Possibly bad number of equations and variables computation

[massimo ceraolo]

When OM R16555 checks the enclosed model it reports 33 equations and 34 variables.
Instead, the model should be balanced (Dymola reports 33 equations and variables).

[Per Östlund]

It seems like there might be some issues with this model. Dymola does say it's balanced when you check it, but the model does not simulate. Dymola gives an error about overconstrained connectors which are connected but that does not have any root defined when you try to simulate it.

[massimo ceraolo]

The model does not simulate because it is a sensor and as such it needs to be connected to a circuit to operate.
I've added a tiny package named "BugPeak", that contains the model "simpleMesh".
This model uses "getUPeak" and checks and runs well under Dymola.
OM complains about equation balance and does not run it.

[Christian Schubert]

As far as I can tell, it has got to do with the connection graph. There are two VoltageSources within the system. Each having the root:

Connections.root(pin_p.reference);

And therefore each source provides an equation for reference.gamma:

omega = 2*Modelica.Constants.pi*f;
omega = der(pin_p.reference.gamma);

Problem is that in the OpenModelica Connection Graph both roots are connected to each other and therefore (indirectly)

U.pin_p.reference.gamma = U1.pin_p.reference.gamma;
der(U.pin_p.reference.gamma) = U.omega; 
der(U1.pin_p.reference.gamma) = U1.omega;

holds, with known U.omega and U1.omega.

The question that springs to mind is if the sources are modelled correctly?

A possible solution could be not to connect two roots with each other, i.e. also break the connection between getUPeak1.negativePin and U.pin_n for example.
Although, in this case, this seems physically dodgy to me as this would allow the user to connect two sources with different frequency. But then there is the assert statement

 assert(abs(reference1.gamma - reference2.gamma) < 1E-6*2*Modelica.Constants.pi, "Reference angles should be equal!");

Which would at least throw an error at runtime.

[Christian Schubert]

yEd ConnectionGraph (without the inductor)

[massimo ceraolo]

I've created a very small model containing the issue.
It contains just one source.
It is in "tiny.mo" that I will upload in a minute.

[massimo ceraolo]

A model containing the issue (model tiny in package tiny)