Compressed-Air Propulsion.

Updated: 30 July 2010
Tsar's car added
Back to Home PageBack to The Museum

This page refers expressly to vehicles that are driven by stored compressed air to provide independent locomotion. In contrast, the German diesel-pneumatic locomotive used compressed air purely as a transmission element between a diesel engine and the cylinders turning the wheels.

The principle of compressed-air propulsion seems very simple. Pressurise your storage tank, connect it to something very like a reciprocating steam engine, and off you go. At least you are spared the difficulties, both technical and medical, of using ammonia, petrol, or carbon disulphide as the working fluid.

Unfortunately there are still serious problems. If you have ever pumped up a bicycle tyre, you will know that the pump body gets uncomfortably hot quite quickly. Compressing a gas generates a lot of heat, and all this energy is lost when you store the air and it cools down. The losses can be reduced by compressing the air in two or more stages, and cooling it between the stages, but they are still substantial.
At the other end of the process, using compressed air to run an engine, the main problem is keeping the system working at all. When a gas expands it gets colder, and unless the stored air is perfectly dry (which it won't be) ice will start forming in the pipework and engine, and things will soon grind to a halt.

Compressed-air systems flourished, insofar as they did, in situations where the smoke, sparks and steam of the much more effective steam engine were not acceptable- in city streets, and down coal mines- and at a time before electricity was a viable means of propulsion. There were several compressed-air tram systems, though none proved very successful, and most were quickly abandoned. Compressed-air locomotives in mines lasted longer, but they too were eventually replaced by electric haulage. Now read on...

The first compressed-air vehicle I have found so far was devised by Bompas, a patent for a locomotive being taken out in England in 1828. There were two storage tanks between the frames, with conventional cylinders and cranks. It is not clear if it was actually built. (Knight, 1880)

The first recorded compressed-air vehicle in France was built by the Frenchmen Andraud and Tessie of Motay in 1838. A car ran on a test track at Chaillot on the 9th July 1840, and worked well, but the idea was not pursued further.

Left: The Parsey compressed-air locomotive of 1847

The reservoir A was filled with air "compressed to as great an extent as was compatible with safety" which fed chamber B, kept at engine pressure by automatic reducing valve C. Pipe D fed the double-acting engine E. At F is the air recharge valve, and G is the safety valve. The locomotive was intended for coal-mine work, but again it is not clear if it was actually built.

In 1848 Barin von Rathlen constructed a vehicle which was reported to have been driven from Putney to Wandsworth (London) at 10 to 12 mph.

At the end of 1855, a constructor called Julienne ran some sort of vehicle at Saint-Denis in France, driven by air at 25 atmospheres. (350 psi)

Compressed air locomotives were use for haulage in 1874 while the Simplon tunnel was being dug. An advantage was that the cold exhaust air aided the ventilation of the tunnel.


Most of the images and much of the information on the Mékarski system are displayed by courtesy of John Prentice, whose stunning exposition of the history of compressed-air trams can be seen at Tramway Information. Do not miss it.

In Louis Mékarski built a standard gauge self-contained tramcar which was tested in February 1876 on the Courbevoie-Etoile Line of the Paris Tramways Nord (TN), where it much impressed the current president and minister of transport Maréchal de MacMahon. The tramcar was also shown at the exhibition of 1878 as it seemed to be an ideal transport method, quiet, smooth, without smoke, fire or the possibility of boiler explosion.
Following this success, Tramways Nord used compressed air locos to pull horse trams on their Route E, Saint Denis to Place Clichy, beginning in February 1879. Air at 25 atmospheres (350 psi) was stored in eight reservoirs 0.3 m or 0.4 m in diameter, mounted transversely under the vehicle. These were in two sets, a main and a reserve set. The two-cylinder engine drove the front axle through the usual cranks set at 90 degrees to avoid stalling at dead centre; cylinder dimensions were a modest 125 mm bore and 260 mm stroke.

The compressed-air locos were soon withdrawn due to a number of accidents, possibly caused by icing in the pipes of the brakes, which were also worked by compressed air. This strikes me as an inherently flawed concept; if you ran out of air the brakes didn't work. A car where the brakes stopped working completely if you ran out of petrol would probably not be a saleable proposition. The servo brakes fitted to almost all cars these days retain enough engine vacuum for at least one serious stop if the engine ceases to run, and when that is exhausted the brakes still work, even if some serious foot effort is required.

