Fix SI units by introducing reference phase impedance ZsRef

[HansOlsson]

Fix the unit for the inductances.
Eliminates non-local unit errors based on modelica/ModelicaSpecification#3266 (combined with other unit-changes.)

[HansOlsson]

Note that there are similar-looking examples of unit-errors in other classes that isn't corrected in this PR, like Modelica.Magnetic.FundamentalWave.BasicMachines.SynchronousMachines

It would likely be possible to correct them as well in a similar way (but then there are messier ones as well - as the SquirrelCage).

[christiankral]

This PR shall actually affect the machine models of three different packages as they all have the same shortcomings:

• Modelica.Electrical.Machines.BasicMachines
• Modelica.Magnetic.FundamentalWave.BasicMachines
• Modelica.Magnetic.QuasiStatic.FundamentalWave.BasicMachines

Addirtionally I would like to discuss some aspects of this PR with @AHaumer first: I see three different "simple" fixes of this issue:

Option 1

Follow the proposal of this PR but change final constant SI.Resistance unitR=1 to final constant SI.Impedance unitZ=1, as the inductances are calculated from impedances, not resistances.

Option 2

Add parameters of the reference phase voltage and current to the machine models, according to the documentation and the original intention of the used paramerterization:

   parameter Modelica.Units.SI.Voltage VsRef = 100 " RMS reference phase voltage";
   parameter Modelica.Units.SI.Current IsRef = 100 " RMS reference phase current";
   final parameter Modelica.Units.SI.Impedance ZsRef = VsRef / IsRef "Reference phase impedance";

Use parameter ZsRef instead of unitZ then.

However, option 2. has the drawback, that additional parameters are used in the machine models. Even though this parameters do not change the model behavior it will most likely happen that a user model then shows the new parameters VsRef = 100 and IsRef = 100 which might not match the actual user parameters. This is a bad side effect which could confuse the users.

The parameters VsRef and IsRef could be made protected, but then it is questionable if there still exists and advantage over option 1.

Option 3

Replace the division by 2*pi*fsNominal by the term 100*pi and thus fully ignore the units in the default paramters.

[dietmarw]

I agree that Option 2 is suboptimal if the new reference parameter has no function other than creating a new ZsRef to fix the unit. Although having them might finally explain to the user better what the Lx calculations in the data records of the examples are based on. So in I think:

• Option 1: Pragmatic solution but seems like a "hack" to please the tool.
• Option 2: Gives a bit more background info to the meaning and origin of the parameter calculations, more user friendly but at the same time also dangerous to be confusing (since Vref and Iref don't "do" anything).
• Option 3: Makes it very hard to understand the equations and their physical meaning.

[HansOlsson]

An option 4 would be to use:
final parameter Modelica.Units.SI.Impedance ZsRef = 1 "Reference phase impedance based on nominal voltage 100 V and nominal current 100 A; per phase";

(Possibly with better description.)

It avoids the extra variables of option 2.
Compared to option 1 it isn't just a magic unit conversion, but properly using a reference impedance (which happens to have the value 1).

[christiankral]

I agree with @HansOlsson. I think option 4 is the way to go.

[christiankral]

In order to provide a consistent solution

• the protected parameter ZsRef is moved into the partial model Modelica.Electrical.Machines.Interfaces.PartialBasicInductionMachine
• the resistances Rs and Rr are also multiplied by ZsRef, as this was the original intention of this default parameter set

[christiankral]

One more thing to think of: In the slip ring induction machine model the rotor locked rotor voltage is based on the stator reference voltage of 100 V:

  parameter SI.Voltage VsNominal(start=100)
    "Nominal stator voltage per phase"
    annotation (Dialog(enable=not useTurnsRatio));
  parameter SI.Voltage VrLockedRotor(start=100*(2*pi*
        fsNominal*Lm)/sqrt(Rs^2 + (2*pi*fsNominal*(Lm + Lssigma))^2))
    "Locked-rotor voltage per phase"
    annotation (Dialog(enable=not useTurnsRatio));

This units of the start value of VrLockedRotor is actually also not "correct" with respect to units and its original inention. Therefore, a much better and consistent implementation were:

  parameter SI.Voltage VrLockedRotor(start=VsNominal*(2*pi*
        fsNominal*Lm)/sqrt(Rs^2 + (2*pi*fsNominal*(Lm + Lssigma))^2))
  ...   

It is, however, not backwards compatible (if VsNominal is changed by the user, and VrLockedRotor isn't changed).

@AHaumer @HansOlsson @dietmarw Any comments here?

[TManikantan]

This pull request requires second reviewer to approve the fix to merge it successfully.
@AHaumer would you please review fix for this ticket? Please also let us know if additional info or summary would help you expedite the review.

[TManikantan]

@AHaumer Would you please review the fix?

[AHaumer]

fine with me.

[TManikantan]

@christiankral would you look into the fix and give a yes so we can go ahead and merge the PR?

[christiankral]

Looks good.