Compound Internal-Combustion Engines.

Updated: 2 Oct 2008
Diesel engine updated
Back to Home PageBack to The MuseumBack to IC index

The use of compounding to increase the efficiency of steam engines was common practice from many years, dating back to Hornblower. But what about internal Combustion Engines? The exhaust gases are red-hot, and appear to be at considerable pressure. How about adding a low-pressure cylinder to get useful work out of these gases? That may sound like a radical notion, but it's not exactly a new one...



Left: The Deutz compound engine: 1879

This water-cooled compound gas engine was built by the German Deutz company in 1879. (Deutz is a suburb of Cologne) It had two high-pressure cylinders, one on each side of the single low-pressure cylinder. The HP pistons moved in and out together; the LP crank was set at 180 to them.

From Geschichte des Deutschen Verbrennungsmotorenbaues von 1860 bis 1918, by Friedrich Sass, pub Springer-Verlag 1962 p71

It is believed that the design was proposed by Gottlieb Daimler because the English patent 3245/1879 is in his name. The Germany patent number 10116 (15th August 1879) is simply in the name of the Deutz company. Gottlieb Daimler famously went on to develop early petrol engines and make early automobiles, as described in Wikipedia.

High-pressure cylinders
High-pressure cylinder exhaust valves
Low-pressure cylinder
Low-pressure cylinder exhaust valves
Valvegear cam-shaft with bevel gears for speed reduction.

Gas inlet was controlled by slide valves, which can be seen at the extreme left of the engine; there was a single slide for both HP cylinders, controlled by the same camshaft c. The speed level reduction of the camshaft is achieved by the bevel gear visible near the crankshaft. The incompletely expanded gases from the two HP cylinders flow through the intermediate exhaust valves d1, d2 into the upper part of the chamber above the piston of the LP cylinder, which is supplied in turn by each of the two HP cylinders. Therefore the LP cylinder is working on a sort of two stroke cycle, in that it produces power twice as often as each HP cylinder. Through the exhaust valve e the fully expanded exhaust gases flow on each inward stroke out of the middle cylinder to the exhaust. On both sides of valve e another valve was installed. In the patent they are called "counter-valves", and they were used to block the passage to the other HP cylinder during the transfer of the hot gases. These valves were actuated by levers, presumably driven from the camshaft in some way. To ease starting the compression of all 3 cylinders could be canceled by a "Decompression" device. The means of ignition is not currently known.

Left: The Deutz compound engine: 1879

This not very satisfactory illustration appears to be a drawing traced from a photograph.

As with the Butler engine below, there are two heavy flywheels- possibly to make hit-and-miss governing practical?

From Geschichte des Deutschen Verbrennungsmotorenbaues von 1860 bis 1918, by Friedrich Sass, pub Springer-Verlag 1962 p71

A single engine was built, and as a precaution installed in the factory a friendly company- the Sugar mill of Pfeifer & Langen in Elsdorf in the Rhineland, whose owners were the principal shareholders in Deutz Ltd. News of any breakdowns could therefore be kept in the family.

The engine worked there for several years; its output was about 60 HP with a gas consumption of 3/4 m3 per HP-hour. Unfortunately it appears to have given a lot of trouble. The temperature of the exhaust gases flowing from the HP to the LP cylinders was above 1000C, and the intermediate exhaust valves had to be cooled to cope with it. (How is currently unknown, but probably by some sort of water-cooling) This cooling apparently caused most of the remaining usable energy (exergy) to be lost, and that remaining was barely able to keep the LP cylinder at idle, ie with no worthwhile power output.

The compound engine was returned from the sugar mill to Deutz Ltd in 1884, where it found a special place in the Deutz factory museum. Unfortunately it was scrapped in 1925 because of lack of space in the museum.

After all this time, The Deutz Company is still very much in existence, and still developing new IC technologies.

My heartfelt thanks to Felix Brun for translating much of the information from the German.


