TODO for December 2022 Release

[lbianchi-lbl]

Summary

There are two distinct sets of changes that should be merged into IDAES/idaes-pse for DISPATCHES, which I'm calling very arbitrarily A and B:

A

• What it is Updates to MultiperiodModel
• What is affected Needed for the nuclear case study (Although all three case studies use MultiPeriodModel, the updates are needed only for the nuclear case study. The updates are backwards compatible, so no changes are needed to the other case studies.)
• How it is now: MultiperiodModel is re-implemented in DISPATCHES (i.e. imported from https://github.com/gmlc-dispatches/dispatches/blob/main/dispatches/models/nuclear_case/flowsheets/multiperiod.py)
• How it should be: MultiperiodModel is imported from idaes.apps.grid_integration (Both renewables and fossil case studies are already doing this. After the updates are merged, the nuclear case study should be able to do it too.)

B

• What it is Real-time bidding
• What is affected
  • Currently needed by the RE case study
• How it is now
  • The required changes are implemented in a separate fork and branch/tag of IDAES (dguittet:2.0.0.0a4)
    • Most of the changes were authored by @xiangao1; @dguittet created the branch/tag to store the changes to be able to include the RE case study in the June 2022 release during Add core infra for RE double-loop simulation #124
    • IMPORTANT these changes are NOT compatible with the rest of the codebase (i.e. just setting the idaes-pse requirement to this ref in setup.py will cause the non-RE case studies to fail)
    • This link shows the differences between dguittet:2.0.0.a4 and IDAES:main, i.e. "if a PR was opened from dguittet:2.0.0.a4 to IDAES:main, what would it look like?"): IDAES/idaes-pse@IDAES:main...dguittet:2.0.0.a4
  • An explicit check in various parts of the RE case study codebase ensures that the required functionality is available by looking at the version of the installed
  • To be able to test the RE notebooks in CI, a fairly clunky workaround is implemented in the CI workflow to run the RE notebooks only after re-installing idaes-pse from dguittet:2.0.0.a4
• How it should be: These changes should be merged into IDAES/idaes-pse together with A so that the same version of idaes-pse works for the entire DISPATCHES codebase

C

• ThermalGeneratorModelData is needed by the RE case study, but missing from the current IDAES
• Fix AGC bounds in grid integration code for thermal generators
• Update NE and FE double loop flow sheets to use above updates

D

• Fix issues with double-loop initial status (Fixing grid integration double-loop issue with initial status IDAES/idaes-pse#1002)

PRs in IDAES/idaes-pse

• (A) Include agc Pmin and Pmax in double loop model data IDAES/idaes-pse#899
• (A) Add additional methods to MultiPeriodModel class IDAES/idaes-pse#908
• (B) Real-time bidding in double loop IDAES/idaes-pse#855
• (B) Generalize LMP forecaster IDAES/idaes-pse#870
• (C) Update grid integration IDAES/idaes-pse#986
• (C) Grid integration: update bounds of AGC bid IDAES/idaes-pse#996
• (D) Fixing grid integration double-loop issue with initial status IDAES/idaes-pse#1002

TODO in DISPATCHES

• Make all the scripts and notebooks in the repo compatible with IDAES 2.0 and switch back to upstream IDAES in setup.py (addressed in Make DISPATCHES compatible with IDAES 2.0 #151). Note: This PR will also include the changes in Update NE double loop notebook for RT-bidding and Backcaster #139, so Update NE double loop notebook for RT-bidding and Backcaster #139 can be closed after Make DISPATCHES compatible with IDAES 2.0 #151 is merged.
• (A) Update the nuclear case study notebooks with the IDAES multiperiod import and delete multiperiod.py in the NE case study folder Update IDAES multiperiod.py and remove DISPATCHES multiperiod.py  #112 (addressed in Update the NE flowsheet to use the new MultiPeriodModel class in IDAES #143)
• (B) Modify RE case study as needed (Ensuring full compatibility between Double loop notebooks and upstream IDAES #154)
• Fix convergence issues and test failures in the fossil case study (Fix discharge flowsheet in fossil case study #153) (Addressed in Idaes2update #156) and Update idaes-pse requirement to 2.0.0.dev4.dispatches.22.11.14 #161
• Add concrete TES unit model to the repo (addressed in Add Concrete TES unit model #100)
• Add RAVEN and TEAL dependencies to DISPATCHES (Need RAVEN dependency for DISPATCHES #76)

Moved to #123

• Add steady state surrogate code from Jordan's fork (addressed in Add steady state surrogates code #142)
• Modify NE case study Jupyter notebook to reflect RAVEN and TEAL integration (addressed in Adding TEAL cashflow and synthetic history generation to new NE multiperiod notebook #150)
• (D) Any changes connected with Fixing grid integration double-loop issue with initial status IDAES/idaes-pse#1002

[radhakrishnatg]

@dguittet From what I understand, you need this branch https://github.com/dguittet/idaes-pse/tree/2.0.0.a4 for the renewable case study. So, are you planning to create a PR in IDAES with these changes? If yes, it would be better to merge these changes as soon as possible.

[radhakrishnatg]

@dguittet @xiangao1 Another question. Are the changes to Bidder, Coordinator and Tracker classes in this branch (https://github.com/dguittet/idaes-pse/tree/2.0.0.a4) valid in general, or are they RE case study specific? @lbianchi-lbl and I were thinking we could do the following. Merge the changes to the above classes that are not case study specific with IDAES main. And, the changes that are case study specific can be included in DISPATCHES. Let us know what your thoughts are regarding this. Thank you!

[adowling2]

My vote is to generalize and contribute back to IDAES to the extent possible.

[ksbeattie]

@dguittet : We need to determine if this is generalizable and contribute this back to IDAES - which would need to happen for the IDAES Aug release.

[lbianchi-lbl]

@ksbeattie I've just updated the issue description with a more detailed description of (what I understand being) the situation. @dguittet @radhakrishnatg, could you verify that the information is correct, and add fixes and/or clarifications as needed?

[radhakrishnatg]

Sorry, @lbianchi-lbl ! I must have missed the notification, and I noticed it just now. Thanks for capturing everything we discussed! Everything looks good. I only made minor changes the description.

[dguittet]

Looks great @lbianchi-lbl

One thing that I haven't seen written here that we've discussed is: in (B), needing to update the nuclear and fossil double loop notebooks to use the changes. Once idaes-pse has been released with the changes in (B), those notebooks released in the current dispatches are going to be incompatible with this newest idaes-pse. The next dispatches release that follows this idaes Aug release should contain changes to all 3 double loop notebooks.