Just wanted to throw what I think is an often overlooked aspect of the situation involving transients (voltage and current) on the model railroad. Thing is, it is mostly below 30v by regulation, so our skin keeps us out of most of this trouble--- most of our pain is the wallet. Of course there is the exception--- the hot wire. But we do have real experience above 30v, namely our 120v home--- we all know the hot wire is the one to watch. Well, the other side is grounded. and the earth is just too big to change. So, many of us see the small railroad in the same light: hot side as source, common side as return.
The thing here is that the small railroad is not the same. The reason for this is that in the small railroad's tracks, we have the electrical age hazard known as the extended conductive metallic surface. So, it has been mandated that the toy train transformer shall be an isolation transformer. That means that no wire of the output to the track, track signals, switches, or accessories is connected to the 120v neutral at the toy isolation transformer (the 120v neutral, "identified" as white or grey, or by having a "fin" on its insulation, is connected to the power system ground, which is most commonly the incoming service neutral. (There are also delta systems, but these can be awkward as service drops in residential areas, and I'd rather not get into these weeds.) 
The other problem associated with the 120v neutral is the transient voltages which it may carry from time to time. In my experience (limited) pulses just shy of 400v are most common, and associated with nearby electronic drives for electric motors running off either 460v or 277v. You can rent a widget which will go through a whole roll of narrow paper tape printing out the voltage of each peak. But unless you live above a car-wash, this is probably over cautious. The postwar ZW had 1-mil shellac varnish insulation, and Lionel was proud of pointing out that this would withstand 1000v on the input (not to mention 105-C). But not to worry--- the UL has raised this to 2000v for the power-limited Class 2 under 30v transformers, and already extended the reduction on these from 32v to 30v and to 660 VA to the toy transformer (UL697). The really critical wire in many locomotives is that which may run as a connection between single roller pickups under motorized trucks at each end of the engine. Lionel learned this lesson in 1954 with its Trainmaster H-66, quietly asking its service stations to replace the usual thermoplastic interconnection wire with a heavier wire having rubber insulation, from the triplex switch control harness. The rubber insulation would char, rather than melt into one's skin.
The bottom line is that with the isolated wiring carrying alternating current, when the short recurs (per the string of sparks) each time, it is driven primarily by the collapse of the magnetic field in the 180w brick. In full size practice, it is assumed that the motor load will contribute another 25%. In this case the combined weight of 4 Pullmor motors could be compared to that of the 180w brick to estimate the added effect. To protect say a TIU, from the overvoltage pulses, all its input and output wires wires should have TVS protection, with say four or five times as much protection facing the the brick (transformer). The phase that the voltage is in (positive or negative) will determine the direction from which the voltage peak will come. That direction will determine from what collapsing magnetic field it is coming; whether it begins at the beginning crossing of the current zero (the so-called zero-sequence fault) will determine the relative effect at the source; the ohms impedance along the direction of wave travel will determine the reduction in voltage level along the path. Worth noting that the ZW could produce a transient voltage of 39v, due to significant leakage reactance. That reactance is significantly reduced in the 180w brick, I would say by a factor of 2. The speed of electrical impulse over typical Lionel/MTH layout track and wiring is about 0.6 to 0.7 time the speed of light. At 60 cycles, the source effect can be assumed to be simultaneous over the layout, provided the source effect is not steep-fronted. Well, I think that covers it--- my head is hurting, a sure sign that it is mostly covered. Well, a last reminder that to a certain extent, many of these TVS are sacrificial, and will eventually fail at the worst time.
As for the current wave, that damages by its heating effect, and duration. But that involves breakers, fuses, and other devices, and those are under discussion in another current thread here.
I think the Gunrunner has already mentioned the singular importance of putting a TVS on both wires of a pair whenever you decide one wire needs one.
---Frank
