Home History My Recent Efforts Generating the Steam
Fuel Designs for aircraft Hydroplanes Practical
Skylark I.C. Engines Compressed Air

Compressed Air Model Aircraft

Compressed air engines were, with the use of Rubber motors the pioneer source of power for model aircraft until the advent of model Internal Combustion engines in the mid 1920's. Less common was Flash Steam no doubt because it was, and indeed remains a very demanding way of getting a model aloft. I speak from personal experience!

At that time there were combined contests where Rubber and CA. (compressed air) competed on equal terms and of course it improved the breed of both. Eventually the contests were separated and the last contests incorporating CA. as a class was I think in UK about 1928/30.

I have for many years had an interest in performing a few experiments with CA using modern materials particularly for the pressure tanks which often represented more than 60% of the weight of the complete machine. In those days brass tanks made of thin shim material 0.004-0.006" (0.1-0.15 mm) were the normal method used for making the tanks. Brass was preferred probably because it was easy to join using soft solder.

Lighter Tanks for more Performance

I have an original copy of the 1931 Joe Ott Book "Model Airplanes- Building & Flying" this Gem of a book is part of every aeromodeller's heritage whether they know it or not! On page 316 is a compressed air model the suggested weights being 7 Ozs (200 Grms.) for the tank and motor. For the rest of the Model he suggests a 4 Oz (114 Grms) limit. This I am sure is typical of CA. models at that time, the tank and motor weighs nearly twice that of the Tailplane, Undercarriage, Fusalage and Wings put together. I am confident that modern materials can help and perhaps even reverse that ratio such that the Tank and Motor Weighs half that of the rest of the machine. My mind turned towards Kevlar and Carbon laminates, however the contruction of such a tank is not straightforward unless weight is not a major consideration, which rather defeats the object! A Kevlar tank 2" diameter to take 100 psi could be as thin as 0.004 to 0.006" (0.01-0.15 mm) at which thickness it would be almost as delicate and flexible as a paper tube of similar thickness, unless of course it were pressurised. For the same reasons it is predictable that fitting wings, engines and undercarriages to a tank built as lightly as that could be difficult and end up weighing more than my target weight.. I toyed with the idea of moulding the laminate in one application over an expanded polystyrene core but that too required special consideration. Getting the core out completely and avoiding excess absorbtion of the resin by the polystyrene was the end of the whole idea. The thought of dissolving it out with thinners was considered too but I thought that is something of a fire risk and a possible chemical risk to freshly cured Epoxy resin as well, so that idea was dropped too.

Recently (May 2008) a small sub-group on the yahoo FFML (free flight mailing list) talked of using beer cans and other low weight freely available alluminium low pressure containers adapted to serve as lightweight tanks for CA. powered model aircraft.. This is where I am now, trying to use beer cans twice! Starting with a humble beer can, you have to use several of them to make a tank of any real size and the obvious way of doing that is to join them end to end. So I made a very simple tool of alluminium with two diameters formed on the overall diameter, one being of such size to allow insertion (just) into the can then a conical section to the outside diameter of the identical can. There is only about 0.008-12" difference in the two diameters. It is arguably a waste of time bothering with a tool but it does make a fair job of it and the result is more consistant. In fact I did not form my first tank of Beer cans, instead I used cans of a "EURO" form of Red Bull the personal Stimulation drink; after all I am after more performance!

These are only 2" diameter (52 mm). As soon as I had made the tank as far as the last can I turned an alluminium adaptor to fit through the shallow domed end fixed it with Epoxy and then stuck it in position at the business end of the last can. As soon as it was cured I was in the workshop where I connected the compressor and at about 80 psi I was rewarded with this little event!

I know I should not have just pressurized it willy nilly! Next time I will hydro test, I have the kit here and it is daft not to use it. I thought the dome was a bit shallow and the next stage was to spread the longitudenal stress into the body of the tank and prevent the blow out in future. The whole idea is to make a much lighter tank of some intrinsic strength for handling the model unpressurised, and to enable the attachment of wings, wheels and the tail. However I also want eventually to produce a simple to make structure; such that it will withstand much higher pressures than is possible with pop (soda) bottles or the old brass tanks.

