Experience with Subaru at 358 Hours

[David-aviator]

This thread has gone too deep for most pilots - seems like one ought to be able to climb into an airplane and push the throttle up for a launch without wondering if all the chart curves are aligned properly or still even on the page. That's one reason I dumped the Subby H6 and moved on to a Lycoming, an engine intended to be in an airplane without such concerns.

I asked Allen Barrett about a conservative operating approach the last time I picked up an engine from their shop and he commented - don't worry about it, you can run this engine wide open as much as you like. And that is a true statement - I don't worry about it. I did with the H6.

That's not to say the subject is not interesting for some, but I do wonder at its value in accomplishing the RV mission, which is to blast off and have some fun flying.

My experience with an auto engine left me with the distinct impression the only way to get past all the questions raised here is go back to the drawing board and properly design an engine with an internal PSRU, as was done before WWII, and be done with it. The technology of this business certainly could accomplish that objective and do so safely. This business of taking a stock auto engine and trying to make it work with gear reduction unit hung on its forward end simply won't work for all the reasons expressed here which, quiet frankly, most of us do not understand, except to know it won't work in the long run.

 

[rv6ejguy]

This thread has gone too deep for most pilots - seems like one ought to be able to climb into an airplane and push the throttle up for a launch without wondering if all the chart curves are aligned properly or still even on the page. That's one reason I dumped the Subby H6 and moved on to a Lycoming, an engine intended to be in an airplane without such concerns.

I asked Allen Barrett about a conservative operating approach the last time I picked up an engine from their shop and he commented - don't worry about it, you can run this engine wide open as much as you like. And that is a true statement - I don't worry about it. I did with the H6.

That's not to say the subject is not interesting for some, but I do wonder at its value in accomplishing the RV mission, which is to blast off and have some fun flying.

My experience with an auto engine left me with the distinct impression the only way to get past all the questions raised here is go back to the drawing board and properly design an engine with an internal PSRU, as was done before WWII, and be done with it. The technology of this business certainly could accomplish that objective and do so safely. This business of taking a stock auto engine and trying to make it work with gear reduction unit hung on its forward end simply won't work for all the reasons expressed here which, quiet frankly, most of us do not understand, except to know it won't work in the long run.

My mission has never been to blast off and fly cross country primarily, it has been R&D. The airplane was built as a testbed for our EFI systems really and I wanted to use an auto engine too as I never had an interest in a Lycoming, not to mention I saved a bucket load of money in the process- about $35K in my case now. I've always been curious about things mechanical, like designing, machining, welding and fabbing things. So for me, this has worked out well but it is not for most pilots and builders here on VAF. At the same time, why is this thread so popular- because it is interesting. Even failure is interesting or perhaps entertaining?

Few here can appreciate the feeling of accomplishment and satisfaction of designing and building everything FF. To me that is important. As we were flying down to Reno a few years back, droning along, I was thinking wow, this is cool with the Sube purring along up front, I built all that stuff.. No problems the whole trip. Vans has done a great thing here for the masses who want a fun, reliable, versatile aircraft, no doubt about it but I have never been part of the masses. When friends had Z28 Camaros, I was blowing them away with Weber carbed Corvairs and turbocharged 1200 and 1600cc Corollas. I remember the purpose of experimentals was for education purposes, well I have received a lot of education from this project and the RV10 and it continues still as I refine the cooling system and other systems as well. I love to learn even if it is a bit painful at times.

Can't be successful? I think not. While my creation has not lasted as long as I hoped and still has TV issues, many others have lots of trouble free hours. I know of two with 2000+ hours and several race planes with many hundreds, the installations have never been touched. The 600+ RAF Gyros used Subaru power and had accumulated in excess of 125,000 flight hours as of 2006. They are very successful.

It is important to realize that not everyone flies the typical RV mission. Some of us use these great airframes for other purposes.

 

[DanH]

Well, all Subaru aircraft installations with light flywheels seem to have similar F2 frequencies, no matter what drive or prop as I have feedback from dozens of users from Flysoob and Subenews as well as a bunch of users I polled by email.

Then the subjects all have similar inertia and stiffness values.

If you remember about 4 years ago I did do a bifillar suspension test on a scrap 2L crank and sent you the results. I believe you said we'd need to calculate or model the stiffness to fill in the equation. We kind of stopped there.

We stopped because an accurate model requires the inertia and stiffness values for the Marcotte gearbox. All you had was a drawing sans dimensions, and you were not interested in taking it apart for measurements.

However, even a fictional model with typical numbers is illustrative. Let's lump inertias into three elements and then make changes to the crank/flywheel inertia and the stiffness connecting the crank/flywheel and the gearbox inertias. I'll use some values taken from an early Suzuki model, with pure guesswork for Marcotte inertia and propshaft stiffness.



Units are inertia in kg m^2 and stiffness in N m/rad.

Aluminum flywheel and stiff coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 1.898e+004
Inertia of disk 3 = 0.07

Natural Frequencies
f1 = 0 Hz
f2 = 67.76 Hz
f3 = 207 Hz

F2 mode shape
1
-1.618
-4.878

Steel flywheel (2.5x alum flywheel), stiff coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 1.898e+004
Inertia of disk 3 = 0.175

Natural Frequencies
f1 = 0 Hz
f2 = 48.28 Hz
f3 = 202.5 Hz

F2 Mode Shape
1
-0.3293
-2.172

Steel Flywheel (2.5x alum flywheel), soft coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 2576
Inertia of disk 3 = 0.175

Natural Frequencies
f1 = 0 Hz
f2 = 22.08 Hz
f3 = 163.1 Hz

F2 mode shape
1
0.7221
-2.352

The first model approximates what you have now. We know it's not an accurate model of your current Subaru system because the predicted F2 is 68hz, meaning it would hammer at about 2000 RPM, not the 1250 ballpark you report. Still it is useful to compare with models 2 and 3.

#2 swaps the flywheel. I've merely increased inertia 3 (the crank/flywheel) by a factor of 2.5

#3 is the heavy flywheel plus a soft connecting element between the flywheel and the gearbox.

