The old-style couplers used in Europe are too weak for longer trains. "The buffers and chain coupling system has a maximum load much less than that of the Janney coupler. They allow around 3,000-4,000 (metric) tonnes total train weight depending on the how they are constructed. The Janney coupler sometimes is built for 32,000 tonnes (31,000 long tons; 35,000 short tons)."
The EU attempt to standardize on an automatic freight coupler across the entire EU was a flop. There's a new project under way to test a new generation of freight coupler, one that connects the mechanical, air, and electrical connections.[1] But it's just getting started.
[1] https://uic.org/com/enews/article/germany-rail-freight-trans...
So for container trains at least, a programme of loops and sidings lengthening would be the more pressing requirement, before actually and absolutely having to upgrade the couplers, but that kind of infrastructure modification doesn't come cheap or necessarily easy, either.
Another difference to consider is that in Europe, passenger traffic is much more widespread and common than on North American railways, and also commonly has priority over freight trains.
While in principle, longer trains are indeed more efficient than an equivalent larger number of shorter trains, past a certain point they'll eventually become too unwieldy to fit inside the schedule demands of passenger traffic. There's certainly scope for moving beyond the current 750 m maximum, but the multi-kilometre range would probably be pushing things a little too far without completely dedicated infrastructure.
a) even ignoring electrification and only at a guess, there are probably on average more overbridges and tunnels per route mile than on the major North American freight corridors, therefore making any attempt to implement the loading gauge necessary for double-stacking much more expensive
b) this I know for certain – electrification is much more widespread than in the US (especially if you're only looking at trunk lines), and that poses a continuous obstacle to implementing double-stacking without expensive modifications.
Probably the big differences are that a) the US already had a comparatively roomy loading gauge compared to Europe before conversion was necessary and b) the West Coast-to-Midwest routes were probably more likely to be double-stack-compatible before the projects started and provided a seed route to develop the concept before the expensive loading gauge conversions in the east had to take place.
Electrification doesn't affect double-stacked routes other than needing (I think) an extra ~2 ft of height compared to non-electrified track. The cost of relaying catenary cables isn't going to be all that high: electrifying track from scratch in the first place appears to cost about $1 million per km, and a lot of that cost will include the substations which don't need to be replaced; actual costs of replacing the gantries is likely to be less than half that. That's of course assuming that there is actually room above the track to support the new height.
Regarding electrification, yes, it's not rocket science, and the cost of raising bridges and tunnel clearances will usually dominate, but because electrification on the main lines covers basically every metre, I'd think that the costs would still add up significantly.
[1] https://www.bathnes.gov.uk/sites/default/files/sitedocuments...
Electrification is an aspect also, though I don't know if EU mixes passenger and freight as much as the US does (and also, more of the EU is relatively easy accessed by ocean going ships).
That's a list of projects that are undergoing expansion from single-stack to double-stack, although that's not the main purpose of CREATE, and the Cross-Harbor Rail Tunnel is a perennially-proposed project to link NJ with Long Island that's unlikely to come to fruition. Furthermore, my understanding is that most of those projects are already completed; Wikipedia isn't always timely on updating these kinds of topics.
It isn’t the only issue; apparently double stack has issues with third rail electrification, which is the norm on New York City area commuter rail, so even if you built the tunnel, a double stack train may not be able to go much further.
1) Loading/unloading. Double stack requires cranes of some sort and is inherently dangerous when lifting large heavy mass into the air. Single stack could be more loaded/unloaded cheaper, faster, safer with less equipment, and keep trains moving and making money rather than sitting around being loaded/unloaded.
2) Maintenance cost and wear/tear on rail infrastructure. Wear and tear on rails (and roads) is proportional to the axle load to the power of 4 based on AASHTO testing [0]. Double stacking containers will roughly double axle loads, so damage and maintenance cost increases by ~16x. Keeping axle loads down keeps costs down. As an aside for roads, that means a fully loaded semi trailer (80000 lbs) is about 9600 passenger cars (4000 lbs) worth of road wear [1].
Re Longer trains: they are generally better, but there are practical limits. Trains have to be assembled and disassembled in rail yards. If the train is extra long it makes things more complicated and time consuming. If there are grade crossings on the route (there usually are), extra long trains result in long gate down time (5-10 mins) which can constitute a public safety risk (emergency vehicles not being able to cross tracks), and generally frustrates all other road users including pedestrians and cyclists.
Older railway axle counters used for train control and signalling systems in Europe and elsewhere had 8 bit integer logic controllers. That's 256 axles per train, or 64 rail vehicles assuming 4 axles per vehicle. Any longer and it will overflow, and the train might not be detected. So that is a hard limit without upgrading those systems or risking a catastrophic failure on a safety-critical system, that prevents 2 trains from occupying the same section of track at the same time [2].
[0] https://higherlogicdownload.s3.amazonaws.com/IPWEA/c7e19de0-...
For someone who lives in the US, this seems not-state-of-the-art. Most long-distance train traffic in the US has shifted to double-stack container trains.
> Not sure what the train length limit is. That's set by the length of sidings.
From what I can tell, Europe tends to prefer far smaller trains than the US does. The typical train length in the US is >100 cars (I want to say 120-ish is the current average, but I don't have firm statistics), but in Europe, it seems to be more like 40 cars.