Torsional Dampening

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
That's why I'm interested in the planetary; I'm pretty sure it's lighter than a stiff system, and it should be significantly cheaper, with mostly off the shelf parts.

Charlie
I will point out the importance of optimizing systems vs optimizing components. If it doesn't make a better airplane, you have made a worse airplane with the efforts.

While the planetary gearset may look lighter than the stiff system gears, it requires some other hardware to work. The soft element is not trivial, requiring structure and significant spring mass. If it has elastomeric springs, they can be bulky, particularly if they are to see low enough loading for survival and to stay cool enough for long life. Then there is the issue that a two mass isolator has poor isolation with one mass much larger than the other. The engine has much smaller inertia than the prop, and the isolation performance (f/fn ratio) improves with added inertia at the engine side. Yeah, you could further reduce spring rate but your spring mass (and associated hardware on both prop side and engine side) will go up with the stored energy which goes up rapidly with reduced rates. Ross at SDS has talked about the benefits of added flywheel inertia on his bird. Optimal weight should be pursued in the design space because WEIGHT IS THE ENEMY.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
Sorting out the rotary's processes is a bit hair-pulling for me, too, and I've been paying attention since my father bought an RX-4 in 1974.

The key, I think, is combining 7-6, the text, and 7-7. I wish he'd have included a graph overlaying the individual rotor curves, before showing the mean torque in 7-7. Here's a link to a later edition with overlaid individual rotor curves, plus the summed torque:
http://foxed.ca/rx7manual/manuals/REbyKenichiYamamoto-1981.pdf
Fig 1-32 on pg 8

Note that he mentions that it takes 1080 degrees (3 rotations of the crank) to complete the 4 'strokes' of one cycle, for a single face of one rotor. Divide by 4 & you get 270 degrees, more or less (probably closer to 250 for actual power), for each 'stroke', instead of a <180 degree power stroke in a piston engine. But at the 180 degree point, the 2nd rotor fires.

To be honest, I'm still struggling with 7-6, myself. It almost looks like chamber 1's torque line morphs into chamber 3's. Perhaps it's because when chamber 1 is in its 'compression stroke', chamber 3 is firing. Maybe it's supposed to be a composite curve for a single rotor(?).

I'm looking forward to Billski's report once he dives a bit deeper into these docs.

Thanks, Billski,

Charlie
Charlie,

Go back to the same pages, and look at the tangental torque plot. Positive torque happens for 180 degrees on each chamber, is only really high a little ways after zero degrees. While the instantaneous torque is greater than the mean torque, the eccentric shaft is accelerating, and while the instantaneous torque is below the mean torque, the engine is decelerating. In a single rotor that is easy to see, and the speed changes are big with firing happening at 1x times rotation speed. A twin rotor superimposes a second plot 180 degrees out from the first, which does moderate the magnitude of the speed changes, but they are still there. Firing with a two rotor is 2x rotation, and you still get vibe inputs at 1x, 2x, and 4x firing frequency that you have to manage...

Bill
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
There are 3 faces per rotor, the diagram on page 85 figure 7.6 is clearly labeled as a single rotor with 3 faces. If you superimpose an identical graph delayed by 180 degrees to represent the second rotor (also with 3 faces) you will see the negative torque peak is under a high positive torque area of the second rotor.
Looking at the charts, they do not plot the negative torques from the exhaust and induction strokes - they just show them as zero through there. Let's talk about what is missing, and how they likely got that.

On dynamometers, where data like this would be taken, it is common to run suction on the exhaust pipes to prevent filling the dyno cell with exhaust gases. This can result in artificially reducing pumping losses during the exhaust strokes, make the engine torque look better than it is, and the vibe behaviour look better than it is;

The incredible noise produced by Wankel type engines necessitates a substantial muffler. Even well designed mufflers have some back pressure extending to the rotors, and contribute negative torque (not zero as shown). At medium to high engine speeds, this can be substantial;

Induction with a normally aspirated engine results in somewhat lower pressure in the combustion chamber than atmospheric. The larger the engine speed the more vacuum during induction. While this may be fairly small at WOT, at any reduction in throttle, the negative torque during induction will be substantial.

In short, the data presented is a best slanted to tell the story that their new engines are great, and everyone should be buying them.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
There are 3 faces per rotor, the diagram on page 85 figure 7.6 is clearly labeled as a single rotor with 3 faces. If you superimpose an identical graph delayed by 180 degrees to represent the second rotor (also with 3 faces) you will see the negative torque peak is under a high positive torque area of the second rotor.
Richard,

Correct, but be careful about taking a sales pitch at face value. In another post I talked about how they are representing pumping (exhaust and intake work) as zero, while real engines with mufflers will have exhaust work and real engines with less that WOT will have substantial intake work too. Not shown...

