mailto:davylawyer1@gmail.com
More articles by D. LawyerCopyright 2003 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.
The author is an ardent environmentalist and is even more opposed to the automobile than mass transit. The purpose of this article is to inform the public why mass transit, as currently implemented and utilized, unfortunately doesn't save energy.
Ever since the energy crisis of the early 1970's, the myth that mass transit saves energy and reduces pollution has been widespread. Tens of billions of dollars of public funds have been spent (and mostly wasted) on subsidizing mass transit. An objective appraisal shows that mass transit today is little (if any) better than the automobile as far as energy use and pollution are concerned.
It wasn't always this way. At the start of the 1970's mass transit was about twice as energy efficient than the auto. Since then, the energy efficiency of the automobile has shot up by over 50% while the energy efficiency of mass transit has declined by about a third. (This is a 2-fold change since 1.5 (a 50% increase) divided by 2/3 (a 1/3 decrease) equals 2.)
Thus one might make the dubious conclusion that mass transit did actually have significant energy saving advantages 25 years ago but due to changes in technology since then, these supposed advantages are now a myth. However, even 25 years ago it was not clear that mass transit would save energy and reduce pollution. See my article: "Warnings on Mass Transit," Environment, Sept. 1975. Such "warnings" were few in number and mostly ignored.
When a new rail system is built or more buses added, road traffic congestion is lessened. This reduced congestion encourages more people to take more and longer trips by automobile and to live farther away from where they work. Also, if the new (or expanded) mass transit system is a pleasure to ride on, as compared to driving, even more people are encouraged to both ride on it and live far away from where they work. The expansion of mass transit thus engenders more and longer trips which in turn tends to increase energy use and pollution even if mass transit is more energy efficient and less polluting than the automobile. Tall buildings are often built in a downtown hub served by rail transit far away from where the people who will work in these buildings will live, casting in concrete and steel excessive travel distances and energy use. Building freeways has had the same effect. Thus expanding mass transit may well result in almost the same amount of automobile use as previously plus the added energy use and pollution of the new mass transit. Today, with little difference in energy efficiency between the auto and mass transit, it is more clear than ever that expanding mass transit wastes energy.
Mass transit was much hyped in the early 1970's as a means to save energy during the "energy crisis". But it was well known then that most autos were gas guzzlers so that improving their energy efficiency would be easy while doing the same for mass transit would be hard to do. Some imported autos at that time (such as the Volkswagen from Germany) already obtained about 25 miles/gallon as contrasted to about 13 miles/gallon for US-made autos. Thus any calculation at that time should have included taking into account the future improvements in automobile efficiency as contrasted to the more limited possibilities for improvement of transit efficiencies.
Why is the energy efficiency of mass transit today not much better than the automobile? There are two things to examine: the technology of the vehicles and the ridership in terms of the number (or percentage) of seats occupied by passengers.
First let's compare the bus to the automobile. To make a long story short: the technology of both vehicles (rubber tires and internal combustion engines) is about the same. The percentage of seats occupied by people (load factor) is also about the same for urban buses and the auto. Thus the energy efficiency in terms of passenger-miles per gallon of fuel is about the same. Now for the details of the bus/automobile comparison along with some history. The bus technology should be somewhat more energy efficient than the automobile for 3 reasons:
1. The bus wheels are easier to roll (less rolling resistance) than the automobile because the tires are inflated to a high pressure of almost 100 pounds. This also results in more damage to the pavement. In rare cases, an auto tire may be just as efficient if it's both worn smooth (less rubber to absorb energy) and overinflated. Auto rolling resistance could be reduced by designing auto tires for higher pressure.
2. Most buses use efficient diesel engines which also tend to be noisy and polluting. At full torque (when the gas pedal of an auto is pushed to the floor) the efficiency of a diesel is somewhat better than a gasoline engine. But at lower torque the diesel is much more efficient. So this tends to make the bus more efficient (but more polluting). Many bus lines have been coverted to using natural gas as a fuel so as to decrease air pollution. These engines that burn natural gas are just like gasoline engines and have no efficiency advantages. Automobiles have the same engine options as buses and thus there is no basis to argue that diesel engines make the bus inherently more energy-efficient.
