=3.25 g/m3
Riga experiment 8/11: modeling results
Effect of the velocity and B-field on the wall and bulk concentration
BC :
D
C
 K (C0  C )  0
n
Riga group: C0=6.26 g/m3, K=4.27E-05 m/s
Riga experiment 9/11: modeling results
Effect of the velocity
- no magnetic field
- Hartmann wall
Effect of B-field
-Hartmann wall
Wall effect
- with magnetic field
Riga experiment 10/11:
modeling results
Development length > 10 m (B=1.7 T, U=5 cm/s)
Wall concentration
Bulk concentration
BC :
D
C
 K (C0  C )  0
n
Riga group: C0=6.26 g/m3, K=4.27E-05 m/s
Riga experiment 11/11: conclusions
• Riga experiment on EUROFER-PbLi corrosion has been
successfully modeled (not including grooves)
• Higher corrosion rate of EUROFER samples in a
presence of a magnetic field can be explained by the
steep velocity gradient in the Hartmann layer
• Boundary condition at the solid-liquid interface is still an
open issue. Saturation concentration at the wall can be
used as a first approximation
• Uncertainty in experimental data on transport properties
(e.g. saturation concentration) severely limits modeling
predictions
• If to extrapolate to LM blanket conditions - the mass
transfer development length can be more than 10 m
Tritium transport, 1/6
DCLL Geometry (not to scale)
z
y
x
Outflow
RAFS wall 5 mm
thick
2 mm
gap
y
2.0 m
2.26 m
• DCLL DEMO blanket
conditions (outboard)
• Poloidal flow in a front duct
with a 5-mm SiC/SiC FCI
• HIMAG is used to simulate
MHD flow, assuming fully
developed flow conditions
• CATRIS is used to
simulate tritium transport in
the multi-material domain,
including PbLi flow, SiC
FCI and Fe wall
• Goals: (1) T permeation
into He; (2) sensitivity
study
B
z
211 mm
FCI
SiC wall 5 mm thick
Inflow
207 mm
0.3 m
231 mm
•Neutron wall loading (peak): 3.08 MW/m2
•Surface heating: 0.55 MW/m2
•PbLi Tin/Tout: 500/700C
•Flow velocity: 6.5 cm/s
•Magnetic field: 4 T
•Inlet T concentration: 0
•T generation profile: 4.9E-09 Exp(-3y), kg/m3-s
Tritium transport, 2/6
Physical properties
Pb17Li
RAFS
Solubility
D
Solubility
D
Solubility
D
σ
mol/m3/Pa0.5
m2/s
mol/m3/Pa0.5
m2/s
mol/m3/Pa0.5
m2/s
S/m
[1,2,3]
Low
High
1.
2.
3.
4.
5.
6.
SiC FCI
0.0005
0.1
[4]
1.0 ×10-9
7.0 ×10-9
0.0025
[5,6]
1.5×10-8
0.117
5.0×10-16
5
500
Mas de les Valls, E., Sedano, L.A., Batet, L., Ricapito, I., Aiello, A., Gastaldi, O., Gabriel, F. (2008) Lead-lithium eutectic
material database for nuclear fusion technology. J. Nuc. Mat. 376, 353-357.
Reiter, F. (1991) Solubility and diffusivity of hydrogen isotopes in liquid Pb-Li. Fusion Eng. and Design. 14, 207-211.
Aiello, A., Ciampichetti, A., Benamati, G. (2006) Determination of hydrogen solubility in lead lithium using sole device.
Fusion Eng. and Design. 81, 639-644.
Aiello, A., Ciampichetti, A., Benamati, G. (2003) Hydrogen permeability and embrittlement in Eurofer 97 martensitic
steel. ENEA Report SM-A-R-001.
Causey, R.A., Wampler, W.R. (1995) The use of silicon carbide as a tritium permeation barrier. J. Nuc. Mat. 220-222,
823-826.
Causey, R.A., Karnesky, R.A., San Marchi, C. (2009) Tritium barriers and tritium diffusion in fusion reactors.
http://arc.nucapt.northwestern.edu/refbase/files/Causey-2009_10704.pdf
There is a considerable degree of uncertainty in the physical properties,
particularly for the solubility of T. That is why sensitivity study is needed.
Tritium transport, 3/6
The electrical conductivity of FCI may have a strong effect on the T
transport via changes in the velocity, especially in the 2-mm gap
Side-wall jets in the bulk
Hartmann-wall
gap flows
=100 S/m, Ha=15,900
Side-wall
gap flows
Tritium transport, 4/6
T concentration (10-6 kg/m3) for cases with low (0.001 mol/m3/Pa0.5)
and high (0.05 mol/m3/Pa0.5) solubility of T in PbLi
X=0.5 m
X=1.5 m
Low solubility
X=0.5 m
X=1.5 m
High solubility
Tritium transport, 5/6
Total tritium loss in the front duct
Fluxes of tritium through the
steel. S= 0.001 mol/m3/Pa0.5, units
are 10-9 kg/m2/s
More T permeation occurs from the
Hartmann gap, where velocity is low
#
D
S
10-9 m2s-1
mol m-3Pa-
σ
T leak
Ω-1m-1
%
1/2
1
1
0.01
5
1.30
2
2.54
0.01
5
1.40
3
7
0.01
5
1.35
4
2.54
0.0005
5
2.08
5
2.54
0.001
5
1.99
6
2.54
0.005
5
1.65
7
2.54
0.05
5
0.60
8
2.54
0.1
5
0.35
9
2.54
0.01
50
0.36
10
2.54
0.01
500
0.06
Total T leakage < 2%
Tritium transport, 6/6
• Due to very low diffusion coefficient of T in SiC, FCI can
be considered as a T permeation barrier
• All tritium generated in the bulk flow remains there.
Tritium permeation occurs mostly from the gaps,
especially from the Hartmann gap, where velocity is very
low
• Electrical conductivity of the FCI has indirect effect on T
transport via changes in the velocity profile: higher  smaller leakage
• Total T leakage into He can be estimated as 2% of all
tritium generated in the blanket (not taking into account
pressure equalization openings and 3D flow effects)
• More accurate databases for physical properties are
needed