Rotary-Valve Internal Combustion Engines.

Updated: 7 Jan 2010
RCV engine added
Back to Home PageBack to The MuseumBack to IC index

The Frayer Rotary Valve
The Minerva Rotary Valve
The Aspin Rotary Valve
The Cross Rotary Valve
The Froede Rotary Valve
Felix Wankel and Rotary Valves
The Mellors Rotary Valve
The Alpha Rotary Valve
The Norton Rotary Valve
The Lotus Rotary Valve
Other Rotary Valves
Rotary Valves today
Rotating Cylinder & Valve engines New
List of Rotary Valve patents
The Museum contains several other IC engines with
rotary valves, but this is not their main unusual feature.
See them on their respective pages:
The Almen axial engine
The Bristol Axial Engine
The Butler Compound Engine

For almost all its history, the preferred valve for the Internal Combustion engine has been the poppet type. (the etymology of the word "poppet" is an interesting study in itself, but not now) However, several inventors were attracted by the apparent simplicity and uniform motion of rotary valves of one kind or another. However, as with both steam and IC rotary engines, this simplicity was more apparent than real, and the engineering problems were daunting.
The basic problem, which still remains essentially unsolved, is that the pressures in the cylinder of an internal combustion engine are high, due to both the compression stroke and the explosion of the fuel-air mixture. This produces large forces on the valve system, however it is contrived; the beauty of the poppet valve is that such forces simply push it harder against its seat, and have no effect at all on the valve-actuating mechanism.
However, the geometry of rotary valve systems is inherently different; in the Aspin concept below, the vertical valve cone is pushed up axially against the cylinder head, while the horizontal Cross valve is pressed up against the top half of the bearing surfaces. In both cases this can cause excessive friction and seizure, the root of the problem being that enormous forces are acting on the valve while it is moving.

There were a large number of rotary valve schemes tried, most of which exist only as patents. The first US patents tracked down so far are by Frayer & Howard in 1907 and 1908. Another early patent was in 1911 by Vallillee (Patent 983328) who used a rotating disc. Several other versions by other people were patented later; see the patent list below. In Britain the two systems which gained the most attention were the Aspin vertical rotating valve and the Cross horizontal rotating valve.

Rotary valves were also tried in steam engines, without conspicuous success. See The Paget Locomotive and the Carel engine on the High-Speed Steam Engines page.


Left: The Frayer & Howard rotary valve.

This is the earliest internal-combustion rotary valve patent so far unearthed by the untiring staff of The Museum. Note however that Butler was apparently using rotary valves in 1904.

The valve is mounted horizontally across the top of the cylinder, like the Cross valve, and driven by chain and sprocket from the crankshaft at half-speed.

Frayer and Howard do seem to have had some idea of what they were up against. The valve, shown in Figs 3 and 4, was designed not to touch the surounding metal, but maintain a 2-thou clearance, the gap being flooded with lubricating oil. The valve rotated in ball bearings at each end, protected from the exhaust gases by two sets of sealing rings. (19,20,21,22) It sounds plausible, but I wonder how much oil would be lost into the induction and exhaust passages.

Nothing is currently known as to how much success Frayer and Howard had with their invention.


This rotary valve was used by the Minerva company, who made luxury motorcars in Belgium. There is some information on the company here; (external link) it says they made Knight double-sleeve valve engines under licence, but strangely fails to mention rotary valves at all.

Left: The Minerva rotary valve.

According to Newton & Steeds:

"It has shown under severe test conditions remarkable reliability and freedom from any tendency to seize up or give trouble through scoring."

From The Motor Vehicle by Newton & Steeds, pub Iliffe, date unknown but post 1921, p71

The rotating valve-block V ran horizontally across the top of the engine, and had ports R which allowed the inlet manifold I and the exhaust manifold E to communicate with the cylinder at the right time. Each valve-block V was pressed against its seating by two segmental saddles, on the top which bore a wedge W through two steel balls to ensure uniform distribution of the pressure; the spring-loading on W could be adjusted by screw and lock-nut S.

For a six-cylinder engine two valve blocks were provided, each serving three cylinders. Because of the multiple ports spaced at 120 degrees, the valves only had to rotate at one-sixth engine speed, which must have reduced problems with wear. Drive was by a double roller chain from a gear-driven countershaft running at one third engine speed.

This rotary valve concept by Frank Aspin has a vertical valve rotating above the cylinder.

Left: One version of the Aspin rotary valve. Note the conical thrust bearing C, intended to absorb the upward thrust of high pressures in the cylinder. This version has no rotor cooling.

