BOLTING WOOD PROPS - INCLUDING THE USE OF BELVILLE SPRING WASHERS
Paul Lipps (an innovative prop designer referenced in a previous issue of Contact Magazine) contacted me after reading the treatise on bolt preload. He indicated that he was employing an assembly of spring washers for installation of his composite over wood propeller He solicited my comments on this approach, and suggested that this would be a good basis for some analysis and an informative article. Since I had previously of some reference to this practice, I agreed with him on both counts, and started searching out reference materials
It is always a source of amazement for me, how things which are very simple on the surface, have a very complex nature when examined closely. Propeller bolts are just such an enigma. For a metal prop, things are virtually that simple. Just follow the usual bolt practice, and torque them up just short of yield, and the just keep them from backing out. For a wood (or primarily wood core composite) propeller, the “wicket” gets stickier than the proverbial tar baby. The crushing strength, and “crushing” modulus of most woods used are relatively modest, and the usual bolt torque tension loads would severely damage the wood structure. This is further aggravated by the dimensional changes of the wood as moisture is absorbed and released.
PROPELLER LOADS
Just how tightly do we have to secure a propeller? What are the forces and loads that are trying to take the prop off your airplane? The first thing you think of is the thrust forces that are pulling (or pushing) the plane around the sky. These loads are the least of our problems. Another, more troubling load is the gyroscopic precession as the plane direction is changed by pitch and yaw. With the lighter weight wood props, this is seldom a serious problem. However a big metal constant speed prop can react a load of as much as 600 ft lbs with a yaw rate of one radian per second. This can be a serious problem if one engages in violent aerobatics. The fabled Lomchavok uses this precession force to turn the stalled airplane end over end (and broken crankshafts have been know to occur).
The big need is for the clamping force, which acts much like the clutch disk in a manual transmission automobile. Although the classic prop hub has the drive lugs, the primary drive force is still this “clutch” action. If it were not for this friction, the prop would cyclically slip back and forth in the hub. The situation is further aggravated by the large displacement four cylinder engines typically used in aircraft. With only two power pulses per turn the peak torque values are higher than the rated steady values, and are actually cyclically reversed twice each turn. Once the shrinkage has reduced the preload, the cycling can induce alternating slippage at the flange face. The resulting heat further dries the wood, and a totally charred prop hub can result.
I have personally seen the result of just such a scenario. The Continental IO-240 has a small prop flange designed for metal propellers, aggravating this situation. On a flight to a local fly-in, engine roughness was noted as the destination approached. An expedited landing was initiated without problems, but as the engine was cut the propeller looseness could be visually seen. The hub of the wood prop was charred, and delaminated. A replacement was borrowed for the trip home, and the damaged prop now hangs on the wall as a visual reminder. This application normally employs a 4-inch prop extension, and an extension transitioning from the Continental hub to an S.A.E. number 2 flange was mandated for all subsequent installations.
The preload on a wood propeller must be moderated to avoid a crushing failure of the wood. The crushing strength of wood varies with species and density, ranging from about 1700 psi for maple down to about 840 psi for spruce. The rather aptly named “crush plate” for most props has about 18 square inches in bearing. Most of the wood varieties selected for propellers are on the high end of this range. Staying a bit below the high end at a target value of 1000psi, this would equate to a total clamping force about just under 18,000 lbs, or about 3000 pounds force for each of the six bolts. As stated in a previous article on preloading bolts, and as a general truism, determining preload on a bolt using a torque wrench is a very inexact measure. You might at first think that this is a relatively simple treatment of the analogy to driving a force up the inclined ramp representing the pitch of the thread. Sorry! No cigar. The component of the effective ramp angle is so obscured by the other friction forces, that it is totally ignored in the usual prediction, As most of you are aware, the coefficient of friction varies widely with surface finish, and degree of lubrication, as well as the properties of the two materials in rubbing contact. The usual assumption in this case is smooth steel to steel, lightly lubricated. Lightly lubricated generally means that you wiped off any visible liquid, but did not clean with any degreaser, which is about what you would do to avoid rust. Torqueing a bolt involves at least two surfaces turning against friction. The thread , of course, and the washer face of the bolt. The thread friction has a multiplier because of the vee angle of the thread, which is a much larger driver than the lead angle of the thread. Lumping all these forces and coefficients together for a 0.3 to 0.4 at the radius “arm” at the washer face of the bolt gives us a pretty good WAG estimate. The attached table of suggested torque value for the different classes of bolts in automotive use, is probably targeting about 75 percent of the allowable yield strength in the thread roots. These would also be typical of the values used for metal props, but would vigorously crush a wood prop.
