Cannot manage lots of FMU inputs/outputs in a SSP model with OmSimulator

[jean-philippe.tavella@…]

At EDF we have some co-simulation graphs with FMUs having hundreds of inputs/outputs.
These graphs can be managed with our tool DNG (​https://bitbucket.org/simulage/daccosim/wiki/Home).

It appears more difficult with OMSimulator as:

• each input/output is graphically distinguished on the left side (for inputs) or the right side (for outputs).
• each FMU-to-FMU connection only carries one variable (FMU.output --> FMU.input).

I suggest as a solution to allow transmission of several variables per connection.

[casella]

Wow. I'm not sure we want to get there now, first we need to make sure that we can manage co-simulation graphs with a handful of objects in an efficient and robust way.

I am a bit suprised by this requirement, what use case do you have that requires so many inputs and outputs? Are they really representing sub-system interaction, or are they just used to fetch external inputs (e.g. active and reactive loads in a sub-grid)?

[jean-philippe.tavella@…]

Exactly active and reactive values for loads in an energetic systems composed of weakly coupled FMUs coming from BuildSysPro (buildings), PowerSysPro (distribution grid), and MixSysPro (heating network).
About 720 buildings, that is to say 1440 inputs for the electric part.
About half of the buildings are connected to the heating grid.

We would also like to add ReqSysPro requirements on that in the future...

[casella]

OK. But does it actually make sense to have a graphical representation of all these connections?

[jean-philippe.tavella@…]

It depends on situations. In some cases I think yes.

[casella]

OK, then the question is if we want OMEdit to actually be able to support this kind of SSP models.

Maybe what makes sense is a solution whereby you can build and edit such model graphically using your dedicated tool, and then you can run them with OMSimulator, without the need of representing them graphically.

What do you think?

[jean-philippe.tavella@…]

Probably a good idea. Should be discussed in a more global context.

What I can tell is that EDF will probably increase >2021 its funding contribution in the OpenModelica consortium but under the condition OMSimulator is improved and gets closer to main DACCOSIM NG features...

I'm waiting for some internal decisions to go on this direction.

[casella]

Replying to jean-philippe.tavella@…:

Probably a good idea. Should be discussed in a more global context.

What I can tell is that EDF will probably increase >2021 its funding contribution in the OpenModelica consortium but under the condition OMSimulator is improved and gets closer to main DACCOSIM NG features...

That would of course be good for us.

I'm waiting for some internal decisions to go on this direction.

OK, thanks for notifying us.