Monday, April 29, 2013

Long distance trains don't work?

Everyone seems to have figured out exactly why long distance trains don't work, and can't work, in the US. I wrote before about the efficiency arguments that Jarrett Walker and Bruce Nourish make against them, and a whole raft of people weighed in in the comments with well-constructed cases against routes longer than 750 miles.


The problem with these arguments is that they're too good. They ignore the fact that in many places outside the US, long distance trains are still popular, even with competition from subsidized planes and high-speed rail. The Trans-Siberian Railway is the most famous, but it is just one line in the large network built by the Russian Empire and the Soviet Union. Wikipedia lists the networks of India, China and Malaysia, as well as Germany, Italy, Morocco and the United Kingdom. There are even still a few long-distance trains in Argentina and South Africa.

This is not a question of population density: there are parts of Morocco, China, India and the former Soviet republics that are essentially uninhabited. So what do they have that we don't, and how can we get it? I'll leave this to all the commenters, since they know what's going on so much better than I do...

47 comments:

LetsGoLA said...

I have some longer thoughts on this that I'm trying to pull together but I'll offer some quick things for consideration on these countries...

Morocco, UK, Italy, Germany: you'd have to go end-to-end in these countries to approach 750 miles. They don't have anything analogous to, say, Chicago to LA. To get that in Europe you need something like Madrid to Warsaw. I'm sure it exists; does it make money? Even that analogy is not great, since there is a hell of a lot more between Madrid and Warsaw than between Chicago & LA.

Former Soviet Union, China, & India: most people can't afford to fly, so the question is, as those nations grow wealthier, will the popularity hold up? For example in China, many migrant workers use long distance rail to visit family during holidays. If they could afford to fly, they might do that to get more time with family. Of course, that brings up the question of to what extent will the Chinese govt will subsidize air travel.

Trans-Siberia is an interesting comparison, though nothing in the east of Russia is near the size of LA or SF. Lots of freight on the line, which is good for comparison to US. Six days Moscow to Vladivostok, comparable to US long haul speeds. Electrified - interesting for a route that long and different from US. Anyone know how much ridership it gets?

Ok, sprawling comment without any real answers, but maybe food for thought for others...

Helen Bushnell said...

Long distance trains do work. Long distance trains in the US regularly sell out. But it is a mistake to think that most people on them are traveling long distances between major cities. The last time I took the California Zephyr, there were large numbers of people taking the train from Grand Junction to Fort Morgan. People in Iowa and Montana regularly take trains to travel within those states.

As of three years ago, the cheapest airfare out of Grand Junction was $350, and each ticket sold had an average subsidy of $300. We could upgrade the mountain rails for much less.

We have to remember tax money pays for the trains in India. Tax money pays for the trains in Malaysia. Tax money pays for the trains in South Africa. The routes available in Argentina vary as the government money put toward that purpose varies. Tax money pays for the trains in Russia. Tax money from Kazakhstan subsidizes rail outside of that country, both passenger and freight, because that helps grow Kazakhstan's economy.

Basic infrastructure costs money. This desire by privileged white boys to get something for nothing is destroying our country.

Adirondacker12800 said...

Everyone seems to have figured out exactly why long distance trains don't work, and can't work

Siberia is like Alaska. Population clusters and lots of of a whole lot of nothing that makes North Dakota look crowded.

If you priced fares at nuisance fee levels.... there aren't enough people to have a train a day crossing South Dakata and a train a day crossing North Dakota. If you made the train free and ran two trains a day there would be empty seats.

Wikipedia says there are 40 million people in Siberia. There are ten million in the cities served by the Trans Siberian not including Moscow. Apparently The Trans Siberian runs once a day.

The biggest city along the line is Novosibirsk. Population of 1,473,754. The population of Idaho is 1,567,582.
The population density of Novosibirsk is 7,600 per square mile. I suspect that most places in Novosibirsk have bus/tram service to the train station. It's 381 miles from Boise to Couer D'Alene. Or 387 miles from Novosibirsk to Omsk, population 1,154,116. Or 474 miles from Novosibirsk to Krasnoyarsk, population 973,826. If the Trans Siberian can host one train a day for the cities with a station the five states of ND, SD, WY, MT and ID should get a train every other day. One every four days arriving or departing Fargo and one every four days arriving or departing Sioux Falls. Lets be generous and say they should be getting a train twice a week to get the same level of service the population along the Trans Siberian gets..... it's population...

Alon said...

If only tax money went to subsidizing intercity trains in Rhode Island.

LondonKdS said...

While some German trains run very long distances, my experience is that almost all the passengers are using them for short to medium journeys. I assume that they're still advertised and operated as continuous journeys for historical reasons or for the company's convenience.

Paul Druce said...

