Defining a Custom Weather File in Boptest – FMU Initialization Issue

[babak-mohajer]

Hi everyone,

I am considering using Boptest for my PhD research. As part of this, I want to define my own weather file (.mos) to have more control over the weather conditions used in the simulation. To achieve this, I wrote a script that:

1. Generates a weather file in .mos format.
   

2. Uses IBPSA.BoundaryConditions.WeatherData.ReaderTMY3 to read this weather file within OpenModelica.
   

3. Compiles an FMU from OpenModelica for use in PyFMI.
   

4. Runs the simulation in Python to generate weather data over a full year.
   

Current Issue

1. The FMU compiles successfully and loads without issues in PyFMI.
   

2. However, when running the simulation, it fails during initialization (fmi2EnterInitializationMode), producing the following error:
   

FMIL: module = Model, log level = 2: [logFmi2Call][FMU status:Error] fmi2EnterInitializationMode: terminated by an assertion. pyfmi.fmi.FMUException: Exit Initialize returned with an error. Check the log for information (model.get_log).
What I Have Tried

1. Changing the solver (defaulting to CVode, since IDA is not available).
   

2. Adjusting initial conditions for key variables (weaBus.TDryBul, weaBus.relHum, weaBus.pAtm).
   

3. Increasing the simulation start time (t=10s or t=100s) to avoid singularities.
   

4. Reducing solver step sizes (maxh=0.1, minh=1e-6).
   

How You Can Help

1. Are there specific constraints on how the .mos weather file should be structured for IBPSA.BoundaryConditions.WeatherData.ReaderTMY3?
   

2. Could the error be related to how the FMU is compiled in OpenModelica?
   

3. Are there recommended solvers/settings for this type of FMU?
   

4. Is there a debugging approach you would recommend to better understand why the FMU fails to initialize?
   

5. Does Boptest have any internal solutions for defining custom weather files?
   

Any insights or guidance is highly appreciated!

Thanks in advance,

Babak

[piotre13]

I have a similar problem I would like to use a different weather file within one of your test cases and I do not completely understand which are the steps to follow, can you please help me for this issue? which are the proper steps to do this? where can I find some documentation on this? thanks a lot

[EttoreZ]

Dear @babak-mohajer thank you for your interest in BOPTEST. I will try and answer questions 1 to 5

1. Are there specific constraints on how the .mos weather file should be structured for
2. IBPSA.BoundaryConditions.WeatherData.ReaderTMY3?
3. Could the error be related to how the FMU is compiled in OpenModelica?
4. Are there recommended solvers/settings for this type of FMU?

Your error seems to be due to an ill conditioned .mos file due to the fact that it does not initialize.

5. Is there a debugging approach you would recommend to better understand why the FMU fails to initialize?

Have you tried running the model directly in openModelica or other modelica tools? The asserts coming from IBPSA library usually have a log that helps you understand what is wrong with it.

6. Does Boptest have any internal solutions for defining custom weather files?

BOPTEST provides a standard set of building test cases for control comparison. Changing the weather files undermines that original intent, since you wont't be able to compare your results with anyone else. So as for today there is not a script / tool within BOPTEST that allows you to "easily" switch between weather files. That said if the only thing you want to change in a test case is the weather file you can just swap the ".mos" (for simulation) and ".csv" (for forecasting) within the "wrapped.FMU". FMUs in the end are just .zip files and their file structure is similar to the test case file structure highlighted in the docs. I.e. by exploring the file structure of bestest_air wrapped.fmu you will see something like this:
wrapped.fmu (parent directory)
|── binaries
|── resources
| ─── weather.csv
|── 0(modelica boundary)
└─ DRYCOLD.mos
|.....

The updated wrapped.fmu will then be loaded and simulated in BOPTEST (of course this works if BOPTEST is deployed locally). This should answer also @piotre13 question. The full documentation is available here documentation. The note I would like to make is that the test case building is tailored for the weather that is present (envelope, baseline controller, equipment sizing). Changing only the weather should be done with caution. If the Modelica model itself is changed, then you can follow the developer guideline in the documentation. If you think the new test case is different enough from the existing ones, you can think of contributing it to BOPTEST core set of test cases by submitting an application BOPTEST_TAF, and join IBPSA Project 2 Task 3 for test case development Task3, where the community gathers once a month to discuss new test case developments and help contributors with their test cases (Modelica questions, BOPTEST questions, etc...).

[dhblum]

I would like to just emphasize @EttoreZ's point: "Changing only the weather should be done with caution." There is no auto-sizing in Modelica models, so the existing HVAC system is very unlikely to be sized appropriately for the new climate.