looking at recommendations in engine manuals and propmakers, we see torque values for the typical 3/8in prop bolt in wood props ,from about 150 to 250 inch pounds. Using 200, and our above stated scheme that would be about 200/(0.22r*.35f)= @2600 lbs (not to far from my straw man case above). In spite of the “drive lugs (bosses) “ on the SAE hubs, the primary drive forces, and especially the torque reversals from four cylinder engines, are taken in friction. The clamped wooden propeller acts very much like the friction clutch in your “stick shift”. Sensenich uses birch for their wood props, and produce a high quality product with more experience in this field than almost anyone. One of their recommendations for torque is the number of degrees turned after initial “bottoming”. They are seeking a compression of 0.006 inch per inch of hub thickness. The modulus of elasticity for compression perpendicular to the grain varies greatly with wood species, and even at various points within the same log. Working values for the birch used by Sensenich would appear to be between 200,000 and 300,000 psi/in/in. This would suggest a preload of roughly 1500 psi under the “crush plate” which is pretty close to the crushing limit stated for birch in the files of wood properties. With the beam deflection in the crush plate, the effective area under each bolt would be less than 2 square inches, or a bit under 3000 lbf per bolt. Pressure on the “clutch plate” driving the prop would be on the order of 36000 pounds. With about a 0.4 friction and a 2.5in radius this would drive about 3400 ft lbs of torque without slipping – well over the steady state value for most likely engines,
However, the result of this situation is a spring loaded system, the primary spring being wood. A 0.006 in. per in shrinkage with moisture change would completely relieve the spring force. Actually it is even worse than that, since wood will notoriously “take a set” further reducing this spring load value
SPRING WASHERS
As mentioned earlier, Paul was installing his prop with spring (Belleville) washers. This is an approach that I had heard of being applied to wood props on the McCulloch drone engines used on gyrocopters, and one that I had often thought was a good idea for retaining clamping forces in use. The spring washers can provide high spring forces in a very compact package, and with a variety of stacking techniques, provide tailored combinations of force and deflection. They also provide visual monitoring of the preload on the bolt. The spring washers he chose, were from McMaster Carr (a well known supplier of industrial hardware). Two p/n 9712k32 . and one 9712K29 washers were stacked in parallel under each bolt. In parallel, the spring forces are additive. The specifications of these washers are:
Gardner# I.D. O.D. height thick deflection load flat load McMstr#
1187-105 0.406 1.188 0.125 0.105 0.016 1950 2812 9712K32
1000-105 0.406 1.00 0.118 0.105 0.010 1830 2657 9712K29
With 25 ft-lb torque (300 in-lb), the washers were not “flat”, which is as it should be. The specified total load for flat should be almost 8300 pounds, and the bolt load for 300 in-lbs torque would be expected to be about 3600 pounds, or bit less than half the bottoming load. and even a bit short of the 5700 rated displacement load. It would seem that this stack is a bit on the “stiff” side. A tailored stack of washers in combinations of parallel and series arrangement would provide the desired preload with greater deflection possible without losing too much preload. A selection of hardened flat washers should also be included to protect the softer aluminum “crush” plate.
Checking a recent catalog from McMasters-Carr, I ordered an assortment of the spring washers with ID values suitable for 3/8 inch (typical propeller bolt size) bolts. I think That I will be excused from any copyright usage I printing the selected portion of the McMasters- Carr cataloge showing the variety of possible choices for our 3/8 inch bolts. If you are using 7/16 inch or ½ inche bolts there are a similar number of choices 0n the subsequent page. Prices, are of course, subject to change, but are quite reasonable for the potential benefits derived.
Specifications of typical spring washers are:
ID OD Thick height deflect load flat load qty part# price qty part#ss price
The units selected for testing – shown to scale on a 3.\/8 inch bolt
And the force/ deflection characteristics are”
The spring washers may be “stacked” in various combinations to match the force and deflection characteristics that may be desired. Stacking in parallel = like Dixie cups. Will increase the loads for a given deflection. The forces are directly additive for this stacking arrangement. To provide more working travel for a given force change, a series stack can be used (point to point or “flare” to flare) The series stack tends to be a bit unstable, with the points and flares slipping out of alignment,O>C>Baker one of our KIS bulders suggested a larger washer inserted at the flare to flar intersection to stabilize the assembly (attached figures show this arrangement'. The force to flatten the discs is the same, but the travel to get that load is now doubled.
Two washers in series with the central stabilizing washer.
A four Belville stack combining series and parallel stacking.
With combinations of assemblies, a great number of force to displacement ‘curves” can be provided. For our purposes, we decided that a fairly large displacement, with a rather ”flat” force curve would be desired. What this would mean is that the wood could change dimensions (shrink or swell) over a fairly wide range without a large change in clamping force. For a birch prop with 3/8 in bolts, the characteristics of four number 428 washers in a series/parallel setup appeared the most promising. The total assembly would probably consist of a flat steel washer against the aluminum, the first two bellvilles point down, the second two points up, and a flat washer under the bolt head. This system appears to be nearly flat with roughly 200 inch pounds of torque, and close to 3000 pounds of clamping force for each bolt. Total deflection would be over .050 inch, and even a 010 change in shrinkage would result in little loss in preload, Different bolt diameters, and different wood hardness values would probably favor a different selection.
The local bearing forces of the sharp edges of the Bellville washers dictate the use of steel washers to protect the aluminum “crush plate”, and probably also the washer face on the associated bolt head or nut. This total “stack” would add fairly measurably to the selected bolt length for your installation.
These curves were generated using wrench angle as an indication of inches of compression. and reading wench torque value at each point. This is admiditly a bit crude, but the direction and rough magnitude can be seen. The four Belville stack of two 423 Belvilles in parallel on each side of the stabilizing washer shows a great travel range with low fall of in clamping force.
The result of such an installation would appear to be clearly a win, win situation. Weight and cost is quite minimal, and there do not appear to be any significant failure modes that have been increased by this system. Comments and suggestions are solicited.