Start with basic numbers.
Your typical semi carries a payload of a little over twenty tons. It requires a full-time operator, has a 500 HP (370 kW) engine, and gets 5 to 6 miles to the gallon.
A “standard” boxcar carries a payload of 125 tons. A train of 100 cars thus carries 12,500 tons of payload, has three engines each of 4000 HP for a total of 12,000 HP (9,000 kW), and requires two operators. Its fuel use is hard to discover — nothing really apposite turns up on the Internet.
Fortunately the Federal Railroad Administration has commissioned a beltway bandit [ACTHUNG: .pdf] to reduce the numbers to something that makes them easy to compare. Trains nowadays use about 2.5 gallons per 1000 Revenue Ton-Miles, so our 12,500 tons of freight burns 31.25 gallons per mile or 0.032 miles per gallon. From the same report, trucks nowadays are getting around 120 ton-miles per gallon, which is 8.3 gallons per 1000 ton-miles. The train is thus a little over three times as efficient as the truck in fuel alone. The train also requires only two operators, thus 6,250 tons per person. At 20 tons each, it requires 312 (and a half!) truck drivers to move the same load.
So we ought to be using trains for everything, right?
Not so fast, bub.
Look back at the numbers. The truck has 25 horsepower per ton, because it needs it as it goes up hill and down dale on rubber tires. The train only has 1 horsepower per ton of payload — that’s how it gets its fuel efficiency: it’s moving a much bigger load with a comparatively much smaller engine. This is why we have railroads in the first place. The first prime movers were steam engines, and steam engines are heavy and not very powerful compared to internal combustion, and use more fuel. In order to move any useful load at all, they had to reduce rolling resistance as much as possible and eliminate as many hills as they could. The lowest possible rolling resistance comes from a hard wheel on a hard surface, thus steel wheels on steel rails. (There are harder materials, but they’re expensive and have drawbacks like brittleness. Steel on steel is the best compromise.) As for hills, the “ruling gradient” — the steepest hill for “normal” train tracks — is 1.5 to 1.8 percent, that is, less than two feet of rise per 100 feet of travel, and the maximum that’s ever been used regularly without special provision is less than 6%. By contrast, the ruling gradient for Interstate highways in hilly terrain is 5% with short exceptions up to 8%, and there are a few places where the road goes up ten feet per hundred feet. The trucks don’t like it and use a lot of fuel going up, but they manage without the extraordinary mechanical systems trains need for steep grades.
So, increase train horsepower, right? Two problems: the train only has a three-times advantage in fuel use to start with; doubling the horsepower doubles the fuel use and halves the advantage. There’s also a problem: friction, or, rather, not enough friction. Steel on steel has very low friction, which means it’s hard to convert horsepower to forward motion without the wheels slipping. Even with one horsepower per ton the train has to dump sand on the rails to get enough friction to climb a fairly steep grade, or even to start all those tons of load moving. A train with even two horsepower per ton would probably just spin its wheels, converting both wheels and rails into scrap — and would only have a 1.5 times advantage over the truck in fuel usage per ton moved.
All that means train tracks are hideously expensive. The rails themselves are cheap, and the ties they sit on are hardly less so, but that’s not where the expense comes in. The expense is in moving all those tons of dirt to flatten out the hills. It can’t be just plain dirt, either. The train is so heavy that ordinary soil would just squeeze out from under it; the fill material for a train track has to be crushed rock that’s not only heavy, it has lots of corners to hang on to the next piece. Even then, it still creeps out from under. The railroad has huge, expensive machines that literally pick the track up and shove more “ballast” underneath it to compensate for the movement, and to gradually fill low areas so as to eliminate as many grades as possible.
And it gets worse. The pressing need to eliminate grades, because the train has such low power, means that there are only a few places where train tracks can be built at all — and those places have already been found and put into intensive use. We literally cannot build any more railroads, because there’s no place to put them where there isn’t already one! In fact, if we had to start over we couldn’t build the ones we have now — imagine the Sierra Club’s reaction to a plan calling for dynamiting huge chunks of the mountains to level them out for a rail line. For one thing, every little dell and alp in hilly terrain has a different microclimate and thus different trees, insects, etc., living there, which some environmentalist is going to define as a different species. We probably won’t be building any more major roads, for the same reason, but there’s a lot more excess capacity in the roads we have now than in the rails.
Railroad usage is totally constrained by the slowest train using it, because trains can’t pass (“overtake”) one another without expensive and elaborate provisions. At this point we hit friction again: low friction between wheel and rail means long braking time and distance. The absolute minimum interval between trains is the distance needed to stop if the train in front has problems. Stopping distance is proportional to the square of the speed, and (approximately) the square of the gradient — double the speed, four times the stopping distance; if it’s coasting down a 2% grade, multiply by four again. That’s a long, long way, and given the necessity to follow contours to minimize grades the train’s driver is almost never able to even see the train in front, much less judge whether something a mile or more away is moving fast enough that it isn’t necessary to stop, or if a stop is possible in time. Very few trains ever go over about 70 MPH/110 KPH, and a lot of train tracks go through urban areas where their speed has to be held down to as little as 25 MPH/40 KPH or less; even then, the occasional driver of a car or school bus finds out that trains don’t stop very fast, even for children or pretty girls. Slow trains can bunch up at shorter following distances, but it doesn’t make sense for trains to run at high speed in open country if they then have to stop and wait for the train ahead to clear the town.
