Possible tiny improvement

[ceraolo]

During a cross check that a PhD student of mine is doing between DigSilent and PowerGrids, we considered whether parameters are better expressed in components' masks in PU or in SI units. PowerGrids does as follows:

• generators in PU
• transformers in ohm and siemens
• lines in ohm.

Digisilent does as follows:

• generators in PU
• transformers PU
• lines in ohm/km (and probalby siemens/km, not sure)

I am rather agnostic for lines.

Regarding transformers, I think it is better if PowerGrids adapts to DigSilent's way, both because of homogeneity with what is done for synchronous machines, and because parameters expressed in PU of machine basis are much more significant for engineers than in ohms

[casella]

I guess we could have both, with a parameter puParameters to select which ones should be considered.

[ceraolo]

Are you sure? Not always adding choice enhance usability. For Synchrnous machines I don't think anybody would use ohms in masks. I'm afraid this is the same for transformers. We are thinking of power engineers for which a PU reactance of 10% means something, a reactance of 3.55 ohm nothing. Note that this could also be the root cause of #167.

[casella]

We can discuss this with @joyelfeghali and @marcochiaramello. It's a non-backwards compatible change, so we need to be careful.

[joyelfeghali]

We can discuss this with @joyelfeghali and @marcochiaramello. It's a non-backwards compatible change, so we need to be careful.

I agree with having the parameters value of the transformer in pu. If we want also to keep the values in SI units it could be defined as a final parameter, for example ?

[casella]

We can discuss this with @joyelfeghali and @marcochiaramello. It's a non-backwards compatible change, so we need to be careful.

I agree with having the parameters value of the transformer in pu. If we want also to keep the values in SI units it could be defined as a final parameter, for example ?

Absolutely. The transformer models are variants of the PI network base class, which is a physical model with currents and voltages in ampere and volts, not in p.u. So, we can make the current parameters final and add new ones in p.u. to parameterize the transformer model.

In fact, this could even allow to put reasonable default values for the p.u. impedance, if their variability in real life is limited. What do you think?

[ceraolo]

In fact, this could even allow to put reasonable default values for the p.u. impedance, if their variability in real life is limited. What do you think?

Transformer values in PU are very coherent and rich of information. Larger Xcc's mean lower short circuits currents, but larger voltage drops. The compromise is around 4-6% for LV/MV transformers, up to 15-20% for very large AAT units. The Xcc/Rcc ratio is typically very large, over 100; the larger this value the more efficient the transformer. So, if users input transformer data un PU, they know what they are doing, and this prevents mistakes, such as the (probable, I still wait confirmation from someone that knows about it) mentioned in #167.

I would not recommend a default for these values. Even though they are very meaningful and do not vary too much, there is still a huge difference between a transformer with 4% and 20% Xcc.