Compare the examples and you see what you can expect, in very general terms:

Increased flywheel inertia is good. A reduction in connecting stiffness is more effective at lowering frequency. The mode shapes change; models 2 or 3 greatly reduce crank oscillation at the accessory end (remember your beat-up vac pump drive?). The F3 in model 3 is creeping into the top of the operating range; it may resonate at 163 hz, or 4890 RPM for your 4-cyl 4-stroke.

Sure beats cut and try

 

[rv6ejguy]

Then the subjects all have similar inertia and stiffness values.



We stopped because an accurate model requires the inertia and stiffness values for the Marcotte gearbox. All you had was a drawing sans dimensions, and you were not interested in taking it apart for measurements.

However, even a fictional model with typical numbers is illustrative. Let's lump inertias into three elements and then make changes to the crank/flywheel inertia and the stiffness connecting the crank/flywheel and the gearbox inertias. I'll use some values taken from an early Suzuki model, with pure guesswork for Marcotte inertia and propshaft stiffness.

Units are inertia in kg m^2 and stiffness in N m/rad.

Aluminum flywheel and stiff coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 1.898e+004
Inertia of disk 3 = 0.07

Natural Frequencies
f1 = 0 Hz
f2 = 67.76 Hz
f3 = 207 Hz

F2 mode shape
1
-1.618
-4.878

Steel flywheel (2.5x alum flywheel), stiff coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 1.898e+004
Inertia of disk 3 = 0.175

Natural Frequencies
f1 = 0 Hz
f2 = 48.28 Hz
f3 = 202.5 Hz

F2 Mode Shape
1
-0.3293
-2.172

Steel Flywheel (2.5x alum flywheel), soft coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 2576
Inertia of disk 3 = 0.175

Natural Frequencies
f1 = 0 Hz
f2 = 22.08 Hz
f3 = 163.1 Hz

F2 mode shape
1
0.7221
-2.352

The first model approximates what you have now. We know it's not an accurate model of your current Subaru system because the predicted F2 is 68hz, meaning it would hammer at about 2000 RPM, not the 1250 ballpark you report. Still it is useful to compare with models 2 and 3.

#2 swaps the flywheel. I've merely increased inertia 3 (the crank/flywheel) by a factor of 2.5

#3 is the heavy flywheel plus a soft connecting element between the flywheel and the gearbox.

Compare the examples and you see what you can expect, in very general terms:

Increased flywheel inertia is good. A reduction in connecting stiffness is more effective at lowering frequency. The mode shapes change; models 2 or 3 greatly reduce crank oscillation at the accessory end (remember your beat-up vac pump drive?). The F3 in model 3 is creeping into the top of the operating range; it may resonate at 163 hz, or 4890 RPM for your 4-cyl 4-stroke.

Sure beats cut and try

Indeed, wow, thanks Dan. Very interesting!

I was somewhat worried about changing the soft element part and the flywheel together, possibly moving F3 into the operational range of the engine which would make the whole airplane useless. Right now, I am assuming it is somewhat beyond 4500 rpm. I have operated up to 5000-5100 for speed runs years ago. Super smooth there but I'm unlikely to be able to detect much without instrumentation at those frequencies also.

A friend with a 6 cylinder EG33, steel flywheel (about 16 lbs) and same M300 gearbox, 43 pound prop has a slightly noticeable period at about 800-900 rpm. Everywhere else is super smooth. He runs up to about 5100 rpm for the races and has 450 hours on it now I believe, bushings not hammered.

The 4 cylinder guy replaced the 7 lb aluminum flywheel with a 17 lb. steel one after about 150 hours because the TV scared him, bushings were pretty toasted after only 50 hours. The second time around was night and day he said. Way smoother in the old problem ranges. M200 gearbox and very light Warp drive prop.

 

[chaskuss]

snippedI noticed that the bottom of the piston crowns had fried Mobil 1 varnish on them and Mobil 1 cokes at something over 500F.
snipped

The economics work if you don't count the 400 extra hours of build time to do the conversion and 200 hours of debugging and doing stuff over again.

Ross,
I've just gotten to this thread. Very interesting topic. I'm wondering if the coked up oil layer insulates the pistons. This would reduce the effectiveness of your oil squirters and would add to other things working to ruin the temper of your piston rings. How often do you change the engine oil? Perhaps reducing oil change intervals would help to prevent/slow the layer on the under side of the pistons?

Don't feel bad about your time. Aviation in general is a black hole of time and money. Just remind yourself that this is a hobby. [ A very time intensive one]

You've probably already thought of this. [or perhaps someone has already mentioned it in the upcoming posts] Those damaged rod bearings are going to allow a lot more oil to be sprayed up onto the cylinder walls. The bearing damage has to be a factor [possibly the major one] in your increased oil consumption. Is there more oil coking on the cylinders with the damaged bearings? My apologies if I am stating the obvious.
Charlie

 

[rv6ejguy]

Ross,
I've just gotten to this thread. Very interesting topic. I'm wondering if the coked up oil layer insulates the pistons. This would reduce the effectiveness of your oil squirters and would add to other things working to ruin the temper of your piston rings. How often do you change the engine oil? Perhaps reducing oil change intervals would help to prevent/slow the layer on the under side of the pistons?

Don't feel bad about your time. Aviation in general is a black hole of time and money. Just remind yourself that this is a hobby. [ A very time intensive one]

You've probably already thought of this. [or perhaps someone has already mentioned it in the upcoming posts] Those damaged rod bearings are going to allow a lot more oil to be sprayed up onto the cylinder walls. The bearing damage has to be a factor [possibly the major one] in your increased oil consumption. Is there more oil coking on the cylinders with the damaged bearings? My apologies if I am stating the obvious.
Charlie

Oh the varnish on the underside of the piston crowns was not even .001 thick, could see through it, and small in area- about 1 inch square. The rod bearings only had centralized and tiny damage, additional oil throwoff would be minimal as the rest of the bearing clearance was stock. Oil changes at 35 hours generally. Also all pistons and chambers appeared to have pretty equal deposits on the top, not just the 2 holes with bearing damage. All 4 top rings were relaxed so I am guessing that is where most of the blowby came from.

I am not too disappointed, I had aimed for 400 hours to take the thing apart 9 years ago, ended up just a bit short. I did not think it was ever going to go 1000 hours- too many unknowns.