Aspin's first experimental engine was a 250cc machine, based on a Rudge 250cc motorcycle engine. It was built in 1933. Aspin claimed that it produced 18 bhp at 7500 rpm, 31 bhp at 10,000 rpm, and was capable of a maximum speed of 14,000 rpm. It used a 14:1 compression ratio, implying great resistance to pre-ignition, given the low-octane petrol of the time, and this was ascribed to the absence of a hot exhaust poppet valve in the cylinder.
All these figures were startlingly high for the engine technology of the time, and it appears that they were greeted with some disbelief. A recurring theme in the Aspin story is impressive claims for performance, without backing from independent testing.

An Aspin modified Velocette 350cc engine was tested without much success by the Velocette factory in Birmingham in 1936.

By 1939 Aspin had decide it was necessary to liquid-cool the valve rotor to prevent distortion. This change did not greatly reduce the heating problems but it did require the spark-plug to be moved away from its central location, and it was placed at the side of the cylinder, placed so the rotating valve uncovered it at the right time for ignition. The valve rotor at this point was made of case-hardened nickel steel, rotating in an aluminium-bronze housing. Performance was once again claimed to be exceptionally good, but lubrication problems persisted, and the engine required careful warming up since heavy-footed use when cold would be certain to cause seizure of the valve rotor.

At some point Aspin tried a variable-speed drive to the rotor, ie changing speed during the combustion cycle but still in sync with the crankshaft; the idea being that the rotor would be moving relatively slowly during the high-friction parts of the cycle. It does not appear to have been a success, and would certainly have added complications to what is (in principle) a beautifully simple design.

In 1944 the magazine Autocar tested a family car with a four-cylinder Aspin engine.

In the 1960's an Aspin-converted 8.6 litre Leyland lorry engine was used to power a bus operated by The Northern Ireland bus company. At this point Aspin was using a lead-beryllium coating on the rotors to reduce friction. The bus ran successfully for an unknown period, recorded only as "many thousands of miles" until oil consumption abruptly climbed as the special coating on the valve rotors wore off. No further testing was done so the results cannot have been too impressive.

Aspin developed his rotary valve concept over more than thirty years, and there is not space here to trace the detailed evolution of his ideas. In essence, Aspin could build an engine with high oil consumption and good reliability, or with normal oil use but having a distressing tendency for the rotary valves to seize up solid. He never resolved this problem, and many others also failed.

Aspin's last rotary valve patent was taken out in 1977 (US 4033317) and I believe he died shortly afterwards.

There is a very fine site on the Aspin valve and its engines here.

Roland Cross was the great British exponent of the horizontal rotary valve. He began work on rotary valve engines in 1920. The project lapsed in 1945, squeezed out by the demand for components from his company for conventional aero-engines. Roland Cross died in 1970.

Cross evolved a process for the manufacture of high performance piston rings from wire for his rotary valve engine. Remarkably, the company he founded to make these still exists. See the Cross Manufacturing Company, and look at "history" under the Overview tab.

Left: The Cross rotary valve.

Unlike the Aspin, the Cross valve was always driven at constant speed from the crankshaft.

The "reaction bridge" shown in the picture at left absorbs the upward forces on the horizontal valve assembly, and is supposed to have reduced the gas forces on the actual valve. At the moment I'm not quite sure I understand how it worked.

One problem was fouling of the spark plug by oil sprayed from the rotating valve; wiping seals were added to later versions of the engine to control this.
The split valve housing permitted a balanced loading system to control the forces on the valve. Wear was compensated for, lubrication and oil recovery were good, and it is reported that there was no danger of seizure. The sealing problems that other approaches suffered from were solved, and there seems possible that Cross had an effective and practical design which under other circumstances could have reached quantity production; it would probably have been more profitable than the Wankel if it had. On the other hand, Norton had very little success with a very similiar concept- see below.

Left: Roland C Cross looking confident.

Cross entered two rotary valve motorcycles in the 1935 TT races on the Isle of Man, but one machine proved too slow to qualify, while the other withdrew after two laps with "sparkplug trouble" which may well refer to the problems with plug oiling mentioned above.

Left: The cover of a booklet the Cross company published to publicise their rotary valve system. Date unknown, but probably around 1930.

Note that Combe Down phone numbers only had three digits in those days.

Left: One of the inner pages of the booklet the Cross company published.

The performance and fuel consumption figures claimed here were remarkably good for their day. Whether the engines were ever subjected to impartial testing is not currently known.

Walter Froede was the head of the German company NSU's motorcycle racing program; at the time, NSU was the largest builder of motorcycles in the world.

Left: Froede rotary valve in a 125cc racing motorcycle engine. Valve rotor coloured pink.

The engine is tilted over to the left, and the piston is shown at TDC. The rotary valve was driven at a constant 1/4 of crankshaft speed. Although initially successful in racing, it was not repeated, probably due to the demise of NSU.

Later NSU introduced the famous Wankel rotary IC engine, but its poor reliability and short service life finished off the company and NSU was absorbed by Audi.


