Minutes of March22, 1998 meeting (Atlanta, GA)


Sunday, March 22, 1998 8:30 am - 2:00 pm Atlanta 2 Room Radisson Hotel Atlanta, Georgia


1. Alan Glasser 2. Carl Sovinec 3. Scott Kruger 4. Jim Callen 5. Eric Held 6. Ming Chu 7. Rick Nebel 8. Alfonso Tarditi 9 Sergei Galkin 10. Tom Gianakon 11. Alan Turnbull 12. Jean-Noel Leboeuf

1. Status of Code

Carl: Bug fixes, didn't make much difference Cure for the jaggies: improved boundary conditions elecd -> 0 consistent with ideal MHD boundary conditions

Alan: Jaggies: figures illustrating the effect of improved boundary conditions Packed grids: dispersion relation showing truncation error grid packing algorithm and figures

Carl: need for parallelization over Fourier modes inefficiency of parallization over grid blocks, caused by degradation of preconditioner

Tom: Neoclassical terms CGL pi_parallel addition to Ohm's law Callen: why include pi_i? results inconclusive so far

Carl expressed appreciation of Jim Callen

2. Status of GUI, Tom

Changes here and there; lack good French wine to lubricate process Ming: GUI very useful; need to run on different machines

3. Status of Graphics

XDRAW discussion Tom: request to do EPS rather than PS Ming: stride and label for stepping through family of curves, comparison of eigenfunctions Alan: existence and usefulness of xdraw.ini to control features

XDRAW progress, Alan vector plots now working, though not yet perfected xdraw now works on T3E source code release

ANALYZE: Alan analyze now produces DX output and Poincare plots improved efficiency of Poincare plots by factor of 4, now 1 minute per frame on Octane

NIMPLOT: Carl now calls xdraw from with code

4. Status of Validation Plan

Ideal modes, Ming took Carl's demos and looked in more detail 3 dissipation coefficients: resistivity, viscosity, bdivd works well for ideal internal kink S > 1e6

Resistive modes, Scott, Carl, Alan, Jean-Noel boundary condition, jaggies S=1e4 rattiness in velocity profiles increased my from 47 to 67 helps at beginning but not by end gamma*taua=1.9%, FAR=2% effectiveness of small viscosity to reduce rattiness not an equilibrium problem Jean-Noel, DIII-D deeply reversed case S scaling

2-fluid: Rick APS meeting: whistler resuls DEBS semi-implicit operator NIMROD now exceeds whistler CFL by 1e4 convergence to freqency and growth rate need about 100 time steps per diametnetic rotation period for accuracy pleasant surprise: NIMROD works much better than DEBS

Nonlinear evolution: Carl comparison on saturation of RFP tearring mode in cylinder toroidal RFP cases as well as cylindrical comparison to DEBS on cylindrical

5. Linear Solver

C90 beats T3E better efficiency of parallel preconditioners Ray Tuminaro, Aztec, general preconditioner extra layer, parallelization over Fourier modes Alfonso: Isis, diagonal better than others bl_diaga, 1D direct solve along grid lines, like ADI but cruder averages over two directions vectorizes over perpendicular direction no news on new person at Sandia

6. Discussion of GA Acceptance Plan

Ming, meeting 2/5 at GA, follow up on 3 things more detailed validation of cases

excitation of side bands NCS relaxation

more detailed validation of cases slightly different eqdsk files

two classes of cases NCS disruption (cf Popov) neoclassical seed islands

Nimrod should reproduce FAR case

7. Next steps/plans

Alan: tearing modes

Carl: ideal modes, numerics & physics doing well, good results Main concern: speed 2D grid integral part of code wouldn't be inconceivable to have Fourier region

Alan: fblocks (Fourier in theta)? Carl: dealiasing in theta direction would cause problems

Carl: publications, toroidal RFP, aspect ratio scaling spheromak, R=0, sustainment

Rick: helicity injection current drive, Hall effect Hall MHD for RFP

Scott, tearing S=1e4, 1e5, high aspect ratio

Jean-Noel, no money from Curt, runs FAR eventually interested in running NIMROD numerical tokamak, Lew Hedrick, closures Paul Snyder, working with Ahmet

Tom: neoclassical terms seed island, parameter scans, internal kink, scaling with S; long term neoclassical closure, nonlinear destabilization, cf reduced MHD, anisotropic thermal conduction

Sergei: no special plans 2-fluid effects, cf Altok-E, linear adaptive grids

Alfonso, Alice, German T3E

parallel preconditioners new diagnostics

Short meeting planned Tuesday afternoon to discuss resistive modes Alan, Ming, Scott, Jean-Noel

8. Other old and new business


9. Time and Date for next teleconference

April 9, 1PM EDT

10. Adjourn

Meeting ended early, around 11:30