 

[rv6ejguy]

M300

Hey Dan, I sent you an email with some dimensions and drawings of the Marcotte box. Hope that may be enough to get a closer model worked out.

 

[DanH]

Hey Dan, I sent you an email with some dimensions and drawings of the Marcotte box. Hope that may be enough to get a closer model worked out.

You do the grunt work. I've broken the gearbox drawing into component parts to show you which assemblies get hung, and which sections get calculated stiffness.

First step is an accurate calculation of torsional stiffness for each of the two shafts.

Then you'll need to hang each of the two shaft assemblies for bifilar measurement of inertia.

Next you need inertia for the propeller.

You need the stiffness value for the coupler. You'll need a physical measurement of force vs angular deflection that can be translated into lbs-ft/radian or similar units.

Last you need inertia for the crank/flywheel assembly including rod small ends and harmonic balancer (stupid name).

I've included a white paper for shaft stiffness. If you can't find your bifilar white paper let me know. It's on my other computer.

 

[flyvans.com]

guys, please keep this discussion going (even if it only affects the vast minority of rv builders)! interesting stuff to follow.

even though i have no clue about either mathematics or mechanical engineering, i've followed the whole alternative engine/psru etc... topic for years...

understanding the challenges is the first step to an eventual solution/viable alternative to the old-fashioned and expensive (but reliable, simple, working, relatively light etc..) aviation engine.

unfortunately, due to the various small or one-man "inventor" operations with the need to generate money quickly, most developments have been trial and error banana products that ripened at the customer or early adopters...

hats off to all the engineers that try to put science before beliefs and measure wishes against reality! only those will succeed in the long run (5+ years)

good luck, bernie

 

[DanH]

A friend with a 6 cylinder EG33, steel flywheel (about 16 lbs) and same M300 gearbox, 43 pound prop has a slightly noticeable period at about 800-900 rpm. Everywhere else is super smooth. He runs up to about 5100 rpm for the races and has 450 hours on it now I believe, bushings not hammered.

This is off-topic, but illustrates a good point.

Same gearbox, probably the same pinhead coupler (), but installed on a 6-cylinder and incorporating a higher inertia flywheel and a high inertia prop. Resonant RPM is in the 800-900 area rather than 1250 or so like yours. Why?

The flywheel and prop make a contribution, but don't assume they are entirely responsible for the different resonant RPM. Doing so would miss the effect of more cylinders.

Let's assume you bolt the same flywheel/coupler/gearbox/prop assembly on three different engines of 4, 6, and 8 cylinders. Further assume when mounted on the 4-cyl it resonates at 1250 RPM, just like yours. It's a match of natural frequency (an F2 of 41.6 hz) and forcing frequency (torque oscillation due to firing events). Those pulses are banging the system at a rate of (RPM x #cyls)/120 for a 4-stroke engine, or 1250 x 4 /120 = 41.6 in your case.

OK, bolt the system on the 6-cyl engine. To predict the resonant RPM we just turn the equation around:

Natural frequency x 120 / #cyls = resonant RPM, or 41.6 x 120 / 6 = 832

Now bolt it on the 8 cylinder. It will resonate at 41.6 x 120 / 8 = 624 RPM

Same drive, but merely bolting it on an engine with more cylinders lowers the resonant RPM.

 

[Denase]

Very interesting thread

I have been following this thread. Dan your profile shows you are a commercial car and truck dealer, are you also an engineer?

 

[LoopDLoop2]

Welcome here and I appreciate your interest. The piston is JE as described on the web page. The oil ring tension, well, frankly never paid attention to this as I have never had an oil consumption problem before- tried to duplicate the same same successful recipe I've used to build engines professionally for quite some time. So, the answer is- I have no idea.

I don't own a blowby meter but I could show the the belly of the aircraft- blowy at the end was "severe". But when built originally, the belly was completely dry after 35 hours and stayed that way for over 200 hours.

Chrome got me home for 35 years building turbocharged engines... and was the choice of the Fuji engineers who designed this engine. Perhaps old school but it still works very well in this application.

This was honed at a pretty flat angle which I have found works fine too.

CK10 also dated like me but again, gave good results with a good operator, engines worked fine.

Subaru engines do not use head bolts into the block deck like most other engines. A torque plate would have minimal effect IMO. Of course "round" and "straight" are relative terms and I agree, the rounder and straighter, the better. With old equipment and measuring stuff, I do the best I can.

What is TBO on a Pro Stock engine? Curious. What BMEP at torque peak and power peak?

TBO on a NHRA 500 CI Pro Stock Engine is about 16 to 18 runs. The make peak torque at around 7900 to 8200 depending on manifold and peak the power around 9800 to 10100 depending on the same. They make in the 3 hp per ci.

I looked at the bearing pictures and if I had to say right off what caused the flaking it would by detionation. This would corolate with the loss of ring seal. If you are going to use the same top ring part # you could measure the free gap. This would be what the end gap is on the bench after you fit it to the bore. Then measure the one that came out of the motor. Just because the piston is not burnt up it does not mean it is not beating it self to death.

 

[DanH]

I have been following this thread. Dan your profile shows you are a commercial car and truck dealer, are you also an engineer?

I have little talent for advanced math, thus did not pursue engineering as a profession. Sometimes I regret it, sometimes not.

This particular subject became interesting after bits of a drive system went streaming past the cockpit in the summer of '98. One thing led to another....education and recreation run amuck.

 

[SvingenB]

Then the subjects all have similar inertia and stiffness values.



We stopped because an accurate model requires the inertia and stiffness values for the Marcotte gearbox. All you had was a drawing sans dimensions, and you were not interested in taking it apart for measurements.

However, even a fictional model with typical numbers is illustrative. Let's lump inertias into three elements and then make changes to the crank/flywheel inertia and the stiffness connecting the crank/flywheel and the gearbox inertias. I'll use some values taken from an early Suzuki model, with pure guesswork for Marcotte inertia and propshaft stiffness.

Units are inertia in kg m^2 and stiffness in N m/rad.