3. The bus has less wind force (aerodynamic drag) per seat. This force is mostly due to air (the headwind) striking the front end (including the windshield). The front end shields many more passengers than an auto's front end. However, the front end of a bus is both larger and less streamlined than an auto. As a result, an extremely well streamlined auto is not much worse than a bus in aerodynamic drag per seat. Furthermore, aerodynamic drag is not very significant at low speeds under 30 mi/hr and is of minor significance for urban buses on city streets.
Since automobiles could also enjoy the advantages of 1 & 2 above, only the 3rd advantage (less aerodynamic drag) is an intrinsic advantage of the bus over the auto. In spite of these 3 "advantages" the typical urban bus only gets about the equivalent of 3 miles/gallon (of gasoline) and carries only 9 or 10 persons. This is 27 to 30 passenger-miles per gallon, about the same as a typical automobile in urban use getting 28 passenger-miles/gallon (1.4 persons x 20 miles/gallon). Actually the urban bus obtains 3.5 miles/gallon of diesel fuel. This figure is reduced to 3 for two reasons: 1. To compensate for the fact that diesel fuel has a higher energy content than gasoline. 2. To account for about half of the fuel used in driving the bus during non-revenue service (called "deadheading").
One reason for the low fuel economy of urban busses is that they consume much energy in the frequent stops they often make. Long nonstop freeway runs are an exception. Air conditioning also uses energy and it keeps running even when the bus is stopped. It wasn't always this way. In 1960 urban buses got 5.5 miles/gallon of diesel fuel (vs 3.5 today).
But when one looks at intercity (long distance) buses one finds they are about 2 1/2 times the energy efficiency of the automobile (in intercity use) for obvious reasons. There is no frequent stopping, the percentage of seats occupied is significantly higher than the auto, and the lower aerodynamic drag results in a significant savings at high speed. Also, they are not subsidized and thus have to either operate efficiently, with a high percentage of seats occupied, or go out of business.
Twenty five years ago most American cars were gas guzzlers. Efficiency had been ignored in favor of power, performance and rugged design using overly heavy components. Thus it was easy to make very significant gains in fuel economy. Not so for buses. Bus lines had been run by companies concerned about the bottom line. Fuel economy was important to them, resulting in buses designed for efficiency with little room for future improvement. Government mandates resulted in long overdue improvements to automobile efficiency while government subsidy to buses resulted in the purchase of more luxurious (but less energy efficient) buses. Subsidy permitted the operation of buses with low ridership. This partly explains why urban buses, which were formerly over double the energy efficiency of automobiles, are today no more efficient than the auto.
The technology of rail transit using electric motors and steel wheels on rails is quite different than that of highway vehicles with their energy-absorbing rubber tires. The rubber tire has several times the rolling resistance of the steel wheel on a rail. Since this is so, why do the statistics (see US Dept. of Energy: "Transportation Energy Data Book") indicate that rail transit is not much more energy efficient today than the automobile? Here are the reasons for it:
While the rolling resistance of high speed rail is inherently low, the resistance is increased due to the heavy weight of the trains. Aerodynamic drag, while also inherently low, is still substantial due to the higher speeds. This drag is proportional to the square of the velocity. So if the speed of a train doubles, this energy used per mile is four times greater.
At the same speed as an auto, rail has far less aerodynamic drag since each rail car shields the car behind it from the headwind. But in underground tunnels a train acts something like the piston of a compressor pushing a long stream of air ahead of it. This significantly increases the aerodynamic drag. In spite of all this, rail transit's rolling resistance and aerodynamic drag are much lower than that of the automobile (especially if compared at the same speed).
