Myth: Public transport doesn’t really save energy
Fact: Well-used public transport, despite using a lot of energy in manufacturing, uses only a small fraction of the energy consumed when driving a car—internal combustion or electric. Even at modest levels of patronage, the energy savings are significant.

In countries like Australia, where public transport use has been undermined by decades of road-centred transport policy, it’s easy to harbour lingering doubts about whether public transport is really ‘worth it’ on environmental grounds. The idea that one can save energy by running trains – which anyone can see require a lot of energy to run and take a lot of energy to make – can seem a little paradoxical at first. Perhaps public transport really is just another high-energy transport mode, like cars and aeroplanes.

Some environmentalists have thus been led to the belief that in a ‘post-carbon’ future where energy is scarce and expensive, walking and cycling will be the only transport modes available for regular use. This can lead to the idea that public transport is an irrelevancy as it (supposedly) can never achieve the same level of energy savings.

THE right to travel when and where we please will be eroded over the next 50 years as the shortage of cheap oil and environmental concerns force us to lead more local lives, according to a government report. Every journey will have to be justified and face-to-face contact with colleagues, friends and relatives will increasingly become a luxury….

In the bleakest scenario, an acute oil shortage and lack of affordable alternative energy source trigger a global depression….People survive in increasingly isolated communities….with most journeys made by bicycle or horse.

The Times, 27 January 2006

But this is all just another myth – one which the road lobby has turned ingeniously to its advantage. In Melbourne, better footpaths and new bike lanes are often used to put a friendly face on road projects. But when such facilities are built as a mere accompaniment to road expansion, all that happens is that new walking or cycling trips are outnumbered by new car trips, with an overall negative result for energy use and the environment. Between 1996 and 2001, for example, Melbourne’s path and bike lane network increased in length by 50 per cent. The number of people walking to work increased by 1,900 per day, those cycling to work increased by 2,200 per day, but those driving to work increased by 95,000 per day.

Compare this with what public transport improvements achieved over a comparable period. Between 1991 and 1997, improvements to Perth’s rail system led to an additional 15,000 people going to work by train each day, and to 60,000 new train journeys overall – and Perth’s population was only one-third that of Melbourne!

(In fact, it’s no coincidence that Western Australia not only brought about a renaissance in public transport, but also overtook Victoria as Australia’s premier cycling state: 14 per cent of WA residents now cycle, compared with 12 per cent of Victorians.)

Rest assured that those extra passengers were doing a world of good for the environment by taking the train instead of driving cars. In fact, switching from cars to public transport can more than halve household energy use, even if nothing else is done to save energy.

Of course, the key to really big energy savings with public transport is the high occupancies which follow from employing world’s best practice. Failure to employ best practice, both in route planning and operation, is the reason behind the very poor energy figures seen in some US cities where public transport has only a marginal role. Fortunately, even though Melbourne’s public transport lags well behind best practice, it still manages to do quite a bit better than cars, as we explain below. With proper service improvements to attract passengers away from cars, particularly outside peak times, it could do much better. And of course, notwithstanding any of the figures given, the energy cost of carrying one extra passenger on any existing public transport service is close to zero.

The following table provides range estimates of the energy use per passenger-kilometre (pkm) for various Melbourne transport modes. For the public transport modes, the lower figure applies during peak hour and would apply more generally under ‘best practice’ operation, while the higher figure is closer to the current Melbourne average. The detailed calculations supporting these figures can be found below.

To get an idea of the magnitude of energy consumption, one megajoule (MJ) is the energy consumed by one 15 watt lamp over a full 18 hour day, or by a 500W radiator in 33 minutes. A kilowatt-hour (kWh), the unit of consumption on electricity bills, is equal to 3.6MJ. Household energy consumption in Australia is typically in the order of 50-100MJ per person per day, excluding transport, but varies a great deal from one household to another and between summer and winter.

(Our greenhouse page provides estimates of comparative CO2 emissions.)

Life-cycle efficiency of transport modes
Energy use
(MJ / pkm)
Car (internal combustion)Operating2.53.4
Car (electric)Operating0.50.7
Bus (internal combustion)Operating0.281.1
Bus (electric)Operating0.060.25

Source: Australian Greenhouse Office. National Greenhouse Gas Inventory: Analysis of Recent Trends and Greenhouse Indicators 1990 to 2002, and Australian Methodology for the Estimation of Greenhouse Gas Emissions and Sinks 2002: Energy (Transport). Industry figures for public transport power consumption and PTUA calculations (see below).

To take one practical example: suppose Jim drives 10km each way to work each day but shares a ride with a colleague once a fortnight. This will consume an average 80MJ of energy per workday. By switching to an electric car Jim can cut this in half or better, to around 36MJ daily. But suppose Jim switches to catching a bus 2km to the nearest railway station and taking the train 10km (a larger overall distance since the station isn’t on the way to work). Then on conservative assumptions, his daily energy use falls to under 8MJ—a drop of 90 per cent relative to the petrol car, and more than three-quarters relative to an electric car.