There are two major problems with the current long distance train network:
1. Sleepers/diners.
2. Fares

Sleepers and diners add immensely to the cost of a train. It's no surprise that the all-coach Palmetto is the cheapest train in the entire long distance train network, by a significant fraction. Comparing it to the Silver Meteor, a coach seat-mile costs $0.164 while the cost for a sleeper appears to be $0.394 per mile (2.4 times the expense of the coach). This does attribute all diner costs to the sleeper.

Meanwhile, fares are ridiculously low on the long distance trains. The Palmetto, despite a lack of sleepers, is tied for highest revenue per passenger mile with the Crescent (ignoring the Auto Train which has significant extra revenue from hauling the automobiles). Average fares on the long distance trains can range down as low as 11¢ per passenger mile for coach and sleepers generally bounce around the 20s.

The long distance trains have the potential to work, but in order to work, they need to raise fares. Look at the international examples; all of them have higher fares per mile than does Amtrak (indeed, the Australian long distance trains can run over a dollar per passenger mile).

Steve Stofka said...

What Paul said.

I suspect that the best way to go about long-distance trains would be to price fares at market clearing levels and then see whether or not further subsidization is needed. If it is, tie Essential Service subsidization to the trains instead of this Essential Air Service nonsense. That could then be used to either (a) reduce the total amount of Essential Service subsidy needed, or (b) greatly extend the coverage of Essential Service services.

GeekGuyAndy said...

There are so many factors as to why trains just don't work for most people. Other commenters are right on with the subsidy - other countries can discount the per mile rates significantly where the US focuses more on flights. The time is a struggle for me too, since if I only have 3-4 days to go see family, spending nearly 2 full days in transit versus 2 half day of driving or flying is a better use of my time. What I find the most complicated is the connections to other modes though. Here's a specific example:

From upstate NY to northeast Ohio, I would need to drive/bus to Syracuse which is about an hour away on the highways. Then the train leaves in the afternoon and gets to Cleveland at 3am. Then I'd need to find a ride to my destination which is about another hour away by highways. For a longer time trip than driving, that costs significantly more (especially as 2 people in a car doesn't cost more, versus two train tickets doubling the cost), and requires either a very pricey 3am taxi ride or convincing family to forego sleep to pick us up, the train just isn't going to work for that type of trip.

Even if there was a train in both the start and end point cities, if it takes longer than a day, the sleeper car is far more expensive than flying and getting a hotel.

Helen Bushnell said...

No one has actually proved that trains don't work.

Steve, trains are already selling out in peak periods which is when people think of trying them.

Adirondacker12800 said...

"Then the train leaves in the afternoon and gets to Cleveland at 3am."

If the train arrived at 9PM it would be much more useful. The LateForSure Limited takes 20 hours to get from New York to Chicago. Cut that down to ten and there would be a lot more riders. Enough so that 4 trains a day could run instead of 1. Cut it down to 8 and there would be a train every hour during the day. Maybe two, one from New York and one from Boston. Chicago to New York in 10 hours implies Syracuse to Cleveland in three and half and Chicago to New York in 8 hours implies Syracuse to Cleveland in two and half. Cut it to 8 between Chicago and Boston it implies Syracuse to Boston in two and half. Or Syracuse to New York in two and half.
...10 hours Chicago to New York is four and half from Buffalo to New York or Boston and five and half to Chicago. 8 hours Chicago to New York is three and half to New York or Boston and four and half to Chicago.
8 hours Chicago to New York and 2 hours for New York to DC makes it possible to get from Syracuse to DC in under 6. That's faster than changing planes in to get to DC.

busplanner said...

Adirondacker hit the nail on the head. Most trains are too slow (which also means that crew and equipment are not as well utilized as they might be) and frequency on most routes is once a day (which means that trains are not arriving or leaving at reasonable times. Cities like Fargo and Cleveland have their service in the middle of the night, not at all attractive.)

So, the real issue is not whether long distance trains fill up on certain days or whether some people like to ride them; but what problem are we trying to solve by providing long distance train service.

The Auto Train clearly provides a service for people who don't want to drive between Florida and the Northeast, but wish to have their car with them (suggesting an extended stay at the non-resident end where renting a car is not economical).

But otherwise, why rail? Here are some possibilities: a. To relieve airport congestion for distances up to about 750 miles where decent speed rail is competitive with flying. b. To relieve congestion on roads which otherwise would have to be expanded to meet growing demand.

Thus, investment in the Northeast Corridor with a huge population and overlapping markets is a good investment in rail travel. San Francisco to Los Angeles is another intermediate corridor with potential.

But otherwise, government investment in passenger rail does not make a lot of sense.

Run deluxe trains for sightseeing tourists (though prior efforts in the U.S. have generally failed.) Run buses during daylight hours for distances up to about 500 miles and subsidize air service for longer trips if essential.

neroden@gmail said...

What do they have that we don't have?

The train systems are nationalized, and they are operated to give priority to passenger transport.