The reason that the rail energy efficiency is lower than one would expect is due to:
Similar to the case of the bus, 25 years ago rail transit was reported to be over double the energy efficiency of the auto. Since then, the auto energy efficiency has increased while that of the rail transit has decreased. The decrease is due to many reasons: the luxury of higher speeds, more lighting and air conditioning, and declining ridership of older systems. The energy use by stations is almost independent of ridership so with fewer riders there is more station energy use to be allocated to each rider. Some of this decrease may exist only on paper due to overestimation (25 years ago) of travel on the New York City rail transit system. Although this is an oversimplification of what happened, they formerly assumed that a passenger boarding a train was equally likely to travel a long distance as a short distance which likely overstated the travel on the system, thus overestimating its energy efficiency.
While there is not much potential for improvement in the energy efficiency of buses, there are obvious ways of significantly improving the energy efficiency for new urban rail systems: Build them on the surface, operate at lower speeds, use the motors as generators for braking, and coast before braking. The problem with the above is that for an urban area that is already built up there is no room for rail on the surface. Lower speeds will discourage ridership. Thus improvements in rail transit energy efficiency often comes at a significant cost.
Even if a vehicle is very energy efficient in terms of vehicle-miles per gallon (or the equivalent in electricity), it will not produce many passenger-miles per gallon unless it has good ridership. The number of people per auto has been steadily dropping over the last 25 years. For trips to work in 1970 the average number of persons per auto was about 1.2 but today it's roughly 1.1 in spite of efforts at ridesharing. Thus the number of persons other than the driver has fallen in half for such trips. Thus while auto energy efficiency has increased by roughly 60% in the past 25 years, the energy efficiency of moving people by auto has increased only by roughly 50% due to a decline in automobile occupancy by about 10%.
Serious errors in statistics on automobile occupancy happened in the "Nationwide Personal Transportation Study" of about 1970 by the Federal Highway Administration. They may have simply asked people how many people were in an auto with them and obtained a simple average. This seemingly correct method is actually biased sampling and resulted in an erroneous figure of 1.4 (should have been 1.2) persons/auto for trips to work in 1969. A possible reason for this error is double counting. If two persons are together in an auto, then two persons witness an event as compared to only one witness (the driver) for a person driving alone. If there are double the number of witnesses for an event, a simple sampling of people will result in twice as many people reporting it. It appears that some of this error still existed in 1977 and 1983 but was likely fixed in 1990. Also the 1969 figure was later revised downward.
The above type of sampling error also biases public opinion of mass transit ridership. Most people observe a higher percentage of seats occupied in mass transit than is typically the case. Many tend to use mass transit during rush hours when buses and trains are nearly full. These events are observed by a large number of people. The cases when buses and trains are nearly empty have almost no witnesses. Even on half-full train, a rider may erroneously think the train is full. I have boarded a train near the center of the platform where most people congregated and found the cars that stopped there to be nearly full. Little did these crowded passengers realize that the cars at the ends of the train were almost empty.
Some errors and illusions have already been mentioned regarding ridership but there are many more. During the 1970's, as a result of the "energy crisis" and no gasoline at some gas stations, there was much written about energy efficiency. The government sponsored a number of studies on the energy efficiency of mass transit (and comparisons with the automobile). These various writings came up with seemingly different results and for the most part, overestimated the energy efficiency of mass transit. In at least one case a writer just guessed and then someone else cited this guesstimate without mentioning that it was just a wild guess.
Most people have no idea how much energy transit uses. Reports that made exaggerated claims could not easily be put to test. However most people own automobiles and many would not accept unreasonable auto efficiency estimates as credible. Another reason for the errors was that most transit systems then did not measure passenger-miles but only kept track of the number of passengers using the system. Thus estimating the number of passenger-miles was a guessing game that could yield widely varying results. When transit agencies were asked to submit estimates of it in the late 1970's they tended to make quick and inaccurate estimates. Some simply ask bus drivers what they thought it was on their routes. A few transit agencies just "assumed" that their buses were always full. Then around 1980 the US government started to require more accurate reporting of it and came up with sampling methods of estimating it. Passenger-miles figures became available enabling more accurate estimates of energy efficiency.
