Carbon fibre and fibreglass is a better laminate for props. Its stiffer, lighter and CF tow is cheaper too. The only advantage I can see for Kevlar is impact resistance.

From : Don Stackhouse

Rick's posts are usually quite accurate, but as a former engineer for a full-scale prop company that was (and still is) the industry expert with Kevlar props, I have a few points of disagreement in this case.

First of all, props are one of the few situations where Kevlar is usually the BEST choice among the common advanced composites. More in a moment on that...

Kevlar is the lightest of the common advanced composites. The typical density of Kevlar/epoxy is about .055 lbs/in^3, carbon/epoxy is about .060, and both "E" and "S" type fiberglass are about .076 pounds per cubic inch.

Kevlar has the highest tensile strength of the common composites, even higher than carbon. However, its compressive strength is its Achilles' heel, only about 40% that of plain old garden-variety "E" glass. This is also a problem in bending situations, since a bending load normally causes tensile stresses on one side of the part and compressive stresses on the other side. I've been told that the Kevlar molecule actually has a kink in it, and the molecules actually buckle under compressive loads. They also tend to have poor bond strength with most matrix materials such as epoxy. The net result is that the Kevlar fibers tend to dump compressive loads onto the epoxy matrix at a relatively low compressive load level, causing the poor overworked epoxy to crumble. The Kevlar fibers are left intact by this failure, and when the same structure is then put under a tensile load again it will have most of its original tensile strength. However, the now unsupported Kevlar fibers will have little or no stiffness in compression. In addition, the individual fibers will chafe against each other when subjected to vibration, which will eventually cause failure of the fibers. This truly awful compressive strength is the main reason why Kevlar is usually a poor choice for most structures.

In a typical propeller the tensile stresses due to centrifugal forces on the blade usually dominate the bending stresses due to thrust and torque, so that the entire blade is under tensile stresses. In this situation Kevlar is stronger and lighter than an equivalent carbon or fiberglass blade.

One exception is in extremely large slow-turning props for very high horsepower applications. The low RPM keeps centrifugal force relatively low, while the high power increases the bending loads. In these cases one side of the blade can see substantial compressive stresses, which could make carbon (with its compressive strength nearly equal to its very high tensile strength) a better choice. For example, the composite blades on planes like the new Pitts Special, the Beech 1900 commuter airliner and the Casa 212 are Kevlar/epoxy, while the big six-bladed prop on the Dornier Do328 regional airliner are carbon/epoxy.

Another exception might be the situation of a non-folding, stopped propeller blade being used as a "landing gear" in a belly landing. A Kevlar blade might tend to get a buckling failure on its aft face, while a carbon blade might survive. OTOH, the tensile strength of the Kevlar blade would be nearly the same, while the relatively more brittle carbon blade could suffer hairline cracks that would cause a loss of both tensile and compressive strength. Cracks in Kevlar tend to grow extremely slowly in fatigue, and tend to leave a surrounding fuzz of loose fibers. Cracks in a more brittle material such as carbon tend to stay much less visible, right up to the point where the blade flies off. Sometimes the nature of a failure can be more important than the basic strength of the part.

In my experience Kevlar is cheaper than carbon by a good margin. The cost of "S" type fiberglass is similar to or a little less than Kevlar, and "E" glass is of course the cheapest of all.

Kevlar has a very high structural damping coefficient. In other words, it would make the world's worst church bell, tending to damp out vibrations quite effectively. If you knock on an aluminum or a carbon blade, or even a fiberglass blade, you get a very clear ringing sound. Knock on a Kevlar blade and all you get is a dull "thunk". This can be a big advantage in a propeller by limiting the severity of vibrations.

For example, propellers normally have to be carefully tuned so that all of their significant resonant frequencies fall outside of the frequencies the prop will be subjected to by the airframe and engine characteristics. Normally the standard method is to make the prop so stiff than the prop's resonant frequencies are all higher than the operating range it will see in service. However, sometimes there are some very low prop frequencies, and if you stiffen the blades to drive the prop's other frequencies up above the airplane's frequencies, you can pull these lower frequencies up into the danger zone (where they would otherwise be below the operating range). One common case is the typically low-frequency "reactionless mode" frequencies that occur in props with four or more blades.

