mailto:firstname.lastname@example.orgMore transportation articles by David Lawyer
Copyright 2002-7 by David S. Lawyer. Feel free to make copies but commercial use of it is prohibited. For example, you can't (except to an insignificant degree) combine it with advertising on the Internet. Please let me know of any errors or suggestions for improvement.
Between 1936 and 1963 the Interstate Commerce Commission (ICC) collected detailed statistics from all class I railroads in the United States related to energy efficiency. See Appendix: ICC Data Sources. From this information one may estimate the energy-efficiency (fuel economy) of various types of passenger trains (diesel, steam, electric, etc.). One key statistical publication which permits this is entitled Fuel and Power ... but unfortunately, it ceased publication in 1963.
To the author's knowledge, no one else has published such estimates for this time period. For coal-powered trains prior to 1940 see Hultgren: Steam Train Pass-Mi/Ton-Coal, but it's low by about 20%. Obtaining my estimates was not straightforward since the raw data needed to exactly derive pass-mi/gal (passenger-miles per gallon) is simply not available. But by making some approximations and assumptions, the pass-mi/gal can be estimated for various types of train power: diesel, steam, and electric. Old versions of this article cover only diesel locomotive trains but steam coverage was added starting in late 2002.
Note that most of the links are just to notes within this document since the documents cited are, unfortunately, not on the Internet. If you just want to jump now to the results see: Diesel Locomotive Train Results and Steam Locomotive Train Results
There are two major problems in attempting to estimate the energy-efficiency (in pass-mi/gal) of various types of passenger trains. One problem is that passenger trains hauled a great deal of "freight" such as mail and express parcels. As much as 30-40% of the energy used by passenger trains needs to be allocated to the transportation of freight and not allocated to passenger transportation. How can this be done? A second problem is that reported pass-mi (passenger-miles) were not differentiated by fuel. For example, we simply don't know the pass-mi on say trains powered by diesel locomotives.
The period from 1936 to 1963 was that of transition from steam to diesel power. Diesel trains consumed diesel fuel while steam trains used either coal or fuel oil. Some lines were electrically powered, mainly in the Northeast Corridor (Washington DC to New York, etc.) and for several hundred miles in the Far West from Montana to the Pacific (with a gap). See Rail Electrification. Trains could either be pulled by locomotives or have motors under the passenger railcars. The later were called "motor-car trains" and were either electric (like rapid transit) or diesel, but a few were gasoline powered. Thus there were various types of trains being operated.
Between 1936-1963 fuel and power were reported by train type (such as electric motor-car trains, diesel locomotive trains, steam locomotive trains using fuel oil, etc.). But the pass-mi (passenger-miles) were not reported by train type. So fuel economy couldn't be "determined" for each type of train since one doesn't know the pass-mi for each type of train. But it's possible to roughly estimate the pass-mi for each type of train as will be explained later on.
Passenger trains carried both passengers and "freight", with much of the "freight" being mail and not normally called "freight". In a few cases, freight cars were even put into passenger trains. Some rail cars (called "combination passenger cars" were designed to be shared by both freight and passengers. But most of the freight was carried by passenger trains in separate baggage, express, and mail cars. Express cars transported parcels in boxes, etc. for the Railway Express Agency, etc. Separate mail trains were run for just carrying mail but they were classified as passenger trains. Their fuel statistics, etc. were not reported separately. Thus many "passenger trains" didn't carry any passengers.
A major problem in these fuel statistics is how to allocate fuel use between passengers and the freight (including the mail and express) carried by passenger trains. One can't neglect this since about 30% to 40% of passenger train energy was used to transport such "freight".
The ICC started collecting data on rail fuel consumption in 1889 but since they lumped fuel used for both freight and passenger trains together, there was no way to estimate passenger fuel efficiency ( Bukovsky table 3, p.6). The same comment applies to the US Census of 1880.
Then starting in 1918, fuel consumption was reported separately for passenger trains ( Bukovsky table D2, p.100.). But the output in passenger-miles or in car-miles was not differentiated by the type of fuel. Regarding electricity for electric trains, Bukovsky says "Data not available or not tabulated" (prior to 1936). But the ICC has this information in their annual reports starting in 1927.