As I talk about elsewhere, the issue is not about whether torque goes to zero, it is about the eccentric flange speeding up and slowing down.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
Looking at the rotary's eccentric (crank?) shaft, one possibility that looks plausible is to extend a quill shaft through almost its entire length, which would make for a compact and lightweight soft system reducing the transmissability of the forcing vibrations to a very small percentage. I see no reason why the oil supply to the rotor bearings could not flow around the quill; it would need a seal or a tight annular aperture (a bearing on a quill journal?) at the end...
I like it! There are several challenges:

Can we get a low enough spring rate on that length of quill shaft and maintain durability?

Can we remove enough of the inside of the eccentric shaft to put in that quill shaft while maintaining durability?

Can we support the now output end of the quill shaft adequately to run the gearbox?

Can we seal in the pressurized oil paths in the eccentric shaft that has to lube the shaft bearings and lube/cool the rotor system?

It does present another issue. If viscous damping is needed, you won't supply it this way, all you will have is elastic capability with a clearly defined natural frequency for the system.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
OK, I took the time, speed read much of the Yamamoto book, read it carefully where it talks about torques.

It looks like a combination of brag and encouragment to other companies to buy ToyoKogyo's technology.

In the torque variation, they presented highly filtered low rpm WOT data, and skipped any indications of exhaust back pressure or induction vacuum. That improves their story and does make their engines look better for vibe than they are. Yeah, they have about 270 degrees of eccentric shaft power stroke available, but for much of it, pressures and thus torques are way low, so it is not like they actually use all of the long stroke to make power. They also did not publish their mean torque - any torque less than mean torque means the engine is decelerating then. How convenient...

They also talk about amplitude of firing order vibration using the same inertia in several different engines. I could not tell if they included counterweights needed in the inertia or not. If they did not include the needed counterweights, they were comparing higher rotating inertia on their engine against lower inertia on the others. Convenient, eh?

I am still trying to wrap my head around the vibe content, and the book did nothing for us there. More on that after I sleep on it. I was hoping it would help with insights into vibe spectrum.

In total, I still believe that when listening to someone's opinions, you have to know what they are selling.

Billski
 

fredoyster

Well-Known Member
Joined
Mar 20, 2012
Messages
333
Location
Monterey Bay, CA
I wonder why no one has proposed making a dog clutch damper as found on Rotax gearboxes. For those not familiar, this is not the multiplate "slipper clutch" intended to protect the crank in case of a prop strike and/or lessen starting kickback. Every 912 has a two piece dog clutch tensioned by a stack of Belleville washers, with the object seemingly to decouple the output shaft on torque peaks, just enough to lower the Q at resonance. Anyone else looked at that?
 
Last edited:

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
I wonder why no one has proposed making a dog clutch damper as found on Rotax gearboxes. For those not familiar, this is not the multiplate "slipper clutch" intended to protect the crank in case of a prop strike and/or lessen starting kickback. Every 912 has a two piece dog clutch tensioned by a stack of Belleville washers, with the object seemingly to decouple the output shaft on torque peaks, just enough to lower the Q at resonance. Anyone else looked at that?
I am not familiar with this detail. Is it used on 2-strokes? If it is truly a dog clutch and is never disengaged, it requires no springs to keep it engaged. Are the springs engaging a friction device or two? Common implementations of a brake with a lash device have several components:

A low force spring (usually a wave spring) engaging a brake between the driving and driven parts of an isolater, putting known amount of friction between the elements as it oscillates under firing vibration. While it can control vibration, it also sets a minimum vibratory torque that is transmitted - vibratory torque less than the brake torque will all be transmitted, and the lowest vibratory torque that the device will isolate is equal to the brake torque, so it must be set pretty low in an automotive setting. On a PSRU, maybe not;

A high force spring (usually Bellevilles) engaging a brake between the driving and driven parts of the isolator. In series with the brake is a lash device designed to have more free lash than the engine vibration at some low rpm/low power setting. This device only adds drag when the relationship of the driving and driven plates change - which happens during starts and stops, when mean engine torque is changed, or under extreme conditions, like sidestepping a clutch, misfiring, rev limiter actuation, etc. This reduces the swing of a low rate spring device and can reduce impact.

Neither of these devices are directly in the torque path from engine to downstream devices, they are between the driving element (rotates and vibrates with the engine) and the driven element (rotates and vibrates with the downstream components), with the soft spring elements between them. These devices are primarily to help with engine isolation at idle, low power, and during changes in engine torque. They are compactly built around centerline, inside the diameter of the spring sets on clutch discs. It is there not to carry engine torque but to keep vibration under control between the elements. Automotive manual trans clutch plates and hybrid dampers do use these devices. Rest assured that the slip torques and lash are tuned for known flywheel and downstream inertia, rpm, and desired torque changes, and it is a safe bet that no one selected them for your airplane powertrain. Also, these devices are small low energy devices, air cooled, and not designed to handle much energy. Typical torques are 3-10 ft-lb of drag on the continuous device and 15-30 ft-lb on the lag device.

My wife has a 912 in her bird, and it has spring loaded slip clutch set well above operating torque to truncate torque as the system passes through resonance during startup and shut down. Checking slip torque is part of annual inspection.