In such cases it may be necessary to put a placarded zone on the tachometer to prohibit continuous operation at an RPM corresponding to that frequency. However, I know of at least one case of a commuter airliner that was fitted originally with aluminum blades and later converted to Kevlar blades. Both types of prop had a reactionless mode frequency, and the aluminum props were placarded against running at that RPM. However, the structural damping of the Kevlar was enough to limit the amplitude of the reactionless mode vibrations to well within the allowable stresses. Thus, even though that prop had a reactionless mode resonance at that frequency, it did not require a placard in that installation. A carbon or fiberglass blade would act like the aluminum blades in that case and would almost certainly require a placard.

Prop spinner domes (those bullet-shaped fairings that cover and streamline the propeller hub) can be interesting beasts. They typically see a lot of flexing and bending, so they have to be able to handle both compressive and tensile stresses. The original spinner dome for a certain turbine-powered all-composite pusher I worked on was made from Kevlar, and suffered some cracking problems around the blade cutouts. It was a classic compression+fatigue failure like I described above. Somewhere I have some nice photos of it. I redesigned it in carbon fiber, but added a layer of Kevlar to the inside surface. This improved the impact resistance (in an impact from the outside, the inside surface of a laminate sees a tensile load, the very thing that Kevlar excels at), but also added some of Kevlar's damping ability, damping out some of the higher frequency buzzing and flexing, which made life easier for the carbon that formed the primary structural shell. This combination worked very well, completely solving the cracking problem.

Carbon is quite a bit stiffer than Kevlar, but Kevlar is a little bit stiffer than S glass and much stiffer than E glass. Given the lower weight of Kevlar and it's structural damping characteristics, it can give carbon a serious run for its money in cases where stiffness is needed for vibrational requirements and where there are little or no compressive loads (such as in most propellers).

Kevlar has other interesting traits and requirements, such as problems with machineability for repair work and with water absorbtion along the fibers (both of which can be easily handled through proper initial design). OTOH, its toughness is legendary. So is its fatigue resistance. I recall one example of a blade that was birdstrike-tested in the lab in a manner that was far more severe than it would see on the aircraft. The tip of the four-foot blade was bent over about 2 feet by the initial impact, and several large cracks opened in the structural shell of the blade near the shank, as well as a few other places. We marked the ends of all the cracks, then installed that blade on a prop on a ground test engine. The prop was then run for over an hour, most of that at takeoff power. There was no measurable growth in any of the cracks!

We had another example of a Kevlar prop on a turbine pusher that suffered a power turbine rotor burst during a test flight. The guts of the engine went out the exhaust pipe and through the prop.The propeller continued to power the aircraft (the aircraft used two engines driving a single prop) for the rest of the flight, over an hour. They discovered after shutdown that the entire prop was covered with little nicks, from being peppered with large chunks of broken red-hot nickel alloy turbine blades. Only two of the nicks were more than purely cosmetic, and the prop was easily repairable. A metal prop under the same circumstances could have had problems with low-cycle fatigue under the same conditions. We elected to save that prop for our "museum" just as it was, as an example of what properly designed composite props could shrug off with amazing impunity.

All advanced composite materials have their own "personality" and set of quirks. One of the keys in good design with these materials is to understand these quirks, and using that understanding to get them to be good "team players" with the rest of the structure.

Even though carbon fibre is more dense than Kevlar, you can achieve a carbon layup that is stiffer and lighter than Kevlar.

Note: we're talking about pure compressive (i.e.:crushing) strength here, not buckling. Buckling failures are a function of material stiffness (once again a strong point of carbon) and to an even greater extent the part geometry, especially the laminate thickness and how it's supported. Folks often refer to carbon and other composites as having poor compressive strength, when (except for Kevlar) what they're really seeing is buckling failures, a different animal entirely.

As far as props are concerned, too much stiffness can be just as bad as not enough. Either one can put a resonant frequency in a bad place. In addition, from a strength standpoint the tensile strength to resist centrifugal forces is usually a bigger concern, which is also helped by Kevlar's lower density. The vibration damping effects are also very important, especially on model props where we don't do proper vibrations surveys on each new prop/motor/airframe combination. Kevlar's very slow fatigue crack growth, plus its greater tendency to show fatigue cracks instead of hiding them add a very significant amount of extra safety.

My first Kevlar canoe was far lighter than the fibreglass of the time, but carbon canoes are another 10% lighter. The very lightest are Kevlar, carbon and foam combining the best attributes of all materials.

There is data, such as some generated by Frank Weston some years ago, that suggests that Kevlar fuselages are stronger in landing impacts than fiberglass, in apparent defiance of the fact that the loads involved are essentially compressive, and that the compressive strength of even E glass is more than twice that of Kevlar. I think there are two obfuscating factors involved in these cases. First of all, a failure of Kevlar generally leaves the structure looking intact, even though it has lost most of its compressive strength and stiffness. The part has actually failed, but folks don't recognize it as having done so.