Felix Wankel, the originator of the famous rotary engine, was involved from 1930 to 1945 in the development of a rotary disc valves for use in aircraft and torpedo engines. His primary expertise was in the difficult business of sealing. During this time he worked at BMW, DVL, Daimler-Benz, Lilienthal and Junker Aircraft.

Various sources state that he was involved in developing the "Daimler-Benz DB601 V-12 rotary disk valve aircraft engine" but other sources say that this engine had four conventional poppet valves. This mystery currently remains unresolved. He certainly worked on the Junkers Jumo KM8 torpedo engine, which had to fit into a cylindrical casing that had no room for conventional valvegear. The short working life of a torpedo engine meant that rapid wear of the rotary valve disc was not an issue.

Left: Cross-section of a Junkers Jumo KM8 disc-valve engine, enclosed in its torpedo casing.

The disc valve is visible on the right side of the engine, just above the piston.

The engine had eight liquid-cooled cylinders of 90mm bore by 85mm, stroke arranged as a V-8 with a 90deg angle. The total swept volume was 4.34 litres and the compression ratio was 6.6 to 1. Output was 275 HP at 3650 rpm. It ran on a mixture of petrol, oxygen, and its own exhaust gas- the latter presumably to dilute the oxygen to a manageable content.

A production order for 100 engines was issued towards the end of WW2 but was never completed. A prototype was examined by British and American intelligence engineers, who concluded it was "a progressive trend in automotive development." It would appear they were wrong.

Above: Longitudinal section through one cylinder bank of a Junkers Jumo KM8 disc-valve engine.

The disc valves are fitted between upper and lower cylinder heads. They had teeth on their periphery, and formed two gear trains without the need for idlers or extra gearwheels. The train of valve discs was driven by bevel gears from the crankshaft, with splined connections to allow valve timing to be altered.

In 1951 NSU became interested in Wankel's rotary engine project, and he joined them. It seems very likely he was involved in the design of the Froede rotary valve shown above.


Left: The Mellors maltese-cross valve drive system.

The Mellors rotary valve system had two conical valves, one for inlet and one for exhaust. They moved not continuously but in 90-degree steps, driven by a maltese-cross mechanism. When stationary the valves were pressed against their housing by helical springs and so were cooled. Just before each movement they were lifted a few thou from their seatings by a face cam. The valve was thus able to turn freely, and then was released back to its rest position when it stopped.
The use of the maltese-cross or Geneva mechanism was first suggested by Mueller in a 1914 rotary valve patent.

It appears that the valve drive shaft at the bottom of this picture rotates once per cycle, ie at half crankshaft speed. The maltese cross mechanism would have been subjected to some severe accelerations, and I am not too sure how well this would have worked at high rpm.

Left: The Mellors valves in cross-section.

Note the hinged sections and the two little levers at the top. This may have been a method of applying sealing forces to the valves, but I suspect it might have been a misinterpretation of the two levers that lifted the valves when they were due to turn.

Ted Ambrose Mellors was an international motor cycle racer; he became European Champion in 1938. He designed and patented his rotary valve system in the early 1940's, during WW2. See Patent 559830, March 1944.

He died in 1946: see Ted Mellors Biographical Notes (External site)

Left: The Mellors rotary valve system fitted to a 250cc New Imperial engine.

The two bolted-down covers on the side of the cylinder head show the valve locations.

Left: Mellors' prototype rotary valve engine today.

A close-up of the maltese-cross mechanism. The two maltese crosses marked A are driven by pins on opposite sides of the central disc B.

Image kindly provided by John Wood.

In the early Sixties, Alpha Bearings of Dudley in the West Midlands produced an engine in which the flywheel acted as a rotary valve. All I know about it comes from the very informative website here so I will send you straight there.

It is heartening to report that Alpha Bearings are still very much in business.

Joe Craig, Norton development engineer, spent two and a half years working on a horizontal rotary valve proposed by Laurie Bond. (The originator of the Bond three-wheeler) The valve design closely followed Cross technology.
There were the usual problems with lubrication,sealing, and plug fouling, and the resulting engine was less powerful than the standard version. Definitely not a success.

Left: The Norton Rotary cylinder head fitted to a Manx Norton motorcycle engine.

The cylindrical valve at the top is driven by bevel gears from a vertical shaft, which is driven from another vertical shaft via two small pinions. The second vertical shaft is driven by bevel gears from the crankshaft. Why there are two vertical shafts is currently unknown to me; it may just have been a matter of alignment, allowing the bottom end of an existing engine to be used.

Image kindly provided by John Wood.

Left: The Norton Rotary cylinder head seen from the underneath.

The bearing for the vertical drive shaft can be seen at top right.

Image kindly provided by John Wood.