Then starting in 1936 (and lasting thru 1963), the ICC reported both fuel consumption and car-mi by train type. The car-mi data is found in the ICC statement: Passenger Train Performance while the fuel (and electricity) is in: Fuel and Power .... As a result, for each type of train, one could now simply divide the fuel used by the car-mi to get, for example, gal/car-mi for diesel locomotive type trains. One could also find the kwh/car-mi (kilowatt-hours per car mile) for the two types of electric trains (motor-car or locomotive hauled). In fact, the ICC made these calculations and reported the results in Fuel and Power ...
While car-miles were differentiated by the type of train, they were not further subdivided by the type of car (such as sleeping car, mail car, etc.) If this had been done, such data could have been used to roughly estimate the energy used by sleeping cars, mail cars, etc. by assuming that energy use is proportional to car-mi.
However, car-mi statistics by type of car were kept, but this wasn't differentiated by fuel. For example, we have the car-mi of passenger coaches for for all locomotive trains regardless of whether the locomotive was steam, electric, or diesel. Unfortunately, the number of categories of car types were small and various types were lumped together. For example, an "other" car-mi statistic is reported for the combined category of "mail, express, and baggage cars, and combination cars other than passenger". There is, unfortunately, no separate statistic for say car-mi of baggage cars.
In spite of the problems previously mentioned there is a way to roughly estimate the energy efficiency of various types of trains. From 1936 on, the ICC was reporting the energy use per car-mi for each type of train. If we only knew the average number of passengers per train-car (pass-mi/car-mi) for each type of train then we could divide it by gal/car-mi to get pass-mi/gal. Problem solved!
The ICC did have overall statistics for passenger-miles per passenger-car-mile (the average number of passengers per car). See Revenue Traffic Statistics. But they didn't have it by type of train. But as a rough estimate, one can just assume that the average number of passengers per car was the same for all types of trains. But not so fast! The number of passenger per car as reported by the ICC was not really the number per car as is explained in the next section.
Since the passengers-per-car statistic only counted real passenger cars (with sold seats for revenue passengers), the problem regarding cars used for "freight" is seemingly solved. But there's still another problem.
The pass-mi per car-mi as reported by the ICC is only for strictly passenger cars and excludes the car-mi of club, lounge, dining, observation, and baggage cars. Most of the energy used to haul these types or rail cars should be allocated to the transportation of passengers. One way to do this is to estimate the car-mi of such cars and allocate them to passenger transportation.
Fortunately, some statistics were kept related to the car-mi of such cars (see Passenger Train Performance). Unfortunately, baggage cars were included in the "other cars" category. This "other" category is mostly mail and express cars used for freight. Another problem is that part of the space in the baggage cars was sometimes used to transport freight. Now if we could somehow estimate the baggage car-mi and multiply that by the percentage of the baggage car utilized for carrying ordinary passenger baggage.
To estimate baggage car-mi I assumed that only 1/3 of a baggage car was used per train to transport the ordinary checked baggage of the passengers. But perhaps a third of the train-miles had no baggage cars at all since they were mail or commuter trains. So assuming 1/3 of a baggage car per train is like assuming that the typical passenger train that had baggage cars, used half of one baggage car to transport the luggage of the passengers. Excess baggage, for which an additional fee was charged, is considered to be "freight transportation" and is not allocated to passenger transportation. Baggage cars were sometimes also used for mail and freight in addition to the baggage of the passenger. (Prior to version 1.6, I assumed 1/2 of a baggage car, instead of the current 1/3, for the passenger luggage transport.)
There is still another possible problem. Does moving a mail, express, or baggage car (all used for transporting freight) use the same energy per car-mile as a passenger car? While such "freight" cars may be heavier and thus use more energy per mile, there are two reasons why they might not use as much fuel as expected. One possible reason is that mail trains may have made fewer stops (especially when compared to commuter trains). They also likely used less energy for heating, lighting and air-conditioning.
Thus, based on all the above discussion there is a crude way to estimate the passenger miles per gallon. Exactly how this is done will be illustrated by the following example for diesel locomotive trains.
