As has been mentioned by many rail bloggers, FRA regulations needlessly add extra weight to American trains. As basic physics would tell us, unless the power to weight ratio is maintained, the heavier train will accelerate far slower. A slower acceleration means a longer period of high power demand and so it is unsurprising that fuel and energy consumption will be rather higher for American trains compared to European or Japanese trains. The degree to which it is higher, however, means that hundreds of millions of dollars are wasted every year by Amtrak and the various commuter rail agencies on additional and unneeded fuel expenses.
Amtrak's 2011 Annual Report shows the following consumption figures on page 36.
Seat-miles per gallon of diesel fuel: 143.7
Seat-miles per kWh of electric traction: 6.85 (.09 kWh per seat-km)
The diesel figure isn't terribly useful for purposes of international comparison; nearly half of all diesel miles are from long distance trains which, with the exception of the all-coach Palmetto, have a rather poor seating density due to sleepers, diners, crew dorms, and baggage cars, although the Palmetto doesn't have a terribly large passenger capacity itself.
Amtrak's monthly reports (such as this April 2012 one) indicate that on a train-mile basis, they average 2.3 gallons of diesel per mile. Combined with the seat-miles figure, our average Amtrak train should have about 338 seats. Studies in Minnesota have shown similar figures, at 2.42 gallons per train-mile, but others have greatly differing figures. A study on restoring passenger rail service between Los Angeles and Las Vegas showed approximately 1.66 gallons per train-mile using two locomotives and four Surfliner cars (Appendix 6A) while the MBTA reports their average fuel consumption is 2.8 gallons with 6 cars and on some lines, that can rise as high as 3.29 gallons per train mile (with their lowest consumption at 2.23 gallons).
It's a rather stark contrast with the fuel and energy efficiency that we see from lighter British train sets, data on which can be found in these two reports.
With diesel trains, Amtrak might well be contrasted to the IC125, a 125mph capable diesel train set with an engine on either end and seven to eight cars in between. Where Amtrak has a figure of 2.3 gallons per mile (and the Vegas Surfliner 1.66), the IC125 with eight cars and 617 seats has a measured fuel consumption equal to 1.78 gallons per mile (346 seat-miles per gallon). While that's a touch higher than the Vegas Surfliner study, it's with nearly twice as many seats and at significantly higher speeds. A 9-car Class 222 Meridian diesel multiple unit, which can reach 110mph in under 3 minutes, burns 2 gallons per mile for about 239 seat-miles per gallon (by contrast, the notional Vegas Surfliner is ~216 seat miles per gallon).
MBTA's Fitchburg Line, which holds the dubious record of being their single most fuel inefficient line, presents a special case. With 17 stations in 50 miles, it really is more suitable for light rail operations than diesel push-pull commuter trains and I can't find a British DMU with similar station spacings (although there are several EMUs with similar stopping patterns). Capital MetroRail and Westside Express, however, both have similar stopping patterns using DMUs; SMART quotes them as having fuel consumption of 0.57 and 0.65 gallons per mile respectively (page 15). With the Capital MetroRail's fuel consumption and $3.25 per gallon, MBTA could save $4 million annual (with 2010 schedule; with longer station spacings, some British DMUs have even lower fuel consumption however, the Class 150 is 0.517 with 3 power cars and 16 km station spacing as an example). In all honesty, such a line needs to be electrified and equipped with EMUs with even greater fuel efficiency.
An interesting detail emerges looking at British and international electric trains and multiple units and that's that there's a fairly consistent trend of using 0.03 kWh per seat-kilometer for trains under 125mph. With electric trains dominated by corridors and Acela rather than long distance trains, it's rather hard to understand why Amtrak uses three times the electricity of any comparable train and indeed averages 33% higher energy use than European high speed trains (Japanese high speed trains have energy consumption figures comparable to sub-125mph European trains on a seat-km basis). Since single level coaches do not suffer terribly much, if at all, in terms of weight penalty due to FRA regulations, my suspicion is that the Acela, a horribly overweight train set, is also extremely energy inefficient. With its low number of seats, the lowest number of any high speed train in the world, this will have an outsized effect upon Amtrak's efficiency. Somewhat astonishingly, Amtrak's average energy intensity is worse than that of the old Metroliners which, at 7.81 kWh per car-mile, averaged out at 0.07 kWh per seat-kilometer. Fortunately, electricity is cheap, unlike diesel fuel, so Amtrak's energy expenses are not ruinous, but this still represents tens of millions of wasted dollars.
I should note that Amtrak's electricity consumption numbers are sufficiently bizarre that it is entirely plausible that they are completely wrong (they do, for instance, proudly state a rate of 50 kWh/seat-mile in another document). One possible answer is that they decided to account for inefficiencies in generation and transmission of the electricity. Alternatively, 50 kWh per seat-mile was a typo and the number meant was actually a train-mile figure. Combing the two sets of figures would give us 342 seats, a small figure for the Northeast to my knowledge. An old study indicates that an Amfleet of such length (about 6 cars assuming one business class and one cafe) would only use 25.8 kWh/train-mile at 120mph top speed (page 25) which does put us in the unfortunate position of assuming absolutely hellacious energy use by the Acela.