That's the big difference, full stop, end of story.

neroden@gmail said...

busplanner:

We are providing train service because
(1) many people CANNOT drive (for medical reasons)
(2) many more people DO NOT WANT TO drive
(3) many people CANNOT fly (for medical reasons)
(4) many more people DO NOT WANT TO fly (for reasons such as the TSA).

All such people are served by train service.

And that's not all.

Investment in the New York-Chicago corridors (by way of upstate NY, and by way of Pittsburgh, and by way of Southwest Ontario and Michigan and by way of Ohio and Indiana), with huge population and overlapping markets, makes a good investment in train travel.

You seem to be one of those who haven't noticed that there are lots, lots, lots, lots, lots of parts of the US which could support a dense network of trains.

Yes, Wyoming isn't one of them. And neither is Alaska. We're not talking about Wyoming or Alaska.

Services across the Rockies, doomed to be slow by the mountain crossing, will probably always be "lifeline" services for those who cannot fly or drive -- and those services may as well remain at their current minimal levels. But that's also not typical -- most of the country has *far* more potential for rail service than that.

There is plenty of demand for a route as long as Denver to Chicago to make a good passenger train corridor (not on the current route, however -- on the route which stops at the major population centers in Iowa).

The Trans-Siberian stops at the major population centers on the way. And the entire route runs several times per day. At good speeds. That's how you do it. There's no magic distance limit.

And if Denver-Chicago is viable and good value for money (which it is), how much more valuable is NY-Chicago?

neroden@gmail said...

Paul Druce wrote:

" There are two major problems with the current long distance train network:
1. Sleepers/diners.
2. Fares

Sleepers and diners add immensely to the cost of a train."

Wrong as usual. This is a common, and stupid, misconception. DINERS add immensely to the cost of the trains. Sleepers pay for themselves, but diners don't.

The diners are necessary -- yes, even on all-coach trains -- because the trains run too slow.

Speed up the trains, you can get rid of the diners. (You will probably still retain the sleepers, since they are profit.)

neroden@gmail said...

"This desire by privileged white boys to get something for nothing is destroying our country."

Looking back at the last several years, during which major banks looted the government treasury, while stealing people's homes, while the bank executives got their taxes cut to record low levels (15% for capital gains!)...

I tend to call those people who want to get something for nothing "bankers".

neroden@gmail said...

"They don't have anything analogous to, say, Chicago to LA."

This is the analysis mistake which infuriates me and makes me see red. Because it's a dangerous mistake, one which potentially hurts routes you aren't even thinking about.

Chicago to LA -- whatever. I've used it, but hey, it's questionable, there's a lot of empty space in between.

However, every attack on Chicago to LA is *always* converted rhetorically into an attack on "long-distance" trains.

Which means an attack on the extremely valauble NY-Chicago, NY-Florida, NY-New Orleans, New Orleans-Chicago, Chicago-Minneapolis, Denver-Chicago, etc. etc. service., where there are LOTS of cities in between. All of these routes simply need to run on time, run faster, and run more frequently.

neroden@gmail said...

"Steve Stofka said...

What Paul said."

Paul doesn't have a clue what he's talking about. As usual, he's mixing apples and oranges; fares on the Lake Shore Limited are a very different matter from fares on the Sunset Limited.

In short, Amtrak already charges the market-clearing price. That price is pretty low on crappy trains like the Sunset Limited, which runs at a very low speed three times a week while avoiding cities.

That price is much higher on the trains which are more useful.

neroden@gmail said...

busplanner: to anticipate your response, you'll probably propose buses.

Well, that might work SOME places. Wyoming, perhaps.

However, east of the Mississippi River, those buses will spend most of their time in traffic, and significantly slower than the trains; and the tax-funded road maintenance which they will require will be more than the funding required for the trains, including track maintenance. Plus they're less pleasant.

Adirondacker12800 said...

"Most trains are too slow (which also means that crew and equipment are not as well utilized as they might be)"

It wouldn't work out this neatly but if you can cut the time it takes to get from New York to DC or NY to Boston in half you can carry the same amount of passengers with half the staff and half the equipment at the same fares. Or twice the passengers with the same amount of staff and equipment at half the fares. Or the same fares and make so much money that you can afford to electrify to Albany and Springfield getting them to break even and then have them making money which gives you enough money to upgrade DC to Richmond... a virtuous circle. When it gets to Buffalo and Pittsburgh the people in Cleveland and Columbus are asking rude questions about why they have to take those slow planes to Chicago or New York when they could be taking fast trains that make money. The piece between Cleveland and Buffalo is going to be really cheap which then gives you enough money to build the expensive bits between Boston and Albany and Harrisburg and Akron... which gives you the money to build to Montreal. Montreal to Philadelphia faster than flying is very very attractive.... And then enough money to build from Richmond to Charlotte. eventually there is high speed track from Minneapolis to Miami and San Antonio to Portland Maine. Alon has a fantasy where Amtrak is carrying a billion people a year with something a bit more fleshed out.