The pass-mi/gal is obtained by multiplying together the following three statistics: 1. car-mi/gal 2. pass-mi/car-mi 3. factor F. F is to account for cars in the train which serve passengers but were not counted by pass-mi/car-mi. 1. and 2. above are respectively from the ICC statements Fuel and Power ... and Revenue Traffic Statistics.
F is about 0.8 but varies from year to year. It's calculated from the car-mi/train-mi statistics (see Passenger Train Performance) for various types of passenger cars. Here's an example of calculating F for 1936: The ICC reports 2.08 car-mi/train-mi for passenger coach cars. This means that the typical train had 2.08 coaches in it. Here "typical" means what observers would observe if stationed along the track so that long-distance trains would thus count for more than short-distance ones. The observer would also see mail trains with no coaches so the typical train with passengers would actually have more than 2.08 coaches.
For 1936, per the ICC there were also 2.30 sleeping and parlor cars in the "typical train". This gives 4.38 (2.08 + 2.30) cars on which the car-mi (in the ICC pass-mi/car-mi statistic) is based on. But the number of cars/train used by passengers is larger than 4.38 since there were total of 0.62 club, lounge, dining, and observation cars in the "typical train". There is also assumed to be 0.33 baggage cars resulting in a grand total of 5.33 ( = 4.38 + 0.62 +0.33 ) cars which are earmarked for passenger use. Now we would like to convert the pass-mi/car-mi figure to include the car-mi of club, lounge, dining, observation and baggage cars. How do we do this? Well, we multiply the car-mi for 1936 (used in the pass-mi/car-mi statistic) by 5.33/4.38, or what amounts to the same thing, we multiply pass-mi/car-mi by the inverse of this: 4.38/5.33 = 0.821. This is the just the factor F mentioned above.
Thus for 1936 we multiply 0.821 by 13.33 pass-mi/car-mi to get 10.94 pass-mi/car-mi and then divide by 0.193 gal/car-mi to get 56.7 passenger-miles/gallon (diesel). Then by multiplying by 0.901 we get 51.1 passenger-miles/gallon (gasoline), since a gallon of diesel fuel contains more energy than a gallon of gasoline.
Here are the results along with more recent data reported for Amtrak (includes electric trains). PM = Passenger-miles. gal = gallons of gasoline equivalent. The energy-efficiency is shown as PM/gal (Passenger-Miles per gallon of gasoline equivalent) while the energy-intensity is shown as BTU/PM (British Thermal Units per Passenger-Mile).
Diesel locomotive trains| Amtrak, both diesel and electric trains per David Lawyer | per Transportation Energy Data Book PM/gal BTU/PM | PM/gal BTU/PM 1936: 51 2,440 | 1975: 34 3,677 1940: 41 3,080 | 1980: 39 3,176 1945: 85 1,480 | 1985: 45 2,800 1950: 42 2,950 | 1990: 48 2,609 1955: 41 3,010 | 1995: 48 2,590 1960: 41 3,0l0 | 2000: 38 3,253 1963: 40 3,160 | 2005: 45 2,784 | 2002 4,137 Reported by Amtrak but Retracted | 2003 4,830 " " " " "
Regarding Amtrak data see Amtrak vs. Auto. The high value of 85 pass-mi/gal during the last year of World War II (1945) is due mainly to gasoline rationing but also is due in part to the curtailment of automobile manufacturing during the war. Even for the non-war years prior to 1963, rail was significantly more energy efficient than the automobile. The "Highway Statistics" books (by the US Federal Highway Administration) estimates that autos got about 15 miles/gallon back then. So even with 2 person per auto (30 pass-mi per gallon) the railroads did somewhat better.
But today (2000) the automobile is a lot more energy efficient than it used to be. In intercity use it gets about 25 mi/gal and has about 2 people in it, resulting in about 50 pass-mi/gal (gasoline). Thus today's autos are more energy-efficient than diesel trains of the mid 20th century (except for World War II years). Amtrak's energy-efficiency today is a little less than the auto. See Amtrak vs. Auto.
