<<<<<<<<<<<<<<<<--->>>>>>>>>>>>>>>>
         <<  Lancair Builders' Mail List  >>
         <<<<<<<<<<<<<<<<--->>>>>>>>>>>>>>>>

Posted on behalf of Jack Kane <Epijk@aol.com>:

Gentlemen:

A few days ago I submitted this article to the LML. Apparently, it was somehow truncated during transmission. In addition, due to my ignorance of the LML restrictions, the format in which it was submitted did not readily translate into an acceptable format, resulting in lots of garbage-characters being included in the text. I apologize for any inconvenience you may have experienced, and offer the following cleansed 7-bit re-submission, with a few sentences made more clear.

Thank you.

             SOME PSRU DESIGN ISSUES

This forum has hosted several discussions of various theories and intuitions about PSRU's. Therefore, I think it would be appropriate to offer up some facts on the subject, based upon the design, implementation and testing of one of the few geared PSRU's which has demonstrated its performance and reliability, both on the test stand and in the air.

The EPI PSRU is a part of the EngineAir Power Systems 440 HP liquid-cooled, turbocharged V8. That engine system powers (among others) Tom Zedaker's Lancair-4-P which won Grand Champion at Sun n' Fun this year.

The prototype EPI PSRU was in the aircraft which won the Kittyhawk-to-Oshkosh race in 2000 (and which nearly won in 2001). That prototype has over 650 hours of flight time in Lancair-4 aircraft, driven by turbocharged engines producing up to 500 HP. It weighs 71 pounds and includes the integrated control plumbing and gear-drive for a standard Woodward-Hartzell prop governor.

DESIGN SUMMARY

The EPI PSRU was designed using a system approach to the problem, based on a severe-service aircraft load model, and taking into account the following loads:

(a) the bearing, shaft and housing loads produced by:
torque,
gear separation,
thrust, and
gyroscopic forces,
(b) the static gear tooth loads from MEAN engine torque,
(c) the dynamic gear tooth loads resulting from the
    stiffness and vibration characteristics of the system
       (these CAN greatly exceed the static loads),
(d) the cooling load imposed by the power transmitted,
         and
(e) the lubrication requirements.

The PSRU subjects (briefly) addressed here are:

     VIBRATION FUNDAMENTALS
     ENGINE TORSIONAL VIBRATION
     PSRU and PROPELLER VIBRATION
     GEARS AND BEARINGS
     LUBRICATION AND COOLING
     GYROSCOPIC LOADS
     CONCLUSIONS

(A more complete treatment of the broad subject of PSRU's, including belt, chain and gear reductions, will be available in a few weeks on our website, www.epi-eng.com.)

VIBRATION FUNDAMENTALS

Everyone knows that any piece of metal has mass. Any piece of metal also demonstrates the properties of a spring. That is, if you apply two equal and opposite forces to opposite sides of it, it will deflect. Sometimes that deflection can be seen; sometimes it is so small that it can't be measured with a micrometer. That depends on the size of the applied force and the dimensions and properties of the piece of metal. The amount of deflection caused by a specific force determines the "spring rate" of the metal piece.

Any system which has mass and spring rate will vibrate at it's resonant (natural) frequency (like a tuning fork) when struck ("excited"). If it is repeatedly excited at or near its resonant frequency, the vibrations will increase in magnitude until something breaks. As the excitation frequency is increased beyond the resonant frequency, the vibrations become smaller and smaller until they virtually disappear. At frequency ratios above six, the vibration amplitude is less than 3% of the excitation.

ENGINE VIBRATION

An even-firing 8 cylinder, 4-stroke engine produces a power pulse once every 90° of crankshaft rotation. Therefore, the waveform of the instantaneous full-throttle torque output (at the crankshaft flange) has four torque "peaks" which are over 200% of the instantaneous mean torque output (the torque which the dyno measures), and four torque "valleys" which are approximately 25% of mean torque. (Fortunately, that waveform is approximately sinusoidal.)

A crankshaft, like a plain torsion-bar, has mass and a torsional spring rate. That means that the crankshaft system has it's own torsional resonant frequency. The instantaneous torque peaks and valleys described above cause portions of the engine crankshaft to deflect forward and backward (relative to its rotational motion) while it is turning. When those pulses (excitations) are near the crankshaft resonant frequency, they can cause the crank to vibrate ncontrollably, causing potentially high stresses which can lead to fatigue failure.

