Here's a preprint from 2025: https://arxiv.org/abs/2507.00100v1
(has played sim city thousands of times since 1993) "I got this."
Without getting behind the paywall, the abstract says that how far you have to travel determines how far you have to travel??
This is not quite a spherical cow and there is something to be learned here, but really, I'd like to see this with
- non-square-grid cities with highways, public transport, …
- realistic synthetic populations doing non-random destination selection (nobody randomly goes to a supermarket at the other end of a city, after all)
- actual traffic simulation with jams (or at least using real traffic data)
and then evaluate both the energy use and travel times. That being said, doing this kind of thing, e.g. with ActivitySim, even for one real city, is quite a bit of work.