18 November 2018

I Wish That High Speed Rail Was More Economically Viable In The U.S.

Rail fans have a proposal for a four phase high speed rail construction program for the United States. But, while I wish this would ever make economic sense, but it really doesn't, outside some selective niche corridors in the U.S.

A transcontinental high speed rail line, for example, does not make economic sense unless you can get speeds of about 6000 mph in an advanced hyperloop type system with construction costs per mile similar to existing high speed rail line estimates, speeds that are about eight times as fast as currently protoype hyperloop design.

Amtrak Fails Because It's Too Slow To Attract Enough Traffic To Make Economic Sense

It is no wonder that Amtrak is hemorrhaging cash outside the Northeast corridor. 

Our existing Amtrak passenger rail system in the U.S., outside the Acela corridor, with average speeds of 50 mph and post-subsidy costs per passenger mile somewhat higher than luxury buses, isn't competitive economically even if you skip interstate highways entirely and travel only on state highways by luxury bus. The operating expense subsidy per passenger mile is much greater than the cost of a commercial airline ticket between the same destinations in many case, with a flight time that is 10% of the Amtrak travel time. Unless the comfortable and scenic trip across the country is the destination, this doesn't make any economic sense.

Even the Acela corridor in along the Atlantic coast from D.C. to Boston more or less, which just barely breaks even economically, is just barely competitive with driving, since it averages about 75 mph in a region where the interstate highway speed limit is 55 mph and traffic frequently reduces your average speed below that, and an Acela ticket costs considerably more than driving or taking a bus. Also, traveling by Acela is only time competitive if both your initial place of departure and your destination are pretty close to train stations served by Acela service.

Of course, with dedicated 220 mph high speed rail lines in the Northeast Corridor from Boston to the District of Columbia, the system would probably win a much larger market share of the market for travel between those destinations, relative to buses, driving and flying on commercial aircraft, even at prices somewhat higher than those for existing Acela service. In the Northeast Corridor, and in a number of other medium length, high traffic corridors, as discussed below, high speed rail does make economic sense, even though there is probably no other niche where it makes sense in the U.S.

High Speed Rail Isn't Competitive With Existing Commercial Aircraft For Long Trips

Even at 220 mph with additional time spent stopped to let people on and off, rail cannot compete with airplane for long distance travel, since commercial aircraft can fly at 450-600 mph, from airport to airport, with no intermediate stops along the way, at a cost per mile that is only slightly more expensive (even after immense federal Amtrak subsidies per passenger mile on most non-Acela routes, and even with no rail infrastructure costs) than Amtrak travel at 50 mph for the same distances.

But, the sunk cost of the interstate highway system and our domestic commercial airline system, as well as population densities and traffic levels much lower than in places where high speed rail is used now (mostly Europe, Japan and China) and the immense costs per mile to build high speed rail lines, suggest that high speed rail will never make sense in most of the U.S. until jet fuel prices increase from 1/7th of the expenses of an airline today to something on the order of 15 or so times that much in inflation adjusted dollars (i.e. when oil starts selling for around $750 a barrel in 2018 dollars up from $40-$50 a barrel today). And, with jet fuel at those prices, high energy density batteries for electrically powered planes start to make economic sense.

Realistically, even at 220 mph, high speed rail is never going to be seriously competitive with commercial aircraft travel in the U.S. for long trips.

High Speed Rail Isn't Competitive With Cars and Buses For Short Trips

Our quite fast interstate highway system, coupled with intracity transit options that are anemic by international standards, also means that a rail system with train stations far enough apart to make sense for a 220 mph train, can't compete with cars and buses going point to point over short distances. This is because, with a high speed train, like an airport, you have to travel a considerable extra distance to get to the stop and then from the stop to your final destination, and this eats into the time you save by going 220 mph instead of 65-76 mph on the fastest highways.

High Speed Rail Can Make Economic Sense For Medium Length High Traffic Routes

Instead, the only niche where high speed rail is economically competitive in the United States is on medium length trips that have a high volume of traffic, and only if their speed is much greater than the 75 mph you can drive on an interstate highway. 

Rail only makes sense if it makes economic sense for people to ride it given the time and cost comparisons to driving or flying, to the extent that you can move a lot of passengers along that route by train on an average day, day after day for decades to come.

