LM-MHD Simulation Development and Recent Results by pptfiles


									         LM-MHD Simulation
       Development and Recent
       Presented by Sergey Smolentsev (UCLA)

with contribution from:
R. Munipalli, P. Huang (HyPerComp)
M. Abdou, N. Morley, K. Messadek, N. Vetcha, D. Sutevski (UCLA)
R. Moreau (SIMAP, France)
Z. Xu (SWIP, China)
    MHD and heat/mass transfer considerations are
   primary drivers of any liquid metal blanket design

• The motion of electrically conducting
  breeder/coolant in strong, plasma-
  confining, magnetic field induces
  electric currents, which in turn interact
  with the magnetic field, resulting in
  Lorentz forces that modify the original
  flow in many ways. This is a subject of
  magnetohydrodynamics (MHD).
• For decades, blankets were designed
  using simplified MHD flow models (slug
  flow, core flow approximation, etc.). The
  main focus was on MHD pressure drop.
• Recent blanket studies have shown that
  the MHD phenomena in blankets are
  much richer and very complex (e.g.,
  turbulence, coupling with heat and
  mass transfer, etc.) and need much
  more sophisticated analyses.
 MHD Thermofluid issues of LM blankets

            MHD related issue / phenomena                  S-C      DCLL      HCLL

1. MHD pressure drop                                       ***        **       **

2. Electrical insulation                                   ***        **        *

3. Flow in a non-uniform magnetic field                    ***       ***       **

4. Buoyant flows                                            **       ***       ***

5. MHD instabilities and turbulence                        ***       ***        *

6. Complex geometry flow and flow balancing                ***       ***       ***

7. Electromagnetic coupling                                ***        **       ***

8. Thermal insulation                                       *        ***        *

9. Interfacial phenomena                                   ***       ***        *

  *- not applicable or low importance; ** - important; *** - very important
    Where we are on MHD modeling for fusion?

• No commercial MHD CFD codes
• Modification of existing CFD codes (Fluent,
  Flow3D, OpenFoam) – no significant progress yet,
  results are often obviously wrong
• Many 2D, Q2D and 3D research codes – still limited
  to simple geometries; other restrictions
• Development of specialized MHD codes for
  blanket applications (e.g. HIMAG) – good progress
  but there is a need for further improvement to achieve
  blanket relevant conditions: Ha~104, Gr~1012
    MHD modeling and code
development at UCLA/HyPerComp

 • HIMAG (along with HYPERCOMP) –
   ongoing work on development of 3D MHD
   parallel MHD software for LM blanket
 • 2D, Q2D and 3D research codes to
   address particular MHD flows under
   blanket relevant conditions
             In this presentation:
• New modeling results for “mixed convection” in poloidal flows
• Study of hydrodynamic instabilities and transitions in MHD flows with
  “M-shaped” velocity profile
• 3D modeling of Flow Channel Insert (FCI) experiment in China
               TITLE                 Presenter        Oral/Poster
3D HIMAG development progress        R. Munipalli         oral
Study of MHD mixed convection in      N. Vetcha          poster
poloidal flows of DCLL blanket          UCLA
Modeling China FCI experiment        D. Sutevski         poster
LM-MHD experiments and PbLi loop     K. Messadek          oral
progress                                 UCLA
           Mixed Convection (MC)

• In poloidal ducts, volumetric heating
  causes strong Archimedes forces in PbLi,
  resulting in buoyant flows
• Forced flow ~ 10 cm/s
                                                 In the DCLL blanket conditions,
  Buoyant flow ~ 30 cm/s
                                                 the poloidal flows are expected
• MC affects the temperature field in the FCI,   to be hydrodynamically unstable and
  interfacial temperature, heat losses and       eventually turbulent
  tritium transport – all IMPORTANT!
 How we attack the MC problem
• Full 3D computations using HIMAG: limited to
  Ha~1000, Re~10,000, Gr~10^7; the code does
  not reproduce turbulence
• Spectral Q2D MHD code (UCLA, Smolentsev):
  captures MHD turbulence but limited to simplified
  geometry and periodic BC
• 1D analytical solution for undisturbed flow
• Linear stability analysis to predict transitions in
  the flow – see poster presentation by N. Vetcha
• Experiment – see presentation by K. Messadek
     3D modeling of MC flows
  Ha=100        Ha=400            Ha=700           Ha=1000


            g                 g                g

Tendency to quasi-two-dimensional state as Ha number is increased has
been demonstrated for both velocity and temperature field
   3D modeling of MC flows
                                • Pronounced entry/exit
Ha=400         Ha=700   Ha=1000 • Reverse flow bubble at
                                  the entry
                                • Accelerated flow zone
                                  at the entry
                                • “Hot” spot in the left-top
                                • Reduction of entry/exit
Ha=400         Ha=700   Ha=1000   effects with B
                                • Near fully developed
                                  flow in the middle
MC: comparison between 3D and 1D

     Full solution       Wall functions BC             Wall functions BC

  Ha=400                Ha=700                    Ha=1000

                            1D analytical solution
      Fully developed

                            -Flow is Q2D
                            -Flow is fully developed

                            Major assumptions of the 1D theory
                            have been verified with 3D modeling.
                            1D/3D comparison is fair
                           MHD turbulence, instability and transitions

                                                                                                                                           •   All liquid metal blankets fall on the
                                                                                                                                               sub-region below the line
                                                                                                                                               Re/Ha~200 associated with the
                                                                                                                                               turbulization of the Hartmann
                                                                                                                                               layer. Here, MHD turbulence
                                                                                                                                               exists in a very specific quasi-two-
                                 Molten salt self-cooled