3sigma said...

Night trains are a dying breed in Europe. The total number of night trains was reduced by 80% since the mid-1990s, due to both low-cost airlines and faster daytime trains.

They are a niche market of services that take, at night, routinely double the times it takes for daytime travel. The number of passengers the remaining night trains carry is a statistic insignificance on the total market of long (>400km) distance rail travel. And they are not subsidized at least within EU.

This fixation with multiple-night rides doesn't do any favors to a rational, economic-centered discussion of rail service.

arcady said...

Multiple-night rides are indeed not very useful, but a single overnight journey can get you a long way while you sleep, and could be perfect for medium-distance corridors like NYC-Chicago, where travel time can be cut to something like 8-12 hours. As an example from Russia, there is Moscow-St. Petersburg (the country's two biggest cities by far), which was traditionally served by overnight trains with a running time of 8 hours or so. Relatively recently, a "high speed" day train service was also introduced, taking around 4 hours from end to end, using high speed trains on existing upgraded tracks. But, if you look at the actual service offered, there are still plenty of night trains. In fact, there are currently 24 night trains between the two cities, 7 "high speed" day trains, as well as one slower day train. And tickets on the night trains cost about the same as on the high speed trains, but you get to sleep on a bed in a 4-person compartment, and even get a "continental breakfast", which is not a bad deal for $100 if it saves you a night in a hotel.
If you want to check out the timetables for yourself, they're available through bahn.de, though without the ticket price information.

letsgola said...

@neroden we are specifically talking about long distance routes with nothing in between - essentially the routes that travel between the Mississippi River and the West Coast. So yes, we are talking about Wyoming. I would not put NY-Phila-Pitt-Cleve-Chicago in the same category as Chicago to LA because there's a lot of stuff in between and it's not even half as long a route.

And if we really care about rail transportation, climate change, etc., nationalizing the UP and the BNSF and prioritizing passenger service would be one of the worst things we could do.

Adirondacker12800 said...

.. but a single overnight journey can get you a long way while you sleep,

30 passengers per car and a car that goes out of service in the morning and then sits in the yard all day. So lets say half the density and twice the cost per car. The fare is going to have to be a lot higher than coach to cover those costs.

busplanner said...

@neroden -

I think you missed my point. I have nothing against trains if they are part of the solution to a real problem getting people from point A to point B.

But to support train routes that have infrequent service (once a day or less), are slower than air alternatives (most distances over about 300 miles), and are often empty (the Empire Builder approaching Seattle that was hit by a mudslide in Washington had 90 people on it according to news reports) does not make sense.

So, let's identify where rail investment makes sense and focus on those corridors:

1. Distances under about 500 miles in length where higher speed rail (Acela speeds)can compete with air travel.

2. Cities with congested airports (Chicago, New York, DC) where adding more high speed, frequent passenger rail can lead to the reduction of short distance flights).

3. City pairs with congested highway links where rail expansion makes more sense than highway expansion.

However, no individual train should run for more than about 12 hours, even with overlapping markets between endpoints and should have no more than comfortable coaches and a partial car cafe. Thus, on a corridor like New York to Chicago, there might be overlapping train segments - New York to Cleveland; Syracuse to Chicago, each train leaving at 7:00 AM with additional trains providing more frequent service New York - Albany or Cleveland- Chicago or wherever else there is demand running only 8 or 4 hour trips; such that there is a train every two to four hours between 7 AM and 7 PM over the route. However, this is a major committment; so it needs to be addressing a major problem and truly be serving a major demand. Unfortunately, outside of the Northeast Corridor, there are very few places where this level of service is needed currently.

And as for the person in Minot, North Dakota who cannot drive or fly for medical reasons, I guess that person (if he/she can travel at all) will need either to take a bus or hire a private ambulance, just as someone in Bismark, North Dakota, or Pierre, South Dakota, or Evansville, Indiana, or Nashville, Tennessee has to do today.

Paul Druce said...

Paul doesn't have a clue what he's talking about. As usual, he's mixing apples and oranges; fares on the Lake Shore Limited are a very different matter from fares on the Sunset Limited.

You'll note that I compared Silver Service to the Palmetto, very similar trains. As far as fares: LSL coach paid 11.9¢ per passenger mile, nine cents less than coach passengers on the Palmetto. Sleeper passengers on the LSL pay 31.9¢ per mile, but passengers on the Empire Service averaged 31.4¢ per mile in the same fiscal year.

letsgola said...

One other thing... can't forget that US rail equipment is heavier thanks to the FRA. Extra fuel costs add up.

arcady said...