Since mass transit operating agencies were receiving subsidy and one of the arguments to subsidize mass transit was that it saves energy, they had incentive to turn in statistics showing high efficiency. Later, a policy was established where the amount of subsidy depended in part on the amount of passenger-miles they generated. This provides even more incentive to overestimate passenger-miles. There are many ways to intentionally overestimate passenger-miles, especially if it is "determined" by sampling. One way is to sample bus or train runs that are known to have high ridership, although sampling is supposed to be done at random. Another way to cheat is to take a random sample of bus runs but discard it and do it over if the results show too low a ridership. It's entirely possible that true random sampling will just happen to yield results that are far too high and thus it becomes almost impossible to prove cheating. There is an inherent conflict of interest when the agency that is to receive money based on the results of sampling is permitted to do the sampling. The correct way to do it is by an outside agency that has no stake in the outcome. The transit system being sampled should not even know when it is taking place so that they will have no possibilities of influencing the results.
The studies of the 1970's also made mistakes in estimating the energy efficiency of the automobile. The erroneous figures on automobile occupancy mentioned above lead to an overestimation of automobile energy efficiency. However stating with the early 1970's there was a massive shift to smaller autos which were more energy efficient. This shift was not picked up by the statistics which continued to report erroneous dismal automobile energy efficiency for many years. As a result, automobile energy efficiency was often underestimated. To some degree, the two errors above partially canceled each other.
The reason for the low estimates on automobile miles/gallon prior to 1990 is because of incorrect estimates of vehicle miles in "Highway Statistics". The data used is supplied by each state which reports its gasoline sales and estimated vehicle-miles. The reported amount of gasoline sold is reasonably accurate but states sometimes estimated the amount of miles traveled mainly based on the amount of gasoline sold. To get miles some states simply multiplied the gallons of gasoline sold by an estimate for miles/gallon obtained from "Highway Statistics". The result obtained gives a value of miles which, when divided into the number of gallons reported, will give back the exact number for miles/gallon as reported by "Highway Statistics". If every state had done this, the reported miles/gallon would never have changed, regardless of improvements in automobile efficiency.
Neither the automobile nor mass transit will significantly reduce our energy consumption in urban passenger transportation. There is no magic technical fix for the problem, but there is an obvious solution: Simply greatly reduce the amount of travel. This can be encouraged by reducing population so as to provide more available housing so that people can live nearer work. Land use controls should prevent the construction of workplaces unless there is plenty of available housing nearby (within walking or bicycling distance).
********************************************************************************
This includes short titles and acronyms. The subject name for "mass transit" is "local transit" in many library catalogs.
Data Tables = Data Tables for the ... National Transit Database (or Section 15 Report Year)/ U.S. DOT, FTA, Audit Review and Analysis Division, Office of Capital Formula Assistance. Annual (a government periodical). Exact title and author varies but title always starts with "Data Tables". Before 1990: National Urban Mass Transportation Statistics, Annual Report, Section 15 Reporting System. The data includes pass-mi estimates and fuel consumption.
A summary of some info. in above "Data Tables" may be found in:
NPTS = Nationwide Personal Transportation Survey (or Study) /FHA & others; 1969, 1977, 1983, 1990. This reports automobile occupancy (persons per auto). The published reports of these surveys are dated a few years later.
TEDB = Transportation Energy Data Book" of Oak Ridge National Labs. (annual) table: "Energy Intensities of Passenger Modes". Used edition 14, 1994, Table 2.14. While this table has some errors in it, it does correctly indicate that the automobile and mass transit are about equal in energy intensity. However, the figures for both modes should be higher as will be explained later on.
Steam Bus Project = California Steam Bus Project Final Report, Assembly Office of Research, California legislature, Sacramento, Jan. 1973; see p. 31: Table VIII, Characteristics of Urban and Suburban Bus Driving Cycles (for ordinary diesel buses).