                                                                  TURBULENT FLOW
                                                                                                                                               dimensional (Q2D) form.
                                                                                                                                           •   The Q2D turbulent structures
Reynolds number

                                                                             PbLi self-cooled                                                  appear as large columnar-like
                                                                                                                                               vortices aligned with the field
                  105                                                                                                                          direction. This Q2D MHD
                                                                                                                                               turbulence is mostly foreseen in
                                                                                         DCLL, DEMO OB

                                                                 DCLL, ITER TBM
                                                                                                         DCLL, DEMO IB
                                                                                                                         Li self-cooled

                  104                                                                                                                          long poloidal ducts resulting in a
                                                                                                                                               strong modification of heat and
                  103                                                                                                                          mass transfer.
                                                                  HCLL, ITER TBM
                  10   2
                             LAMINAR or                                                                                                    •   We do some analysis for MHD
                             Q2D TURBULENT FLOW                                                                                                instability and laminar-turbulent
                  101                                                                                                                          transitions for flows with “M-
                           101                             102         103         104                                               105       shaped” velocity profiles, which
                                                           Hartmann number                                                                     are typical to blanket conditions
MHD turbulence, instability and transitions

   Direct Numerical Simulation of Q2D MHD turbulence
                                                    Side layer
                                                     shear layers

                       The next few movies will illustrate
Type I                 major findings, namely:
         Type II
                        •How the instability starts
                        •Two types of instability
                        •Primarily instability (Type I):
                         inflectional instability
                        •Secondary instability (Type II):
                         bulk eddy/wall interaction
                        •How MHD turbulence eventually evolves
      MHD turbulence, instability and transitions

  Type I (primarily) instability (Re=2500, Ha=200)

Transition from Type I to Type II instability and evolvement of MHD turbulence
    Modeling FCI experiment in China
      M.S. TILLACK, S. MALANG, “High Performance PbLi Blanket,”
      Proc.17th IEE/NPSS Symposium on Fusion Engineering, Vol.2,          Poloidal duct of the DCLL blanket
      1000-1004, San Diego, California, Oct.6-10, 1997.                   with FCI and helium channels
   Sic/SiC FCI is used inside the DCLL blanket and also in the
    feeding ducts as electrical and thermal insulator allowing for
    ΔP<2 MPa, T~700 C, >40%

   Possible thermal deformations and small FCI displacements are
    accommodated with a ~ 2-mm gap also filled with PbLi

   Tritium and corrosion products in the gap are removed with the
    slowly flowing PbLi

   There are pressure equalization openings in the FCI, either in the
    form of holes (PEH) or a single slot (PES), to equalize the
    pressure between the gap and the bulk flow

   The FCI surfaces are sealed with CVD-SiC to prevent “soaking”
    PbLi. The sealing layer can also serve as a tritium permeation

   The FCI is subdivided into sections, each about 0.25-0.5 m long.
    Two FCI sections are loosely overlapped at the junction, similar to
    roof tiles

   The FCI is thought as a purely functional (not a structural)
    element experiencing only secondary stresses, which can be
                                                                            Two overlapping FCI sections
     Modeling FCI experiment in China
Flow of InGaSn in a SS rectangular duct with ideally insulating FCI made of epoxy
subject to a strong (2 T) transverse magnetic field

                                                       1000 mm

Picture of experimental MHD facilities in the
Southerstern Institute of Physics (SWIP), China.

  Courtesy of Prof. Zengyu XU, SWIP
Modeling FCI experiment in China
                  Dimension           Notation            Value, m
         Half-width of the FCI box       b                 0.023
         Half-height of the FCI box      a                 0.027
         FCI thickness                  tFCI               0.002
         Thickness of the gap            tg                0.005
         Thickness of the slot           ts                0.003
         Thickness of the Fe wall        tw                0.002

         •2 mm FCI made of epoxy provides ideal electrical insulation
         •Maximum magnetic field is 2 T (Ha=2400)
         •Uniform B-field: 740 mm (length) x 170 mm (width) x 80 mm (height)
         •Outer SS rectangular duct: 1500 mm long
         •FCI box: 1000 mm long
         •Pressure equalization openings: slot (PES) or holes (PES)
         •Measurements: pressure drop, velocity (LEVI)
          Modeling was performed under the experimental conditions
          using the fully developed flow model first (2009) and then
          in 3D (2010) using HIMAG
 Modeling FCI experiment in China
S. SMOLENTSEV, Z.XU, C.PAN, M.ABDOU, Numerical and              Previous 2D computations show
Experimental Studies of MHD flow in a Rectangular Duct with a
Non-Conducting Flow Insert, Magnetohydrodynamics, 46, 99-111     MHD pressure drop much smaller
(2010).                                                          than that in the experiment
            Pressure drop coefficient
                                                                Current 3D computations demonstrate
                                                                 good match with the experiment
                                                                These suggest 3D axial currents
                           3D modeling, NEW !

                           2D modeling, previous                    2

                                                                  In this figure jx (axial current) is plotted:
                                                                  1 – axial current in the gap, just above the slot
                                                                  2 – return current
           Concluding remarks
• In the recent past, the main focus of MHD studies for fusion
  applications was placed mostly on MHD pressure drop.
• Although MHD pressure drop still remains one of the most important
  issues, current studies are more focusing on the detailed structure of
  MHD flows in the blanket, including various 3D and unsteady
• These phenomena are not fully understood yet. For example, the
  mass transport (e.g. tritium permeation, corrosion) is closely coupled
  with MHD flows and heat transfer, requiring much better knowledge
  compared to relatively simple pressure drop predictions.
• Therefore, the key to the development of advanced liquid metal
  blankets for future power plants lies in a better understanding of
  complex MHD flows, both laminar and turbulent, via developing
  validated numerical tools and physical experiments.

To top