If Jim’s share of household energy consumption excluding transport is 50MJ per day, then his share including transport is 130MJ if he drives the petrol car to work, 86MJ if he switches to an electric car and 58MJ if he takes public transport. In other words, switching to public transport reduces Jim’s share of household energy consumption by 55 per cent, and the transport component from 60% of total consumption to 14%. In fact, the energy saving is greater than if Jim completely eliminated all other forms of energy use!

All things considered, one could do worse than heed the advice of the Stern Review on climate change:

Higher energy prices and rising congestion require central and municipal planners to develop mass transit systems to cope with inner city and suburban traffic…. Such systems lead to large gains in energy efficiency and reduced emissions as passengers transfer from private cars to public transport.

—Sir Nicholas Stern. The Economics of Climate Change. HM Treasury (UK), 2006.

Operating Energy

Burning one litre of petrol in a car releases around 34MJ of energy. Based on the average fleet efficiency of about 11 litres per 100 kilometres (see our efficiency page), it follows that driving 1km in a car uses around 3.8MJ. The ‘high’ figures in the table assume an average car occupancy of 1.1, as is observed for real traffic in peak hour. The ‘low’ figures assume occupancy of 1.5, which is the average taking into account all off-peak and leisure travel. (Long-distance family holidays help raise the average here.)

These figures are for internal combustion engines using petrol, still the most common by far. Both diesel and LPG cars have slightly higher energy consumption than petrol cars based on fleet averages (about 4MJ rather than 3.8MJ per km—see our fuels page).

A typical electric car in the same class as the typical petrol car, meanwhile, consumes some 200 watt-hours per kilometre, or 0.7MJ—around one-fifth the energy. (The absence of a heat engine and the simplified drivetrain account for much of this.) The ‘high’ and ‘low’ figures are otherwise calculated based on occupancy as above.

The energy calculation for trains can be found in the PTUA publication It’s Time To Move. The power consumption of a six-car Comeng train—at full loading, with the air-conditioning running—is approximately 800kW, and its average speed is 40-60kph (assuming well-managed operations). Putting these figures together gives an energy consumption of 48MJ to 72MJ per train kilometre. Consumption per passenger kilometre depends critically on patronage. The lowest energy consumption is achieved under crush conditions of 1200 people per train (at the expense of passenger comfort) and is equal to 0.04MJ per pkm. Under light loading conditions (400 per train) and power consumption at the upper end of the range, the energy consumption rises to 0.18MJ per pkm. Well-used train systems will have energy consumption ranging between these two values. But even under very light loading conditions of 100 per train, as seen outside peak times in Melbourne’s underutilised system (though never seen at all in systems run according to world’s best practice), train passengers consume less than one-fifth the energy of car drivers.

While these calculations have been done for electric suburban trains, energy consumption for diesel trains (such as used in Adelaide) will be similar. Train services in regional and rural areas are generally designed to operate with lower passenger loadings, and so will have higher energy use per passenger than urban services. The Swiss Federal Railways, for example, report consumption of 0.33MJ (0.092kWh) per passenger-km.

The calculation for buses is similar to that for cars. According to the National Greenhouse Gas Inventory, the average energy consumption of Australian buses in 2002 was 10.7MJ per vehicle-km – a little less than three times the figure for cars. (This has reduced slightly from 11.8MJ per vehicle-km in 1991 due to the increased use of buses running on natural gas.) Applying pro-rata scaling based on the figures for electric cars, an electric bus is expected to use (conservatively) 2.5MJ per vehicle-km.

At light loadings of 10 passengers per bus (typical of Melbourne buses at present), energy use is approximately 1.1MJ per pkm for a conventional bus or 0.25MJ for an electric bus. On well-used systems the occupancy is closer to 40 passengers per bus, reducing energy use per passenger by a factor of four.

The energy performance of trams is covered extensively on another page, where we investigate the erroneous claim that trams have higher greenhouse emissions than cars. For trams the National Greenhouse Gas Inventory figures are not much use, since they lump trams in with systems like the Sydney monorail in a larger ‘light rail’ category. For reasons given on our tram emissions page, we estimate the energy consumption of trams as similar to that given in the Inventory for buses in 1991: around 12MJ per vehicle-km. (This is around 30% more than the observed energy use of a Combino tram, so is relatively conservative.) Using a ‘low’ average occupancy of 20 passengers per tram as estimated for Melbourne in the 1980s, energy use is around 0.6MJ per pkm, around half that for buses and less than 20 per cent that for cars. Well-used tram systems have average occupancies of around 80 per tram, and the figures reduce accordingly.

We have included motorcycling in the table as it is also considered a low-energy alternative to car use. The average fuel consumption rate for motorcycles is 5.7 litres per 100km, around half that for cars. Motorcycles vary greatly in weight and engine power: a 1000cc bike has a rate of energy consumption about 50 per cent higher than that of a 250cc bike. Taking a simple assumption that the fleet average represents half of each type, and applying a similar calculation as for cars, gives the low and high estimates in the table. For most forms of urban travel the occupancy of motorcycles is not significantly greater than 1.