Left: A bouillotte mounted on the front of a tram built by Mékarski and used in early trials in Paris. (Drawing 1875)

One solution to the freezing problem was the use of the bouillotte which apparently can be translated as either "hot-water bottle" or "hotpot". This was a vertical cylinder mounted on the front platform and was 0.35 m in diameter and about 1.5 m high. It was 3/4 full of water at about 180 degC, and at about 7 atmospheres of pressure. This warmed the air and also saturated it with water vapour. As the water vapour condensed on expansion, it gave up its considerable latent heat and limited the temperature drop in the engine cylinders. The condensation may also have helped to lubricate the cylinder walls of the engine, but I would suggest it diluted the lubricating oil and caused more friction than it saved.

Note the two pressure gauges, one showing the pressure of the stored air and the other the pressure of the air leaving the bouillotte and going to the engine.

According to La Nature, the storage capacity was 2640 litres, holding kg of air at 80 kg/cm2. This weighed 262 kg at 15 degC. The range was about 16 kilometres, by which point the storage pressure had dropped to 12 kg/cm2.

Left: The air control valve on top of the bouillotte.

The water in the bouillotte cooled quickly, and was initially reheated by blowing steam through it when the tram stopped to recharge the air tanks. Later a certain Monsieur Bonnefond introduced an internal coke firebox to provide continuous reheating of the water. This approach appears to have only been used in the Paris operations. See below.

Left: The Bonnefond bouillotte.

The Bonnefond system is all very well, but you immediately lose the main advantage of compressed-air power; ie the absence of smoke, ashes and potentially hazardous sparks. There is also the stimulating possibility of a boiler explosion. Considering that the driver already had his hands full dealing with the uncertain braking system, to expect him to stoke and supervise a small steam-boiler as well seems rather rash.

Note the butterfly valve at B to close off the chimney while dropping coke into the firebox.

According to La Nature again, the Bonnefond bouillotte consumed 0.6 kg of coke per kilometre.

Mékarski went on to run an extensive compressed air tram system in Nantes, opening in 1879. The first trams had ten steel storage cylinders between the frames and were charged to 30 atmospheres, reduced to 4 to 6 atmospheres at the engine, which was very similar to the Paris engine. An additional 32 trams were bought between 1898 and 1900; these were more powerful than the first series, with air storage at 60 atmospheres (840 psi) and with both axles driven to improve adhesion. The compressed-air trams were replaced with electric trams in 1911.

Left: A Nantes tram recharging with air and blowing steam through the bouillotte.

The bouillotte, with its distinctive handwheel, can be seen to the left of the driver.

Mékarski system tram networks were also built in other towns in France: Vichy (1895), Aix-les-Bains (1896), La Rochelle (1899), and Saint-Quentin (1901).


Compressed-air trams were tried in East London, Wantage, the Vale of Clyde, Liverpool and Chester. Various designs were used. None of the trials lasted long; the cost of operation proved excessive.

The Wantage experiments used two Mékarski-type trams built by the Compressed Air Engine Co Ltd, of 19 St Swithins Lane, London EC. The air was preheated by some sort of bouillotte, which raised its temperature to 312 degF, doubling its volume. Compressing plant costing Ł2000 was installed at Wantage Town tram terminus. A single-acting compressor pressurised six large air receivers to 450psi. Recharging a tram took 15 minutes. Quite why it took that long to fill the compressed air tanks is not easy to understand- for me anyway.

The range of the compressed-air trams was barely sufficient for a round trip. Matters were not eased by the 1:47 gradient at the end of the route, when the air pressure was at its lowest. The first compressed-air journey was made on Thursday evening, 5th August 1880

Mr. G. Stevenson, the line's engineer, estimated that the compressor used about 24 cwt of coal per day, but on such a sparse service this was far too high, as steam locos would only use 5 cwt per day.


In 1878 the Second Avenue Railroad of New York City tested and then operated for a period in 1879, five tramcars built by the Pneumatic Tramway Engine Company. They were designed by Robert Hardie, who had General Herman Haupt, a civil engineer, as an enthusiastic backer. Haupt wrote extensively and plugged compressed-air trams at every opportunity. The tram engine had one stage of expansion, and is said to have had a more advanced type of preheating than the Mekarski system, though it still seems to have used hot water; the details of this are currently obscure. The storage pressure appears to have been 1000 psi, but the engine working pressure is currently unknown. Regenerative braking was introduced; using the engine as a compressor to slow the tram, hot air could be forced back into the storage tanks, increasing the range, enhancing overall efficiency, and alleviating if not eliminating the problem of running out of air for braking. The Hardie trams were supplied with air at 1000 psi by the 1500hp steam-powered four-stage compressor used by the Manhattan Elevated Railway; this also powered the Hoadley-Knight pneumatic locomotives mentioned below. Note the use of multi-stage compression to reduce losses.