Aluminum flywheel and stiff coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 1.898e+004
Inertia of disk 3 = 0.07

Natural Frequencies
f1 = 0 Hz
f2 = 67.76 Hz
f3 = 207 Hz

F2 mode shape
1
-1.618
-4.878

Steel flywheel (2.5x alum flywheel), stiff coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 1.898e+004
Inertia of disk 3 = 0.175

Natural Frequencies
f1 = 0 Hz
f2 = 48.28 Hz
f3 = 202.5 Hz

F2 Mode Shape
1
-0.3293
-2.172

Steel Flywheel (2.5x alum flywheel), soft coupler:

Inertia of disk 1 = 0.39
Stiffness = 2.7e+004
Inertia of disk 2 = 0.03
Stiffness = 2576
Inertia of disk 3 = 0.175

Natural Frequencies
f1 = 0 Hz
f2 = 22.08 Hz
f3 = 163.1 Hz

F2 mode shape
1
0.7221
-2.352

The first model approximates what you have now. We know it's not an accurate model of your current Subaru system because the predicted F2 is 68hz, meaning it would hammer at about 2000 RPM, not the 1250 ballpark you report. Still it is useful to compare with models 2 and 3.

#2 swaps the flywheel. I've merely increased inertia 3 (the crank/flywheel) by a factor of 2.5

#3 is the heavy flywheel plus a soft connecting element between the flywheel and the gearbox.

Compare the examples and you see what you can expect, in very general terms:

Increased flywheel inertia is good. A reduction in connecting stiffness is more effective at lowering frequency. The mode shapes change; models 2 or 3 greatly reduce crank oscillation at the accessory end (remember your beat-up vac pump drive?). The F3 in model 3 is creeping into the top of the operating range; it may resonate at 163 hz, or 4890 RPM for your 4-cyl 4-stroke.

Sure beats cut and try

Interesting. I assume the f3 mode shape is one of + - + ? What happens if you keep the original setup, but decrease the stiffness of the coupling between the flywheel and marcotte?

 

[rv6ejguy]

TBO on a NHRA 500 CI Pro Stock Engine is about 16 to 18 runs. The make peak torque at around 7900 to 8200 depending on manifold and peak the power around 9800 to 10100 depending on the same. They make in the 3 hp per ci.

I looked at the bearing pictures and if I had to say right off what caused the flaking it would by detionation. This would corolate with the loss of ring seal. If you are going to use the same top ring part # you could measure the free gap. This would be what the end gap is on the bench after you fit it to the bore. Then measure the one that came out of the motor. Just because the piston is not burnt up it does not mean it is not beating it self to death.

Interesting. Don't take this the wrong way but how does a TBO of 125 seconds relate in any way to prime mover engines especially ring "life" and how do you quantify oil consumption with a total engine life of 2 minutes?

Hmmm, detonation damage on the bearings? I think not. I've been building turbocharged engines for a long time and we can/ have encountered detonation very frequently. NEVER seen a rod bearing damaged by detonation- EVER, pushing 3.5hp/cubic inch and the engines are going 25-50 hours before overhaul and we are doing this down around 7500 rpm so the BMEP is very high, about 50% higher than the Pro Stock engine at power peak. First thing to fail is either top ring lands or head gaskets and that pretty well signs the engine off right there before anything else can be damaged. Clearly this is cavitation damage. On prime movers, if you encounter heavy detonation, because of the the continuous high power, the engine will certainly die. Many engines will last though 7 seconds of heavy detonation, very few will take 1 minute or 5-10 minutes of it. I hold takeoff MAP through 500 feet and make a slight MAP reduction for the climb where the engine is often held here for 10 to 15 minutes. There were zero signs of detonation on this engine and I'm very familiar with the results of this.

The rings were going south for over 150 hours but the oil consumption never got above 1 qt in 6 hours ( about the same as lots of Lycomings I flew). This is not nearly enough oil to significantly lower the octane rating of the fuel to cause detonation.

If you read the web page I already stated that ring gap was the same as when the engine was assembled 9 years ago- in other words no wear on the rings or bore.

 

[Mike S]

....education and recreation run amuck.

Much better mental image than the potato

 

[DanH]

Interesting. I assume the f3 mode shape is one of + - + ? What happens if you keep the original setup, but decrease the stiffness of the coupling between the flywheel and marcotte?

Side-by-side comparison, with gearing included. The gear correction means multiplying the faster-rotating inertia and stiffness values by ratio squared. In this case I've assumed 2/1 gearing. So, as compared to previous models, the coupler and crank/flywheel values have been increased by a factor of 4.



Remember, this a fictional system, an example for purposes of illustration. It is similar to but not exactly what Ross is flying. We don't have the real stiffness and inertia values, and the above is a "lump" model (for example, the entire flywheel, crank and accessory sections of the engine are represented by a single inertia). A really good model would use carefully measured values and break the system into several more inertia/stiffness pairs. The only point in posting these values is to demonstrate how applying known principles/procedures is better than randomly changing hard parts.

 

[David-aviator]

Side-by-side comparison, with gearing included. The gear correction means multiplying the faster-rotating inertia and stiffness values by ratio squared. In this case I've assumed 2/1 gearing. So, as compared to previous models, the coupler and crank/flywheel values have been increased by a factor of 4.

Remember, this a fictional system, an example for purposes of illustration. It is similar to but not exactly what Ross is flying. We don't have the real stiffness and inertia values, and the above is a "lump" model (for example, the entire flywheel, crank and accessory sections of the engine are represented by a single inertia). A really good model would use carefully measured values and break the system into several more inertia/stiffness pairs. The only point in posting these values is to demonstrate how applying known principles/procedures is better than randomly changing hard parts.

When MT came to Florida to do the vibration survey of the MT-7 and the Egg H6 engine, they attached strain measuring devices to the engine and prop.

I believe your computer generated data with various components would have to checked in a similar manner to prove its worth.

For sure an airplane can be designed, built and safely flow based on computer data, Boeing does it successfully but flight tests are necessary to verify the end result. As far as I know, the FAA has not certified an airplane without actual flight testing.

Same can be said for your efforts in determining vibration data or whatever you are attempting to determine. It needs to be flight checked.

Unfortunately in this business, flight checking alone is not the answer. Engine-psru-prop combinations are being assembled with little science, briefly flight checked, and off to market we go. The end result can be a less than satisfying experience for many pilots.

