I’ve thought about this problem some too.
I think the goal with dead rail is to be able to just have two rails, and not worry about track cleaning or signal strength in the tracks. Conversely, I think the primary concern about battery power is running time and managing the charging of batteries between running sessions. For me, combining aspects of both systems would make for an ideal scenario. This is where a partially powered track and an onboard charging circuit would come in.
First to simplify wiring and allow for reversing loops with two rails, only sections of the layout would be electrified. What makes the most sense to me would be sections where locomotives are parked, spurs, sidings, and mainline blocks between turnouts only. Picking the longest or most speed restricted section of the layout would maximize charging time and reduce the number of charging on/off cycles.
With any dead rail scenario, I think it’s safe to assume that the communication is always wireless via Bluetooth or some other RF standard. Therefore this should be independent of the charging state or section of the layout.
The charging circuit would need to first detect a current coming from the wheels and then after some appropriate time begin to charge the battery, switch the internal power circuitry to an output of the rectifier. As soon as that circuit loses power, the system should immediately default to battery power, much like an on-line UPS system for a computer. There would also need to be some sort of hysteresis to prevent the whole system from oscillating over dirty sections of track. It should be opportunistic, but give up easily so that the loco doesn’t stall in the charging mode at a powered/unpowered transition.
As for handling shorts... if the chassis shorts the rails, then power will flow through the path of least resistance which will definitely NOT be the battery. Shorting is a problem that must be dealt with in any electrical circuit, and adequate circuit protection is a must. However lithium-ion batteries are definitely dangerous when they are shorted, so it seems to me any system that uses them should already have some protection against that. I would argue that in this hybrid battery-track-charging scenario that nickel metal hydride would potentially be a better choice. Although its not necessarily the best in high current draw situations, the half to full amp or so that my engines draw I think it would be fine depending on pack voltage. Alternatively, nickel cadmium could be used, which are superior in that regard, although are not well suited to intermittent charging due to memory effect.