An article in the French journal La Nature states that compressed-air locomotives on the Hardie system were giving satisfactory results on the Elevated Railways of New York, though there are few details of the working and it is not clear from the text whether the locomotives were on trial or in regular operation. If anyone can provide more details I would be most grateful.

Above: Illustration from La Nature of a New York air loco: 1882.

The locomotive carried four steel storage cylinders 91 cm in diameter and 13 m3 in volume, pressurised to 42 kg/cm2. The air passed through a vertical bouillotte which heated it to 90 degC, and went to the engine cylinders via a throttle and a reducing valve designed to keep the cylinder pressure at 8 - 9 atms. How the bouillotte was kept hot is not revealed. Meyer valvegear was fitted, and regenerative braking- when slowing down the engine worked as a pump, pushing air back into the storage tanks. The range was given as 13 km.

I have found a reference to Rufus Gilbert's Elevated Company, chartered in 1872, which was to run along 6th Avenue to 59th Street in New York on compressed air power. It was apparently "stalled by the financial panic of 1873". This is some years ahead of Mékarski.


Left: A Compressed-Air Tricycle for mail delivery.

Nothing is currently known beyond the information in the New York Times article. No US patent appears to have been taken out by either Hartley or Stoll.

If the compressed-air-tank was installed "under the handle bars" this seems to indicate it wasn't very large, and it seems surprising that such a tricycle could cover thirty-three miles. Perhaps in reality it couldn't.

Thanks to Bill Levine for bringing this to my attention.


The Hoadley-Knight system (devised by Joseph Hoadley and Walter Knight) was the first to incorporated a compound (two-stage expansion) engine. This would have improved efficiency directly, as compounding did for steam engines, because it gave the opportunity for reheating the air between the HP and LP cylinders, and also have reduced icing problems. The Hoadley-Knight patents suggest hot water was used to heat the air before the HP cylinder, and also reheat it between the HP and LP cylinders. The system was trialed in New York from 1894 to 1899, but without lasting success.


Charles B. Hodges invented a two-stage engine employing a reheater between the two piston stages to warm the partially expanded compressed air. This air was passed through a heat exchanger, which was warmed by the ambient air, drawn through it by using the exhaust air in an ejector. Simiiar ejectors (blown with steam, not air) were commonly used to generate the vacuum used by locomotive braking systems. This ingenious development eliminated the need for bouillottes and little coke fires, and introduced no new moving parts. Air was the only fluid employed. A substantial gain in range was attained; up to 60% appears to have been possible. The H K Porter Company of Pittsburgh bought the rights to Hodge's US patents and sold hundreds of locomotives so equipped to coal-mines in the eastern USA, in the period 1896-1930. They were used extensively in gassy mines where explosions were an ever-present danger. No doubt the cold exhaust air was once more welcome to supplement the mine ventilation.

A typical Porter stored air at 800 psi, throttled down to 150 psi at the cylinders.

Above: Triple-expansion mine loco engine with reheat from ambient air. Note that the medium and high-pressure cylinders are combined.

The diagram above shows a triple-expansion version of the Hodges system. The source is currently unknown but is probably Porter literature. Air is stored at 150 atm (2100 psi) and dropped to 25 to 30 atms (350 to 420psi) by a reducing valve. It is further reduced (by the throttle? though it looks like an ordinary hand valve) to something below 15 atm (210 psi) and enters the HP cylinder, which has a safety valve set to 15 atm on the "steam chest" which is a pardonable error on the part of a draughtsman no doubt more familiar with steam locomotives. The IP cylinder pressure is not shown, but the LP cylinder has a safety valve set to 5 atm. (70 psi) All the references to working Porter air locomotives that I have seen indicate they were double-expansion, and it is not currently clear if triple-expansion was actually used in practice.
2100 psi sounds like a frighteningly high pressure to me, given that steam locomotives rarely exceeded 250 psi, but of course there was no fire or scale build-up to cause erosion of the metal. Even so, I can't help wondering if any of them blew up. It would be rather important to check the inside of the storage tanks for corrosion, I suspect. Note the inspection manhole on the end of the Porter tank below.

Left: Porter double-expansion locomotive No104 with reheat from ambient air.

This example used a storage pressure of 800psi, and an engine pressure of 250psi. The reheater is on the far side of the tank, and the cone of its exhaust ejector diffuser can be seen to the top right, above the tank manhole. The small cylinder visible on this side of the tank is a reservoir for air at a pressure of 250 psi, which has been through the pressure reducer from the main tank at 800 psi. The throttle valve is at the front, operated by a handle and linkage from the drivers compartment. Fittings include a pressure relief valve, a brake lever which applies brake shoes to the steel wheels, air operated sanders to maintain traction, and a driver's compartment. (missing- it would have been at the left, where the control levers can be seen)

Note the large number of very big rivets required to hold the air storage tank together, compared with a steam locomotive that worked at a much lower pressure.