The torsional resonant frequency of the crankshaft system is primarily a function of:

(1) crankshaft length;
(2) crankshaft torsional stiffness;
(3) crankshaft stroke;
(4) bobweight mass;
(5) moments of inertia of rotating items
    attached to or driven by the engine.

Because of the nature of the coupling between the engine and its load, the torsional vibration characteristics of the engine crankshaft need to be addressed with the characteristics of that coupling taken into account. If you believe that a crankshaft can live for long without an effective torsional attenuator on the free end, look to the experience of the Nissan folks. The crankshaft in the very early 240-Z inline 6-cylinder engine didn't have an effective attenuator, and therefore lasted only about 100 hours in automotive service {i.e. VERY LIGHT DUTY}.

Previous discussions have mentioned the vibration attenuating devices on the free end of an engine crankshaft. Often, these are incorrectly referred to as "DAMPERS". In most cases, they are ABSORBERS. (That's not semantics. A damper DISSIPATES energy, typically as heat. An absorber alternately STORES and RELEASES energy to counteract vibration.)

The elastomeric ("metal-ring-on-rubber-spring") devices used by the automotive industry (as well as by Teledyne-Continental Motors on the GTSIO-520)are ABSORBERS which are tuned to counteract vibration at the frequency where the particular engine generates its worst torsional excitation.  Continental and Lycoming also use internal ABSORBERS in all their high-output engines. Those absorbers consist of pendulous counterweights attached to the crankshaft cheeks by loose pins in hard bushings. The clearance between the pins and bushings establishes the torsional order which each counterweight absorbs.

One aftermarket device, the "Fluidampr" (trademark) really is a DAMPER. It dissipates energy by transforming it into heat by shearing action in a high viscosity fluid. Some race car people seem to like it, but it is banned from Winston Cup racing.

That type of damper is mildly effective over a wide range of excitations, but contrary to intuition and hype, it is significantly LESS effective in reducing vibration in a specific, targeted frequency range, the exact situation you have in an aircraft engine. (There is ample research in the engineering literature showing exactly that fact.)

PSRU VIBRATION

The engineering literature is rich with information on the subject of gearboxes. An American Gear Manufacturer's Association (AGMA) publication summarizes the problem as follows:

"The gearbox is one component of a system comprised of a power source, gearbox, driven equipment, and interconnecting shafts and couplings. The dynamic response of this system depends on the distribution of the masses, stiffnesses, and damping. In certain cases, a system may contain a torsional natural frequency close to an excitation frequency associated with an operating speed. Under these resonant conditions, the dynamic gear tooth loads may be very high, and operation near a system resonance is to be avoided."

Clearly, a PSRU cannot be treated as an isolated entity. It is one critical component in a vibrating system having at least two natural frequencies (in addition to the natural frequencies of major engine components such as the crankshaft system) and two sources of excitation. Any change to any part of the system can significantly alter the loadings imposed on the other parts of the system.

The torsional vibration properties of the engine have already been covered.

A propeller produces torsional excitation which varies with rotational and translational speed, flight attitude, airframe characteristics, and the properties of the engine torsional excitation which are applied to the prop.

Worse yet, each prop blade has more than one resonant frequency. The one excited by thrust variations is different from the one excited by torsional variations. Prop blades are especially susceptible to destructive vibration if it is excited near a resonant frequency. Certified prop manufacturers go through extensive analysis and testing to be sure that a particular prop will survive the fatigue environment produced by a particular engine on a particular airframe.

Get it wrong, and blade pieces will eventually be departing the aircraft.

So there is the fundamental problem: both the input and the output of a PSRU are attached to torsional excitation machines. That suggests there are two primary functions of a PSRU:

(1) To reliably transmit engine power to the
    propeller;
       and equally important,
(2) To isolate the gears from the engine and
    prop excitations and from each other.

Most of the intuition about PSRU's suggests that substantial DAMPING is necessary for successful operation. In fact, the exact opposite is true. The mathematics which define the system make it abundantly clear that, if your goal is to minimize the torsional vibration transmitted to the gearbox (and thus to the prop), then you must minimize the TRANSMISSIBILITY (the ratio of the amplitude of the applied vibration to that of the resulting vibration).