For 2002-3 Amtrak reported sharp drop in energy-efficiency but later retracted it. What is going on? Amtrak did introduce heavy and overpowered electric trains and the reported increase was for electric energy. More investigation of what actually happened is needed.
Why did energy-efficiency (pass-mi/gallon) drop from 51 pass-mi/gal to 41 from 1936 to 1940? It's not due to fewer passengers per car but due to an increase in gal/car-mi from 0.19 in 1936 to 0.25 in 1940. The reasons seem to include the introduction of air-conditioning and the introduction of diesel service to trains that made more frequent stops and thus used more fuel in accelerating. In 1936 only 0.35% of car-mi was by diesel locomotives (the rest was mostly by steam locomotives but about 5% was electric). So the figure for 1936 represents mainly long-distance express trains that made few stops. By 1940 5% of car-mi was by diesel. During World War II it was about 8% but by 1950 it reached 55%. See Bituminous Coal Annual
As time went on the gal/car-mi continued to increase, reaching .35 by 1963. This was partly due to the discontinuance of long distance trains which made few stops resulting in greater significance of suburban trains which used more energy by making many stops. However, this led to a relative decline in sleeping car travel resulting in more passengers per car which in turn compensated for the worse fuel economy per car-mile.
During World War II the reported average number of passengers per rail-car more than doubled from about 14 during the 1930's to over 30 during the war. At the same time, due to longer trains there were fewer dining cars, etc. per passenger. With gasoline rationing, people flocked to the railroads and many were turned back for lack of space. People are unlikely to tolerate such train crowding in peacetime. The wartime record does show that under conditions of long trains full of passengers, rail transportation can be very energy-efficient (provided it doesn't go too fast since aerodynamic drag increases with the square of the velocity).
Between 1936 and 1963, the train appears to have been somewhat more energy efficient than the automobile, assuming 15 miles/gallon and 2.2 passengers per auto (33 pass-mi/gal). One could make the claim that rail was even relatively more efficient than this: If train passengers went by auto, there would be less than 2.2 passengers per auto (since the 2.2 figure includes the presence of family and friends who were less likely to accompany someone on a train trip). In the same vein, one can argue that if someone didn't go by train, there's a good chance they would go by bus which was much more energy efficient than the train. So just how much (if any) fuel was saved by travel by train is not at all clear.
Automobiles today get better mileage and are somewhat more energy efficient than the trains were then (except during the war years of 1942-1945 the train was outstanding in energy efficiency). There are no satisfactory estimates of the average automobile occupancy (assumed to be 2.2) for 1936-1963 (for trips similar to rail trips). Data on automobile occupancy reported in the "Nationwide Personal Transportation Study" for 1970 was in error.
Another consideration is that sleeping car use was substantial since car-mi of sleeping cars was roughly equal to the car-mi of coaches during the 1930s. Thus, one could argue that the extra energy for "hotel services" (heating, lighting, and air-conditioning) for sleeping cars should not be counted. If a person were traveling via automobile and stayed in a hotel (or motel), "hotel service" energy would be used for the overnight stop (which is not counted by the pass-mi per gal by auto). Taking this into account would make rail even more energy-efficient as compared to the automobile.
Using the same methodology one can estimate the passenger-miles per gallon for steam engine trains powered by coal (and fuel oil) by using the fact that the heat value of one gallon of gasoline is about the same as the heat value of 9.5 pounds of coal (per US Bureau of Mines, 1950). The F value was assumed to be 0.8 (due to lack of time to obtain the actual data). The data on car-mi per ton of coal (or coal equivalent) was obtained from Bituminous Coal Annual. See also Equivalent Fuel Dilemma
Passenger-miles/gallon (gasoline equivalent) for steam engine trains 1920: 8.0 1930: 5.9 1940: 7.0 1950: 8.1 1925: 7.0 1935: 5.4 1945: 15.5
The low values during the depression years of 1930 and 1935 are due to poor ridership (low numbers of passengers per car) and not due to any decrease in locomotive efficiency. In fact Bukovsky on p.72 shows that in 1916-18 trains consumed about 19 lb. of coal per car-mi which decreased to 15 lb. per car-mi by 1929 and stayed close to 15 thru 1943 (last date reported).