Curves follow the same rule. Doubling the speed gives four times the side force, requiring four times the curve radius to keep the train on the tracks. In hilly areas, where the train has to follow the contour to minimize the gradient, there are sharp curves where the train may have to go as slowly as 15 MPH/23 KPH so it doesn’t turn over, and the gradients are steep enough that it may not be able to slow down for the next curve if it speeds up between. Train speeds aren’t controlled by horsepower. They’re controlled by stopping distance, curve radius, and reaction time.
In song and story, trains used to go faster. That’s true, but not by all that much — 100 MPH/160 KPH was about the limit, roughly a square-root-of-two difference from today’s maximum. That was only for passenger trains, and meant that the freight trains, which topped out at 50 MPH/80 KPH, had to either use a different track or stop altogether on a siding to let the “varnish” through. It was also in an era when people didn’t care as much as they do now about accidents, and tended to assign responsibility to the victims — if you got run over by the Twentieth Century Limited, it was because you weren’t paying attention. There’s no way that attitude would be possible today.
Which brings up accidents. If a truck falls over, we’ve lost twenty tons of freight and the road is closed for a few hours. If a train derails, we lose twelve thousand tons at one blow, and can’t use the rail line again until it’s rebuilt, which can be weeks or months — and the trains can’t go around, because there are a limited number of tracks, all of which are already close to capacity.
A limited number of tracks means limited access to rail shipping. The railroad can’t go everywhere; it’s too expensive and too dangerous. Loads, or passengers, have to be taken from their origin point to the rail terminal, removed from that transport (probably a truck or an automobile), and loaded onto the rail car; when the train gets to its destination terminal, they have to be offloaded from the train, loaded onto local transport, and taken to their actual destination, where they have to be offloaded again. More subtly, very few people need to ship twenty tons at once, let alone 125, and a passenger is of course much less than that. Even with trucks, there’s a big, complex “consolidation” and “less than load” industry devoted to combining small shipments into truckloads for efficient carriage. If the minimum unit for transport is six times as big, the matter becomes even more cumbersome, inefficient, time-consuming, and expensive.
So trains are somewhat more efficient than trucks, but not all that much more, and there can only be a few of them running in highly restricted corridors. There aren’t going to be many more railroads, because building them is ferociously expensive both in money and in environmental degradation. They aren’t going to go much faster than they do now, because getting them to do that would require even more ferocious expense to straighten out curves and get rid of interaction between trains and people, cars, etc., in urban areas, and even if we did that the problem of stopping wouldn’t go away; in fact, it would get worse. A single accident brings the whole system down, which may mean nothing moves at all for a long time. Limited access means complex and expensive means to consolidate loads and get them to and from the rail terminal. If we had it to do over from scratch, if railroads hadn’t been invented in the first place, no one with any sense would propose such a stupid idea.
“High-speed rail” for passenger service is all that, squared. We do have the existing railroads, and given that they’re here already they do yeoman service in transporting large loads long distances. Converting them for high-speed passenger use would displace the existing freight onto the highways, because there’s no way for 50 or 70 MPH freight trains to share tracks with TGVs. Passengers would have to go from their homes to a strictly limited number of rail terminals, and again from the rail terminal to their real destination — and they already have to do that to go by air at twice the speed. A single accident would shut the whole system down for an extended period; a car or truck wreck hardly impacts the system as a whole, and even an airplane crash only slows air travel down in the worst case, and only for a few hours. High-speed rail has to be completely separated from everything else: no grade crossings allowed, because the train isn’t going to stop and people going across wouldn’t have time to see the train coming before being squashed, and elaborate fencing the entire length of the line, to keep animals and fools off the tracks.
Most proposals for high-speed rail posit entirely new rail lines, to avoid the displacement problem. It’s clear that that hasn’t been thought through. If there’s a satisfactory place for a rail line, there’s already a rail line there; building a new one requires Herculean effort, and if it has to be level enough and have large enough curve radii to accommodate higher speeds, it’s going to mean moving a lot of rock and dirt. The Californians think they can build 800 miles of high-speed rail for $60 billion, or
$750 $75 million per mile, by consolidating it with existing rail corridors. The only possible explanation for that is too many bong hits per bureaucrat. They’re already fighting the Sierra Club over how to get it through the hills to the San Francisco Bay area; when the good folks in Los Altos and Mountain View discover that it means replacing the existing rail line — which they already hate; it’s ugly and disrupts traffic — with a ten-foot rock and concrete wall topped by chain-link and razor wire I, for one, expect fireworks. That’s nothing, though, compared to the reaction when the Greens find out how they’re going to get it through the Tehachapis and into Los Angeles, which could easily cost three-quarters of a billion $75 million per mile just for lawyers.
[ed. note: There was a time I could do arithmetic. Commenter Murgatroyd has pointed out that that is no longer the case. Corrections above.]
It’s a stupid plan. Or it’s a different plan. It’s already been suggested that the goal isn’t high-speed rail, it’s a big-money project that’s an easy target for graft. If you’re moving sixty or a hundred billion around, losing a million here and a million there is easy to ignore, and even a few tens of millions can be covered up in the overhead. If that’s the real point, the projects make sense from the point of view of the people proposing them; whether or not any benefit would actually accrue to the population in general is then irrelevant. Is that cynical enough?