 

[DanH]

When MT came to Florida to do the vibration survey of the MT-7 and the Egg H6 engine, they attached strain measuring devices to the engine and prop. I believe your computer generated data with various components would have to (be) checked in a similar manner to prove its worth.

Absolutely. Math models are design and diagnosis tools. Live measurement of the flight article validates the model...the vital reality check.

 

[rv6ejguy]

I'd mention that the drive ratio on this aircraft is 2.2 to 1 and I just want to thank Dan again for his time to help educate us in this field.

I will be doing inertia tests on the prop and any other parts before I re-install then on the aircraft and try to do some reasonably accurate calcs on the internal gearbox parts as far as stiffness and inertias.

The last chart laying everything out side by side shows the relative effects of changes in certain elements. It will be cool if I can gather enough accurate data to be able to have the math model agree closely with practice. We'll be able to have some more confidence in finding a better solution to all this.

It will take a while to do this because of my work and plane work schedule. At the moment, I am working on electrical components, airframe inspection and the new radiator/ cooling system layout. Once I get the flywheel back, I can do some tests and get the engine back in. This is a fairly big project in itself- something over 300 hours I am guessing.

 

[LarryT]

I have survived...

...43 years and counting as a practicing mechanical engineer with nothing more than Algebra II. Despite the fact that I was forced to take 7 semesters of college math starting with Calculus, then diffyQ etc., in my entire career I used calculus one time! That was a canned program to integrate lining force to offset the piston and outboard disc brake shoe to prevent tapered wear in a single piston sliding caliper disc brake. Maybe you gave up early. What I read here convinces me that you have more intellectual curiosity and ability than most of the engineers I worked with.

LarryT

I have little talent for advanced math, thus did not pursue engineering as a profession. Sometimes I regret it, sometimes not.

This particular subject became interesting after bits of a drive system went streaming past the cockpit in the summer of '98. One thing led to another....education and recreation run amuck.

 

[rv6ejguy]

Dan is a very talented guy on many levels I think. His creative skills and workmanship are simply amazing.

 

[DanH]

Ross, you're welcome. Larry, thank you.

One additional illustration, the previous table with frequency translated to RPM and vectors drawn as mode shapes. The three disks, from left to right, represent the prop inertia, the gearbox inertia, and the engine inertia. The connecting lines are the propshaft (left) and coupler (right).



The aluminum flywheel/stiff coupler example is pretty bad. When in resonance at 1653 RPM (a common operating range) the prop inertia rotates in opposition to the gear and engine inertias. A large torsional strain is applied to the propshaft, which is acting as a powerful undamped spring. As a practical matter, the strains would make shaft detail design critical. For example, any section change would have strong potential as a stress riser.

Couple the aluminum flywheel to a soft coupler and things improve for the propshaft, but the engine inertia is now oscillating in opposition with significant amplitude.

OK, switch to a steel flywheel and go back to the stiff coupler. What you get is a less severe version of the first (bad) example.

Steel flywheel and soft coupler. The prop and gearbox inertias now oscillate together; propshaft stress is low. The big twist is now found in the soft coupler, a component designed for the task. The resonant frequency is moved to a point lower on the RPM range, meaning lower on the engine's torque curve, a reduction in the power of the forcing frequency. And it may now be outside the operating range. Remember, this system is geared 2/1. We want a prop RPM below 600 or so, meaning an engine RPM of 1200, so 817 is now an RPM we only pass through for a short period during startup.

Ok, enough about torsional mechanics. Let's return to the original point of the thread. Ross, here's a little brain teaser. How could severe crankpin oscillation (like the first and second examples) lead to cavitation damage? Hint; think about the mechanics of hydrodynamic lubrication. Spitballin' a little on this one, but I'm curious if you might start having the same thought.

 

[Lars]

Thoughts coming to my mind are oil column resonance and effects of increased local oil pressure (at the rod journals, compared to the main bearing journals) due to centrifugal acceleration. Possibly interacting with the torsional vibration.

Dan, these bearings are hydrostatic (pressure lubricated) as opposed to hydrodynamic (lubricated due to motion and geometry of the bearing/journal). Similar, but different. Nice discussion of the two concepts here.

 

[Randy]

Dan and Ross,

There are quite a few previous Egg customers out here probably wondering how to apply some of this information to our situation.

Can either of you comment more on the dual mass flywheel and it's use as a dampener in our situation?

We do have some data indicating that the dual mass flywheel provides some relief, as in wear on the spline adapter between the flywheel and the gearbox, but some of this effect could be due to alignment issues. It appears from the data that proper alignment and use of the dual mass flywheel add up to less wear on the spline shaft and connecting parts.

From my own experience with with the earlier Gen 2 gearbox and solid flywheel, it obviously had some TV issues at lower engine idle RPMs and the cure was to idle the engines up around 1500 or so to stay out of the "rattling range". Switching to the Gen 3 box and dual mass flywheel stopped the rattling almost completely. It shakes pretty hard on shut down, but I do not notice any strong vibrations in operation otherwise. I can idle at "normal" idle speeds with no obvious adverse effects.

I guess the one thing we have going for us is the hours that continue to build on use of light props in front of the Gen 3 gearbox and both 4 and 6 cylinder engines behind them. I understand that does not mean there is not a problem there but it is nice to know that there have been no catastrophic failures that I know of. Some welds have cracked and a few bearings have failed. No gear failures that I know of.

Also, what sort of failure mode would be expected from torsional vibration? Gear teeth breaking? Shaft? Splines? Probably all of the above?

I appreciate all the information shared so far on this thread and I would sure like to find a way to put it to good use...

Randall Crothers

 

[DanH]

Dan, these bearings are hydrostatic (pressure lubricated) as opposed to hydrodynamic (lubricated due to motion and geometry of the bearing/journal).

It's going to be a big surprise in certain circles...

