The article says this version sends it back out on the same power lines, either used by other trains or pushed back into the grid. Seems like a perfect job for supercapacitors, although I have no idea about the feasibility of that solution. I imagine that batteries having huge input/output cycles like that wouldn't be healthy for them. Again, pulled from my imagination because I don't really know much about battery wear/use.
If it were 1995, the optimal solution would have been NiMH batteries at the side of the track. They can do massive currents in and out, and a few tons of batteries would be enough to fully store the energy of a passenger train stopping from 60 mph to nil inside 30 secs.
NiMH could have been attached directly to the rails.
Today, lithium batteries win for Watts per dollar, and perhaps custom made packs could also be attached directly to the rails.
But a cheaper solution is probably bidirectional inverters, allowing the DC generated by the trains to be fed back into the 3 phase national grid.
Unfortunately, all trains in London today cannot regen into the grid - they can only regen into the rails and hope that some other train on the same rail is accelerating at the same time to use the energy. By my estimates, that only happens less than half the time.
Generally, trains are scheduled so that one train decelerating roughly coincide with another accelerating. You can plan ahead and orchestrate all you want and overcommit capacity as much as you want. Leftover that didn't cancel out is fed back to hydroelectric dams for gravity storage which do require cleaning but are immune to chemical degradation.
By the way, implementing regen on synchronous motors is relatively easy, IIUC, command a positive torque to the inverter and it draws current and line voltage gets pulled down. Command negative and opposite happens.