Low transmissibility is ONLY available above the crossover frequency, and the lowest transmissibility is achieved with ZERO damping. That is how the EPI system is designed and implemented.

The EPI engine coupling system removes over 99% of the torsional excitation which the engine produces at takeoff power, which makes the input to the gearbox nearly as smooth as that of a turbine, and provides the prop with an operating environment which is essentially devoid of engine torsional excitation.

That is accomplished by designing the system so that the first mode resonant frequency (with respect to engine-order excitation) is below idle speed and the second mode resonant frequency is over four times above max RPM. That provides a frequency ratio greater than 9.0 at cruise and a transmissibility from engine-to-PSRU of approximately 0.01. The transmissibility from prop-to-PSRU is roughly 1.04.

As a PSRU purchaser, you should be aware that those critical resonant requencies undergo potentially significant change whenever any of the following properties of the system are changed:

(1) the reduction ratio;
(2) the propeller mass moment of inertia;
(3) any property of the engine which changes
    its mass moment of inertia (accessories,
    bore, stroke, etc.),
(4) any property of the gearbox which alters
    the torsional stiffness of ANY shaft.


GEARS AND BEARINGS

The gears in the EPI PSRU are straight-cut (spur gears), made from a specially heat-treated E-9310 (manganese-nickel-chrome-molybdenum) alloy, which have a face width of only 1.5 inches (in the model suitable for 600 LB-FT of engine torque). The gears have special design features which greatly improve their resistance to both hertzian and bending fatigue, as well as to prevent the edge-loadings which are so destructive to highly-loaded gears.

The design of the EPI system allows the gears to operate in an environment in which the tooth dynamic loads are only slightly greater than the static mean-torque loads, with a safety factor of over 2.0 using conservative allowable stress levels.

(HOW CAN THAT BE, you ask, if the peak torque values are over 200% of the mean torque?? You already know the answer, because you just read the VIBRATION section above.)

Intuition often leads to the claim that helical gears, having a contact ratio in excess of 2.5, are superior to our spur gears, which have a contact ratio of slightly over 1.6. However, most gear engineers know that contact ratio is not the major discriminator between designs. Again, the mathematics and engineering of gear design show that, while helicals do have a greater contact ratio than spur gears of the same of the same pitch diameter and diametral pitch, helicals suffer from the inherent problem of highly-asymmetric tooth loading (edge-loading). They also produce substantial thrust loads, but that is less of an issue. The only advantage helicals have over similarly designed spur gears is less noise (hardly an issue in an aircraft engine application!). The disadvantages they have are significant.

Rolling element bearings which are sufficiently robust to provide satisfactory calculated life factors in a realistic PSRU load model are unacceptably large and heavy. The EPI PSRU does not use rolling element bearings where the large loads are (on the input and output shafts).Therefore, the design bearing life at severe load levels is conservatively calculated to be well over 2000 hours. In-service inspections suggest a much longer life.

LUBRICATION AND COOLING

Obviously, gears and bearings must be lubricated. Intuition would suggest that high-viscosity gear lubricant is required in a PSRU. Experience has proven quite the opposite. However, an equally important (and often ignored) function of the lubrication system is COOLING of the gears and bearings.

Why? Because involute gear tooth contact is SLIDING motion (not rolling motion, as some believe). For well-lubricated gears of high quality, the amount of power which is converted to heat by tooth friction is approximately 1/2% per mesh. Low quality gears generate quite a bit more heat. The temperature of highly-loaded, carburize-heat-treated gears and of rolling element bearings should never exceed 250 deg.F.

Those cooling requirements suggest the need for a continuous flow of relatively cool oil. It should be clear that a high-power PSRU with its own separate oil supply cannot survive without a significant oil cooling system, which implies at least one pump and one more heat exchanger (complexity and COOLING DRAG).

