It wouldn’t be particularly difficult to design a vehicle that never had to stop, ever. But that would depend upon being able (allowed) to change the various components whilst it’s on the move. For example have at least two engines in it, with only one needed to run to keep the vehicle moving, so you could have a service vehicle drive beside it, to swap over the old engine for a new one. That wouldn’t be needed very often with a modern diesel as you’d expect at least a million kilometres out of one.
Much the same for the gearbox/drivetrain.
It’d also be easy to design the suspension so that when a tyre got a little thin (measure rpm’s Vs road speed to determine the diameter, Tesla’s can do that right now) it could be lifted up clear of the road to allow for the tyre & wheel to be changed.
Refuelling would, of course, be a simple exercise.
But assuming you aren’t allowed to do any of that, the idea of having towing several trailers is the way to go. I’d also equip the engine with a cylinder deactivation mechanism, (as used in many cars right now) so that as the overall load decreased from the fuel burn & subsequent release of the fuel trailers, less power would be needed to keep it all moving. Perhaps even design the engine so that it’s in multiple sections, so that as the load decreased you could shut-down & decouple that part of the engine to eliminate drag.
Determining the optimum speed is a balance between engine efficiency & aero/tyre/drivetrain drag. It’s not difficult to design the engine to run at a specific rpm, then use a gearbox with lots of gears to vary the speed as needed as the tyre drag & overall mass decreased. Only maybe 5% ratio change between each gear would be enough I reckon.
Some kind of energy recovery from the exhaust would help, as would perhaps recovering a bit of heat from the radiator. Exhaust energy recovery on a diesel is normally just a regular turbocharger fitted to the exhaust manifold. Again, since the engine would be running at a very constant rpm you could tailor the turbo to be right in the middle of its best efficiency island.
Tyres are the next engineering challenge. You’d want them to be as thin as possible to reduce drag but not so much that even when pumped-up to, say, 200 bar they wouldn’t deflect too much from the mass they are carrying. There’s no real reason why you couldn’t just use metal wheels that had a very hard rubber ring around them.
You’d have to go super pedantic on the aero drag as well. I mean really anal. Things like having no windscreen, just a small bump or hole that a camera lens sticks out of so the driver can see where they’re going on a screen. The entire vehicle & trailers would have to be one smooth round sectional profile to minimise surface area. And yeah a pointy nose and even longer pointy tail as you typically get more aero drag from a blunt tail than a blunt nose, due to the larger amount of turbulence coming off the rear. A bit of an engineering challenge to do so because of the trailers dropping away at various intervals, but far from impossible. The tyres would be a bit tricky but a pretty conventional fairing would work okay. A bit more difficult to incorporate steering as the entire fairing would have to move with the wheel’s change in angle – perhaps have the fairing split into two so it pops open whenever more than about 0.5° wheel defection is needed to keep it clear, then shuts again after the wheel comes back straight.
Fuel is another consideration. There’s no doubt that brewing up a special blend would also add several percent to the range. I know that very special types of fuel were tested in some US jet aircraft to improve range & performance but they were pretty nasty to be anywhere near as they had radioactive boron particles in it. Zip fuel it was called.
Anyway if you did all that you could easily double the efficiency of a typical semi vehicle, probably even far better than that.