WIMAS-2 Weekly Meeting

Wednesday 3 Jun 2020, 10:00 → 11:10 Europe/Berlin

https://kth-se.zoom.us/j/7764082514

Thomas Jonsson

WIMAS-2 Weekly Meeting are technical meetings to plan and coordinate the development of the ETS-6 workflow.

 

Remote participation:

https://kth-se.zoom.us/j/7764082514

Minutes from discussion around Jorge's presentation on the modelling of Ohmic scenarios with the ETS-6

• {R0,B0} is different in equilibrium and core_profiles.
  • Seems like CHEASE is updating R0 from a definition different from the one in the input.
  • ACTION 1 on Thomas: Discuss with Olivier if we could add an option in CHEASE not to modify R0.
  • Consistency in {R0,B0} between IDSs should be check by a ETS-6 validation tools.
  • Proposal: Let either ETS-init or an equilibrium-database actor set {R0,B0} from the TF IDS.
    • ACTION 2 on WIMAS-2 team: Decide what we want.
• Proposal from Thomas and Jorge: Update the database_plasma_boundary actor into a equilibrium-database actor
  • In this actor we should be able to select what part of the equilibrium want to update, e.g. {B0,R0}, the boundary, the profiles etc.
  • After this actor we can choose to run CHEASE or not. Thus, it enables us to run with an interpretative B0, boundary, while the rest of the equilibrium is predicted. Or we could run a full interpretative equilibrium. 
  • ACTION 3 on WIMAS-2 team: Decide what we want.
• Proposal from Jorge: We should add a sawtooth actor in the ETS-6 during 2020.
  • ACTION 4 on WIMAS-2 team: Decide what we want.
• The branch IPramp in the ets-core repo, should be merged into develop. (Jorge should first rebase, then Thomas will merge).
  • ACTION 5 on Jorge: Rebase IPramp from develop.
    • COMPLETED
  • ACTION 6 on Thomas: Merge IPramp into develop.
• Jorge has a list of issues that needs to be addressed.
  • ACTION 7 on Jorge: Write an informal report on these issues.
• NCLASS crashes if E-parallel is not filled on input. It looks like it modifies the data internally, which causes ids_deallocate to crash in the wrapper.
  • ACTION 8 on Jorge or Par: Debug TCI
• In order todo more modelling with impurities we add impurities with the ion_mix script. Possible additions required.
• Sawteeth are very important for the high current ohmic scenarios. A reference was provided by David Coster
  • "A new explanation of the sawtooth phenomena in tokamaks featured", Physics of Plasmas 27, 032509 (2020);
    https://doi.org/10.1063/1.5140968
• Alexei proposed that someone should study the correlation between Zeff and Ip, like they have already done at AUG
  • On the dependence of Zef ~ (I/n)^3: Nucl. Fusion 57 (2017) 126067 (9pp)
    https://doi.org/10.1088/1741-4326/aa8e32  

Improvements needed in IMASgo!

• ACTION 9 on Jorge: Add Li, Vloop and Wdia
• ACTION 10 on Jorge: Add option to select the impurity species to match Zeff

What's next...

• Find a better scenario from JET
• WEST, Philippe
• ITER, ramp with e.g. DINA
  • ACTION 11 on Nathan: write to Alexei and Sun-Hee
• Add impurities and re-run JET discharge.
• TCV:
  • ACTION 12 on Thomas: Ask for Ohmic reference

1 Actions from last meeting

https://docs.google.com/spreadsheets/d/1Ps-cGQyKNsYTObCr5Kqu58bXBNQdH45Fjpz5QD0hLB0/edit?usp=sharing

Speaker: Thomas Jonsson

2 Active branches

Here we will go through and discuss the ongoing developments on different branches in the ets-core repository.

Speaker: Thomas Jonsson

New active branches

• ...

Old active branches

• IPramp - see presentation from Jorge later on today.
• revise_impurity - Denis.
• pinchVelocity - Still not done the testing...

Merged branches

• intentETSInit - All routines in ets_ini.f90 use intent-statement.
• quasineutralSources - check that all particle sources are quasi-neutral.
• quasineutrality_with_DNE_and_DNI - add derivatives in output from quasineutrality routine.
• database_profiles - changed the sign of gradient-boundary condition.
• sanitise - minor addition to what was committed last week.
• transport_combiner - add an equilibrium port to the transport-combiner actor; first step towards calculating the flux.
  • NOTE: To use latest ets_core/develop you need to use ETSrc2.xml (not yet in dressed Kepler).
• Cherry-picking from IPramp - mostly cosmetic changes.

3 Full ohmic scenario modelling with ETS6

Speaker: Dr Jorge Ferreira (Instituto Superior Técnico / Instituto de Plasmas e Fusão Nuclear)

ohmic_ets_6.pdf

4 Any other business

Need to discuss:

1/ questionnaire

2/ move default to 3.28.1

3/ empty input IDS's

Hoenen Olivier added a comment - Yesterday

Not sure to understand Owsiak Michal, currently we never 'put' empty IDS from the AL point of view (empty means ids_properties/homogeneous_time is not set, in this case the put operation returns with no actions taken), this is not depending on the backend at all.

Now just to make sure we do understand the feature request : Coster David this feature is basically to ensure that the selected IDS are NOT copied from input run to runwork is that correct?

Permalink

Palak Bartek added a comment - Yesterday

Hoenen Olivier: How behaves GET? I see no checks of homogeneous_time... It just go through all fields returning default values (because they don't exist) independent of BE?

Permalink Edit Delete

Coster David added a comment - Yesterday

We would like to have the same behaviour as if we had attempted to read the data from the input, but there was no data present.

The problem we are trying to address is that continuing an ETS run from a previous ETS simulation can have a very expensive start-up time as all of the IDS's that are present as the output from ualInit are copied to runwork, even if we know we are never going to read from them.  These include distribution and distribution_sources, both of which can be big.  We would still like to have the Kepler IDS identifier to be present on the various connectors, but pointing at an empty object.

Permalink

Hoenen Olivier added a comment - Yesterday

Thanks for the clarification Coster David. 

From the description of the problem you are trying to avoid here (without impacting the direct solution proposed by Owsiak Michal), I believe that this would be solved nicely by simply no copying at all data to runwork at the UALInit stage. This was proposed already in IMAS-2687, so maybe you could vote for this development such as we would increase its priority? 

Permalink

Owsiak Michal added a comment - 2 hours ago

I will wait for a final decision here before proceeding with development.

• If we want to return empty IDSes - by not copying them, I will implement new way of passing data from UALInit by skipping some of fields while producing runwork
• If we want faster mechanizm for copying whole files, I suggest to mark this one as duplicate of IMAS-2687.

Till we have final decision, I am moving it to backlog and switching to something else.