Issue1080 ROM soil temperature and interzonal heat exchange

[PGorzalka]

Finally, I was able to finish the work on issue #1080
@DaJansenGit & @FWuellhorst thanks for your support!

Changes mainly encompass the work mentioned in this paper:

• building up on changes already implemented via IBPSA merge, some adaptations for non-constant soil temperature
  • Option selector added to AixLib.BoundaryConditions
  • Options added to ZoneBaseRecord
  • Replaced TSoil constant in ThermalZone by option selector
• Interzonal heat exchange with new FiveElements RC model
  • Integration of that model into ThermalZone
  • Interzonal connection of zones in Multizone. For OM compatibility, I had to add an additional NzSplitter model and parameters to make sure arrays are initialized without errors.
  • Added ASHRAE test case 960

Additional minor stuff:

• made some parts of the ThermalZone model replaceable, so boundary conditions can be adapted more easily
• two missing eachs in AixLib.Fluid.Pools.BaseClasses.AirFlowMoistureToROM caused OM to fail -> fixed

Missing:

• Test Case 960 was not actually tested against standard results because the weather file provided (AixLib\Resources\weatherdata\ASHRAE140.mos) does not contain the weather data provided by the standard.

See also PR 785 in TEASER

Looking forward to your feedback on the changes!

[FWuellhorst]

@PGorzalka , nice, thanks for adding this nice feature to AixLib! We have an extensive CI which does a lot of groundwork of the review, but it only runs for internal branches. Thus, I added an internal branch based on your fork. Now, we see if naming guidelines and regression tests all run. Depending on the changes, you either can work in your fork and merge your fork into the internal branch, or we could add you as a developer so you can work directly in AixLib @DaJansenGit ?

[DaJansenGit]

@PGorzalka thanks for contributing!
@FWuellhorst Lets see what the CI outputs. If there are bigger changes that need to be addressed, I would add Philip as developer. Thanks for creating the CI branch!

[FWuellhorst]

@PGorzalka : You can find check and naming errors here:
https://rwth-ebc.github.io/AixLib/PGorzalka_issue1080_ROM-soil-and-zone-borders/index.html
With regard to the naming violations, please just address new code lines/files you added.

[FWuellhorst]

The regression plots are displayed here: https://rwth-ebc.github.io/AixLib/PGorzalka_issue1080_ROM-soil-and-zone-borders/charts/ThermalZones/AixLib_ThermalZones_ReducedOrder_Examples_MultizoneInterzonalsFixedHeater.html
Are the fluctuations in zone 1 intended? I had to change to cvode solver, as radau gets timeouts in CI.

Some other results seem to have changed translation statistics. Without interzonal exchange, the change in soil temperature model could cause this, or?

[PGorzalka]

Thank you for checking the results!

Regarding fluctuations: this is indeed intended. Zone 1 is an unheated attic with high infiltration.

Regarding translation: yes, this could be caused by the soil temperature model. Additionally, the FiveElement model has replaced the FourElement in the Multizone.

[FWuellhorst]

Ok, it seems to only affect this model:

*** Warning: AixLib.ThermalZones.ReducedOrder.Examples.MultizoneMoistAirCO2: Translation statistics for simulation changed for linear, but results are unchanged.
 Old = 4, 0, 4, 0, 4, 0, 4, 0, 4, 0
 New = 3, 0, 3, 0, 3, 0, 3, 0, 3, 0

I will update these statistics, as smaller is better. Then you can update your fork with the main and the aixlib-internal fork branch.

[PGorzalka]

Thank you! I merged the changes back into this branch. The CI branch still seems to have some issues though?!

[FWuellhorst]

@PGorzalka it seems that your model takes more than 300 s to solve in CI, where multiple models run in parallel. Running it locally in docker on a single core, it works fine. 4 processors work fine as well, but I guess this is the problem, because various models get timeouts which worked fine before.
So, my idea would be to decrease the simulation time to maybe 3-4 days? As long as it is a phase with interzonal exchange, it should be sufficient for regression test purposes.

[PGorzalka]

@FWuellhorst i changed the T_start in the example so it requires less time to initialize. Simulated time is now 140 hours to have at least a few of the daily dynamics included. I think I removed the reference results, but I'm not sure. Can you have a look if it works now?

[FWuellhorst]

@PGorzalka Thanks. Now, the CI runs. @DaJansenGit From my side the changes look good, do you have open issues?
@PGorzalka , you have to merge the new reference results into your fork if we merge your fork.

[PGorzalka]

@FWuellhorst @DaJansenGit thanks! Did that. I also added our paper to the publications list.

[DaJansenGit]

Everything looking fine now. I just updated the publications.md in your branch as it was faulty formatted. Thanks again for the extensive change to the model.