In early 2003, a paper entitled Comparison of Emissions from the Public Transport System and Private Cars, by RMIT engineers Ed Boyapati, Astrid Hartono and Joel Rowbottom, appeared in the 8th Cairo International Conference on Energy and Environment. It purported to show that Melbourne’s trams emitted 1.23kg of carbon dioxide (CO2) per passenger kilometre, whereas cars emitted just 0.25kg CO2 per passenger kilometre. Trains were found to be just as bad as cars; only buses came out ahead.
The findings were reported by several media outlets, and were soon picked up by the ABC Science Unit, where inquiring minds stumble upon them to this day. They are now obsolete of course, as a consequence of the fact that from 2019 the electricity from Melbourne’s trams is purchased from two solar farms in northern Victoria, ensuring that trams are practically a zero-emission form of transport. Nonetheless, we provide this analysis of the pre-2019 situation for the sake of the curious.
By the time the ABC picked up the story in 2003, Dr Boyapati and his coauthors had moderated their finding to 0.74kg CO2 for trams compared with 0.25kg for cars. This was still an alarming figure, and for some time an ABC ‘Greenhouse FAQ’ carried a highly pessimistic claim about tram travel as a result:
[T]he Melbourne researchers showed that in greenhouse terms trams truly suck….Increasing occupancy up to about 80-90% would improve the story for trains and buses, but even that won’t help the figure for trams – they really should only be making a one-way trip to the depot.
—ABC Science: PlanetSlayer Greenhouse FAQ, 2004
So, on what evidence did Dr Boyapati et al. reach such a provocative conclusion? The truth is we simply don’t know, because their actual paper didn’t provide any!
What did the paper actually say?
The six-page paper contained two major sections. The first, Comparison of GHG Emissions from Various Modes of Transport stated that the emission figures were obtained by surveying the loading conditions on trains, trams and buses at various times and calculating the emissions per passenger-kilometre using published figures for the vehicle’s energy consumption. The final figures were then provided. No actual survey data or calculations were given, even in summary form, from which it would be possible to determine whether their sample was an accurate reflection of actual tram use patterns. Usually when such studies are reported, the researchers carefully outline the experimental design and provide a summary of the survey data. No such information was provided here.
The second major section, Flow On Effects of Trams on GHG Emissions, was devoted entirely to rehashing old 1950s arguments against trams, most of which we deal with elsewhere on this site. The authors asserted (again without any evidence) that trams add to pollution by delaying cars, that trams cost six times as much as buses, that trams take valuable road space away from cars, that tram wires pose a safety hazard, that tram stops in the middle of roads are ugly, and so on. The only claim for which any evidence was presented is that roads with trams have more casualty crashes than those without – a claim we deal with elsewhere.
It is unclear why a research paper in 2003 would devote so much space to airing all these 50-year-old arguments without offering any new perspective – unless what is being intended is simply a gratuitous tram-bashing exercise dressed up to look academically respectable.
While the Cairo Conference appears to no longer operate, a check of its website in 2004 revealed that its papers were not subject to the full peer review that is the normal process for checking the correctness of research results. (In the early 2000s it was still common for some academic conferences to accept papers after reviewing an abstract only—250 words in this case—essentially providing just a basic check for relevance.) To the best of our knowledge, Dr Boyapati’s findings have not appeared in any peer-reviewed journal or conference.
Is the paper’s conclusion true?
Vehicle occupancy
Short as the paper was on detail, it did advance (for 2003) a plausible argument. It is certainly true that burning brown coal to make electricity to run tram motors is less energy-efficient than burning fuel in an internal combustion engine directly. If trams carried very few passengers on average, it may be that the greenhouse emissions per passenger were so high that even driving a car was benign by comparison.
This is far from the first study to show how easy it is to make public transport look bad in energy terms, simply by making sufficiently pessimistic assumptions about patronage. A background study for the 1991 Victorian Bicycle Strategy (produced by RTA, the predecessor of Vicroads) implied in its results that tram passengers generate slightly more greenhouse emissions than car occupants; it did so by assuming an average occupancy of just 10 passengers per tram. (It’s important to note that in energy terms the tram passenger still came out ahead in this study, using about half the energy of a motorist; the problem is not with trams per se but with coal-fired electricity, which generates higher emissions per unit of useful energy.) The RMIT study presented a worse conclusion than this older study, so even though we don’t know what occupancy figure they were using, it must have been even lower than 10 per tram.