In order for a coolant to be able to carry a significant amount of heat energy away from an object with a modest coolant flowrate, the entry temperature of the coolant must be substantially less than the temperature of the hot object to be cooled. If that is not the case, then the coolant flowrate must be correspondingly higher, and the net effectiveness of the heat transfer will generally be lower. No matter what the flowrate is, the temperature of the items being cooled can never be less than the exit-temperature of the coolant, and usually are quite a bit higher than that exit temperature.

As an example, consider a two-mesh gearbox with high quality gears, transmitting 400 HP. The contact between the gears generates a heat load of about 10,000 BTU per hour. Depending on the specific heat and specific gravity of the lubricating oil, and the oil temperature rise deemed acceptable, the oil flow rate required just for cooling can exceed 3.5 GPM.

Bottom line: Ask to see test data that verify gearbox oil flowrate as well as the entry and exit temperatures, measured after a half-hour of steady-state operation at the rated power level. Also consider that if 3.5 GPM (or more) of oil goes into the PSRU, there must be some suitable method to get it back out again. Gravity? Try to pour 3.5 gallons of warm ATF through a 3/4" hose in one minute.

GYROSCOPIC LOADS

The dynamic loads imposed on a PSRU come from several sources, and can be very large. In addition to dynamic tooth contact and bending loads, gear separation forces, and propeller thrust, it is sometimes forgotten that gyroscopic moments can impose severe loads on not only the housing, but on the propshaft and the bearings which support it, as well as to the engine mounting structure and the points where it attaches to the fuselage.

Consider, for example, a particular (certified) 600 HP V8 engine/gearbox with a particular (certified) 3-blade metal prop. If the prop on that powerplant is turning at 2057 RPM and the powerplant is subjected to the FAA Part-23 (14-CFR-23.371-a-2} yaw-rate, the resultant gyroscopic moment of the propeller alone will produce over 3580 LB-FT of bending moment. That is over 5 TIMES GREATER THAN THE ENGINE TORQUE and over 2.3 times greater than the PROPELLER TORQUE (the torque which the engine mount is designed to handle).

With typical PSRU bearing placement, that moment alone imposes over 7000 pounds of force on each propshaft bearing. When coupled with the over-1400 pounds of gear separation force generated by 600 LB-FT of engine torque (on 20 degree PA gears), the bearing loads can be high enough to severely reduce the expected life of a rolling element bearing. (Recall that the EPI PSRU does not use rolling element bearings on either the propshaft or the input gear shaft.)


Another gyroscopic issue often ignored (regarding the engine mounting structures used to install V8 engines on aircraft) is the analysis for WHIRL MODE.

Whirl mode is a divergent oscillation of the engine and mount about an axis parallel with the static thrustline.

It occurs because of the positive feedback loop which occurs when a large, sudden gyroscopic moment is applied to an engine mount structure which is insufficiently STIFF (as differentiated from STRONG).

The feedback loop occurs because a gyroscopic moment, which is produced by a pitch or yaw velocity, is applied in a plane which is 90° to the angular (pitch or yaw) velocity which caused it. If the excitation mode couples with an engine mount resonant frequency, the prop-end of the powerplant can begin to whirl around an axis parallel to the static prop centerline, describing a cone of ever-increasing base diameter.

For example, suppose that an aircraft, which has an engine mounting structure with insufficient STIFFNESS, encounters a violent vertical gust which causes it to pitch-up rapidly. If the prop rotation is clockwise, the upward pitch excursion produces a prop gyro-moment which tries to deflect the engine mount structure to the right. If the mount is sufficiently flexible, it would deflect rapidly to the right (yaw), which would generate a downward gyroscopic moment on the engine mount. The flexible mount deflects downward (pitch), causing a yaw-left moment. That 90-degrees-out-of-phase excitation continues, and if these excitations and deflections occur at some harmonic of the natural frequency of the engine/mount system, the deflections can develop into a whirling deflection of the engine structure of increasing amplitude until something breaks.

Early in the days of turboprop-powered aircraft, there were several inflight breakups caused by engine nacelle departure from the aircraft because of whirl mode. It happened because the engines were much lighter and much longer than the radial engines of comparable (or less) power. The longer mounting structures, which were designed to carry the g-loads appropriate for the powerplant weight, were relatively more flexible than those designed to carry the heavier and shorter radial engines.