With this in view I quickly realised that I had cut away most of the radial strength of the Red Bull type tank because I had removed BOTH ends of all the intermediate cans in the four cans I had used. To get over the problem of the blow out I removed the end of that can, recovered the original alluminium adaptor and added another complete can and the same adaptor. I think in future I will cut away much of the intermediate can's bases but leave a ring as shown by the beer can (left). This will provide a series of hard areas along the cylinder to relieve stress and allow direct attachment of wings etc. It is easy to imagine that the ends of the can are not going to be very heavy, a judgement based on their area. In fact if you do the experiment you will find that removing just both ends of any beer can, actually halves its weight! Saving weight is vital to develop the performance of CA. powered model aircraft up to the level that I believe is possible and safe.

This is how the Red Bull type can looked before I began fitting the balsa end plugs, these were made of three plies, grain at 120 degrees of medium/soft balsa fixed into position using epoxy, filled with micro ball filler, such that it would still reasonably easily squeeze out of the joint. I fitted the plug carefully making a former that fitted the female dome accurately.

I tried finishing the dome ends first covering them with Aramid fibre and epoxy, this was I think a mistake, next time I will cover the tank lengthwise first and add the dome material last. It will I imagine look better and I will be able to optimise the overlaps more effectivly and save a few grams.

I am a little bit disappointed by the weight which ended up at 70 grams including the white plastic adapter to fit direct to the AH motor. The cans alone epoxied together with the alluminium air feed stub firmly fixed in place with a threaded collar sealed with epoxy weighed 40 grams. The balsa plugs added 3 grams = 43 grams. The Aramid fibre and epoxy added another 18 grams and the white plastic AH adapter 9 grams = 70 grams altogether, however I have retained the virtue of an easily fixed motor which can be used on several different models with a minimum of fuss.

Laying up the fibre took less time than I had imagined with 40 Mins pot life I need not have concerned myself at all. Very few bubbles appeared and caused no trouble in that they were removable with the course bristle brush, most of the bubbles appeared at the joints between the cans. The outer layer at the joints is prominent, this is to be expected and I am glad I thought to fill these tiny 'steps' and smooth them over before laying up began. I will follow the work done so far by building a test tank of the same materials, made just two cans long and hydro-testing to distruction, I fondly hope that the single layer of Aramid will remain intact up to at least 200psi (14 Atm). I intend to use the real tank up to half the rupture pressure. As far as I can see there is little doubt that this very simple contruction could produce pressure vessels of very high performance indeed just by adding more layers of Aramid Fibre. Below are pictures of the complete cylinder, the plastic adaptor between tank and the standard AH motor is fairly easy to spot. The modification to the crankshaft allows me to attach any propeller I want as long as it can be drilled out to 2.5 mm; (the diameter of the crank} which I threaded ISO standard 2.5 mm course pitch. The Steel is fairly tough and hard and it needed a freshly sharpened button die to cut it nicely

Assembled Unit
Assembled Unit
Assembled Unit

Tank Testing

Having built a fair sized tank I had to test it and in any case the basic technique of making tanks in this way had to be tested before pressurising with air for flight. I had a suitable Pump and gauges and so I decided to make a readily usable test set so a series of tests could be carried out with minimum delay. I won't go into much detail about Hydro Testing pressure vessels suffice it to say that it is essential and the principles were established in the steam age.

Why Hydro test?

Water is almost incompressable, gasses are not, they are fluids like water but far less dense and compress with ease. Gasses also EXPAND with equal ease, when a balloon bursts it releases all the compressed gas in an instant. All the energy (after Losses) that went into compressing air into the balloon in the first place is released explosivly, which is why pressure vessels are not subject to test pressures with gasses. Gas is like a spring and stores its compressed mechanical energy, whilst liquids plastics and solids do store mechanical energy in a similar way but to a negligable degree. Hydro testing is essential to reduce accidents.

