Yes, Transit is Green
I’ve just found a post by Brad Templeton arguing that US mass transit is less green than high-efficiency cars, at least when compared per passenger-km. (He agrees that transit is overall better because it is more efficient when used more extensively, as in Europe and especially East Asia.) The analysis of how this can be given the numbers is cogent, but the numbers themselves are suspect, and are worse for transit than other numbers I’ve seen.
Better numbers can be found in this FTA presentation, on pages 10-11; the data is sourced to the National Transit Database. They’re expressed in pounds of CO2 per passenger-mile; if you’re more used to thinking in terms of passenger-miles per gallon of gasoline equivalent, then convert x pounds per passenger-mile to 19.374/x passenger-miles per gallon. The New York City Subway gets the equivalent of 114 passenger-mpg, versus 47 on Templeton’s page. Even FRA-regulated commuter rail does significantly better than cars – the low efficiency of the trains cancels out with the fact that there’s almost no off-peak traffic.
Another piece of evidence Templeton’s transit numbers are too low: he lists JR East’s energy use as equivalent to about 78 passenger-mpg. In reality, JR East claims much lower emissions, about 13 grams per passenger-km (400 passenger-mpg equivalent) or 19 (280), depending on whether one counts the emissions of the company’s buildings or just transportation emissions. It could be that Japanese power generation is that efficient; but given that Japan’s overall per capita emissions are not low by non-US developed country standards, I doubt it.
Finally, although it appears as if technology is about to make cars much more efficient, in reality technology is expensive if you’re a driver and cheap if you’re a transit agency. Take hybrids: the market share of new hybrid car sales is in the single digits, about 300,000 out of 8 million light vehicles sold in the US in 2008, but the market share of new hybrid bus orders was 22% in 2007. Electrified trains are also gaining efficiency, perhaps more slowly but the important thing for them is to transition to low-carbon power generation; if their emissions are nontrivial thirty years from now, then we have bigger problems than transportation to worry about.
Pedestrian Observations 
I wasn’t sure if Japanese electrical sources really were the same (e.g., if they use more Nuclear). For the record Japan is only slightly better than the US in terms of CO2; relevant sources are the Japanese and US Wikipedia pages on energy.
Japan is only slightly better in terms of percentage of power produced, but the average Japanese citizen still puts out around half as much carbon dioxide each year as the average American.
Conservation is no joke and Americans have forgotten the meaning of the term.
I wonder where Templeton got his entirely bogus numbers.
Mass transit, as long as you aren’t running ludicrously empty vehicles, is clearly better on carbon emissions even than full *battery-electric* cars; at that point it’s a simple comparison of loads, i.e. the average passenger load must be high enough to make up for using a larger, heavier bus or train instead of a car. This doesn’t require a very high average load, though. 10 will usually do.
Meanwhile, the outrageous inefficiency of the Otto cycle engine means that gasoline cars are hopelessly inefficient and are never going to win by any fair measurement.
Your point that the new technology is expensive for individual drivers, but relatively cheap for a large mass transit agency, is well-taken. My upcoming battery-electric car is going to have a sticker premium of well over 50% more than a comparable gasoline car (arguably as much as 100%, depending on how you figure it). In contrast, the premium for electric buses over diesel buses (and diesel is already better than gasoline) is usually less than that, from the studies I’ve seen.
Diesel isn’t really better than gas – it’s just energy-denser. The pollution is actually worse, depending on which study you look at.
Hybrids have a 50% premium over regular buses, same as with cars. The reason transit agencies pay the premium is that the main cost of running a bus are in operations, not capital.
And transit agencies buy fuel by the tanker truck at a time. So even small percentages translate to big bucks.
Long Beach Transit studied performance of its post-2005 gasoline-electric hybrid fleet (Southern California uses gasoline because the air quality board bans transit agencies from buying new diesel buses) and its older diesel fleet. The costs of the hybrids are twice as much as a diesel, but the maintenance costs have so far been half of the diesels.
The higher capital costs helped to save on operating costs, especially over the 12+-year life of the buses.
Hybrids don’t really have a 50% premium over regular cars, though; that’s FUD spread by diesel folk. A couple thousand dollars over the comparable base model on vehicles in the $20K-$30K range is more like it.
A few of his numbers are directly from the Transportation Energy Data Book. It only has very large aggregates, no breakdowns by city or even LRT vs. Heavy Rail. Those numbers are BTUs, without any specific concern for which kind of fuel (i.e., hydroelectric is not considered any better than gas by that measurement). I’ve tried to reconstruct his other numbers but have failed, but I’ve also failed to come up with the same numbers as anyone else (including other government reports). At this point we really need a public spreadsheet that doesn’t leave any calculations out.
For diesel trains, usually there are fuel consumption numbers, but they’re not easy to find. A very modern DMU gets about 4 vehicle-mpg; the FTA numbers for (electric) heavy rail suggest similar emission-efficiency, which sounds reasonable to me.
I recall a similar article published in the UK, which showed that a certain diesel train from the mid-70s with a load of 25% emitted more CO2 per passenger-mile than a modern, small, fuel efficient car with 5 people in it. The calculations were correct, but it ignored (1) the average load factor on said train was more like 75% than 25% (2) most people in the UK don’t drive modern small cars (particularly if they are in the habit of carrying 4 passengers) and (3) the average number of people per car is about 1.1 in every study I’ve seen.