Comparing these steam train results the results for diesel trains and automobiles shows that diesel trains were 5 to 6 times more energy efficient than steam trains. The automobile was (except for the World War II years) about 3 to 5 times more energy efficient than steam passenger trains.
In 1936, about 20% of the fuel (in terms of heat value) for steam locomotives was fuel oil. Converting fuel oil and coal to BTU (heat value) shows that for 1936, fuel oil only used 169,000 BTU per car-mi while coal used 196,000 BTU per car-mi. Thus the fuel-oil-powered steam engines were more energy efficient, likely due to being newer. It also shows that to determine the energy-efficiency of steam-powered trains, the coal equivalent of fuel oil should be correctly determined. But unfortunately, the ICC didn't handle this well. See Equivalent Fuel Dilemma
In conclusion, between 1936 to 1963, it seems that overall, railroad transportation in diesel powered trains was about 20% to 150% more energy-efficient than the automobile in peacetime. The high figure assumes a lower automobile occupancy since people traveling by train were more likely to travel alone. In wartime with gasoline rationing, it was roughly 2 or 3 times as efficient due to very high load factors (percentage of seats filled). Due to energy-efficiency improvements in automobile since then, the passenger train today seems to have only slightly better energy efficiency than the automobile. See Amtrak vs. Auto
Diesel trains were thus several times more energy efficient than the steam engine trains they replaced.
In 1973, a study of 3 commuter rail lines in Chicago (all diesel powered) came up with a figure equivalent to 54.5 passenger-miles/gallon (gasoline). Comparing this with the typical figure of 41 pass-mi/gal for all trains (in this article) suggests that some commuter trains are more energy efficient than the typical train of 1936-1963. Such commuter trains have a large number of seats per car (bi-level) and don't have to haul dining cars, baggage cars, and low-capacity sleeping cars. This results in more passenger per car. But commuter trains tend to use more energy per car-mile due to frequent stops. However, the commuter trains of 1936-1963 typically had fewer seats and thus likely didn't obtain such high pass-mi/gal. See Commuter Trains
Note that the ICC figures for gal/car-mi were only given to 2 significant figures for most years (such as 0.25 gal/car-mi in 1940). A project for a future version of this document is to obtain 3 significant figures for all the years 1936-1963 and redo the calculations.
The assumption of 1/3 baggage cars per train needs to be further refined. As time passed, the number of long-distance trains declined much faster than the number of commuter trains. Commuter trains don't use baggage cars so the estimated baggage car space used by passengers should decrease over time.
Of course there's a lot more to historical rail energy efficiency than only covering diesel locomotive trains from 1936 to 1963 and steam from 1920 to 1945. Future projects are to cover more years for steam, calculate the F values for every year, and perhaps calculate separate results for coal and fuel oil. Using the same ICC sources and methodology, one could estimate energy efficiency for other types of motive power between 1936 and 1963. Since passenger-miles was differentiated by coach vs. sleeping/parlor cars, one could obtain separate estimates for coach travel vs. low-occupancy sleeping/parlor cars. This would show that coach travel was much more energy efficient than sleeping car travel.
While motorcar-train fuel consumption was also reported by the ICC, a major difficulty would be estimating the number of passengers per motorcar.
The following sections contain references to the Interstate Commerce Commission (ICC) "statistics" used. Such data was reported by the ICC on a monthly basis for each Class I railroad. However, the statistics used here are the summary ones that were reported for each calendar year and sum the data for all railroads in the US. The ICC statements for each year also show the previous year's data. However, the data shown for the previous year was often corrected and thus doesn't exactly agree with what was previously shown a year earlier. So it's more accurate to use the revised data shown for the previous year.
There were some other publications that copied the ICC data and often presented such data in a time series covering a number of years. Using these secondary sources is more convenient, since fewer publications are needed to extract the data. See Secondary Sources of ICC Data. Few libraries have either the primary or secondary source data.
Note that the ICC each year issued a long "Transport Statistics in the United States, Part 1: Railroads" which is a few hundred pages in long. But the data needed for fuel efficiency is not to be found there.
