Fix incidence for 3-winding transformers

[josephmckinsey]

I noticed this bug, which causes an error in SMART-DS.

We should probably consider testing small smart ds feeder more often.

[tarekelgindy]

Is this supposed to fail silently? If there are more than three busses then the entire component seems to be ignored. Is that the desired response?

[josephmckinsey]

Is this supposed to fail silently? If there are more than three busses then the entire component seems to be ignored. Is that the desired response?

I couldn't come up with a better response for an ordinary incidence matrix. Would it be better to turn every list of n buses into a clique graph? I'm not so sure.

[tarekelgindy]

Oh I mean I think this will ignore a lot of transformers that are ligitimate - namely center tap transformers which may have three busses represented. This has bus1 distinct, but bus2 and bus3 being the same except for the phases that connected.
e.g.
New Transformer.tr(r:p1udt896-p1udt896lv) phases=1 windings=3 %loadloss=0.798816 %Noloadloss=0.472 normhkva=27.5 wdg=1 conn=wye bus=p1udt896.3 Kv=7.2 kva=25.0 EmergHKVA=37.5 %r=0.266272 wdg=2 conn=wye bus=p1udt896lv.1.0 Kv=0.12 kva=2 5.0 EmergHKVA=37.5 %r=0.532544 wdg=3 conn=wye bus=p1udt896lv.0.2 Kv=0.12 kva=25.0 EmergHKVA=37.5 %r=0.532544 XHL=2.4 XLT=2.4 XHT=1.6

Are the phases included in the bus names? If so, we can probably strip that off. Once that's done, we can create a set of the buses and if the size is greater than 2 we can throw an exception.

Does that sound like a sensible approach?

[josephmckinsey]

Sounds sensible as any other choice, although this does throw away some potentially important information. I'll ask Xinyang about it.

[josephmckinsey]

Ok, we discussed it before, but I have what I consider the best compromise. If there are 2 terminals, then it uses the ordinary path. When there are 2 or more bus names, it uses those. When there are 3 or more unique bus names, then it skips in the incidence matrix.