OCT FMU value reference repeats

[DeadParrot]

The 2021/05 OCT update was found to be using the same value references for multiple variables in a number of the Buildings models such as Case600, Case600FF, and Guideline36.

• This is believed to be the cause of some previously unexplained errors in FMI API calls during QSS runs
• QSS has been updated to detect repeat value references and abort the run if any are found
• The value reference repeats were seen independent of whether QSS (including directional derivative support) options were used to build the FMU
• It is not known if this issue affected earlier OCT versions

[DeadParrot]

Versions

The issue was observed with these component versions (may occur with others):

Component	Version	Date
OCT	OCT-r23206_JM-r14295	2021/05
Buildings	9c37781	2021/05/04
QSS	e4a0f62	2021/06/08

Build

Build the FMU with bld from the OCT directory as described in the Status page.

The dependencies can be examined in the modelDescription.xml file.

Run

Run the models with PyFMI or QSS from a console configured as described in the Status page.

PyFMI: Use run under the model OCT directories to generate the CVode simulation.

QSS: Use run under the model QSS directory in a method subdirectory such as QSS2.

[chria]

@DeadParrot Can you mention a few variables that have the same value reference (for Case600FF)?

[DeadParrot]

@chria In my Case600FF OCT build with QSS options enabled I see:

• TRooAir.T, TRooAir.port.T, TRooAnn.u, TRooHou.u, roo.air.convConExt[4].fluid.T, and others have valueReference="2715"
• multiSum.k[1], roo.air.convConWin[1].conFra.u, roo.air.convConWin[1].conWinUns.u, and others have valueReference="58"
• matExtWal.material[1].nSta, matExtWal.material[2].nSta, multiplex3_1.n1, nConBou, and others have valueReference="268437301"

I also see valueReference repeats in the JModelica FMU I built 2 years ago so this isn't a new issue.

[mwetter]

The "2715" are the same temperature, the "58" are in the model Real variables that evaluate to a constant value of 1, and the "268437301" are constants that all have the value 1. So, this looks like OCT handles them as alias variables, which is good.

[DeadParrot]

I assumed some were aliases but I see repeats between different unit quantities. Maybe they are all constants and it isn't an issue. I didn't find any documentation that said valueReference can repeat for the same/aliased quantity but that makes sense for common port values. We just need to be sure that this isn't happening for states/derivatives or non-fixed values that aren't aliased.

[DeadParrot]

It sounds like these are expected and valid aliasing for non-states/derivatives.

I will close this issue and adjust the QSS value reference checking to allow duplicates for non-state/derivative variables with the same variability and causality and start value. This raises a couple of questions:

• Are refinements to this rule needed?
• For non-fixed variables with the same value reference should we report the alias in the QSS log?

Edit: Due to the potential rule complexity for detecting an invalid value reference duplicate I am only checking for duplicates in the derivatives and state variables, where I believe duplicates should not appear despite any model symmetry or constant-valued derivatives.