How to interpret the voltage of the MS side of trafo in LS grid calculation.

[willemijnbrus]

Hi pgm,
We use the pgm on LS grids from connections untill the MS/LS transformer.
Because we dont have information about what happens above the 10KV side of the grid, i think in our loadflow we/pgm automaticall assumes the 10kv as a stable input. However, colleages pointed us to that in real life of course this is a dynamic unit and not a stable. They pointed to us if we cannot take the 10KV as a dynamic component in our loadflow, we should at least then monitor our voltage boundaries at connections at 218,5 and 241,5 V rather than 207 en 253 Volt.
-is my understanding from pgm correct that we have to assume the 10kv side of the trafo as a give as we don't have in scope data on higher parts of the grid
-is this feedback familiar to you in other projects that focus on LS grids and do you know if in other projects then the solution is to use the 218,5 and 241,5 V rather than 207 en 253 Volt?

[nitbharambe]

Hi @willemijnbrus ,
This is a valid concern when conducting powerflow simulation. A power flow calculation is based on a hypothesis of a source voltage and the actual voltage of the grid at the source can vary practically. The resultant voltage results heavily depend on the source voltage provided at input.

Your colleagues' approach of being conservative by restricting limits from 0.9-1.1 p.u. to 0.95 to 1.05 p.u. is a logical one.
But this transitively approaches the problem: "If LV voltage is 0.95 p.u., then it is most likely that no other voltage is below 0.9 p.u."
This can be more restrictive and is a less explainable assumption.

A better way can be to directly tackle this issue: Run a powerflow using u_ref=0.9 and one using u_ref=1.1 for the MV 10kV level source. Both these situations should then lie between 207 and 253 V (0.9-1.1 p.u.).
There is no need to simulate for all possible ranges of source voltage. For a typical LV grid, simulating for a min and max source voltage situation should be enough for limit violation check.
You can also set these u_refs based on policy. I think dutch grids have a tighter restriction on voltage limits at MV level instead of 0.9 or 1.1. Those can also be used instead.

(Tip: PGM now supports multi dimensional batch calculations (https://power-grid-model.readthedocs.io/en/stable/examples/Multi-dimensional%20batch%20calculation%20example.html). Making such batches is hence easier and efficient.)

• For a scope of LV grid study, you indeed have to assume a source voltage to begin with. You can switch to using actual voltage measurements of MV grid if you wish to be more accurate (but I assume that is not under the scope of your study or that data is not available.).
• It depends on the projects actually. This concern only pops up when trying to monitor voltage results. The current / power / loading based results are negligibly affected by whatever source voltage is chosen. I don't remember receiving this exact question but i imagine most LV projects have to deal with this.

[marialame121-blip]

Hi,

Your understanding is mostly on the right track. In LS grid calculations, when upstream (MS/HV side) data isn’t available, tools like PGM generally treat the MS side voltage as a fixed/slack input. So yes, assuming the 10 kV side as a given value is a practical approach when that part of the grid is outside your model scope.

That said, in reality the MS side is not truly stable, it’s influenced by upstream network conditions and especially OLTC operation, which continuously adjusts voltage based on load. Since this dynamic behavior isn’t included in your model, applying voltage boundaries instead of a single fixed value is the right way to approximate real conditions.

Using a tighter range like 218.5 V to 241.5 V (±5%) is often more representative of actual operating conditions, while 207–253 V (±10%) reflects broader regulatory limits. In practice, many engineers use the tighter band for more realistic load flow analysis.

From a transformer/OLTC perspective (as also reflected in how companies like EMR Global approach OLTC applications), voltage regulation is always dynamic and dependent on tap changer behavior, so modeling it as a range rather than a constant improves accuracy.