This remarkable compound engine has five cylinders; four HP cylinders and one central LP cylinder. It is water-cooled and uses spark ignition. It can be seen in CNAM, the Conservatoire National des Arts et Metiers, in Paris.

Left: The Forest-Gallice Compound Petrol Engine: 1888

The LP cylinder is behind the brass plate in the middle of the cylinder assembly. The rather complex camshaft can be seen running horizontally just above the plate. Two connecting rods (to the right of the cylinder block) from the crankshaft drive a gear that drives the flyball governor on top, and engages with a larger gear on the end of the camshaft, giving what appears to be a 2:1 speed reduction. Note the complex pipework at the top of the cylinders. A high-tension distributor with ceramic insulators is mounted on the far end of the camshaft.

There are two HP cylinders on each side of the central LP cylinder. The museum placard states the engine was constructed for Forest on the Fernand Forest-Georges Gallice system, that the engine was reversible, and that each group of two HP cylinders could operate independently. Quite what the last statement means is currently obscure. Does it mean that you could cut out two HP cylinders when running at low power?

The engine was used on Forest's yacht the "Jolie Brise".

Engine is in CNAM, Conservatoire National des Arts et Metiers, Paris. Author's photograph.

Left: The Forest-Gallice Engine Patent, dated 1890

The drawings correspond very closely to the real engine above. Unfortunately the detail is not very clear, but it looks as though the LP cylinder is double-acting, whereas all the HP cylinders are clearly single-acting. The crank angles do not correspond with those of the Deutz engine above, where the HP pistons go in and out together. Here the two sets of HP pistons are at 180 degrees to each other.

The pipes visible above the cylinders are believed to be the HP to LP gas transfer pipes. They are somewhat indirect and completely unlagged, so they probably caused serious thermal and pressure losses, as Rudolf Diesel discovered in his compound engine.

Left: The Forest-Gallice Engine: 1888

The other side of the engine, the opposite side from the brass plate, showing the complicated pipework on top of the cylinders. The pipe with the two valves attached at left is probably the induction tract, but the path of the exhaust from the LP cylinder remains mysterious. There is a small flange just above the central LP cylinder but it looks too small for the exhaust pipe. The branched copper pipe below is almost certainly one the connections for water cooling.

Sorry about the contre-jour effects; flash is not allowed in CNAM.

Engine is in CNAM, Conservatoire National des Arts et Metiers, Paris. Author's photograph.


Rudolf Diesel presented a compound engine in his 1892 patent (No. 67,207) but several years passed before he could see if it actually functioned. He began work on it in 1894-1895 in Berlin. Assembly of the crankshaft and main bearings began in May 1897, but the engine was not completed until June. The first indicator diagram was taken on 31st July, but serious tests did not begin until late October.

Left: Diesel's Series XIV compound engine.

The Diesel compound was a three-cylinder engine intended to give 120 to 150 HP. There were two 4-stroke HP cylinders of 220 mm bore and 400 mm stroke, one either side of a 2-stroke LP cylinder of 510 mm bore and the same 400 mm stroke, in the same arrangement as the Deutz engine above. The gas transfer from HP to LP cylinders was controlled by four water-cooled exhaust valves.

There were many initial problems in testing, concerned with starting and the two-plunger fuel pump, but the most unpleasant surprise was that the efficiency of the engine was not superior- it was much worse than a standard Diesel engine.

Tests revealed that the loss to the cooling water from the four intermediate exhaust valves was a shocking 11.8%, almost half of the heat that one of the cylinders could potentially convert to work. There were also unknown (but suspected to be large) heat losses from the connecting passages for gas transfer, and pumping losses in moving the hot gases from the HP to LP cylinder.

From The Diesel Engine, Cummings, p185

The result was that the best fuel economy achieved by the Series XIV compound engine was 499 gm/BHP-hr, more than twice that of the conventional Series XV engine. It never gave more than 99 HP, well below the 150 HP aimed at.

Left: Diesel's Series XIV compound engine.