Public Transportation Fact Book, American Public Transit Association. Formerly "Transit Fact Book". Lists pass-mi and fuel use, supposedly using "Data Tables" as a source.
NPTS by FHA & others: The erroneous report mentioned was "Report no. 1: Automobile Occupancy" April, 1972. The 1990 NPTS devotes Chapter 7 of Vol II to Vehicle Occupancy. On p. 7-26 it reports persons/vehicle (all types of trips) 1977: 1.89, 1983: 1.75, 1990: 1.64. These figures are passenger-miles divided by vehicle-miles. I have used 1.6 in part due to NPTS having a history of being too high.
For automobile occupancy on trips to work see: "Journey to Work" a report of the 1970, 1980, and 1990 Census of Population. These give the occupancy (in persons per trip) as 1970: 1.17; 1980: 1.14; 1990: 1.09. The discrepancy between this and the above NPTS for 1969 revealed the errors in the latter. NPTS reported (in pass-mi/vehicle-mi) 1977: 1.32, 1983: 1.29, 1990: 1.14. Note that NPTS figures are higher. One reason which partly explains this "discrepancy is as follows:
The Census figure is in "pass/trip" while the NPTS figure is in pass-mi/vehicle-mi. They are not the same units of measurement.
One problem is that the occupancy may vary during the "trip". For example, if you leave home for work, and pick up another person half way towards work, then only half of the trip had 2 persons in the auto. One way to handle this is to count it as two trips. But did the Census do this? If not, the Census figure would count this as 2 person/auto but NPTS would count it as 1.5 persons/auto.
Another problem is that longer trips tend to have higher occupancy. For example, consider two trips: One is a 1-mile trip with only the driver. The other is a 9-mile trip with both driver and rider (2 persons). Now since there is 1 pass/trip in the first case and 2 pass/trip in the second case, the average is 1.5 pass/trip. But there is 1.9 pass-mi/veh-mi (18 + 1)/(9 + 1). The Census would report 1.5 but NPTS would report 1.9. For the Census, all trips are weighted equally but for NPTS, trips are weighed by distance since longer trips generate more pass-mi. So more weight is applied to long trips that tend to have higher occupancy. Similarly, less weight is applied to short trips that tend to have lower occupancy. As a result of this weighting which favors higher occupancy, the NPTS figures are higher than the Census figures due to a "distance effect".
In the example, one can claim that long trips are given less weight also since they generate more vehicle-mi which appears in the denominator. But this denominator is weighting all trips while the components of numerator for long trips (18 in the example) are much larger than the components for short trips (1 in the example).
The method of using passengers/trip is what one would observe if a large number of observers stood by driveways and sampled people as they left home to make a trip. The NPTS results is what a crew of observers would see if they stationed themselves uniformly spaced along all streets and highways and counted the people in passing vehicles.
While the above considerations might explain the 1990 difference (the number passengers is about 50% more for NPTS) it doesn't seem to fully explain why in 1997 there were about 100% more passengers using the NPTS method. Since the US Census covered a huge number of people and since NPTS has a history of past errors, I am biased in favor of the data of the US Census for the years 1970 and 1980 although it must be multiplied by an "distance effect" factor to make it comparable to NPTS results.
In order to compare the automobile with mass transit, one should not just use the energy efficiency of a typical automobile used for all types of driving. When comparing an automobile to a city bus operating on city streets, one should compare it to the lower automobile efficiency obtained on city streets. In urban use, the automobile is not as energy efficient for 2 reasons: It gets poorer mi/gal and it contains fewer passengers. What will be done here is to consider that an urban bus not only operates on congested city streets, but also on uncongested streets and freeways. We will attempt to estimate the energy efficiency of the automobile under the same conditions as a typical urban bus. We will split the automobile mi/gal into a lower figure for urban use and a higher one for intercity use.