Cycling can be regarded as a zero-energy mode of transport. It may be objected that cycling uses muscle energy and requires cyclists to eat more, so that the energy used in producing the extra food must be taken into account. However cycling is also exercise, some form of which is required for good health. The extra food intake required for cycling (if any) is only part of that required anyway to maintain a healthy lifestyle.

The figures calculated here for the various transport modes (aside from electric buses) have been independently validated by a 2008 Victorian Department of Transport study (available here). Table 1 of the study report presents energy use figures calculated from the government’s Melbourne Integrated Transport Model (MITM). With the exception of trains, the figures fall within the ranges stated above. The train figure is slightly higher, but as it has been stated for 2006, does not account for patronage growth on trains between 2006 and today.

Other comparative figures for a wide variety of modes, agreeing broadly with those presented here, can be found in this Wikipedia entry and this page by James Strickland in Canada. Our figures also compare favourably with older estimates obtained by researcher Patrick Moriarty from official Melbourne sources, and published in the journal Road & Transport Research in 1992. We take encouragement from this statement on James Strickland’s page:

The Combino light rail vehicle is about the same energy efficiency as a Porsche Carrera GT; the Porsche seats 2, the Combino seats 67 and can carry 180. The efficiency advantage is huge, though the Porsche obviously accelerates better and has a higher top speed!

James Strickland

Embodied Energy

The manufacture of vehicles uses substantial amounts of energy, and this should be taken into account when assessing the efficiency of various modes of transport. The following calculations are based on the estimate that to manufacture 1kg of metal, plastic and other raw material for vehicles requires 100MJ of energy. Figures comparable to this are found in many sources, including the RTA study mentioned above.

A small car weighing 1 tonne uses 100GJ of energy; if driven a generous 200,000km in its lifetime this corresponds to 0.5MJ per km. Large four-wheel-drives weigh up to 2 tonnes and increase the manufacturing energy in proportion. On the whole, the tendency since the 1970s has been for cars to get larger. For electric cars, an additional factor of 30% is added to the manufacturing energy to account for battery production.

The energy in manufacturing a 200 tonne train appears daunting at first: 20 terajoules (20 million megajoules). But this train will, conservatively, travel an average 300 to 400km in service each day over a lifetime of 30 years. Thus it will have provided around 4,000,000km of service before going on the scrapheap. On a per-kilometre basis, then, the seemingly large energy requirement comes down to just 5MJ per km. Dividing by mean patronage of between 500 and 1200 passengers gives the very small figures seen in the table.

Similar reasoning applies to buses and trams. Both will typically do about 50,000km in service each year. Trams are of course a good deal heavier than (non-electric) buses—buses weigh 10 to 15 tonnes, while trams weigh 20 to 40 tonnes—but this is counteracted by the difference in operating lifetime. When Melbourne’s W-class trams were retired (for specious reasons) in 2001, the newest of them was around 50 years old, and the oldest about 70 years old. Conservatively, then, we estimate tram lifetime as between 30 and 50 years. Buses on the other hand have an operating lifetime of 10 to 20 years, similar to cars. Thus the embodied energy works out about the same for each when the effect of greater embodied mass is balanced against that of longer life.

A ‘light’ 250cc motorbike has a mass of around 100kg. Based on a (fairly generous) 100,000km driving life this gives embodied energy of 0.2MJ per km. A 1000cc bike has a mass closer to 250kg, and so the upper estimate is scaled accordingly.

Indicative figures for bicycles are 15kg mass and 20,000km ridden in the bicycle’s lifetime if well used. This corresponds to embodied energy of 0.075MJ per km, rounded up to 0.08MJ for clarity.


Our comparison of the energy consumption of various transport modes differs from most, in that we consider both current performance and potential “world’s best practice” performance of Melbourne’s public transport.

It can be seen that even now Melbourne’s public transport is more energy efficient than car use. And if services were improved to world standards, the energy savings would be truly phenomenal. Energy figures near the lower limits shown here are actually achieved in some European and North American cities. (Astoundingly, it can even sometimes be more energy-efficient to take the train than to ride a bike – though not by much.)

Of course, a great advantage of the electrified modes of transport—whether electric cars or trains, trams or electric buses—is that their energy source can be made 100% renewable, ensuring that whatever the raw energy use, the carbon emissions (at least in operation) can be reduced practically to zero. (Melbourne’s trams already operate entirely on renewable energy, while trains and many buses are planned to follow before 2030.) But this doesn’t mean we can ignore the energy consumption figure. Even 1.2 MJ/pkm for an electric car is a long way from zero, and when multiplied by the millions of people needing to travel each day, means a lot of electricity demand alongside the traditional needs of households, commerce and industry. So even in a world where everything is electric, the difference between 0.1 to 0.7 MJ per passenger-km in a bus or train and 1.2 MJ in a car weighs significantly in favour of public transport—leaving aside all the other problems with over-dependence on cars, such as traffic congestion and road trauma.

Last modified: 28 March 2022