Invalid comparisons are misleading.
More important than the current CO2 per mile is the MARGINAL CO2 per mile. If I switch to riding transit, the buses and trains I use will have 1 less space, but will be a little more efficient. If I switch from transit or biking to driving, all that CO2 is brand new.
Since most transit vehicle in the US have empty seats most of the time, if transit usage doubled tomorrow we would only need to increase service at rush hour, in most places, to handle the additional load. So from a marginal standpoint, new transit commuters in big cities do add some CO2 emissions, but they will be riding full buses and trains which are at their most efficient. And new transit riders at off-peak times are basically riding for free, from a CO2 perspective.
Another problem with Templeton’s analysis isn’t the numbers, it’s the metric. Per-passenger miles is a way of measuring an individual taking one form of transportation vs. another, holding all else constant. But of course, all other things aren’t held constant.
Automobiles are very good at moving individuals long distances to far-flung locations. Transit is good at moving lots of people to a small number of places. Consequently, automobile-oriented development is development in which people travel longer distances to far-flung locations, while transit-oriented development is development in which people live and work in denser spaces and travel less distance between destinations.
So, if you’re evaluating what you as an individual ought to do given that you already live 20 miles from where you work, per passenger miles is a decent metric. But often, people choose between driving 20 miles into work, or taking the bus 2 miles.
Yes, the source of the numbers is the Dept. of Energy’s factbook. It looks at energy consumed, not CO2 produced. I have some other articles looking at the various ways we try to compare gasoline to electricity, including energy, greenhouse gas emissions, other pollution, perfect conversion (which is what the EPA uses for electric cars) etc. The emissions numbers vary wildly from place to place. If your grid is mostly coal it’s bad. If you have hydro or nukes it’s great. There are arguments for all the metrics (except for perfect conversion which is not possible as far as we know.)
Much easier is comparing electric cars to electric transit in the same area, or comparing fossil fuel cars to fossil fuel (mostly diesel) buses and trains in any area. Riding an electric train in a hydro-powered city is greener than any gasoline car, but the comparison to an electric car is harder. And of course, many people have concerns about the environmental destruction of hydro and the risks of nuclear, but those are harder to quantify.
As noted there are arguments for all methods, and if greenhouse gas is the big on, it is indeed the one to use, but the comparison going into the future will be efficient electric robocar vs. electric transit, and so it gets back to energy anyway.
The reason I’m still skeptical of the numbers is that they don’t really hold even in places with a fairly coal-heavy power mix, such as Japan.
The basic problem with electric cars: they’re expensive, and the only sort of policy that will make their market share nontrivial is a large tax on emissions. In that case, it doesn’t really matter how it does compared to transit.
Hmm, it seems like there are many problems with cars that aren’t related to energy usage / CO2 at all… Besides being heavier per passenger (so less efficient directly), they take so much more space, and are part of a lifestyle that drives expansive development which increases travel distances, which increases dependence on cars, which …
Maybe you could somehow engineer a space-efficient infrastructure oriented around POVs… (I dunno, super-lightweight vehicles with underground robot parking stacks and automatically-driven 3d tube roads or something) but nobody seems to be doing so… It’s pretty much 12-lane surface roads and giant parking lots for the most part.
Space is a much bigger issue than weight. At average occupancy, New York City Subway trains weigh 1.25 tons per passenger, and New York City buses weigh about 1 ton per passenger. The reason they’re more emission-efficient than cars with the same weight per passenger is that, first, bigger vehicles’ emissions rise less than linearly with weight because of air resistance issues, and, second, steel-on-steel has lower rolling friction than rubber-on-asphalt.
What you’re describing in your second paragraph is basically PRT. It’s a fun concept to toy with – I even wrote science fiction once in which a few cities have it (they also have flying cars and FTL travel) – but it’s not that useful for real-world urban design. In any environment in which space costs money, it’s cheaper to build transit than to go with automated car systems.
Well, I’ll reiterate my personal desire for an open spreadsheet where people can both work with the source numbers, and various interpretation of those numbers (e.g., the environmental impact of BTUs attained through different fuels). I’ve seen lots of different interpretations of these specific numbers, so at least understanding how people came to interpret them differently would be very useful. (I started a spreadsheet, but wasn’t able to replicate your numbers and the whole thing quickly became quite murky.)
“Japan is only slightly better in terms of percentage of power produced, but the average Japanese citizen still puts out around half as much carbon dioxide each year as the average American.”
I agree to this, but still we shouldn’t forget that we still need to minimize co2 each year. Fighting environmental issues like this is not easy. But if we unite for one advocacy, then its not impossible to stop these issues.
Levy contents that personal rapid transit is not that useful for urban design. Quite the contrary, it is far more efficient in “footprint space.” PRT is inherently more efficient in the weight/passenger issue. Is there a value to “waiting time?” Have you actually calculated the number of passengers which can be transported in a 30 mph PRT car as against freeway travel including on-off ramps, parking times, etc. Take another careful look at personal rapid transit. Not only is it far more efficient in btu/passenger/mile, but IT CAN OPERATE ESSENTIALLY WITHOUT SUBSIDY. Does the public even know what the average passenger subsidy is per ride? Do we know what amortization costs are? How are these calculated?
Your ideas are intriguing to me and I wish to subscribe to your newsletter.