In comparing the automobile to mass transit it is unfair to mass transit to use the energy efficiency of a typical automobile for comparison. This is because mass transit takes place in an urban area, often on crowded city streets. An automobile on city streets not only gets less miles/gallon than when used on open highways, but it also typically has fewer people in it. On trips to work, about 90% of drivers drive alone. Thus I will attempt to estimate the energy efficiency of an auto in urban use for the purpose of comparing it to mass transit. Not all the driving will be on city streets since an urban trip by auto as a substitute for a transit trip will often include a ride on a freeway.
One might argue that a substitute auto trip on the freeway is apt to be longer since people often drive longer distances in order to utilize freeways. On the other hand, inspection of bus routes shows that they often do not go by the shortest route either and often make detours to access more passengers. Round-about routing is even more likely if one needs to make transfers.
My estimation for autos uses: miles/gal: urban 20, intercity 26; persons/auto: urban 1.4, intercity 2.0. BTU/gal is 125,000. This gives urban BTU/pass-mi = 125,000/(20 x 1.4) = 4,464. intercity BTU/pass-mi = 125,000/(26 x 2) = 2,404 These numbers are estimated based on:
1. For mi/gal: EPA estimates about 22 mi/gal in a recent pamphlet (TEDB used 21.6). "Highway Statistics, 1990" by FHA, DOT in table VM-1 shows about 1/3 of vehicle-miles is non-urban. Looking over EPA tables of mi/gal leads to the estimates of 20 and 26 mi/gal which averages to about 22 when weighted by urban-intercity vehicle-miles. The "average" should actually be done by using the harmonic mean method but the error introduced by a using the arithmetic mean is small compared to other errors.
The "1990 Nationwide Personal Transportation Survey" (NPTS) by FHA, DOT shows 1.64 persons/auto but I'm using 1.6 since these figures are likely still declining. Based on old (1970) studies for Los Angles and San Francisco where the persons/auto-trip was 1.44 (studies agreed) I'm going to use a figure of 1.4 pass-mi/vehicle-mi. I arrived at this estimate by: 1. Increasing the 1.44 pass/trip to 1.66 to convert it to pass-mi/veh-mi (due to the "distance effect"). Reducing it to 1.4 to account for the secular decrease in automobile occupancy. (for work trips per US Census of Population 1970: 1.17, 1990: 1.08). Since I'm using 1/3 of vehicle-miles being non-urban, the 1.6 average requires that non-urban automobile occupancy be 2.0 persons. From NPTS 1990, p. 8-1 for trips over 75 miles one calculates an occupancy of 1.93 persons/trip and 2.35 pass-mi/vehicle-mi. Note the increase in the no. of passengers (besides the driver) by 45% due to the distance effect. Many non-urban trips are shorter than 75 miles and will have fewer occupants. Thus 2.0 pass-mi/vehicle-mi seems a reasonable estimate. Many of such trips will be people on vacation, recreational trips, going visiting, etc. where people usually travel with others.
Sub-headings: Deadheading, Deadheading mi/gal, Unreported Bus Fuel Use, Gasoline vs Diesel Heat Content.
For prior to 1975 see "Trend in Bus Transit Financial and Operating Characteristics, 1960-1975" by US DOT Sept. 1977 (DOT-P-30-78-43) p. 8-7 Table 8-3: "Diesel Fuel Gallons Consumed Per Bus Mile .." This shows steadily declining fuel economy from 5.52 miles/gallon in 1960 to 4.43 miles/gallon in 1975. In 1994 it was only 3.46 miles/gal (including deadheading). See FTA: 1994 Data Tables p. 2-298 and 2-201. As a comparison, the "Steam Bus Project" gives an average of 4.2 mi/gal in 1972 for 6 typical bus routes in California.
Few people are aware that about 13% of bus miles are run "Out of Service" often called "deadheading". In 1994 it was 13% (see FTA: 1994 Data Tables p. 2-298 which shows both bus revenue miles: 1,474 billion and total vehicle miles: 1,690 billion). "Deadheading" is when a bus runs from the bus storage yard to the start of it's first run in the morning and returns empty to storage after a day of service. The fuel used is not normally segregated by deadhead or revenue service use. Thus dividing revenue miles by fuel use will result in a lower mi./gal. figure than if total vehicle miles are used. For example, in 1994 see FTA: 1994 Data Tables p. 2-298 and 2-201 to calculate 3.46 miles/gallon overall and 3.04 miles/gallon if one only uses the miles in revenue service. Which figure should one use? Since the fuel used to get the buses to their scheduled runs is needed so that the buses may serve the public, then perhaps it should be allocated to revenue service even if the fuel is not actually used during revenue service.