Compressors by the way are very inefficient normally wasting about 70% of the energy expended in running them.

The Test rig is pictured here and the proceedure in all such tests is basically the same whether it be a nuclear power plant pressure vessel test or these tiny tanks. Despite their small size when pressurised they could contain enough energy to cause nasty injuries if they burst. Just like high speed model engine propellors can inflict nasty injuries, both risks deserve a similar level of caution.


The Tank and the pump have to be furnished with connectors of such quality and strength to warrent no leaks and sufficient strength, I use vehicle brake pipe and fittings of 10mm x 1mm pitch, simple, obvious and cheap. The tank is filled to the very top with all air displaced and of course with the connector at the top. The tank will have to be tapped a few times to release entrapped bubbles from the it's internal surface.

The rig comprises three main components, the pressurising water reservoir, the pump and the delivery manifold and gauge. Before pressurising the tank the reservoir is topped up and the system pumped clear of air; the connector is loosly fixed to the tank and water pumped for a few strokes, excluding with it any remaining air.

I tried to tighten the coupling on the final stroke of the pump before pressure is increased. I take the pressure up to say in this case 20 psi and check for leaks. If all is well pressure can be gradually increased until the test pressure is reached and held or rupture occurs which was the intent in this instance. If possible as a further safety measure I submerged the tank in water. If the rupture is sudden any displaced shreds will be slowed down in the bucket. It is essential to stay safe ALL THE TIME because elastic energy is still contained in the stressed steel of the boiler or our Kevlar tank and that energy has to be released in a controlled fashion as far as is reasonably possible. I tested my compressor tank as a first test of the rig itself and I did that with the tank in the open but I was round a corner of the building with bricks and mortar twixt me and it. On other occasions I have used lengths of old carpet, it is usually obvious how to stay safe if you think about it first; recovery after an incident can be a long or impossible job. That is enough on pressure testing.

Testing Materials.

Whilst all these thoughts were going on about pressure tests I realised that some strength data was needed in order to calculate what pressures the tanks might tolerate under test and in use. I wrote to companies, you know highly responsible ones like Dupont and Rio Tinto but I did not get as much as an acknowledgement. I needed stress data on alluminium cans by Alcan owned by Rio Tinto and similar data on Aramid fibre called Kevlar by Dupont. In light of their pig ignorant response to my requests I decided to do my own testing. I learnt about testing of materials in my days as a UK government apprentice. The usual stress figure bandied about is the UTS, = Ultimate Tensile Stress, whilst its fun knowing what stress will bring your bridge down, that knowledge is of little help if it is all the data you have. Yield point, elongation and proof stress data contains information of far greater use in designing a safe bridge or CA tank for a model aeroplane. I had nothing definite on anything I had to hand (beer cans and one piece of woven Aramid) but I had heard that Aramids did not stretch much before they broke and I have old data that told me that most alluminium alloys stretched alot before breaking. I cut a piece of my Aramid cloth about 2" (50mm) square and cut a strip of alluminium beer can about 1/4" (6mm) wide. The fibre I laid up with resin and rolled it onto a sheet of waxed glass squeezing out the excess resin. During the following weeks I set up some tensile stress tests on a very small scale which gave very consistent data at very small cost (not pricing in the time!).

The Aramid fibre square I cut into a few narrow strips about 0.4" (10 mm) wide, under my small 40X binocular microscope it was quite easy to narrow the strips down to three lengths of different breakage widths comprising 2 threads width about 0.050" (1.25 mm), 4 and 6 threads so I had a range of breaking points. I needed to check that the threads themselves were of consistent quality and performance. The Alluminium strips were made up of three identical pieces with a width of 1/16" (1.6 mm). The kevlar pieces were the devil to fix without them breaking before I applied any tension. The steel clamps crushed the resin which destroyed the fibres in the process, I succeeded using medium hard rubber sheet pads between the steel jaws and the clamps. The alluminium strips were easy to stretch to a break point. I used two fishing spring scales to stretch the materials and observe to the nearest KG or so when the break occurred. It all worked out very well. The Kevlar stretched very little (none that I could see or measure) before it snapped it was satifying to measure break stress points which very closly had the same ratios as the number of threads 2:4:6 in the test pieces; see line 2 of this paragraph. After working the arithmetic, using typical alluminium test data, some knowledge and a good text book I was able to work out some approximate numbers to guide me in designing the test rig and the break test samples, (test tanks). Whilst this work has taken two paragraphs to describe it actually required about 8 weeks of free fun time to complete.