A similar scenario is possible with a high-powered, relatively light V8. The engine mount sufficient to bear the g-loads might not have sufficient stiffness to prevent whirl. It is less likely to occur with the low mass-moment-of-inertia (MMOI) composite props commonly found on experimental aircraft, but a higher probability exists with a large MMOI metal prop. It is a subject which demands analysis in each installation.

CONCLUSIONS

(1)  Perhaps the most important point I hope to make is that PSRU's, propellers and engine mounts are complex devices. Generally speaking, PSRU's are the incarnation of EXPERIMENTAL. There is no significant body of accumulated experience behind many of them. Treat them as such.

(2)  The life expectancy of your PSRU and propeller (and by extension, you, your passengers and your aircraft) depends on a complex mixture of factors including the metallurgical goodness of the gear material (AGMA GRADE), the accuracy of the gears (AGMA QUALITY), the dimensions, stiffnesses, metallurgical goodness, heat-treatments, and quality control of critical components, the vibration characteristics, the reduction ratio, the propeller, the talent of the designer(s) and builder(s), and other significant factors such as accumulated reliability experience in the actual application environment.

Test-stand operation IS NOT THE SAME AS FLIGHT!.

(3)  Be VERY SUSPICIOUS of any PSRU which allows the use of a light flywheel and/or which has a torsionally-rigid drive between the engine and the PRSU drive gear. RUN (don't walk) away from anything with a one-way clutch in the propeller drive system, or any device with magical (unexplainable) properties. Be careful and be cynical.

(4) Before buying a PSRU, have the seller satisfy you (with data, not allegations) that the SYSTEM will operate correctly with YOUR selection of engine, propeller, and gear ratio. YOUR LIFE DEPENDS ON IT.

(5)  Do not be misled by the false belief that "if you cannot FEEL vibration, then none exists". Destructive vibration can occur at frequencies far beyond the range humans can detect.

(6)  Be cautious of most race-car experts. There are some brilliant race-engine guys out there, but their perspective is usually quite different than yours. Remember, long life for a typical ski-boat engine is 100 hours. Long life for a Winston Cup or Busch engine is 3 hours. (SOME RACE-CAR GUYS THINK LONG LIFE IS 18 SECONDS, APPLIED IN 6-SECOND CHUNKS).

(7)  It is well recognized that the Federal Aviation Regulations defining design and testing standards (PARTS 23, 33 & 35) do not apply to experimental aircraft.

Nevertheless, those regulations provide a rich source of MINIMUM DESIGN STANDARDS, evolved from years of experience (read ACCIDENTS). A designer
would be well advised to draw from that source, especially with regard to structural and reliability issues.

Jack Kane
EPI, Inc.



PS: A recent submission on alternate engines quoted me as having said:

"Turbine props are guaranteed by their makers to fall apart if subjected to the torque impulses and harmonics generated by a V8."

FOR THE RECORD, I'd like to clarify that quote.

What I said about turbine props was:

"The blades have a much thinner root section than those designed for PISTON ENGINE applications because turbines do not produce the torsional excitation that PISTON ENGINES do. Therefore, a turbine blade section is more likely to fail in FATIGUE if used on a piston application.

It is NOT guaranteed to 'fall apart'. It is just significantly more likely to fail in FATIGUE. And it has NOTHING WHATEVER to do with V8's. A 4-cylinder piston engine is likely to  fail a turbine prop faster than is a 6 or an 8.

The fact that the prop on my dyno (off a Garrett TPE-331-10) has not failed yet means essentially nothing, other than the fact that it hasn't failed yet. The only definitive data comes from a vibration survey, done by Hartzell (or McCauley) on the COMBINATION of a PARTICULAR engine and PARTICULAR gearbox and PARTICULAR propeller, installed in a PARTICULAR airframe. All the rest is just calculation and speculation.

The insidious thing about aluminum in a fatigue environment is that it HAS NO ENDURANCE LIMIT (as do most ferrous alloys). Therefore, the fact that an aluminum part has run for x-number of fatigue cycles is NO GUARANTEE that it will not fail during the next hundred similar cycles."

LML website:   http://www.olsusa.com/Users/Mkaye/maillist.html
LML Builders' Bookstore:   http://www.buildersbooks.com/lancair

Please send your photos and drawings to marvkaye@olsusa.com.