In the early 1990's several manufacturers were investigating supercharged two-stroke engines as the way forward. Since a two-stroke has one firing stroke per crankshaft revolution as opposed to one every other revolution, the power-to-weight ratio is potentially much better. The engineering challenges involved in making an engine that can compete with the conventional four-stroke are, however, severe.

Left: The Lotus two-stroke Rotary valve project: 1991

This two-stroke engine was of the "uniflow" type, with exhaust ports uncovered as the piston moves down. Because there is no dedicated exhaust stroke, scavenge air is blown in to clear the exhaust gases out of the cylinder by an external compressor, and then the fuel/air mixture is admitted.

The rotary bearing was of the Cross or horizontal type, rotated by a toothed belt at crankshaft speed as this is a two-stroke. By rotating the stator inside the valve, the inlet timing and the relative duration of scavenge air and mixture admission phases could be altered.

It was commented at the time that the uniflow operation should avoid the overheating and lubrication problems of other rotary valves which handled exhaust gases as well as induction. However, as far as I can see the rotary valve here is still subjected to the full combustion pressures, and I wonder how this was tackled.

Left: End view of the Lotus two-stroke Rotary valve.

When the Lotus project was reported in the press, it was said to be in "the early stages" of development, and no performance figures were given. I can find no further reference to it anywhere, and it is far from certain that even a prototype engine was built.

Interest in two-stroke engines fell away as emission standards became tighter in the 1990's.


The Brown rotary valve was developed by J & E Brown of Wednesbury, Staffordshire. It first appeared in 1938, fitted to a modified 490cc Norton motorcycle. It emerged again after WW2, and was highly praised in The Motor Cycle for Oct 16, 1947. A conversion kit for overhead cam Nortons was planned but apparently never appeared, and the Brown Valve faded from view.

John Anelay introduced a spherical rotary valve system in 1955. Details to come.

Rotary valve systems have not disappeared; it seems that nothing technological ever does.

Here is a contemporary development: The Coates Spherical Rotary Valve System; one of the Coates patents is in the table below. They seem to be serious about what they're doing.
The Coates website claims that the use of this valve "has permitted elimination of both oil lubrication and cooling water flow in the upper portion of the cylinder head." which seems somewhat out of line with the experiences of Aspin and Cross.
However it is probably significant that the Coates valve, which, if the website is to be believed works, is horizontal like that of Cross rather than vertical like the less successful Aspin.

A related technology is the rotating-cylinder-valve engine, in which not just a valve assembly, but the whole cylinder rotates. This has been introduced for 4-stroke operation by RCV Engines Ltd, a UK-based company.

One advantage claimed is that the absence of a hot exhaust valve in the cylinder gives greater freedom from detonation, allowing the engine to be run on kerosene (JP8) with a high compression ratio and hence good efficiency.

Left: Cross-section of an RCV model engine.

The version shown here is for model aeroplane propulsion. The cylinder (green) is geared to the crankshaft through bevel gears (purple) carrying a valve port around with it. The propellor is fastened to the head of the cylinder, and this axial format gives a small cross-sectional area and consequently reduced aerodynamic drag. Note that the finned cylinder jacket and the exhaust system do not rotate.

The concept is being extended to one-cylinder engines for scooters, mopeds and motorcycles. For more info see RCV Engines.

Some rotary valve patents in the UK. (not a complete list)

Patent No
373 660
Horizontal rotary valve
408 756
Horizontal rotary valve
411 010
Horizontal rotary valve
448 368
Horizontal rotary valve
463 412
Mar 1937
Vertical rotary valve
559 830
Mar 1944
Horizontal rotary valve
570 286
Jun 1945
Vertical rotary valve
570 354
July 1945
Vertical rotary valve

Here is a small and almost certainly incomplete list of rotary valve patents in the USA.

Patent No
908 656
Sept 1907
Rotary valve
908 657
Aug 1908
Explosive engine
983 328
Feb 1911
Vertical rotary valve
??? ???
Vertical rotary valve
??? ???
rotary valve
??? ???
Vertical rotary valve
??? ???
Vertical rotary valve
??? ???
Vertical rotary valve
??? ???
Vertical rotary valve
Some more recent US patents:
4 008 694
Feb 1977
Horizontal rotary valve
4 404 934
Sep 1983
Asaka et al
Horizontal rotary valve
4 658 776
Apr 1987
4 773 364
Sep 1988
Vertical rotary valve
4 788 945
Dec 1988
4 852 532
Aug 1989
Horizontal rotary valve
4 944 261
Jul 1990
Horizontal rotary valve
5 000 136
Mar 1991
Vertical rotary valve
5 052 349
Oct 1991
5 490 485
Feb 1996
5 503 124
Apr 1996
5 941 206
Aug 1999
Smith et al
5 967 108
Feb 1996
Horizontal rotary valve
6 257 191
Jul 2001
Horizontal rotary valve
6 293 242
Sep 2001
Horizontal rotary valve

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