But even the figure of 10 per tram is likely far too low—including as things stood before the post-2005 patronage boom on Melbourne public transport. Back in 1987—when tram patronage was just starting to recover from its lowest level on record—detailed surveys were conducted for the ‘Metplan’ public transport strategy, and revealed an average occupancy of 21 passengers per tram (500 million passenger-kilometres divided by 24 million vehicle-kilometres). From then onward (and particularly from the late 1990s), occupancy on Melbourne’s trams has increased at a healthy rate, especially outside peak hours and on weekends when patronage has traditionally been low. Even in 1992, using updated patronage figures, researcher Patrick Moriarty estimated average tram occupancy as 26 passengers. His estimates were published in the peer-reviewed journal Road and Transport Research in June 1992.
Using updated figures from recent State Budget papers of 200 million passengers and 24.6 million vehicle-kilometres per year, together with the earlier estimate (not since superseded) of 4km average trip length from the Metplan discussion paper, we get an average occupancy in round figures of 32 passengers per tram—20% higher than Moriarty’s 1992 estimate and over three times the 1991 Bicycle Strategy figure.
Vehicle energy consumption
Energy consumption estimates specific to Melbourne’s trams are rare; among the only published estimates to be found in anything like recent times are the 1991 RTA study and the 1992 paper by Patrick Moriarty mentioned above.
Moriarty’s figure is almost certainly the most definitive, at least for the trams that were in use in the 1990s, not only because it appears in a peer-reviewed journal but also because it can be reproduced from credible official sources. Until it was privatised in the 1990s, the State Electricity Commission of Victoria published figures for electricity used by trains and trams. The figure for 1990 as reported by Moriarty was 49.5GWh. Based on the PTC’s reported figure of 20.2 million tram-kilometres in 1990, we can reproduce Moriarty’s figure for tram electricity consumption of 2.45kWh, or 8.8MJ, per vehicle kilometre.
The 1991 RTA study used an energy consumption figure for both trams and buses of 22MJ per vehicle kilometre. In both cases this figure is almost certainly an overestimate. 22MJ is more than double Moriarty’s figure for trams, and for buses the early-2000s National Greenhouse Gas Inventory stated that Australian buses in 1991 used on average 11.8MJ per vehicle kilometre. While the ‘light rail’ figures in the Inventory were not reliable estimates for Melbourne trams (because they also included unrelated modes like the Sydney monorail), buses are a fairly standardised technology. This suggests the RTA figure for buses is a significant overestimate, and since the RTA bus and tram figures are identical (and derived in an identical manner), the same is likely to be the case for trams.
The chief barrier to obtaining accurate figures more recently has been privatisation, with its tendency to treat energy consumption and other figures as trade secrets. However, figures sometimes do emerge from third party sources. Tram manufacturer Siemens monitored the energy consumption of its Combino trams (called the ‘D class’ in Melbourne) in actual operation in the cities of Basle and Potsdam, and obtained average consumption figures of 1.53 and 1.84 kWh per kilometre (5.5 and 6.6MJ per kilometre) respectively. Most trams manufactured in the last 20 years use regenerative braking to recover the kinetic energy dissipated when the tram stops—though not yet in Melbourne owing to our reliance on legacy electrical systems. Even without any energy recovery however, the respective figures were 9.4 and 9.5MJ per kilometre – similar to the Moriarty figure of 8.8MJ.
While not specific to the vehicles in use in Melbourne, sources such as James Strickland’s vehicle efficiency page provide figures indicating that trams and buses use roughly the same energy per vehicle-kilometre, which is intuitively correct: while trams weigh more than buses, their steel-on-steel traction is more efficient than rubber-on-bitumen, and so the two differences tend to cancel (a fact Moriarty also pointed out).