@Adirondacker, typical Russian 4-person compartment cars have 36 passengers per car, and with the longer cars allowed in the US, that might be 40 or 44 per car. And of course day-train coaches end up sitting in the yards all night. And fares are going to be more expensive than coach, but that's okay. The night train is not competing with coach, it's competing with a late evening airplane flight plus hotel, or having to get up at 4:30 am to catch your 6:30 am flight in the morning to get to your destination early enough. And in the concrete case of Moscow-St. Petersburg, you can think of the high speed trains as being like the Acela (and charging Acela-level fares). It's much easier to have a night train priced competiticely with that than with ordinary coach trains. Regardless, RZD still sees enough of a demand for night trains that they run many of them on this corridor, many more than the daytime Sapsan expresses.

@letsgola the thing is, there's far from "nothing" in between the Mississippi and the West Coast. And the thing with those smaller cities is that there are still people living there, and those people want to get places, but there are not always enough of them going to the same place to justify a whole airplane flight direct from A to B. So what flights do exist will require a transfer unless you happen to be going to the hub. So for example, Tucson to El Paso is either 6 or 4 hours on Southwest (or 3:40 but that arrives after midnight. The train is 6 hours, with pretty reasonable departure and arrival times, though only 3 times a week.

Adirondacker12800 said...

Regardless, RZD still sees enough of a demand for night trains that they run many of them on this corridor, many more than the daytime Sapsan expresses.

It doesn't work out this neatly but a Sapsan could make two roundtrips during it's service hours. The sleeper makes a one way trip. So a Sapsan with twice the passenger density could make 4 times as many trips as the sleeper. Twice as much staffing but there are twice as many passengers per shift. At roughly the same fares, if the site I found with fares is accurate. Which one costs them less, per passenger, to run?

The 9:30PM departure from Boston makes it into Washington DC just before 7AM. If this is so lucrative why isn't Amtrak running sleepers? Or on the 10:10PM departure from DC that arrives in Boston at 8AM? Or is it that a late flight, a hotel room and a late flight is faster and cheaper?

busplanner said...

@arcady - re El Paso to Tucson

1. I am sure one can find a few rail pairs where air travel is not competitive with rail service work for distances less than 500 miles; but I also suspect that the actual demand for such service is minimal. By chance, do you know what the station-to-station ridership is between El Paso and Tucson?

2. You chose a poor example. I can fly between the two cities (with a change of planes in Phoenix) on US Airways at least three times a day at convenient times (far better than 3 times a week on the Sunset) in about three hours, including an hour connection in Phoenix. If you want, throw in 90 minutes to clear security in El Paso and the plane still easily beats the train.

Steven H said...
This comment has been removed by the author.
Steven H said...

I don’t know anything about Europe or Russia, but I guess I can address at least one of the reasons why long-distance trains “work” in China: most of China’s sleeper trains travel long distances, but not LONG long distances.

The bulk of China’s population is confined to the eastern 1/3 to ½ of the country. Several major city pairs are (or were until very recently) overnight trips: Beijing to Shanghai is 8-12 hours (5 on CHSR), ditto Beijing to Xian or Harbin, or Wuhan to pretty much everywhere; Shanghai to Guangzhou is 15 to 20 hours, same for Chengdu to Beijing; but even some of the longest trips in Eastern China (say Beijing to Kunming, or Chengdu to the coast) max out at less than 36 hours. Even for those longer routes, all of those trips are through densely populated areas, so there’s plenty of passenger turn-over in the coach cars, if not in the sleepers.

There are some super-long train trips in China…but not many. China doesn’t have a West Coast, so there’s almost no rail traffic of any kind west of Lanzhou. There is a line extending out along the old Silk Road to Xinzhang, and a very new line to Lhasa; however, both of those are sustained by politics more than by either passenger ridership or freight traffic. I doubt anyone in Beijing cares if those lines cover their operating cost. And those trips are *still* largely shorter than long distance trips in the States. A trip from Shanghai to Lhasa takes 48 hours; a very hypothetical and unlikely journey from Harbin to Urumqi still only takes about as long as a trip from Chicago to San Francisco on the California Zephyr.

So, in short, Chinese sleeper trains primarily serve large markets at the far ends of regional corridors--the American equivalent would be DC to Boston or Atlanta; New York to North Carolina or Ohio…maybe Chicago to the East Coast—but they serve virtually nothing else.

I don’t want to say too much more (because I’ve written way too much—and I apologize for that!), but I do want to make one other point about China: I don’t think China’s sleeper services work because China has a large poor population—China’s working poor ride coach—but rather because China has a very large middle class that is culturally attuned to single-night train journeys. Among other things, Chinese work and school holidays are generally two to three weeks long, and most middle class folks are accustomed to sleeping and living in hostel-like conditions on a daily basis anyway. As others have said…Chinese trains don’t have dining cars, and hard sleepers are six beds to a berth.