My first real test was to pop a fizzy drink (soda) bottle just out of curiosity. Fallows and Doug Mc hard had already told us what to expect but I don't recall ever seeing any pictures of ruptured bottles and split containers nor definite test results. The bottle concerned was a 1.5 litre slightly tinted silver one picked out of a neibours Trash Bag on collection day! We don't buy pop so I did not have a bottle when I wanted one, I made a 10 X 1 mm threaded adaptor to fit the bottle top and put a 1" pipe clip around the threaded portion of the cap. It burst at 120-130 psi exactly as Fallows and Doug McHard have told us it will. I have now found a source of smooth sided pop bottles which may be a bit easier to stick reinforcement to. I am curious to understand the phenomenon noted by both these very bright guys as to how a bottle reinforced as they described can possibly take twice the pressure, one thing is sure, they did not make it up and it follows intuition but it does not follow the arithmetitic of a normal parallell sided cylinder. The cylinder contains a fluid and a fluid in an enclosed vessel exerts the same pressure on all those surfaces exposed to it. The above three pictures show clearly how the split in the cylinder wall ran along a meandering straight line on the longitudinal axis AND it extended all the way from the domed end reinforcement to the ring of reinforcement around the centre joint; exactly as it should in theory. So what happens to a pop bottle what makes it so special. The force in the bottle per inch of it's length is about three times the pressure per square inch. I will explain that.

Given that the pressure is 130 psi and the bottle diameter was three inches then we have a force of three times the pressure per linear inch over the paralell portion of the bottle, = 390 lbs (186 KG). In working out the total bursting force we have to multiply the unit pressure by the projected side area of the whole bottle. Measured quite carefully the projected area of the bottle was about 25 square inches at 130 psi there is about 3,250 lbs force trying to burst the bottle apart or well over a ton. This force is called the hoop stress and is the basic calculation when considering the safety of cylindrical pressure vessels.

As soon as it splits of course the pressure falls instantly and the applied force drops to zero. However if the bottle were to split instantaneously all around it's perifery the accelleration force would be a function of that force acting upon the bottle's mass which I measured at a mere 1.5 Ozs (43 Grams). That would be 3250 lbs divided by 1.5 Ozs, There are 16 Ozs. in my pounds weight; therefore accelleration equals 3250 X 16 divided by 1.5 G. About 34,000 G no wonder bits fly about if a pressure vessel goes pop. I know this is an exagerated example however there are some serious considerations too. Having watched a pop bottle burst when it split along it's length; I am sure that particular failure mode is typical and inevitable; furthermore it is unlikely to cause much of a safety problem for two main reasons, they are;
1. That the pressure collapses at the instant of bursting so, whilst the force accellerating it is huge,that force is never actually applied.
2. I think the bottle will allways stay in one piece with one big slit along the side.
Damage may be done to the model but I cannot think of much harm being done to people EXCEPT when the failure mode is at the cap and the cap plus perhaps metal items attached to it go flying with it. One metal item I always attach to the cap is a light metal pipe clip around the threaded part of the cap; I have never had a bottle cap blow off, even at burst pressures. In my view,we must restrict pop bottle pressures to about 60-70 psi, just as Fallows and Mc Hard suggested in Aeromodeller magazine 20 odd years ago. I have yet to test reinforced pop bottles but that will be done by and by.