This drawing shows the large LP cylinder (left) and one of the HP cylinders (right).

Note the small pipe running up the centre of the LP piston rod, to spray cooling fluid into the hollow LP piston. Whether this was water or oil is not currently known.

The arrow A shows one of the intermediate exhaust valves; this was water-cooled internally. It looks like a small valve to absorb so much heat.

From The Diesel Engine, Cummings, p185

The engine last ran on 31st December 1897, leaving Rudolf Diesel deeply disappointed. It was quickly dismantled and scrapped- no one wanted it hanging around as a reminder of an expensive and unsuccessful experiment.


These engines were described in an article by Edward Butler in The Electrical Magazine, 21st July 1904 issue. He was as much concerned with uniformity of torque and lack of vibrations as he was efficiency. As with the Deutz engine, there were two explosion cylinders, one on each side of a low-pressure expansion cylinder. The two explosion cylinders had their cranks at the same angle, with the expansion cylinder crank 180 degrees out of phase. The weight of the low pressure piston and rod was balanced by the sum of the two smaller pistons and rods. There were four equally-spaced explosions per cycle of two revolutions; Butler claimed that this gave smoother operation than any other arrangement.

Left: A compound high-speed oil engine.

This engine had three cylinders of 7-inch, 10-inch and 7-inch diameter, with an 8-inch stroke. This represents two high-pressure cylinders, where the combustion occurred, and one low pressure cylinder for further expansion. It had a balanced crankshaft and weighed 16 cwt. (about three-quarters of a ton)

It is not clear whether this engine was single or double-acting.

It drove a dynamo at 400 rpm.

From The Electrical Magazine, 21st July 1904, p21

Left: A 100 HP balanced compound gas engine.

This engine was built to run on either town gas or producer gas. It had variable compression and rotary valves. Unfortunately there is nothing to give much idea of the scale, but if that's a workbench visible through the flywheel on the left, the engine must have been about 10 feet high. That's a lot of engine for 100 HP.

The horizontal shaft at the front of the engine is chain-driven from the crankshaft, and drives a vertical shaft through helical gears. This presumably drove the rotary valves, though it's not clear how.

Three of this engines were built, though it is not clear by whom- probably some sort of consortium led by Butler.

From The Electrical Magazine, 21st July 1904, p23

Left: A 100 HP compound gas and oil engine.

This is very similar to the 100 HP engine above, though whether it was one of the three built is not clear. When running on oil a maximum of 20 HP was obtained from the low-pressure cylinder, out of the 80 HP total. This maybe points to an inherent problem in compound IC engines- an extra big cylinder and piston with associated valvegear that gave only 20% of the power.

From The Electrical Magazine, 21st July 1904, p25

Nothing more seems to have been heard of the idea. Why not? A likely explanation is that the amount of extra power produced from the added low-pressure cylinder was not enough to justify the cost and weight of the extra cylinder (which would of course be substantially larger than the high-pressure cylinder) and the complications of the extra valvegear. The more even torque is less of an issue with small high-speed engines such as we use now.

Left: Design for 5000 HP compound gas engine.

This was described as a "design" so it was probably never built. It was supposed to run off producer gas at what were then called "central stations" ie what we would call electric power stations. To put it into perspective, the first steam turbine power station in Britain began operation with two 75 kW turbo-alternators at Newscastle in January 1890, and the writing was already on the wall for reciprocating engines generating electricity. In 1905 Sir Charles Parsons supplied a pair of 1500 kW turbo-alternators generating at 11 kV to the Frindsbury Power Station in Kent; Butler's engines were obsolete.

Like most compound IC engines, this one had two normal cylinders in which combustion took place, flanking a central low-pressure cylinder in which the gases were further expanded. 5000 HP seems like a large power output for three cylinders, with only two of them actually combusting; from the dimensions on this drawing it appears that the engine was intended to be 31 feet high with no less than two 17 foot diameter flywheels. That is a lot of flywheel for an engine that was supposed to have a uniform torque as one of its major features.

