FUSE meeting notes: Oct 15 2024

[orso82]

Orso

• APS

  • TE and CFS very interested to use/collaborate/contribute
  • Lots of interest by IFE
    • FocusedEnergy, PacificFusion, GA,...
    • They need a data structure, simple (reduced) models, then increase fidelity
    • Opportunity to provide a common platform

• Synthetic generator

  • Lot's of interest by anyone who wants to train NNs
  • Steve Sabbah's DECAF too
  • We need a worked example (we almost have it already)
  • Could be a killer app for FUSE

• Discord

  • please use Name Surname
  • eventually we should move devs meeting from ZOOM to Discrod

• Got access to costing models used in Columbia class

  • Adriana to look into it and work with students

• OMAS + IMAS.jl ?

  • CFS and TE are extensively using OMAS (that's how CFS released their pf/wall/circuit data)
  • Some concerns about future OMAS support
  • Also people were not aware the IMAS.jl can natively save/load native IMAS h5 format
  • May be good to try using IMAS.jl as backend for OMAS: also a way to ease people into Julia

• Projects

  • https://github.com/ProjectTorreyPines/public_project_ideas/issues

• IOS - ITPA

  • core impurity transport -- Nan
  • core-edge coupling... we need a plan
    • Zeff-scan + evaluate controllability?
    • Useful discussions about core-edge integration with Jerome+Giacomo

Tim

• Code-camp
  • Many internationals (8-9 already!)
  • Worried about time we can dedicate to people... especially the ones traveling from far
  • Should we have another one for GA only? @bclyons12 it should be shorter (like 2 days)
  • SET, when?

Brendan

• Grad Hogan solver did not get it to work before APS
  • Luke or Himank to work on a simple shape controller
  • @mgyoo86 implemented feedback control for RAPID-2D
• Should be done before the tokamaks FOA is due (two months?)
• Talked to Chris Hansen, Nate Ferraro, Steve Jardin

Min Gu

• New TroyonBetaNN.jl to predict no-wall ideal beta limit
  • All aspect ratios, different pressure and q profiles
  • Only positive triangularity though
• Will develop new Actor
• There's also another NN

David

• Comparing neutronics in FUSE (NNeutronics.jl) with high fidelity runs
• TBR looks good
• e-folding length for volumetric heating is off by factor of 2 (attenuation in NNeutronics.jl is pessimistic)
• Neutron wall loading is good (reduced model does not consider scattering)
• Will investigate where difference comes from

Adriana

• Lots of interest about risk assessment
• Need to add something about risk associated to SOL challenge
  • Links to Psol/R
  • What's the impact on availability (coolant leak...)
  • Probability of failure?
• Want to write a paper specifically on risk + costing

Nan

• Setting up ITER with W impurity transport

Tom

• Asked dedicated NERSC resources for training of TGLFNN
• Good suggestions for error metric to compare profiles
• Joe will keep focusing TGLF-EP, will want to integrate it into FUSE
  • Erik Bass will try to use TJLF-EP and develop the GPU milestone there