I don’t think cultural differences are a deal-breaker for American rail service--I think if we offered hostel-ized sleeper car service to the Megabus crowd in rail markets that are just a bit too long for bus travel (i.e. DC to Boston, Hoboken to any number of markets from NC to OH, Chicago to Atlanta), then we would find plenty of passengers at a relatively low cost; but I wouldn’t bother with anything west of the Mississippi...and anything more than 15 hours would be iffy. Even then, there are so many regulatory hurdles, train maintenance issues, labor costs, and freight conflicts, that I don’t think that would really be possible.

Anyway, as I’ve argued before, as long as we’re investing real money in regional services, I think the current long distance services are fine for now, but America’s long-distance services aren’t really comparable to Chinese sleepers. Sorry for the long comment!

arcady said...

Why isn't Amtrak running sleepers? My guess is because they don't have any to run. If you look at how many Viewliners Amtrak has, and how many they need for the existing East Coast trains with sleeper service, there really aren't many spares left. When the 25 new Viewliner II sleepers arrive, that may change, and it'll be interesting to see what Amtrak ends up doing with them.

And as far as Sapsan equipment rotations go, a Sapsan trainset coule make 3 one-way trips per day, assuming a 4 hour travel time and some dwell time at the terminals too, as well as reasonable intercity schedules. But in reality, they end up doing an average of one round trip per day. One benefit of night trains compared to higher speed day trains, btw, is that thanks to their slower speed, they have fewer conflicts with commuter trains, and the evening departure times mean pretty much no conflicts at all. Having a largely tidal flow (inbound in the morning, outbound in the evening) means you may be able to get away with a three track line instead of a four track one.

Steven H said...

Oops, gotta edit one thing. In my comment a while back, I meant Xinjiang, not "Xinzhang." The Silk Road went through Xinjiang.

Stupid mistake considering that my favorite restaurant in Beijing served Xinjiang food.

Adirondacker12800 said...

My guess is because they don't have any to run.

The Sunset Limited leaves and is in revenue service for three days. The sleeper leaves Boston, gets to DC in 8 hours and sits around for 16. Three times the Acela fare for it to be worthwhile? How many people are gonna take it at Acela fares? The people who want to pinch pennies on hotel rooms are the same ones who want to pinch pennies on train fares.

Having a largely tidal flow (inbound in the morning, outbound in the evening) means you may be able to get away with a three track line instead of a four track one.

They have a four ( and in some places 6 and 8 in the Bronx) track railroad between New Haven and Wilington except for the bottle neck between Manhattan and Newark. There aren't big thundering herds of commuters going between Newark Delaware and Aberdeen Maryland. Or Mystic CT and Westerley RI.

Steven H said...

"The 9:30PM departure from Boston makes it into Washington DC just before 7AM. If this is so lucrative why isn't Amtrak running sleepers? Or on the 10:10PM departure from DC that arrives in Boston at 8AM?

Presumably a good chunk of those passengers are really going to New York, though, right?

And those trains don't often run full. I've taken the 66 Regional from Newport News to Boston a few time. There are a lot of people sleeping across two seats during the darker part of that trip.

A perfectly comfortable Chinese hard sleeper can sleep somewhere around 50 to 60 people per car. What's the ridership on the 66 Regional between DC and Boston? If it's less than 60 per car, then it might be worth it.. maybe.

The sleeper leaves Boston, gets to DC in 8 hours and sits around for 16.

To be fair, it doesn't have to sit empty for 16 hours. There's no reason it can't still run during the day, but it would obviously have lower capacity than coach.

I'm completely basing all of this on a hypothetical Chinese hard sleeper with American characteristics, though.

Or is it that a late flight, a hotel room and a late flight is faster and cheaper?"

Maybe, yeah. If you don't mind bunking up with 5 strangers, and don't need a dining car, then why not? Particularly if it means you can avoid an extra night in a hotel.

Adirondacker12800 said...

Presumably a good chunk of those passengers are really going to New York, though, right?

The passengers on the train, no matter their origin or destination, are people who are willing to do the ride in coach..... if you are getting off in the middle of the line in the dead of night that probably means that station is "home" You don't need a hotel at home. You may not want to pay the extra to sleep in a bunk for a few hours instead of a seat.


If you are going to run it during the day it has to be fast enough to keep up with the other traffic on the express tracks. So instead of allocating a new set of sleepers to the NEC you are outfitting Acela IIs with sleeping accommodations. And competing with the first class seats.

And if they are running Acela IIs the 10 O'clock from DC or Boston gets into NY at 11:30 while the sleeper doesn't toddle in until... why would someone take a sleeper when they can leave Boston at 10 and be in DC at 1?..... Which one you gonna take?

Steven H said...

I've never seen a sleeper train that didn't have a lot of coach cars, too. People bound for New York, New Haven, Philadelphia, and all number of points in between will probably ride coach. But there are also people on the same train for 6+ hours who might prefer to have a bit of room to spread out if given the option. Of course they're willing to ride coach for 9 hours...but that doesn't mean that wouldn't also be willing to do something else if given the option.

