If the moving friction is very low, it makes sense that not that much energy is used to move the train once it's at speed. Slowing it down would return some of the energy back (with the rest dissipated as heat).
But a commuter train, by definition, starts and stops a lot so it makes sense that a large fraction of the energy used would be for acceleration.
But to address your main point, the rolling resistance of a train in motion is tiny; basically all the energy cost is at acceleration time— it probably cruises at like 5-10% power.
a good comparison would be the invention of the travois https://en.wikipedia.org/wiki/Travois which benefitted from the addition of wheels.
Even a model/toy train wagon can easily coast the length of a large room when given a push.
See how little metal is in contact, about 1cm²: https://www.reddit.com/r/Damnthatsinteresting/comments/17ysd...
There's very little friction.
does this make sense? let's round that 23 up to 25%, and say brake energy regeneration is 100% efficient (it's not): one quarter of the energy put into the system comes back out? city driving in a car, you might have your foot on the brake 25% of the time (yes, that's not how it works, just think conceptually) but a train spends a lot of time running, and comparatively a small amount of time braking, and if you were to detach the engine from the train, one doesn't get a sense the train would roll very far unpowered, so while there's momentum to scavenge, it doesn't seem like 25% of the total
I'm just thinking, how do these numbers make sense?