The ICC publications used for fuel efficiency changed names between 1936 and 1963. In 1936 they were issued by the ICC's "Bureau of Statistics". In 1963 it was "Bureau of Transport Economics and Statistics". In all cases "Switching and Terminal Companies Not Included" appears after the title (in smaller print). Below find the starting name of the publication in 1936 and the ending name in 1963.
1936: "Fuel and Power for Locomotives and Rail Motor Cars of
Class I Steam Railways in the United States" (Statement M-230)
1963: "Fuel and Power Statistics of Class I Railroads in the United States" (Statement Q-230)
Statistic Used: "9. Fuel and power consumed per passenger-train car-mile --road passenger service: 9-03. Gallons of diesel fuel (diesel locomotives)"
For 1925-1935 it lumped statistics for freight and passenger service together and thus is of little value for fuel efficiency use. Ceased publication after 1963. For an overview see: Primary ICC Data Sources
1936: "Revenue Traffic Statistics of Class I Steam Railways in
the United States" (Statement M-220)
1963: "Revenue Traffic Statistics of Class I Line-haul Railroads in the United States" (Statement Q-220)
Statistic Used: "Passenger-miles per car-mile"
For an overview see: Primary ICC Data Sources
1936: "Passenger Train Performance of Class I Steam Railways in
the United States" (Statement M-213)
1963: "Passenger Train Performance of Class I Railroads in the United States" (Statement Q-213)
Statistics Used: "10. Car-miles per train-mile in locomotive-propelled trains: 10-01. Passenger coaches, 10-02. Sleeping and parlor cars, 10-03. Club, lounge, dining, and observation cars, 10-04. Other passenger-train cars"
For an overview see: Primary ICC Data Sources
The name "Railways" was changed to "Railroads" around 1960. It was published by the Association of American Railroads, Washington DC and reports ICC data. Each annual issue covers about a ten-year period. Since there is overlap, one doesn't need all the issues. It covers the Revenue Traffic Statistics and Passenger Train Performance (except that car-mi is not differentiated between locomotive and motor-car trains). This makes it not suitable to estimate Pass-Mi/Gal for locomotive trains (or for motorcar trains). It lacks the data contained in Fuel and Power ....
In July 1944, the ICC released the mimeographed report: "Use and Cost of Railway Fuel and Problems in Fuel Statistics" ICC Statement No. 4428. (at the University of Michigan: TJ648B93). The author is Alexis P. Bukovsky who was perhaps overly honest in documenting various inaccuracies in fuel statistics since the report contains the disclaimer: "This study issued as information, has not been considered or adopted by the Interstate Commerce Commission". It's much more than just a secondary source of ICC statistics on fuel.
The division of passenger train car-miles by fuel for the years 1936 thru 1950 may be found in: Bituminous Coal Annual, 1951, p. 115 (table 49). Table 48 is for freight. Unfortunately, as the use of coal by railroads decreased, later annual issues truncated their coverage. The 1953 issue shows data only for 1952. After that issue, coverage disappeared entirely. Table 51 (in the 1949 issue) shows the car-mi per ton of coal equivalent but it's missing in later issues. The last year shown, 1948 is 127 car-mi/ton-coal. For 1950 I assumed 125 car-mi/ton-coal.
Using ICC-reported amount of "coal equivalent" isn't always too accurate for estimating steam train fuel economy since the coal equivalent included diesel fuel in later years. But prior to 1940 it may be utilized with little error since steam power was dominant then. Even after 1940, using it (until "equivalent" data was discontinued after 1948) doesn't result in excessive error if it's used to estimate energy efficiency of steam trains. Here's why.
Railroads were told to convert non-coal fuel to equivalent coal fuel per their own "experience" and often didn't equate fuel on a BTU basis. Instead they would equate fuel on the basis of how much mechanical work they could get out of it. If all railroads did this, it would show that diesel trains were no more energy-efficient than coal. Bukovsky discusses this problem in Ch. III. Thus using equivalent coal to calculate steam engine fuel economy does not result in large errors (but figures after 1940 may be in error by a few percent).