A respected bearing manufacturer:

http://www.kingbearings.com/files/Engine_Bearings_and_How_They_Work.pdf

A design engineer who happens to like experimental aircraft stuff:

http://www.epi-eng.com/piston_engine_technology/engine_bearings.htm

And a few PhD's:

http://www.substech.com/dokuwiki/doku.php?id=geometrical_parameters_of_engine_bearings

http://lib.tkk.fi/Diss/2009/isbn9789522481627/isbn9789522481627.pdf

In the fun reading category...a roundtable discussion of the state of the art with industry experts:

http://www.enginebuildermag.com/Article/80015/a_view_from_the_summit.aspx

 

[Lars]

hydrodynamic vs hydrostatic

I suppose it's at least partially a matter of semantics. I designed a few hydrodynamic bearings for an offshore application (seawater lubrication) many moons ago. I was repeatedly corrected by several bearing engineers that I consulted with at the time: if it was pressure lubricated it was a hydrostatic bearing, regardless of what was going on with the fluid film.

Point being that hydrostatic bearings require an external source of fluid, supplied under pressure, to function as advertised. Take the external source away and they stop functioning. Hydrodynamic bearings (apparently turbines in some hydroelectric projects are an example) work simply by dint of rotating, while submerged in the fluid.

Didn't get too far into that King Bearings paper before the radar went off. Apparently the author has never seen the inside of a 2 stroke motorcycle engine. Plenty of rolling element bearings to be found on crankshafts there. But I digress...

 

[rv6ejguy]

Dan,

Yes, when you mentioned this earlier in the thread I thought about this for the first time (thanks) but as I said, most Subaru fours have similar, rather serious F2 problems and those which have been taken apart have not shown cavitation damage to my knowledge plus we only operate there for a few seconds each flight. F3 would appear to be above my operational rpm range but we don't know for sure, maybe I am cruising in F3 but amplitude is lower or frequency is too high to detect with the MK I body sensors.

Randy,

If someone with and EGG setup wants to do the inertia and stiffness tests and calcs we could see if the math model fits a second example and could then see how scary things might or might not be. The dual mass is already much heavier than my Marcotte flywheel and it is damped as well which is likely quite helpful. The original solid setup with no damper was likely not so good and could have been the key to some of the failures.

Much to learn on both setups, it would be awesome to first have the math agree with what we can easily feel at F2, then if we could instrument and confirm each resonant point, then we'd really be able to say and design something concrete. I wish I was retired and could concentrate a lot of time an energy on this to fully understand it. It is fascinating.

 

[DanH]

Can either of you comment more on the dual mass flywheel and it's use as a dampener in our situation?

Sure. Short answer....it is not dual mass, nor is it a damper.

Long answer....Jan removed the second mass, but kept the spring system. That makes it (in the context of the previous illustrations) a steel flywheel with a built-in soft coupler....a clever use of available components.

In an engineer's world the term "damper" has specific meaning. It is generally a device which removes kinetic energy from a system, usually in the form of heat. It is often paired with a spring, in parallel. You have four viscous dampers under your car. They are called "shock absorbers", which again is a really wrong name.

Put another way, a spring is an energy storage device, while a damper is an energy dissipation device. In torsional terms, the stiffness connecting two inertias is a spring, not a damper.

From my own experience with with the earlier Gen 2 gearbox and solid flywheel, it obviously had some TV issues at lower engine idle RPMs and the cure was to idle the engines up around 1500 or so to stay out of the "rattling range". Switching to the Gen 3 box and dual mass flywheel stopped the rattling almost completely. It shakes pretty hard on shut down, but I do not notice any strong vibrations in operation otherwise. I can idle at "normal" idle speeds with no obvious adverse effects.

You're describing exactly what should be expected, the difference being coupler stiffness. Sounds like the the F2 for the latter "soft" system is indeed below idle, a great place to put it.

 

[DanH]

I suppose it's at least partially a matter of semantics.

I hear you brother.

 

[scsmith]

detonation damage vs. cavitation damage

This is a fantastic thread and I have really been enjoying it.

I think I can add one tidbit to the conversation from my practical experience in ill-begotten youth.

I have run an engine for prolonged periods of time in MILD detonation. Namely a Volvo B-20 engine with CR=11.5. After about 1500 hrs, the engine was torn down, and all the rod bearings revealed some damage. The nature of the damage was that the bearing material was "wiped" away, indicating metal-to-metal contact over a very narrow band at TDC. This was indicative of very high pressure (enough to collapse the oil film) over a very short duration, i.e. during detonation.

Bear in mind, these old Volvos are tough engines with stiff cranks and large bearings. Of the dozens of them that I have looked inside over the years, I never saw a Volvo main or rod bearing that showed ANY wear at all, except this one.

I have also seen the same characteristic rod bearing wear on an old Triumph motor that was habitually driven hard on low-octane gas. Again, sustained periods of MILD detonation.

My point: The flaked or "popcorned" bearing damage that Ross found is NOT in my mind indicative of detonation.

 

[rv6ejguy]

This is a fantastic thread and I have really been enjoying it.

I think I can add one tidbit to the conversation from my practical experience in ill-begotten youth.

I have run an engine for prolonged periods of time in MILD detonation. Namely a Volvo B-20 engine with CR=11.5. After about 1500 hrs, the engine was torn down, and all the rod bearings revealed some damage. The nature of the damage was that the bearing material was "wiped" away, indicating metal-to-metal contact over a very narrow band at TDC. This was indicative of very high pressure (enough to collapse the oil film) over a very short duration, i.e. during detonation.

Bear in mind, these old Volvos are tough engines with stiff cranks and large bearings. Of the dozens of them that I have looked inside over the years, I never saw a Volvo main or rod bearing that showed ANY wear at all, except this one.

I have also seen the same characteristic rod bearing wear on an old Triumph motor that was habitually driven hard on low-octane gas. Again, sustained periods of MILD detonation.

My point: The flaked or "popcorned" bearing damage that Ross found is NOT in my mind indicative of detonation.

Yes, the old B series Volvo engine had very robust bottom ends and the same goes for many other types which you can rattle lightly almost continuously (Slant Six Dodge comes to mind). Nothing really breaks suddenly on these. Overloading and fatigue bearing failures look a lot different than mine, I've seen a bunch of these developing engines with very high specific outputs and we've had to find solutions. I have a great old bearing book with many color photos of various failure modes. They all show pretty much what I have on mine. Here is a god web based one from Mahle/ Clevite: http://catalog.mahleclevite.com/bearing/ I think section 19 shows the cavitation photos, you can see fatigue failures generally show larger areas of less severe damage as you'd expect from detonation/ overloading.