And look at their other existing options! They could have taken Megabus for $30--but they paid an extra $40 for a more comfortable ride (and not even a faster trip! The 66 Regional is almost 10 hours long)

Those same passengers also could have taken Acela for $280, but they paid $210 less for pretty obvious reasons. We don't know what they would be willing to pay for a sleeper, but there's a lot of room between $70 and $280!

Why would a (predominately coach) train with a couple sleeper cars be any slower than a regional? The regionals don't need to keep up with the Acelas--we run regionals all day!--so why would the train with the sleeper need to do something that the regional doesn't do? Are we going to stop running regionals when the Acela II comes online?

As for whether I would take a sleeper car for a daytime trip: sure! If I'm going to be on the train for almost 8 hours (which I would be if I was taking any of the regionals), then I would like to have that option. Plus, the Acela is still almost 7 hours! Some people (probably not a whole lot) are paying $100 extra for the cheapest Acela, just to save an hour. Perhaps there are a comparable number of people (again, not a whole lot, but maybe 30-60) who are willing to pay up to $100 for a more comfortable ride instead of a faster trip, particularly during the overnight runs.

Again, there's a lot of room between $70 and $280, and we don't know what people are willing to pay right now. There's an existing market for $30 bus rides, $70-$100 8-10 hour-long coach rides, $170-$280 6:50 hour-long coach rides. There aren't any high-capacity/low-frills sleeper car markets in the States right now (and probably never will be), so it's all hypothetical anyway.

And that's just DC-Boston. There are a lot of 8-15 hour coach trips east of the Mississippi, many of which won't have premium HSR service for many years, that are currently served by only coach or low-capacity/high-frills sleepers.

Adirondacker12800 said...

Why would a (predominately coach) train with a couple sleeper cars be any slower than a regional?

Someday the all coaches between Boston and DC are going to be able to do that in three hours. The train that leaves at 9PM gets to the other end at midnight. The all-sleeper that takes ten hours leaves at 9:05 and gets to the other end at 7:05. You can get on the three hour train at 9, take an hour to get home from the station, sleep from 1 to 8 and get up 5 minutes before you would arrive using the sleeper. Or you can leave home at 9 get on the 10:00 sleeper and arrive at your 9 o'clock meeting. Or you can go to sleep at 9, leave home at 5 and get to your 9 meeting.

There are a lot of 8-15 hour coach trips east of the Mississippi, many of which won't have premium HSR service for many years, that are currently served by only coach or low-capacity/high-frills sleepers.

So instead of losing lots of money, on a per passenger basis, on a few high frills passengers, they'll lose less money, on a per passenger basis, on low frills passengers and make it up on volume?
If they charge enough to break even on the low frills passengers I suspect the amount of people willing to pay that premium is somewhere between "none" and "we get so few passengers on the low frills cars that they lose as much money, per passenger, as the high frills cars"

...There's a name for very very low frills. Coach. Which outside of a few corridors, they lose money on.

arcady said...

Fun Fact: China does actually have Acela-speed sleeper trains, the CRH-1E. This seems like a good solution for corridors before they get full HSR, as well as for longer overnight services on HSR lines where those make sense. When Boston-DC travel time is reduced to 3 hours, then yeah, sleeper trains on that corridor make no sense. On the other hand, it brings points south of DC within overnight range of Boston (and closer to NYC), like Charlotte or Atlanta. And it brings points north of NYC (well, Montreal mostly) within range of DC.

Alon said...

Steven, the issue with hard sleepers vs. coach is that a hard sleeper can only be used at night, while a coach can be reused during the day.

Arcady, the problem with the sleepers is that they need to stick to a schedule south of Washington. If there's a slot every 15 minutes or something then they can get away with delays, but if not then a small delay in North Carolina can lead to cascading delays on the NEC. In fact, some of the schedule analyses used for the NEC assume routine delays originating south of Washington, for example the analysis of the benefits of replacing the B & P Tunnel.

Adirondacker12800 said...

well, Montreal mostly

Why would you get on a sleeper that takes as long as today's Adirondack when you can get on an HSR train that only takes 4 hours? Less if they do something serious about Albany-NYC.

IIRC with Albany-NYC down to 1:45 with 150 MPH trains NYC-Montreal is 4 hours. Pity the the NYSDOT took the study down ( or moved it ). Ah you say they can then use them for Boston-Montreal. If they get Boston-Albany down to 2:00 that would imply Boston-Montreal in 4:15 - which was outside of the scope of the study. Do the same thing with Buffalo, Rochester or Syracuse to Montreal...