When fuel oil used in steam locomotives was equated to coal, relatively little error resulted by basing the equating on the mechanical work it could do. A BTU of heat from coal (used to heat water) would be produce roughly the same amount of mechanical work as a BTU of heat from fuel oil used to heat water. Some error was introduced since the fuel-oil locomotives tended to be newer and thus more efficient than the typical steam locomotive.
Although Bukovsky implies that the ICC may have started the doing the conversions to equivalent fuel themselves (using BTU equivalents) instead of letting the railroads do it, this doesn't seem to be the case as a check of 1936 and 1948 data indicated. For 1936, the equivalent coal reported is close to the heat value of the actual fuel used, but there is about a 10% discrepancy in 1948 when diesel fuel was more widely used.
This is a magazine article: "A Statistical Review of Dieselization of the American Railroads" by George W. Grupp in "Diesel Power and Diesel Transportation" February 1944, pp. 153-4. He reports the fuel consumed per passenger-train car-mile for only the 3 years: 1940-42 but shows it for a large number of different railroads. No clue is given as to why there is wide variation in this statistic among the various railroads.
The 1948 book: "American Transportation in Prosperity and Depression" by Thor Hultgren is Number 3 in the series: "National Bureau of Economic Research, Studies in Business Cycles". In chapter 8, "Fuel Economy", p. 223, chart 94, it shows passenger-miles per ton of equivalent coal for passenger trains. Dates are from 1920 to 1940. It seems to have neglected the fact that a substantial part of the passenger train fuel was used to transport freight on passenger trains. The chart is apparently for all passenger trains and claims to include "fuel used by switching locomotives, in stationary power plants, for heating buildings, etc". My research so far neglects such other uses of fuel (and perhaps rightly so).
If Hultgren had just neglected the freight hauled in passenger trains, the results (in pass-mi/gal-equivalent) would have been about 30% lower than mine (since roughly 30% of train energy was used to move such freight during those years). Counting other uses of fuel would make it even more than 30% lower. But it's roughly only 20% lower than my results. The reasons for this discrepancy are not known.
In order to convert fuel consumption to gallons of gasoline, the heat value of fuels must be known or assumed. Since the heat value of fuels changes over time as the specifications of the fuel vary, it seems best to use the heat values commonly used (by knowledgeable people) during the bygone years in question. Except that for gasoline, the comparison should ideally use both the old value and the current heat value so that people may compare it to their personal experience with gasoline automobile fuel economy.
The heat value of gasoline (as purchased at gasoline stations) has decreased a little due to mixing it with alcohol, etc. In this report, I've used the old standard of 125k BTU/gallon (US Bureau of Mines, 1971-2). This is the high heat of combustion. Today (2003) Chevron says 114k BTU/gallon but this is the low heat of combustion. See Fuel Economy of Gasoline Vehicles. The seeming discrepancy between 114k BTU/gallon and 125k BTU/gallon mainly reflects different ways of measuring BTU as is explained next.
There exists two different values of BTU/gallon for the same batch of gasoline. One is the high (or gross) heat of combustion and the other is the low (or net) heat of combustion. The high value is obtained when, after the combustion, the water in the "exhaust" is in liquid form. For the low value, the "exhaust" has all the water in vapor form (steam). Since water vapor gives up heat energy when it changes from vapor to liquid, the high value is larger since it includes the latent heat of vaporization of water. The difference between the high and low values is significant, about 8 or 9 percent. This accounts for most of the apparent decrease in the heat value of gasoline. See Appendix B, Trans. Energy Data Book. In the U.S. the high heat values have traditionally been used, but in many other countries, the low heat values are commonly used.
For gasoline I've used the old value of 125k BTU/gallon. For diesel I use 138.7k BTU/gallon. Both are per the former US Bureau of Mines (1971-2) and are high heats of combustion. So to covert from diesel fuel to gasoline, it takes about 10% more gallons of gasoline to produce the same heat. This remains true even if one had used the low heats of combustion for both diesel and gasoline.