However an automobile also uses fuel when it is taken to a shop for repair or service or even when going out of one's way to purchase gasoline. Recall that buses are serviced at night at the storage yard. To be equitable would thus require not counting the passenger-miles for such uses but counting the gasoline consumed. Autos are likely used well under 10% of the time for such trips as compared to 13% for buses. However, autos are sometimes driven in "taxicab-like" service to take someone somewhere (such as taking someone else to work, and then returning home alone). This is something like deadheading. The driver is driving an auto alone after (or before) providing transportation for others but the driver has no significant purpose in making the trip (except to provide transportation for others). One might call this "ferrying".
In view of this, one might guess that "ferrying" autos is perhaps about as prevalent as deadheading buses. Thus one might use the higher figure of 3.46 mi/gal when comparing the bus to the auto. However buses obtain more mi/gal in deadhead service (no stops for passengers and buses are lighter) and thus use less fuel per mile. Thus the 3.46 figure is too high. If one assumes that the mi/gal in deadhead service is 50% higher (see next paragraph) than in revenue service, then one obtains the figure of 3.29 (about 3 1/3) mi/gal.
Data from the "Steam Bus Project" shows urban buses on city streets use about half of their traction energy in braking. They also use idling energy while stopped. If in deadheading they only stop 1/3 as much (mostly at traffic signals although many signals are set for non-stop flow) then they will get about 50% more miles/gallon. The after-to-before energy intensity ratio is 2/3: 1/2 (drag energy) + 1/6 (brake energy). The energy efficiency ratio is just the inverse of 2/3 or 3/2 (50% more mi/gal).
Urban buses deadheading for the "far" end of freeway bus routes will get little better mi/gal than revenue buses when on the freeway portion of their routes. However a deadheading bus will often travel mostly on a freeway even if the route serviced is on city streets. From "Steam Bus Project" Table VIII and Table VII on p. 30 buses get about twice the mi/gal on freeways as compared to revenue service on city streets. A deadheading bus may select a less congested and more energy efficient route since it doesn't need to stop to pick up passengers. Without any passenger load, the rolling resistance of a 26,000 lb. bus is reduced about 5% (assuming 10 passengers). This only results in about a 2% savings in energy assuming most of the drag energy is for rolling resistance. Thus considering all factors, a rough estimate of 50% more mi/gal seems reasonable.
Today more and more bus service is being contracted out to "Transportation Providers" who do not report fuel usage to the FTA. Thus the miles/gallon will be too high unless one only uses the mileage for buses that do report fuel use (known as "Directly Operated"). It appears that this is being done and thus that the miles/gallon is being estimated fairly.
Still another consideration is that Diesel fuel contains about 10% more BTUs per gallon than gasoline. (BTU = British Thermal Unit, a measure of the heat content of a fuel). Thus to convert to miles/gallon (of gasoline equivalent) one must decrease the miles/gallon (of Diesel fuel) by about 10% resulting in a figure of only about 3 miles/gallon (of gasoline) for diesel buses.
The number of bus trips (unlinked) is higher than reported by Section 15 (FTA) data from 1994 Nat. Transit Summary and Trends, p. 89. The two reports show (in millions of pass):
FTA FACT 1990: 4887 5677 1991: 4825 5624 1992: 4748 5517 1993: 4638 5381 1994: 4629 The 1985 edition of "Transit Facts" does not agree with the 1981 edition for the years 1977-1980 '81 ed. '85 ed. 1977 4949 5488 1978 5142 5721 1979 5552 6156 1980 5731 5837