Modelica.Electrical.Batteries.Utilities.BusTranscription does not respect the specification around expandable connectors

[qlambert-pro]

In Modelica.Electrical.Batteries.Utilities.BusTranscription the connect equation

connect(gainLosses.u, stackBus.cellBus.lossPower);

fails to respect

"One connector in the connect equation must reference a declared component, and if the other connector is an undeclared element in a declared expandable connector it is handled as follows (elements that are only potentially present are not seen as declared)"

as lossPower is an undeclared element in undeclared expandable connector cellBus

This relates to #428 and #2172

[HansOlsson]

For #2172 the main issue was lack of realistic examples, and it now seems we have them.
In particular the array declaration of cellBus is good since it illustrates why following are good ideas:

• Requiring that it is potentially present; as we otherwise wouldn't know if it should be an array of lossPower (as in stackBusArrays) or array of cellBus (as in stackBus).
• Should be able to connect to it when only potentially present.
• Have the feature at all since otherwise we would need a free-floating stack-bus-array-declaration in the model, which seems odd.

[HansOlsson]

Proposed new text:

At least one connector in the connect equation must reference a declared or potentially present component.
If potentially present component it will be marked as being present (recursively if needed).
If the other connector is not even potentially present it must be an undeclared element in a at least potentially present expandable connector and it is handled as follows (after marking the potentially present expandable connector as present):

Non-normative:
The undeclared component in 'potentially present expandable connector' makes it possible to have a 'potentially present array of expandable connectors'.
(Add example.)

Poll using emojis:

• Thumb up if in favor
• Thumb down if against
• Eyes if we need to discuss it more.

[qlambert-pro]

I don't understand the wording. Can you explain what your text is trying to do that specifying that potentially present component are considered declared does not?

[HansOlsson]

I don't understand the wording. Can you explain what your text is trying to do that specifying that potentially present component are considered declared does not?

I see at least the following reasons:

• Allow us to deduce their causality - that may differ from their declared causality. Obviously we could have some other variant for handling that.
• Handle flow-variables in them (if there are any physical connectors inside the expandable connector). The issue is that any flow-variables in expandable connectors need to be added as both inside and outside connectors for the set-handling (otherwise there are too few equations). Adding them even if there aren't any connections seems to make that under-determined in some cases - I have reproduced that below.
• Making it clear that it behaves as if they aren't there.

I'm sure there might be other solutions, but that might require a lot of work and it is not clear that those potential solutions would be easier to understand.

In the package below MOk doesn't have the potentially present variable and thus translates ok. In MBad I added a silly connections to create it (which shouldn't be legal), and that fails. (The reason for the current rules is if a sub-component has an expandable connector without any contents and it is added elsewhere there would otherwise be a missing equation - I could add that as well.)

package P
  expandable connector C
    annotation (Icon(coordinateSystem(preserveAspectRatio=false)), Diagram(
          coordinateSystem(preserveAspectRatio=false)));
  end C;

  expandable connector SubC
    Modelica.Electrical.Analog.Interfaces.Pin pin;
    annotation (Icon(coordinateSystem(preserveAspectRatio=false)), Diagram(
          coordinateSystem(preserveAspectRatio=false)));
  end SubC;

  model MOk
    C c annotation (Placement(transformation(extent={{-110,-18},{-90,2}})));
    SubC subC annotation (Placement(transformation(extent={{94,26},{114,46}})));
  equation 
    connect(c, subC.sub) annotation (Line(points={{-100,-8},{60,-8},{60,36},{104,
            36}},
          color={0,0,0}), Text(
        string="%second",
        index=-1,
        extent={{-6,3},{-6,3}},
        horizontalAlignment=TextAlignment.Right));
    annotation (Icon(coordinateSystem(preserveAspectRatio=false)), Diagram(
          coordinateSystem(preserveAspectRatio=false)));
  end MOk;

  model MBad
    C c annotation (Placement(transformation(extent={{-110,-18},{-90,2}})));
    SubC subC annotation (Placement(transformation(extent={{94,26},{114,46}})));
  equation 
    connect(c, subC.sub) annotation (Line(points={{-100,-8},{60,-8},{60,36},{104,
            36}},
          color={0,0,0}), Text(
        string="%second",
        index=-1,
        extent={{-6,3},{-6,3}},
        horizontalAlignment=TextAlignment.Right));
    connect(subC.pin, subC.pin) annotation (Line(points={{104.05,36.05},{104.05,10},{90,10},{90,36.05},
            {104.05,36.05}},
          color={0,0,0}), Text(
        string="%first",
        index=-1,
        extent={{-3,6},{-3,6}},
        horizontalAlignment=TextAlignment.Right));
    annotation (Icon(coordinateSystem(preserveAspectRatio=false)), Diagram(
          coordinateSystem(preserveAspectRatio=false)));
  end MBad;
  annotation (uses(Modelica(version="4.0.0")));
end P;

[HansOlsson]

However, an alternative might be the following:

1. If a connect-statement references a potentially present component (either as the complete argument or as part of the argument) it will be marked as being present and will be treated as declared below.
2. After that at least one connector in the connect equation must reference a declared component.
3. If the other connector is not potentially present it must be in a declared component and is handled as follows:

So for:
connect(gainLosses.u, stackBus.cellBus.lossPower);
stackBus.cellBus is potentially present and is by 1 transformed into declared
Then gainLosses.u is declared (assuming lossPower isn't), so 2 is ok.
And then since lossPower isn't declared we continue with 3.

But we still have the benefit that if we just drag in an expandable connector and don't connect it we avoid all of the issues, both the problem above for physical connectors, and also the lack of source for a causal one.

[qlambert-pro]

I prefer this alternative but I don't understand the purpose of 3.. There are no potentially present connector then, since 1. made them into declared component. To me 3. sounds like it forbids adding an undeclared signal to an expandable connector. Unless what you mean is:

If the other connector is undeclared it must be a connector inside a declared expandable connector.

[HansOlsson]

I prefer this alternative but I don't understand the purpose of 3.. There are no potentially present connector then, since 1. made them into declared component. To me 3. sounds like it forbids adding an undeclared signal to an expandable connector. Unless what you mean is:

If the other connector is undeclared it must be a connector inside a declared expandable connector.

Yes, that was the idea - I wrote 'not potentially present' intending to mean 'not even potentially present'; but you are right that undeclared is clearer.