The Catenary Curve

Defined this is the curve formed by a perfectly flexible, uniformly dense, and inextensible cable suspended from its end points. OK the pressure forms a unified force similar to Gravity so the curve of a reinforced plastic drinks bottle should be a catenary but it cannot be a perfect catenary because it IS extendable and it is not of really accurate uniform thickness. However because it is suspended at the line of reinforcement it is unlikely to be a comfortable radius of a circle either. In truth it is probably some inscrutable thing someway between the two, no way could it be worked out accuratly because the data is so poor and I doubt if I am up to it anyway. What we can probably say is that within limits the further it stretches the lower the stress is likely to become. EXCEPT as it stretches it gets thinner and eventually it will break because the stress PER UNIT of length/thickness will tend to rise as it stretches and THINS. As the linear stress is reduced by it's stretching that stress HAS to go somewhere which is of course into the reinforcing wire or tows of Aramid or carbon fibre. It remains surprising to me that the pressure to rupture is DOUBLED, but if it is and it is consistent and safe it is a very valuable, weight saving property. For obvious reasons, like a chain it breaks at it's weakest point along its weakest line which is why we get a jagged edge. I imagine that the phenomena of double the bursting point observed by Fallows and Mc Hard must have something to do with the material's great ability to stretch and form outward curves between every line of reinforcement.

My aim is to test uniform type(s) of bottle with a few combinations of fibre around them and test in a uniform repeatable way, and to test Aramid/Carbon fibre reinforced tanks to establish whether or not a compressed air powered Model Aircraft can put up a credible enough performance to make a safe and satisfactory RC model. There is no doubt in my mind that they make excellent FF models, Fallows and Mc hard have aleady proved that.

Testing the Aramid reinforced Alluminium Tank.

This tank was treated in the same way as the pop (soda) bottle, not surprisingly the burst pressure was very much higher at 320psi or 27.7 Atm. In fact I removed the two end lengths of the long cylinder and used them for the test which is pictured nicely. The slit ran ALL THE WAY from one point of reinforcement to the next which says to me that the Catenary was very shallow Vis a Vis. pop bottles, mathematically for any engineering data it can be ignored. It also proved that Aramid stretches but little before failure and most notably perhaps, it infers that the finished Aramid clad format owes little or nothing of it's strength to the aluminium former, which leads me smoothly on to the next stage.

Remove the Aluminium altogether.

In order to do this I will use caustic soda, dissolved in water poured into the tank with the parts of the alluminium cans that I want to keep, being protected with a thin coat of epoxy painted on during assembly. These areas will be mounting points for wings tailplane and undercarriage. In doing this hydrogen is created which burns explosivly, so no smoking and do it outside.

Compressed Air Engines.

All piston expansion engines cannot be anything but very similar in design and layout but I have yet to find a compressed air engine designed for model aircraft that featured poppet inlet valves. Years ago in the UK magazine 'Model Engineer' an extensive comparison of steam engine valve gear was made, it was very objective and used one boiler and identical engine bore and stroke the only major differences being the nature of the valve gear used. Slide vales, piston valves and poppet valves were compared the conclusion was very positive in it's outcome; a poppet valve uniflow engine is by far the most efficient layout of the three tested. I cannot imagine anything that would tell me that the existing designs of model CA.aircraft engines would not be greatly improved if they were built with poppet valves. It is certainly what I will do.

The reasons that poppet vales are best are simple to understand. Slide valves and piston valves leak, as do the traditional slotted crankshaft valves used on most early Brass Engines, as featured in Bert Pond's Excellent tretise on the whole subject of model aircraft expansion engines. Valve Timing, or the lack of it is a feature that mitigates against piston operated (bash valves in USA.). I have one of these and it is clear that the engine run is shortened by the inevitable advanced inlet timing, far worse as a proportion of the available engine run than CO2 engines. The engine I am crisising is the Air Hogs unit, the pump supplied for this engine fills the tank to 80 psi, the engine is stopped by the advanced inlet timing at 30-25 psi so a third of the available pressure drop is lost. I appreciate that no engine will use all of the energy but I would expect useful power down to 10 or 15 psi to be a realistic target.