For example, high speed rail probably does make sense in Colorado from Fort Collins to Colorado Springs along I-25, if it was fast enough and cheap enough, and maybe even from Denver to Glenwood Springs, both of which are medium length, very high traffic corridors. But, it is very hard to justify extending those rail lines from Fort Collins to Cheyenne, Wyoming, or from Colorado Springs to Pueblo, Colorado or Albuquerque, New Mexico, or from Glenwood Springs, Colorado to Grand Junction, Colorado, because, given the much smaller volume of traffic on those parts of the routes, the cost of doing that per passenger mile relative to the cost of running luxury tour buses along I-25 or I-70 at 75 mph, is incredibly more expensive for a pretty modest time savings for the typical traveller.

Also, the main reason that high speed rail is competitive at those distances relative to air travel is that security screening is required for air travel but not for train travel. One horrific train based terrorist incident in the U.S. (and there have been terrorist incidents on passenger rail lines in France, Spain, England, Japan, India and Russia, for example), could easily destroy that advantage if security checkpoints are imposed for train travelers. 

There are a some corridors in parts of their Phase 1 map  of the high speed rail advocates where higher speed rail can make economic sense, like along the Pacific coast of California, Oregon and Washington State (possibly on to Vancouver, Canada and Tijuana, Mexico), or from Fort Collins to Colorado Springs in Colorado, or between some of the major cities in Texas, or in the Northeast Corridor (as discussed above).

In these high traffic, medium length corridors, the concept is even more attractive if you can go three times as fast at the anticipated 760 mph hyperloop system speeds. 

How Fast Does A Hyperloop Style System Have To Be To Make Sense For Long Routes?

For long distance travel on fixed pathways in a train to make sense you need you need the 760 mph plus of Elon Musk's hyperloop proposal, so you can be performance competitive with existing commercial airplanes. This speed limit is basically the speed of sound in ordinary air. To go faster, you need to replace the current design with an air filled tube with a system that sends passengers through a vacuum contained within the tubes.

But, even then, the cost of developing the technology for and building that much infrastructure for long distance routes (like NYC to LA) is probably much more costly than the cost of developing quieter supersonic airplanes, an almost ready for commercial production technology, that would move passengers at speeds of about 1500 mph with no new infrastructure investments and probably only about double the cost of existing commercial air travel per trip. The main reason we don't already use supersonic aircraft for domestic flights is the noise their sonic booms make when they fly over the country (the Concorde, the only commercial supersonic plane service to have ever flown, was used for transatlantic flights from NYC to London and Paris, where noise wasn't nearly as great a concern). Still, even at hyperloop speeds, with infrastructure costs per mile similar to other kinds of high speed rail systems, hyperloop is still only really competitive on medium length trips that have a high volume of traffic, although the economically viable trip length is quite a bit longer than it is for 220 mph rail systems. 

For a hyperloop type system to be competitive with supersonic commercial aircraft on long haul trips, from a travel time perspective, it would need to go faster than 1200 mph or so (a bit slower than the supersonic commercial aircraft because there are much shorter security waiting times), and to make economic sense it would have to have a decisive time advantage over flying to justify the infrastructure costs, so it would need to have speeds of at least something on the order of 2200 mph. Technologically, that isn't impossible. Theoretically, it ought to be possible as a matter of engineering possibility, to get a hyperloop style system to move passengers at as much as 14,000 mph. But, there is a big gap between what is theoretically possible and can be accomplished for a finite construction price, even with an R&D budget in the tens or low hundreds of billions of dollars.

Also, the economic returns to faster speeds show diminishing returns. An LA to NYC flight with existing aircraft is about four hours of flight time, in round numbers. In a 1200 mph hyperloop system, the trip is an hour and forty minutes or so, but significantly more expensive than a next generation supersonic commercial aircraft. In a supersonic plane, you can cut that to an hour and twenty-minutes or so at Mach 2. 

In a 2200 mph hyperloop system, the trip is about 54 minutes and it would also be significantly more expensive than a next generation supersonic commercial aircraft. It wouldn't be surprising at all if economically, if passengers wouldn't be willing to pay significantly more than a next generation supersonic commercial aircraft to shave 26 minutes off of a NYC to LA trip (the time savings would be even smaller if there were any intermediate stops on the hyperloop system along the way).

To cut the travel time for a hyperloop trip from NYC to LA to half the time of a supersonic commercial jet, i.e. to 40 minutes, which people might pay for, although getting the necessary volume of ridership would be challenging, it needs to hit 3000 mph, about four times the speed of the current design.

With a maximally theoretically possible 14,000 mph hyperloop style system, the NYC to LA trip time drops to about 9 minutes, which I'm sure that many business passengers would be willing to pay a substantial premium for, even if it was stretched to 15 or 20 minutes for one or two intermediate stops, or with a direct trip at 6000 mph or so, about eight times as fast as current design speeds.

