add noise + variable for shunt admittance

[frederikgeth]

Series resistance is proportional to length and cross section
Series reactance is a function of the voltage level
Shunt susceptance is proportional to length
Current rating is a function of series resistance

At the very least shunt susceptance should be obfuscated as well.

[afeutrill]

Hi Fred,

I was writing some of this one today. I've been looking at a data sources and some of them have a "shunt" key in the dict, and was using the "gs" and "bs" values in that part of the dict as the b and g. Then added noise to them, and storing the perturbed values back into the dict.

I need to add this as a variable as well. Haven't done anything on this yet.

[frederikgeth]

shunt objects at the top level in the PowerModels data model don't leak such information, they are abstractions of capacitor banks, and their value is independent of branch length. However branch shunts do leak length-related information. The branch shunts are the br_b values in the matpower format (col 5), and are mapped to b_fr and b_to (each of them assigned br_b/2). There is no way to supply branch shunt conductance data through the matpower format, but PowerModels' internal data format does support it, as g_fr and g_to. In normal transmission lines, g_fr and g_to are negligibly small relative to the other branch parameters, so assuming they're 0 is appropriate. Nevertheless, in super-high voltage transmission (e.g. >750kV) it may be significant. Furthermore, branches are often used in the loss model for transformers. In that case, shunt conductance is actually very significant. So, we can also consider anonymising transformer parameters in a similar fashion.

[frederikgeth]

Both series reactance and shunt susceptance are nonlinear functions of the length and the geometric mean distance (GMD) between the conductors. Branch shunt capacitance varies a lot between underground cables and overhead lines (because of the huge difference in GMD).

If you know the voltage level, you can infer length based on current/power rating and then infer GMD and category (overhead/underground).

Therefore, I think we better obfuscate series and shunt values independently, this gives less risk that we leak information.

[frederikgeth]

this is on master, so I can close it