For coal, I equated one gallon of gasoline to 10 pounds of coal, assuming 12,500 BTU/pound for coal, and 125,000 BTU/gallon for gasoline. In 1950, the US Bureau of Mines adapted 13,100 BTU/pound for coal but today reports 12,000 to 12,500 BTU/pound. The U.S. Dept. of Energy claims that typical coal today (2000) is only 12,000 BTU/pound. What was the BTU/pound of railroad coal? In footnote 33, p. 184 of "Energy in the American Economy ..." by Sam H. Schurr, et. al., John Hopkins Press, Baltimore, 1960, it mentions that electric utilities used coal of a lower heat content (12,263 BTU/pound). So railroad coal was perhaps at least of average heat value. The 13.1k BTU/pound value is mentioned in this footnote 33 as well as on p. 317 of "Energy Sources ..." by Eugene Ayres et. al, McGraw-Hill, New York, 1952. Some railroads used anthracite coal which Ayres (p. 317) shows at 12,700 BTU/pound.
For the energy efficiencies see: "Amtrak and Energy Conservation in Intercity Passenger Transportation" by Stephen J. Thompson (Congressional Research Service, Report to Congress) Sept. 3, 1996. (http://www.cnie.org.nle/eng-11.html) See also "Transportation Energy Data Book" (annual) Oak Ridge National Laboratory, tables on energy intensity of passenger modes. See ORNL Transportation Energy and Click on "Chapter 2".
This shows a steep drop in Amtrak energy-efficiency for 2000-2001. In 2000 Amtrak reported a 25% increase in diesel fuel consumption and in 2001 reported a 74% increase in electricity consumption. Yet car-mi and train-mi only increased by 7-8% from 1999 to 2001 which fails to explain the sharp increase in reported energy consumption. However, the high-powered and heavy Acela trainsets were introduced in large numbers in 2001. This tends to explain the increase in electricity consumption. See Transportation Energy Data Book (TEDB), ed. 23, tables 9.12 and A.16. See also Acela Express
Table 2.11 (of TEDB above) shows for 2001, that Amtrak is less energy-efficient than the automobile-SUV (take a weighted average of BTU/pass-mi for autos and "personal trucks"). Prior to 2000, Amtrak is reported (by TEDB) as being about 33% more energy-efficient that the auto-SUV.
But the automobile figure is for both city and intercity driving. To fairly compare it to Amtrak, one must correct the auto figure so it reflects intercity use. For intercity, there are about 2 persons/auto as compared to 1.6 overall, and the auto gets more miles/gallon in intercity use (about 26 as compared to 22 overall). After making this correction, Amtrak is appears to be little more energy efficient that the auto prior to 2000 and significantly less energy efficient that the auto for 2001.
See also CRS Report: 96-22 - Amtrak and Energy Conservation: Background ... This 1996 update by the Congressional Research Service, agrees with the above conclusion that Amtrak energy-efficiency is about the same as the auto. The Amtrak energy intensity they used from "Transportation Energy Data Book" may have been too low due to the failure to consider that each BTU of electricity takes a few BTUs of fuel to generate. This resulted in the energy-efficiency of Amtrak's electric trains being overstated. But while the data may be flawed, the conclusions of the CRS report turn out to be the same as mine.
See vol. 4: Report on Agencies of Transportation (see Table IV pp. 226-245 and Table D pp. 574-581).
A study of diesel commuter train energy efficiency in 1973 was "Per Passenger-Mile Energy Consumption and Costs for Suburban Commuter Service Diesel Trains" by Edward W. Walbridge, August 1974, report no. IL-11-0006-1 for the Urban Mass Transportation Association of the US Department of Transportation. It studied three commuter railroad operations in the Chicago area: Burlington Northern, Chicago and North Western, and the "Chicago Milwaukee St. Paul and Pacific". On p. 29 it shows an average of 2300 BTU/pass-mi which is equivalent to 54.5 pass-mi/gal (gasoline). This is a little lower than the 2760 BTU/pass-mi reported by Transportation Energy Data Book for 2000.
See the article: "Electrification Returns to the Railroads" by Gordon D. Friedlander, Electrical World, Dec. 1982, pp.65-76.