The pistons were water-cooled internally, and were lubricated by means of oil and water dripped down the hollow tail-rods, from the pipes at the top of the drawing.

You will notice that the valves labelled C are of the rotary type. If you have visited the Museum gallery of rotary valve IC engines, you will know that despite a great deal of development, these have never proved very successful. One can only speculate how well they might have worked here.

From The Electrical Magazine, 21st July 1904, p26

Left: Design for 5000 HP compound gas engine.

Section of the low-pressure cylinder, showing water-cooled exhaust valves, with the water being dribbled down the tail-rod into the interior of the valves. All the other valves were of the rotary type; these were also water-cooled.

Butler says that the size of the cylinders was proportioned to give a mean pressure of 65 psi on the HP pistons and 10 psi on the LP piston. Twin sparking plugs in "firing pockets" were specified; these were to be fitted with a stop valve so that a plug could be changed while the engine was running.

From The Electrical Magazine, 21st July 1904, p26

Left: Design for 5000 HP compound gas engine.

Section through two of the rotary valves, at top and bottom of the cylinder. These were pressed into their seats by small pistons in cylinders that connected to the main cylinder. These can just be seen to the right of each valve. " this simple means any adjustment for wear or unequal expansion is automatically compensated for." asserts Mr Butler.

Note the provision of compressed-air starting.

From The Electrical Magazine, 21st July 1904, p26

At the end of his article, Butler notes intriguingly that the compound IC principle had also been tried out by Crossley, Livesy, Daimler, Holt, and others, though "in quite small sizes". The Museum staff are already on the trail of these engines.

Edward Butler was an Englishman who built the Butler Petrol Cycle, which is accepted by many as the very first British motor car although it never went into production. Karl Benz is generally recognised as the inventor of the modern motor car, but Butler was said to have exhibited plans for a 3-wheeled vehicle some two years earlier than Benz in 1884 at the Stanley Cycle Show.

You can learn more about Mr Butler here. (Wikipedia link)

Left: Some thoughts on compounding by W H Booth.

Who seems to agree that the game is not worth the candle.

Attempts to find out who Mr Booth was have so far been unsuccessful.

From an article on the Diesel engine in The Electrical Magazine, 25th April 1905 issue, p343

The idea of the compound IC engine seems to have disappeared for the rest of the 20th century, but, like a surprisingly large number of the concepts on display in the Museum of RetroTech, has recently resurfaced. A report from Oak Ridge Laboratory in the USA called "Advanced Combustion and Emission Control Research for High-Efficiency Engines" issued in 2004, included this rather intriguing table of IC engine power losses and possible ways of reducing them:

You will see that in the second category, that of Exhaust Losses, the "breakthrough path" includes compound compression and expansion cycles. The concept of the compound IC engine clearly lives on, a century later.

You can read the full Oak Ridge Laboratory report here. I am not quite clear whar compound compression is- possibly two stages of compression with intercooling, as used in industrial air compressors to improve the efficiency of compression.

There are other ways of getting more power out of an engine by greater expansion of the exhaust gases, such as The Atkinson cycle. (Wikipedia link)
The original Atkinson cycle engine performed the intake, compression, power, and exhaust phases of the four-stroke cycle in a single crankshaft rotation, and was invented by James Atkinson in 1882 to bypass the contemporary patents covering Otto cycle engines. It had an unconventional connecting-rod/crankshaft linkage that allowed the exhaust stroke to be longer than the compression stroke. Modern versions, such as the engine used in the Toyota Prius hybrid petrol/electric car, do not have four cycles in one rotation; operation is conventional except that exhaust stroke is effectively longer than the compression stroke, because the inlet valve closes late.

Four-stroke engines of this type with this same type of intake valve motion but with forced induction (supercharging) are known as Miller cycle engines. Here is an article on The Miller cycle. (Wikipedia link)

Back to Home PageBack to The Museum EntranceBack to IC indexTop of this page