Interesting little fun fact I ran across a few days ago. If you take the Lake Shore Limited from Schenectady or points west to Boston, you can, if you get off the train in Albany, transfer to the "regular" Empire service train that leaves 5 minutes before the train to Boston, upgrade your ticket to Acela when that train gets into New York on time and take Acela to DC, you can be in DC 25 minutes before the train gets to Boston... that's how bad long distance trains in the US can be ( 200 rail miles to Boston, 367 to DC )

... if they build the tracks for the 150 MPH diesels to 225 mile standards and then someday convert the 175 miles between Saratoga Springs and suburban Montreal to 225 MPH electric trains it's well under 4. If they've make Albany-NY an hour and half by that time NY-Montreal is under three and half.... Do customs and immigration on the trains...

Adirondacker12800 said...

If there's a slot every 15 minutes or something then they can get away with delays, but if not then a small delay in North Carolina can lead to cascading delays on the NEC.

The current long distance trains with the 100 MPH maximum speed suck up slots now. ( if you believe what the foamers say )... versus 135 MPH Acelas and 125 MPH Regionals. One of the reasons they are chronically late northbound is that if they get to Trenton at :14 they conflict with a MidTown Direct. If they get to Trenton at :21 they can be wedged in.... so they get in to Trenton 7 minutes later than they theoretically could.

How many slots do they suck up if they are toddling along at 125 and there's a train leaving Philadelphia or Baltimore that wants to come through at 200 ? You either have to four track from Wilmington to Baltimore or have sleepers that can go 200. Putting two more tracks in that bypass the curves in Wilmington has it's charms but not the nice straight places in Maryland.
... the current long distance trains are 75 percent of the speed of 135 MPH Acelas and 80 percent of the speed of 125 MPH Regional. At 125 MPH they are 62.5 percent of 200 MPH trains...100 MPH trains are 62.5 percent of 160 MPH trains. 125 MPH trains are 75 of 160 MPH... maybe there is a reason why Amtrak spec'd out the new cars for 125 MPH service even though outside of a very few places off the NEC they'll be going no more than 110....

neroden@gmail said...

"So, let's identify where rail investment makes sense and focus on those corridors:

1. Distances under about 500 miles in length where higher speed rail (Acela speeds)can compete with air travel."

Again, confusing distance with time.

Please stop doing this.

We should focus on trips where, with suitable capital improvements, rail can operate under 4 hours, and can compete on time with car or plane. I believe that we can generally string those together to generate trains which run longer than that.

If we string enough together -- which we can -- then I believe that overnight trips of 8-12 hours are also a viable market, and can compete with shorter plane flights.

Above that, I can see running a train for more hours if it's made up of yet more overlapping routes, but you shouldn't really be targeting the market of people travelling for longer than 12 hours. I'm fine with breaking the double-overnights into single-overnights if it will increase the on-time performance.

threestationsquare said...

"However, east of the Mississippi River, those buses will spend most of their time in traffic, and significantly slower than the trains."
NYP-PGH: Amtrak 9h13m, Megabus 7h35m
NYP-BUF: Amtrak 7h55m, Megabus 7h20m
WAS-CLT: Amtrak 7h50m, Megabus 7h50m
CHI-DET: Amtrak 5h18m, Megabus 5h25m
CHI-STL: Amtrak 5h20m, Megabus 5h30m
Outside the NEC proper, Megabus and other bus carriers are usually as fast or faster than Amtrak. They also tend to be more frequent. In my experience they are approximately equally punctual.

Juan said...

Dear Adirondacker, there is no train called "The Trans Siberian". There is the train no. 1 "Russia" from Moscow to Vladivostok which runs every other day. Besides that train, there are a lot of other trains running partly on the Trans-Siberian main line, some of them for example only between Moscow and Ekaterinburg or Irkutsk. Others use the Trans-Siberian line and then go to branch lines, e.g. to Nizhnevartovsk.

Between Omsk and Novosibirsk there are approximately 15 to 20 trains per day. Only one of them is actually running only between the to cities, others coming from and heading to other destinations.

You can all look it up in the German railways' database at bahn.de

neroden@gmail said...

Threestationsquare: you will note that

(1) Megabus is, in fact slower (on schedule alone, disregarding traffic) on half of the examples you picked.

Megabus is usually slower than Amtrak, except where Amtrak takes a far more roundabout route (NYP-BUF). Have you looked at ALB-BUF? All the other bus companies are the same.

(2) Who in their right minds is willing to spend seven hours on a bus? In contrast, seven hours on a train is pretty comfortable. This matters.

Steve Stofka said...

So, to clear up a major misconception in the response to my comment:

100% seat-sales is NOT market clearing rate. As Shoup makes clear in his discussion of parking pricing, a sellout implies that the resource is underpriced relative to desire. Market clearing means the resource is priced such that 85-90% sales rate is achieved (i.e. most people have reserved seats, but there are enough free seats for last-minute straphangers, and to let and alight passengers at each stop).

This means Amtrak can raise fares on its existing LD trains quite a bit, until it reaches the market clearing rate.