FUSE meeting notes: Sep 24 2024

[orso82]

Orso

• Added example to generate POPCON plots (https://github.com/ProjectTorreyPines/FuseExamples/blob/master/popcon.ipynb)
  • ini.hcd.power_scaling_cost_function accepts a function that defines how to scale HCD powers to minimize an objective
  • In POPCON plots one scales powers to match an assumed transport scaling law (eg. H98y2=1)
• Overhauled plotting of core_sources and core_transport to allow visualization of ion particles
• Fixed issue where NBI torque was not multiplied by power
• Overhauled flux surface tracing done in IMAS so it can be easily accessed by actors
• Outreach
  • APS banners with FUSE logo
  • FUSE technical sheet got people interested in FUSE at SOFT
  • Press release drafted
• Marinoni student will start using FUSE

Adriana

• APS poster ready
  • Optimization with risk objectives works!!!

Giacomo

• Various fixes to open surfaces identification
  • Still some work to do with the limited case
  • We should always have the equilibrium grids as large as the wall
• SOL actors should save their data in the dd

Galina

• PAM almost ready
  • For now no drift
  • Working on ODE solver for pellet radius

Tim

• Supporting the Columbia people
• Simplified BOP heating
• Re-running
• Optimization
• Instead of limiting Psol/R now try to maximize Zeff

Brendan

• Working on Grad-Hogan
• Getting coil coupled to the plasma for current diffusion
  • Now can ramp the OH can induce a current in the plasma
  • Or let plasma current decay and see current being driven in the vessel
• The boundary condition of the plasma current used in QED is critical to ensure numerical stability
• Meeting with Himank and control group to benchmark with TokSys
• For APS
  • Ohmic rampup
  • Vertical instability ?
• Found issue with how ohmic power was calculated
  • Working with Joey to benchmark against ONETWO, still ONETWO finds it should be higer
  • We should test it against TRANSP

Nan

• Transport with TGLF-NN trained with ions particles
  • Compared with TGYRO, and FluxMatcher can converge and matches experiments even better!
  • The FluxMatcher error is high, though inspection of transport Vs sources shows everything converged well
  • Probably a normalization issue

Tom

• Poster layout, parsed more linear runs
• Happy with using GKNN only inside of rho<0.5
• Julia to ONNX and ONNX to Julia (https://github.com/DrChainsaw/ONNXNaiveNASflux.jl)