Let us continue the series of green posts on transients, and today let us talk about traction substations. There are at least two known options:
✅ DC 3 kV (the rectifier is located at the substation, and the locomotive has only a DC motor, i.e. it is light enough).
✅ AC 25 kV (the rectifier is located in the locomotive, that is, train carries not only a DC motor, but also a rectifier; such a train turns out to be heavy).

AC 25 kV system is used in the conditions when:
👉 the distance between neighbouring traction substations cannot be made less than 10-20 km;
👉 large freight trains with dozens of wagons are being transported.

Each AC 25 kV traction substation receives power from a high voltage network (let it be overhead lines of 110 and 220 kV). Windings of a traction transformer are connected as star/triangle. It’s interesting that, on the side of the triangle, one phase (say, phase “c”) is connected to the rail and thereby grounded, and on the other two (phases “a,b”), the voltage is increased from phase (25/1.73) to linear (25 kV) and then:
➡️ 25 kV of phase “a” – feeds one direction of travel (say, West).
➡️ 25 kV of phase “b” – feeds another direction of travel (say, East).

Therefore, this traction circuit is called as 2×25 kV. It can be seen that, even in a normal operation, there is a very complex mode:
👉 25 kV side has permanent ground fault (phase “c”);
👉 high-voltage side (110-220 kV) often has an ungrounded neutral.

Suppose we have a short circuit on OHL 110 kV, and switches S1 and S2 disconnected this line from the network. Since even the normal operation of transformers is complex, after the disconnection of the supply line by switches S1 and S2, it is not always possible to detect promptly the problem on OHL 110 kV and disconnect Traction-1 from it. As a result, AC 25 kV voltage from the contact wires (which never loses power, due to Traction-2) goes to the OHL 110 kV.

The process may be called “reverse transformation”. As a result, on the 110 kV side of the Traction-1, there may be:
❌ significant switching overvoltages (at the moment of disconnection of the last of switches S1 and S2), with a factor of up to 3-4 or more.
❌ significant temporary overvoltages, with a factor of up to 2-3 or more.

All this comes as a surprise to the 110 kV surge arresters (SA) installed there to protect the traction transformer from lightning surges.

This and dozens of other examples suggest that transients in networks are so complex that the selection of arrester cannot be reduced to “dividing the operating voltage of the network by the root of 3”. Not to mention the issue of the selection of the energy absorption capability, about which there is not even a formula. The problem of surge arrester selection should be solved differently.

Please, take a look on the brochure “Selection and Application of High Voltage Surge Arresters” which gives one of the possible solutions to the problem of surge arrester selection.