And, all that time, you still have to stay competitive with hypothetical "hypersonic" commercial aircraft that travel at several times the speed of sound.

Now, then again, getting to hypersonic in commercial aircraft does pose so many technical problems that it will probably never happen.

And, realistically, once you transition from air filled hyperloop tubes in which you can reach 760 mph, and hyperloop tubes that contain a vacuum, this one (expensive in construction costs per mile) innovation can probably get you immediately to 6000 mph, while leaving a decent margin of error for the possibility of an near vacuum that is more like the boundary area between Earth's atmosphere and outer space, rather than a true deep space quality vacuum.

So, a line from NYC to LA with stops in Chicago and Denver that takes 30 minutes or so, isn't a completely unattainable possibility from either an economic or a technological matter. But, between R&D expenses and much greater construction expenses associated with a 6000 mph hyperloop system, you are looking at a price tag in the several hundreds of billions of dollars ballpark.

The additional construction expense, however, probably isn't justified by the economic benefits from increased ridership with a 6000 mph system in medium distance corridors, even though a 760 mph current technology hyperloop system might be very competitive with a 220 mph high speed rail system alternative.

Different Options For A San Diego To Seattle Trip And Pricing High Speed Rail

As I noted before, one of the corridors where high speed rail might make sense is along the Pacific coast. All estimates are for one way trips.

For example, on existing Amtrak service, the trip from San Diego to Seattle would take 35 hours and 16 minutes, plus 30 minutes travel time to and from the train station combined, for a total of 35 hours and 46 minutes and cost $133. The trips to and from the train station might end up costing $5 each for a total cost of $143. You also incur greater food costs on a trip this long of about $25, at about $5 per 4 hoursish chunk of non-sleeping time (eight hours in overnight trips) in excess of the 5 hours and 30 minutes for an airline flight, for a total of $168. This also doesn't reflect the substantial federal operating cost subsidy per passenger mile on this trip, which probably makes it more expensive once that is considered that the proposed hyperloop price that would get you there in 7% of the time.

By Greyhound bus, the trip from San Diego to Seattle would take 31 hours and 45 minutes, plus 30 minutes travel time to and from the bus station combined, for a total of 32 hours and 15 minutes. The bus ticket would cost $182 plus $5 each way to and from the bus station, for a total of $192. You also incur greater food costs on a trip this long of about $25, about $5 per 4 hoursish chunk of non-sleeping time (eight hours in overnight trips) in excess of the 5 hours and 30 minutes for an airline flight, for a total of $217. This doesn't receive a government operating cost subsidy.

It would take 18 hours and 45 minutes to drive that distance in a car or truck or charter bus, not counting meal break times. Realistically, it would take 20 hours, point to point, to make the trip unless you are die hard college students who can rotate drivers. At the standard IRS personal mileage rate of 18 cents per mile, the cost would be $226. you also incur greater food costs on a trip this long, estimated to be about $5 per 4 hoursish chunk of non-sleeping time (eight hours in overnight trips) in excess of the 5 hours and 30 minutes for an airline flight, for a total of about $241. A high speed rail line would absorb almost all short duration trip traffic on this route. This doesn't receive a government operating cost subsidy.

If you trimmed stop times at four intermediate stop (LA, one between LA and San Francisco, San Francisco and Portland, OR) to four and a half minutes, you could make the trip from San Diego to Seattle in a 220 mph high speed rail system in 6 hours. If you add the time traveling to and from the train station from a central city destination of 15 minutes at each end, you are at 6 hours and 30 minutes, from point to point for the 220 mph high speed rail trip. If you set the cost to compete with the less comfortable but one hour faster airline flight and considering the cheaper trip to and from the train station, you probably can't charge much more than $200 all in, and this leaves a budget of $190 for the train ticket itself, for this trip that is available to finance this operation, which might pick up half of the traffic volume in that corridor for the full distance trip, but a larger share to the traffic volume on shorter segments. The food costs would be similar to those in an airline flight.

By comparison, the airport gate to airport gate flight time from San Diego to Seattle on a commercial airline flight is about 3 hours, plus an hour on the front end to get through security and to your gate after being dropped off at the airport if you are cutting it close, for a total of 4 hours, and closer to 4 hours and 40 minutes if you are checking a bag after being checked in, in advance. If you add in trips to and from the airport to your respective destinations you are up to 5 hours and 30 minutes, from point to point, since airports are usually further from city centers than high speed rail stations. The plane ticket would be about $167 including the air travel taxes involved. The trips to and from the airport might end up costing $10 each, for a total cost of $187This doesn't receive more than a negligible government operating cost subsidy. This is the benchmark for food costs.