This locomotive dates back to 1910. It was used at one of the coal mines in Canmore, Alberta, and is on display at a museum in Sandon, British Columbia, Canada, which kindly provided the picture and some of the facts. See:

A Porter catalogue talks about reheating: "...but there are cases where it is wise and economical to reheat the air before it enters the auxiliary reservoir on its way to the cylinders. The additional efficiency gained by this reheating varies from 35 to 50 per cent..." from "Light Locomotives" by H.K. Porter, 1900

Left: Another preserved Porter locomotive.

This Porter loco is in the "Storybook Island" children's park in Rapid City, South Dakota. This looks very similiar to the Porter shown just above, but there no signs of double-expansion plumbing. Once again, there are lots of very big rivets.

Not too sure about the paint job...

Picture courtesy of Duane Overholser of Sheridan, OR


This section has been moved here from the "Unusual Steam Locomotives" wing of the museum, on the grounds that air is, strictly speaking, not steam. 24 June 2005

Compressed-air locomotives were and are used in coal mines (where the danger of inflammable gases makes a fire unacceptable) and in food industries and textile mills, where smoke and smuts might spoil the product.

Above: This 0-6-0 locomotive is basically a large tank of compressed air on wheels. It was built for coal mining by the Dickson Locomotive works at Scranton, Pa, USA. The picture dates from 1899.

Storage Pressure
600 psi
Working Pressure
150 psi
Tank capacity
170 cu ft
16 tons

The difference between the storage and working pressure seems to indicate that some sort of reducing valve was used between the tank and the engine cylinders. Note that 600 psi is a much higher pressure than normally used in steam boilers, which rarely exceeded 250 psi. This is why the tank shown above is studded with very big rivets, compared with a steam boiler. This is characteristic of air locomotives.

Left: A slightly more modern pneumatic locomotive, the Jung PZ 20 Preßluft-Grubenlok, or "compressed air pit locomotive", built in 1955.

Pneumatic locomotives often used multiple cylinders for air storage, rather than one large tank. It reduces the stresses in the metal and is therefore more economical to make.
2900 psi
20 HP
5.6 tonnes

Note that the storage pressure has risen by a factor of five compared with the Dickson locomotive above.


There are several on-going projects for air-driven cars; see Wikipedia.

The French MDI Air Car seems to have been around for a while, but it is still a live project. Apart from straightforward compressed-air propulsion, they are also developing dual-energy engines, in which a fuel (petrol, diesel, oil, alcohol or gas) is burned in an external continuous combustion chamber to heat the air and give more range. The amount of toxic gases released is claimed to be very low.

One of the vehicles proposed has the following specs:

Weight Empty
220 kg
Max speed
70 km/h (43 mph)
Range (urban)
220 km (136 miles)
Reservoir Pressure
350 bar (5076 psi)
Reservoir Volume
175 litres
Refill Time
1.5 minutes

The air storage tank is made of carbon-fibre wound on a thermoplastic liner.

As you can see, the top speed is very modest, at 43 mph, and the range less than stunning at 136 miles. The company make the usual claims about the car being pollution-free, which is of course true in actual operation. But since compressed air is being used merely for energy storage, power will have to generated somewhere else. And proponents of air cars never mention that compressing air is inherently inefficient, with all the heat of compression lost.

Apparently MDI (Motor Development International) will be at the 79th Geneva International Motor Show from 5 to 15 March 2009.


The Odd Bicycle gallery of the Museum has a projected compressed-air bicycle as an exhibit.

One of the most prestigious of the early motor cars was the French built Delaunay-Belleville. They built some big cars with 11.8 litre engines for the Russian Tsar in 1909. These were fitted with Saurer compresed-air starting gear, which could also be used to inflate tyres, jack up the wheels, or blow a whistle. However its most important feature was that the air reservoir was sufficient to drive the car for about 400 yards without even starting the engine, allowing the Tsar to make a quick getaway from an assassination attempt.

There is, of course, another and far less benign class of things very often powered by compressed air. Torpedoes.


Compressed air vehicles in general:
Interesting historical material, but some worrying references to what appears to be perpetual motion.

For French references, Google on "locomotive ŕ air comprimé". For example: tramways_mecaniques

Porter locomotive links:

Back to Home PageBack to The Museum EntranceTop of this page