An even bigger factor in these glass vs. Kevlar comparisons is the material density and its effect on material thickness. For example, in Frank's tests he used equal weights of Kevlar and glass in his test fuselages. Since Kevlar is approximately 3/4 the density of glass, a layer of Kevlar will be nearly 40% thicker than an equal weight layer of glass. In addition, Kevlar has a much higher elastic modulus than E glass, and a little higher than S glass. The failure mode in these tests was generally in buckling, which is extremely sensitive to material thickness (about equal to the cube of thickness, in fact!). Anything that increases the thickness of the laminate without significant weight increases will have dramatic effects on buckling strength.

However, an even bigger issue here is that the failures were in fact in buckling. If you have a buckling failure, it also probably means that you are not even coming close to using the material's full compressive strength. You're carrying around a bunch of extra weight because of an inefficient structure. Anything you can do to reduce buckling failures, such as a change in part shape that makes the shell more resistant to buckling, or that improves the effective thickness of the shell (such as the corrugations in a Ford Trimotor's skin, or the use of a foam layer in the middle of a laminate to move the surface layers of the laminate further apart), or that otherwise adds support to the load-bearing parts of the structure to restrain them from buckling, can have potentially dramatic effects on the effective compressive strength. Some of you who are also on the R/C Soaring exchange may recall some discussions a while back of some work on wing spar concepts done by Dr. Mark Drela at M.I.T., essentially an improvement in how the spar shear web supported the spar caps. He was able to demonstrate some dramatic improvements in the strength to weight ratios of his spars in comparison to the ones we were typically using in sailplanes, simply by making the spar caps more resistant to buckling, and therefore better able to use the full compressive strength of the material (carbon/epoxy, in this case). I expect that if Frank had redesigned his laminates so that the failures occurred in pure compression rather than buckling, he would have found a completely opposite result for his glass vs. Kevlar comparison.

Back to props, I am doubtful that Gracindo will see any benefits spending $40 on Kevlar tow that he can't achieve with carbon fibre. Kevlar cloth might be a different matter as a layer in a graphite prop. Hartzell makes carbon/kevlar/foam props but I don't know whether the benefits would scale down.

With few exceptions, in the model world we don't take the quality of our engineering to those extremes. A typical home-made composite prop is likely to be molded in a mold that was pulled from an existing commercial wooden or plastic prop shape, or designed by what might be just "TLAR" ("That Looks About Right") methods. The mold will then be filled completely full with epoxy and fibers, and the weight of the result will therefore depend on just the density of the material (in that case the Kevlar will definitely be about 9% lighter than the carbon, assuming the same fiber/resin ratios). The carbon will be stiffer, but the Kevlar will also be very stiff (both materials much stiffer than glass, wood or nylon props of the same size and shape), and the Kevlar will have better vibration damping and much safer failure modes. The torsional stiffnesses of the two props will likely be similar, since in all probability the fibers will be almost all spanwise, making the stiffness of the epoxy a bigger factor in the torsional stiffness (although the spanwise stiffness of a laminate can influence its torsional stiffness if properly applied, as the spar design in some of our sailplanes has demonstrated, and as we have done in the wing design of our "Roadkill Series" indoor/backyard/park fliers).

The natural frequencies of the Kevlar prop will probably be lower, but much more damped, and therefore much lower in amplitude. Unless the RPMs are very high, and/or the blades are incredibly thin, and/or there are an unusually large number of blades (like four, five or six or more), the natural frequencies of both props are likely to be higher than the vibrational frequencies imposed on them by the airframe and motor.

If Gracindo already has the carbon lying around but has to buy the Kevlar, then the carbon will be cheaper. If he has neither, and shops around for the best prices before buying, then the Kevlar will be cheaper (although the increased production of carbon these days has reduced the difference somewhat in comparison to the approximately 2:1 or more ratio that it used to be).

As far as whether the benefits of Hartzell's concepts of Kevlar or carbon shells over a foam core would successfully scale down to model size, the answer is a definite "YES", but only if you optimize your design for your application. It's like using a tool; you can benefit from buying better tools, but only if you use them in the manner and for the purpose for which they were designed. The very best digital, carbide-jawed, calibrated with the Nat. Bureau of Standards, computer-output-port-equipped micrometer will not be significantly better performing than the cheapest hardware store micrometer if you're just using both of them as C-clamps! If you want to achieve improvements in your designs, you usually need to do your homework.

Don Stackhouse
DJ Aerotech