 

[rv6ejguy]

Ok, this has intrigued me enough, I had a talk with my electronics/ software guru here and he's interested to start a bit of testing and development for a way to instrument these things and do some actual measurements. Not sure how long it will take as it is a spare time project and we are busy now but... he loves new challenges and often does R&D at home after work.

We have some components in mind we have worked with before that could make this pretty easy- I hope.

 

[DanH]

Ok, this has intrigued me enough, I had a talk with my electronics/ software guru here and he's interested to start a bit of testing and development for a way to instrument these things and do some actual measurements. Not sure how long it will take as it is a spare time project and we are busy now but... he loves new challenges and often does R&D at home after work. We have some components in mind we have worked with before that could make this pretty easy- I hope.

Or you could just use standard equipment and start next week

Examples:

http://www.atitelemetry.com/viewproduct-9043.htm

http://www.binsfeld.com/index.php/products/TT10K/

http://www.datatel-telemetry.de/en/Datasheet/en/Auto-Zero_Single_Channel_Telemetry_Transmitter.pdf

.....and more.

Below is 1999....the stub in the prop center is an old WDC analog telemetry transmitter, the lowest of low tech, already obsolete as digital gear had reached the market. My buddy Ron is at the throttle and I'm recording data with a clipboard and pencil. That's an ancient 15mhz BK scope for signal and vibratory waveform (an Air Force castoff that arrived in a friend's scrapyard for free), and a true RMS Fluke for amplitude. Plenty of space on this propshaft thus a big 'ole Wheatstone bridge for torque.

Point is, you don't need to invent the wheel (or have a NASA budget) to know exactly what sort of vibratory torque(s) are present in your system, and at what RPM. The only unknown is finding a good physical arrangement for the strain gauges.

 

[rmeyers]

Fascinating thread. So fascinating that I signed up just to comment. Been visiting this forum occasionally for some time while trying to get the energy to finish my 6A.

I have a question for Dan. In setting up for a Holzer analysis, how does one determine the correct length for the 'equivalent length' shaft?

I have 'Mechanical Vibrations' by Hartog, an earlier version than yours apparently, since I do not have the 4 cylinder diesel generator example. It does have a 6 cyl diesel marine engine directly driving a prop shaft and prop.

In this example, it seems that he uses the cylinder spacing for the equivalent shaft length. Does this imply that we are placing our equivalent mass at the place where the actual center of gravity would be for the real crankshaft piece? That would seem to make sense, but you know how that goes....

 

[Mike S]

Welcome to VAF!

Fascinating thread. So fascinating that I signed up just to comment. Been visiting this forum occasionally for some time----

Randy,

to VAF

Good to have you aboard.

 

[DanH]

I have a question for Dan. In setting up for a Holzer analysis, how does one determine the correct length for the 'equivalent length' shaft?

Sounds like you're attempting a detailed model incorporating each crankthrow as an individual inertia/stiffness pair. I have a reference at home (BICERA) which would be helpful, but here's a path you'll like...

The classic reference work is "Practical Solution of Torsional Vibration Problems" by W. Ker Wilson. Later editions are a multiple (5 or 6?) volume work and cover every detail you can imagine. Out of print, so you look for it in a university library. I have borrowed the 1956 edition from the University of Alabama on inter-library loan. They even have it at Auburn and Georgia Tech

Go here, click "Find in a library", enter your zip code:

http://catalog.hathitrust.org/Record/009911016

That said, hey, this is the internet. The copyright has apparently expired on the original 1935 edition, see below, start at page 79. You can download single page PDF's:

http://hdl.handle.net/2027/uc1.$b530640

Second edition Vol 2:

http://hdl.handle.net/2027/uc1.b4335883

While poking around in a Google search for Ker Wilson I ran across this...nice discussion in Sec 2, and it references both Wilson and the BICERA book:

http://www.engdyn.com/images/uploads/86-torsional_vibration_guidelines_-_tdf&clh.pdf

 

[rv6ejguy]

Or you could just use standard equipment and start next week

Examples:

http://www.atitelemetry.com/viewproduct-9043.htm

http://www.binsfeld.com/index.php/products/TT10K/

http://www.datatel-telemetry.de/en/Datasheet/en/Auto-Zero_Single_Channel_Telemetry_Transmitter.pdf

.....and more.

Below is 1999....the stub in the prop center is an old WDC analog telemetry transmitter, the lowest of low tech, already obsolete as digital gear had reached the market. My buddy Ron is at the throttle and I'm recording data with a clipboard and pencil. That's an ancient 15mhz BK scope for signal and vibratory waveform (an Air Force castoff that arrived in a friend's scrapyard for free), and a true RMS Fluke for amplitude. Plenty of space on this propshaft thus a big 'ole Wheatstone bridge for torque.

Point is, you don't need to invent the wheel (or have a NASA budget) to know exactly what sort of vibratory torque(s) are present in your system, and at what RPM. The only unknown is finding a good physical arrangement for the strain gauges.

We like building stuff... You learn more by building it than buying it and maybe have some commercial applications down the road. Often we can apply some lessons learned in developing new products.

 

[scsmith]

where to measure torque?

Unless you have a prop extension?
I can't imagine a place on my engine-propeller combination where I could put a strain gage.

Maybe somewhere inside the gearbox? I would think they would package it pretty tight so there is no bare shaft exposed anywhere.

 

[DanH]

I can't imagine a place on my engine-propeller combination where I could put a strain gage.

Physical space for the strain gauges is a classic problem.

In the case of the drive system above, I knew I was going to strain gauge it when I designed it. The usual choice for a prop extension is a machined aluminum spool. The steel tube with flanges was much harder to fabricate, but it worked nice.



Something like the Egg Gen3 would be difficult to strain gauge, which is why some of us snorted when Jan claimed to have done it.

The Marcotte gearbox Ross is using should be easy. It has an inner bore in the right place for a pre-made Wheatstone (red):

 

[rv6ejguy]

We are going to try accelerometers rather than strain gauges. Could be challenging... Need pretty high frequency response. I have a 2 cylinder 2 stroke on the test stand at the moment, should be a good candidate to prove the basic hardware and software on.

 

[DanH]

Can you tell us your plan? What do you hope to determine and how will you do it?

 

[rv6ejguy]

Can you tell us your plan? What do you hope to determine and how will you do it?