On a San Diego to Seattle route, a 760 mph hyperloop system trip would take about an hour and 40 minutes of relative comfort, plus about 5 minutes per intermediate stop. If it stopped at LA, San Francisco, one city in between those two, and Portland, in addition to San Diego and Seattle, it would be a two hour trip to go 1255 miles. If you add the time traveling to and from the train station from a central city destination of 15 minutes at each end, you are at 2 hours and 30 minutes, from point to point. Now, shaving 3 hours of travel time from the trip means that you could charge considerably more than a plane ticket and still get the lion's share of the travel volume in the corridor for all segment lengths. I suspect that you could charge $380 for this train ticket and also a $5 each trip to and from the train station and still take most of the traffic volume from the commercial flights with speeds this fast (double the all in commercial airline charge adjusted for reduced food costs). You also incur somewhat less food costs in a trip this short. saving $5. This implies a price for the train ticket alone in the ballpark of $370 with more trips than the 220 mph service. So, this is an extra $180 per full length trip more of revenue than the 220 mph high speed rail option, with a probably significantly high traffic volume (I estimate 60% more passenger volume, based on 50% of market share for 220 mph v. 80% of market share for 760 mph). This implies about 3.2 times as much operating revenue than a 220 mph high speed rail system.

Also, employee compensation expenses (which are a large share of the total operating expense in most transportation systems other than driving your own car), perhaps 30%-50%, would be roughly half that of the 220 mph high speed rail or the airline trip alternatives because the trips would be faster. My back of napkin estimate based upon 50% of the airline ticket cost of $157 before ticket taxes is $75 per passenger for the full trip on airlines, $80 per passenger on the 220 mph train, and $40 per passenger on the hyperloop. This implies an operating profit of about 3.5 times what the 220 mph high speed rail makes. Now, the Acela just barely breaks even on operating costs. So, the profit per full length trip on a 220 mph high speed rail system might be $20, and the profit per full length trip on a hyperloop system would be 3.5 times the 220 mph system's operating profit, so about $70 per full length passenger per trip times 50% of the existing flight traffic between the destinations, and without an operating subsidy.

So, the hyperloop system could support a significantly larger infrastructure cost and still make economic sense. But, the current estimates are that the infrastructure costs for the hyperloop system actually wouldn't be that much greater than the 220 mph high speed rail option

There are 50,000 passengers a day that leave the San Diego airport. If 8% of them were going to Seattle that would be 4,000 per day, times two because most people would fly round trip, times 365 days per year, you get roughly 3 million trips per year, which implies $210 million of profit per year. Using a 3% per annum interest rate on government bonds and an infinite amortization period, the amount of capital investment cost that this kind of profit could finance is about $7 billion. The estimated construction cost is $11.5 million per mile and you would need 1255 mile of line from San Diego to Seattle, a cost of $14.5 billion in capital expenditures. Of course, this doesn't include estimates of the revenues for shorter segment trips and the traffic estimate is not very precise. But, order of magnitude-wise, a hyperlink system ought to just about break even on a San Diego to Seattle route, and if it needed a construction subsidy, it would need significantly less than a 50% subsidy, when an 80% subsidy is the norm in U.S. transit funding. If it got the 80% customary infrastructure subsidy, it would need to support only $2.9 billion of system contribution to the capital costs, which is $5.1 billion less than what it could have afforded, which leaves a cushion for errors from not being conservative enough in andy and all of the other estimated numbers.

And funding this kind of transit might be a good investment for the federal government for other reasons such as reducing highway maintenance expenses and reduce air pollution and reduced demand for foreign oil, that might be well worth a $5.1 billion expenditure amortized over 30 years or so, for an annualize real dollar cost of about $170 million a year in inflation adjusted dollars (before reducing that for costs avoided), a small price to pay for a state of the art high speed rail system that is heavily used and pays for all of its operating costs without government subsidy and pays 20% of the capital investment from its own stream of "profits."

1 comment:

Dave Barnes said...

There are, maybe, 4 places where intermediate distance rail could work in the USA.
(Long distance passenger rail will NEVER work.)
1. Boston-Washington
2. NY-Chicago
3. San Francisco-San Diego
4. Dallas-Houston

Amtrak should abandon service everywhere else.
Congress needs to acknowledge that significant operating/maintenance subsidies will be required. Just ask the Europeans.