A figure of 12MJ per vehicle kilometre is similar to that given by the National Greenhouse Gas Inventory for buses in 1991, the approximate vintage of the B-class trams which still figure prominently in Melbourne’s fleet (and were predominant in 2003). It is also roughly twice the observed figure for a Combino tram, 25% higher than for a Combino tram with no energy recovery, and 36% higher than Moriarty’s estimate from 1992. We therefore take this as a reasonable, but conservative, estimate of energy consumption for Melbourne trams today.
(Note that while the more recently introduced E Class trams are known to have higher power consumption than other tram types, they are also designed for more passengers. Given continued patronage growth on the system, and the fact the E Class will be confined to busier routes for the foreseeable future, it’s reasonable to assume that energy consumption per passenger will remain around the same level, even as the vehicles themselves get larger.)
Putting it all together
The energy consumed per passenger-kilometre is equal to the energy per vehicle-kilometre, divided by the average occupancy.
Taking Moriarty’s 1992 occupancy figure of 26 passengers per tram, and the original very high estimate of 22MJ per vehicle-km from the RTA study, we obtain a figure of 0.85MJ per passenger-km. The equivalent figure for cars in peak hour is 3.7MJ per passenger-km. This means that even with the least favourable assumptions, energy consumption by tram passengers is a little over 20 percent that of car occupants. This makes tram passengers in 2003 responsible for less greenhouse emissions than car occupants, even with the very poor greenhouse record of Victorian brown coal.
Using the more realistic (but still conservative) figure for trams of 12MJ per vehicle-km, and an up-to-date occupancy figure of 32 per tram, the equivalent figure is 0.38MJ per passenger-km. Thus on conservative assumptions, energy consumption by tram passengers is about 10 percent that of car occupants.
When converting energy use figures to greenhouse emissions, it is important to consider the energy source being used and also the ‘embodied energy’ involved in manufacturing the fuel. The way to make trams (and trains) look worse off compared with cars is to assume all electricity comes from Victorian brown coal and to ignore all the energy used in transporting crude oil and refining it into petrol. A more careful analysis needs to consider that much of Victoria’s electricity now comes from renewable sources, and that liquid fuels require a substantial amount of energy to extract and manufacture.
As stated at the outset, Melbourne’s trams have actually been supplied exclusively from renewable electricity since 2019, making this discussion somewhat moot. On current Victorian government plans, Melbourne’s suburban electric trains will also be powered by 100% zero-emission electricity as of 2025, and all Victorian public bus purchases from that date will also be zero-emission vehicles.
Of course, energy per passenger on trams will improve if more passengers are attracted to the system, particularly in the evenings and on the less-patronised suburban sections of routes. in European cities with well-run tram systems, energy use by tram passengers is estimated at less than 5 per cent that of motorists, compared with our 10 per cent. And emissions per passenger in peak hour are a good deal less, even though there are more trams per hour running.
Our estimates here are in keeping with those obtained in credible studies of transport energy use. These come up with widely divergent results because estimating total passenger kilometres travelled by public transport is a notoriously difficult exercise, requiring large sample sizes to get reliable results, and made even more difficult since the introduction of automatic ticketing since there are no longer any staff with day-to-day knowledge of passenger volumes. Unfortunately, no study since Metplan in 1987 has surveyed Melbourne tram users so comprehensively. Nonetheless, it is worth noting that the Millenium Cities Database for Sustainable Transport, compiled by Peter Newman and Felix Laube of Murdoch University, gives energy figures for tram use in Australia and New Zealand of 0.36 megajoules per passenger kilometre. Further calculations based on other sources can be found on our energy use page.
(It’s also nice to see that the ABC did their homework: a Carbon Cops fact sheet on trams produced after 2004 is a fairly accurate reflection of the available evidence.)
But even if our calculations or a well-designed empirical study were to find that the occupancy of trams is so low that energy use per passenger is worse than in cars, this would just be another fact pointing to the need to increase public transport patronage in order to improve the efficiency of the system. We don’t like the idea of running empty trams any more than the dwindling clique of tram-hating road engineers do, but the solution is to encourage more use of trams by providing faster and more frequent services, not to get rid of them and remove all hope of reducing emissions from transport.
Last modified: 3 May 2021