Without going into too much detail in case it crashes and burns, use at least two or three, 2 axis accelerometers to measure nominal and peak G up through the rpm range on the flywheel, drive ring and prop shaft.

Unlike honing cylinders, grinding valves, inspecting bearings and tightening hose barb connections, I have never done this before...

I read an article a few years back on someone trying this method and they encountered a lot of problems. I can't find the article any more unfortunately. This was being applied to industrial diesel engines driving pumps and conveyors if I recall correctly.

I welcome any comments from you or any engineers with experience in this field.

 

[Lars]

Without going into too much detail in case it crashes and burns, use at least two or three, 2 axis accelerometers to measure nominal and peak G up through the rpm range on the flywheel, drive ring and prop shaft.

Unlike honing cylinders, grinding valves, inspecting bearings and tightening hose barb connections, I have never done this before...

I read an article a few years back on someone trying this method and they encountered a lot of problems. I can't find the article any more unfortunately. This was being applied to industrial diesel engines driving pumps and conveyors if I recall correctly.

I welcome any comments from you or any engineers with experience in this field.

We have used accelerometers to determine resonant frequencies- torsional, axial, radial- on the centrifuge at the lab where I work. It's not my area of expertise, but we have some faculty with a lot of experience with the systems. I can ask if you'd like. The issues are similar, just scaled up a bit. Lab website here.

 

[David Paule]

The sensor location is critical for measuring natural frequencies. Position really matters. And unless you have several in appropriate locations, you'll only capture the lowest one or two.

Dave

 

[SvingenB]

We are putting a whole logging system inside a 300 MW turbine to measure strain and pressure pulsations right now. The whole system will be sealed off, powered by battery.

Personally I would use strain gauges rather than accelerometers to measure torsional vibrations. The reason is that torsional strain is the main strain there, and it is straight forward to measure with a strain gauge, and the torsional strain is what you are after in the first place. Using accelerometers you may pick up radial vibration, they may cause lots of accelerations due to imbalance, but little strain. Thus you will just as easily pick up modes that have nothing to to with torsional vibrations. If you want to measure imbalance, accelerometers is the way to go though, or displacement gauges so you can measure orbit.

Positioning is also straight forward with a strain gauge, anywhere along the shaft will be fine (considering a mostly lumped system). With accelerometers you have to hit the torsional modes where the amplitude is large, the problem is you don't know exactly where that is, and may in theory end up positioning the accelerometer where you measure nothing.

Why not just copy Dan's setup?

 

[rv6ejguy]

Thanks for the input and comments here, all useful food for thought.

In my mind, I am actually more interested to see what each element is experiencing and least concerned with the prop shaft itself other than obviously this a great place to instrument with a strain gauge and would show us the basic picture between the exciting forces and major inertias.

We saw the vacuum pump drive coupling wear which is apparently not typical on a Lycoming, we saw the drive bushings heavily distorted so know they are under a lot of stress at times and we see no problems with the IVO propeller which is sensitive to large torque variations as failures have been seen on direct drive engines with them. The IVO uses 2 bolts longitudinally, loaded in shear to retain each blade. The blades are flat section at the hub end, clamped between 2 knurled plates. When subjected to high torque variations, they have been known to shake themselves to bits in fairly short order. The prop has been 100% trouble free and shown zero signs of any distress to date.

So I think I really want to see the dynamics of the coupler, what kind of a job it is doing or not doing and of course to see what rpm ranges we have concerns at and the peak amplitudes.

It would be very hard to instrument the engine/ flywheel element with a Wheatstone, ditto for the drive plate/ input gear shaft.

IF this works, it may give a more complete picture. If it doesn't work, well it wouldn't be the first time I spent some time on something without good results and I'll probably still learn something- plus we have something to fall back on and still get the basic data.

Right now I am pounding in the time- about 30-40 hours a week, working on the ventral rad plumbing, cleaning up electrical stuff, cabin ventilation/ heating etc. The modified flywheel should be back mid week from the CNC shop. I plan to do the inertia tests then on the prop and flywheel for Dan's math model before installing them. I then have the rad scoop to build (big project) the intercooler mount and plumbing because it is being repositioned and some other airframe inspection tasks still remain plus painting the scoop and various cover planes where multiple air scoops used to be. Lots of little machining and welding jobs too. I estimate at least another 150-200 hours.

So, it will be a while before she runs again...

 

[rv6ejguy]

We have used accelerometers to determine resonant frequencies- torsional, axial, radial- on the centrifuge at the lab where I work. It's not my area of expertise, but we have some faculty with a lot of experience with the systems. I can ask if you'd like. The issues are similar, just scaled up a bit. Lab website here.

I would be interested in any info that might be available using accelerometers to measure vibrational parameters.

Also, does anyone have an idea of what G ranges we might be looking at to pick the accelerometers? I have no idea on the forces we might see here. The types we want to try here are available up to 400G. We have a 50G model to try already on the bench and test engine.

The 2 cyl. test engine will run well before the Subaru just to see if the concept is even viable first.

Oh, we do have a really good idea where F2 is already. Believe me, you don't want to stay here long! I would guess the peak amplitude is 20-30 times nominal here.

Would it be accurate to say that high radial accelerations must necessarily produce higher strains?

 

[SvingenB]

Also, does anyone have an idea of what G ranges we might be looking at to pick the accelerometers? I have no idea on the forces we might see here.

From 0 to anything. That is (one of the main) problems using accelerometers for your application. If you take a closer look at Dans calculations, that is exactly what they tell you: http://www.vansairforce.com/community/showpost.php?p=742009&postcount=73

If you put an accelerometer on the coupling you might not even get the F2 (stiff coupling and steel wheel). Now, if you measure strain somewhere on the shaft, you will pick up everything. In addition you will get the exact strain/stress. Using accelerometers you will have to assume you have the correct mode shape (100% correct) and then calculate the forces. Then you want to get F3 as well. You need 2, preferably 3 accelerometers, you only need 1 strain gage and it will do a better job than the accelerometers.

 

[DanH]

......use at least two or three, 2 axis accelerometers to measure nominal and peak G up through the rpm range on the flywheel, drive ring and propshaft .

You intend to mount them to the rotating components?