[jcranmer]

> Acela does the same trip in 3.5 hours.

I just rode it last week, and it actually took 3 hours, not 3.5.

Most of the NEC corridor from DC to NYC has a speed limit of 125mph; from NYC to New Haven, it's generally 70mph (!), from New Haven thence to about Kingston generally 90mph, and from there to Boston it's mostly 150mph. Given the density of the corridor, Amtrak should be trying for 150-220mph speed limits, but even 125mph is generally agreed upon to be the lowest end of HSR.

A surprisingly easy way to make the train go faster would be to redesign the switching sections before large stations to allow trains to go faster through them. You can probably cut around 10 minutes out of the entire length with an investment of less than $100 million just by doing that.

[Reason077]

> “Given the density of the corridor, Amtrak should be trying for 150-220mph speed limits, but even 125mph is generally agreed upon to be the lowest end of HSR.”

In the UK and Europe, 125mph (~200 km/h) is considered the top speed limit of conventional rail. Legally, operating speeds beyond that require full in-cab signalling, positive train control, upgraded safety and structural requirements, and whatever else is required for HSR. Further, all the trains operating on a section of line need to be upgraded to those standards if any of them are to run at speeds > 125mph.

The UK does have some sections of conventional line that are capable of > 125 mph running, and even have done so in the past, but this is no longer allowed.

I’m not sure if the US has similar rules, but it wouldn’t surprise me if so!

> ”A surprisingly easy way to make the train go faster would be to redesign the switching sections…”

Yes, generally speaking, fixing the low-speed bottlenecks will typically yield the biggest benefits for the cost in terms of overall journey time.

[djvdq]

> In the UK and Europe

UK is part of Europe. It's like writing "In California and the USA", or "In Mexico and North America".

[adhesive_wombat]

No, it's like writing "Mexico and America". If you're going to add geographic qualifiers like "North", the equivalent would be more like "UK and West Europe".

Without such qualifiers, the meaning is clearly "continental Europe", which is a meaningful distinction when it comes to the rail, because they have two almost entirely disjoint rail systems.

Ok, yes, technically the European rail system does actually physically connect to the UK main lines at a few places on HS1 between London and the tunnel at Dover, but no public train uses such connections. Functionally, European and UK rail are essentially entirely separate in terms of operation, regulation and technology. Whereas on the Continent, trains regularly cross between countries and the whole system is much more-but far from entirely -integrated, politically and physically (see https://en.wikipedia.org/wiki/European_Rail_Traffic_Manageme...).

[willyt]

Uk does actually follow EU regulations and standards because all the technology is designed and manufactured in the EU. HS2 is being built to an EU standard loading gauge with EU standard electronic signalling, likewise crossrail. When routes are upgraded to electronic signalling they are upgraded to an EU standard signalling technology, likewise with electrification. Loading gauge is typically smaller but when new bridges are built they will try to aim for the largest EU standard loading gauge that’s practical. The reason why you can’t easily run a night train from Edinburgh to Paris is not technological it’s because the customs and border requirements necessitate an expensive building like an airport to house them. It’s not true that the EU rail system is homogeneous either, e.g Belgium, Germany and France all have electrified lines running at different voltages which can require bi or tri mode trains for through running.

[adhesive_wombat]

> is not technological it’s because the customs and border requirements

Probably why I mentioned political integration in the sentence then.

But at this point I'm akchually-ing an akchually to an akchually!

[sofixa]

> Whereas on the Continent, trains regularly cross between countries and the whole system is much more-but far from entirely -integrated, politically and physically

That's not true. There are some local connections and integrations, but outside of high speed rail which is generally built using the same standard, each country has their own load gauge, electrification standard, signalling system, etc. Cross-border trains are usually special traninsets built to multiple standards to be compatible.

There is progress on more integration and standartisation, and pretty much all new lines are built to the same standards, but the vast majority of rail is existing.

[Reason077]

Quite right. I should have said “UK and continental Europe” to satisfy both the pedants and the politically sensitive ;)

[bluGill]

https://pedestrianobservations.com/2023/11/24/curves-in-fast... is a good analysis for anyone who doesn't want to believe jcranmer or wants more detail. (what I linked is a second best approach, but it links to other low hanging fruit like switches)

[paddy_m]

I read that post, and I have been working slowly on a set of jupyter notebooks, with ipyleaflet (mapping) integration to make building train maps (with max speeds) easier. Looking for colaborators

https://github.com/paddymul/train-calculator

[jcranmer]

And the first link in that post goes to https://pedestrianobservations.com/2019/02/08/fix-the-slowes..., which more directly speaks about the impact of slow speeds in station throats.

[kaliqt]

Could cut it probably but this isn't a private company we are talking about here. The regulations and grifting would be massive, especially in New York which is quite like California with their wastefulness: half to the garbage can, half to their pockets, maybe a cent or two for the actual rails.

So maybe it'd be a few billion at least. Not to say they shouldn't try. But I expect it'll be over budget and behind schedule, it would never happen like Brightline where they broke ground ASAP and just kept building until it was done.

[jcranmer]

I don't share your pessimism, and that's mostly because I've followed a lot more of the research into why US infrastructure costs are unreasonably high. To put it simply, excessive costs tends to come from a combination of overdesign (in particular the need to add lots of goodies to buy off stakeholders who can otherwise arrest the project), extra overhead in design costs, extra overhead in the way contracts are let, and incompetent management of contractors. But this is the sort of project that doesn't have the design stages to let that scope creep come in--it is pretty much "order off-the-shelf part number 42341 instead of 23421, then do routine maintenance tasks to replace old parts with new ones".