More info on the exact configuration of the device being mentioned is helpful.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
Is the rotary engine externally balanced?
The rotors are dynamically balanced when produced. There is a balance weight on the front of the eccentric shaft and a designed imbalance in the flywheel. These are designed to counter the axial imbalance of the eccentrics and rotors moving with them. I do not know if this is adjusted on them as an assembled engine, but should not need to be. I have to believe that designed in imbalance is adjusted when the eccentric shaft and flywheels are finished, just like any other crankshaft and flywheel. The flywheel is external to the engine proper (outside the oil seals), while I believe the eccentric weight is internal.

These balance weights can completely balance this type of engine in vertical, lateral, pitching , and yawing axes. Neat feature and results in the incredible smoothness available from such an engine. Torsional vibration has been found to be another deal...

If one were to go with lighter or heavier rotors than standard, these counterbalance weights would need to be adjusted.

Billski
 

Hot Wings

Well-Known Member
HBA Supporter
Log Member
Joined
Nov 14, 2009
Messages
6,765
Location
Rocky Mountains
More info on the exact configuration of the device being mentioned is helpful.

Billski
The Rotax "C" gearbox has a centrifugal clutch that engages at around 2500 rpm. But I think the Rotax being referenced is the 9xx that uses a clutch pack that appears to function like a slipper clutch in a motorcycle. I'd like to put one in my Triumph some day...;) Seems to me that it would have similar limitations as using a sprag clutch?

[video=youtube;Rjc990AB1TI]https://www.youtube.com/watch?v=Rjc990AB1TI[/video]
 
Last edited by a moderator:

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
Hot Wings,

That device is the overload clutch. Usually, they are held engaged with Bellevilles. It is "On" all of the time, and holds below a certain torque. If resonance builds during start-up, shut-down, or low idling, it will slip and limit the growth of the torque swings.

This does not appear to be what FredOyster was talking about... Fred, have you got any pictures or exploded views of the device you were talking about?

Billski
 

rv6ejguy

Well-Known Member
Joined
Jun 26, 2012
Messages
3,749
Location
Calgary, Alberta, Canada
Some Rotax gearboxes have an overload clutch to protect against prop strikes primarily and a dog/ spring ramp setup to help alleviate TV however they still have a bad period from about 600rpm to about 1100, depending on the prop and they simply say to set idle above 1200rpm to avoid this. I can attest to this after extensive running of a 912 on my test stand during EFI development. The whole stand went into resonance at about 800 rpm. Was amazing how far steel could flex and the drive was just hammering there. Very unpleasant and scary to pass through that range.

We tested with 2 different props (2 and 3 blade) and one 4 blade test club. These changed the worst TV rpm down low slightly due to different MMOIs. Interesting stuff but even the well financed Rotax 912 still has TV issues down low which is not unusual for 4 cylinder engines with redrives. Seen it many times on various different projects I've worked on.

rotax108.jpg

rotax109.jpg

This might clear up some confusion which many people have about the purpose and design of these 2 different parts: http://legacy.rotaxowner.com/si_tb_info/serviceinfo/3kul97.pdf
 
Last edited:

plncraze

Well-Known Member
HBA Supporter
Joined
May 11, 2006
Messages
1,714
I believe the shut-down shake on a 4 cylinder Lycoming is from torsionals. It passes through the bad range more quickly on start-up.
 

Hot Wings

Well-Known Member
HBA Supporter
Log Member
Joined
Nov 14, 2009
Messages
6,765
Location
Rocky Mountains
This does not appear to be what FredOyster was talking about... Fred, have you got any pictures or exploded views of the device you were talking about?

Billski
Yes, I know it is the overload (not Sprag) clutch and is what I think FredOyster is thinking about. After looking at the video again it apparently does not act like a slipper clutch in a motorcycle. A motorcycle slipper clutch uses the cam action to depress the Belleville springs to limit the torque - ion one direction only. This one doesn't seem to use the cam for anything but coupling to the prop and the torque is strictly limited by the preload?

I've seen/heard of using both the Sprag* and slipper clutch as the only method of trying to tame TV. Trying to use the Sprag alone is apparently limited by the slow response time for the clutch to actually release during the reverse torque? Trying to use a slipper clutch would have the same time related problem in that it does in fact limit torque but won't decouple fast enough unless it is used in a soft system?

*BD-5
 

fredoyster

Well-Known Member
Joined
Mar 20, 2012
Messages
333
Location
Monterey Bay, CA
I am not familiar with this detail. Is it used on 2-strokes? If it is truly a dog clutch and is never disengaged, it requires no springs to keep it engaged. Are the springs engaging a friction device or two?
See the drawings at docusearch.flyrotax.com/files/pdf/d03185.pdf They refer to it as a "torsional load absorber" incorporating a "dog hub" with "helical dogs." It's on every 912 and breaks free between 600 and 800 N-m. Many 912s also have a "slipper clutch" which is a friction disk pack that breaks free at 15 to 45 N-m.

The 2-strokes have a rubber giubo.

My wife has a 912 in her bird, and it has spring loaded slip clutch set well above operating torque to truncate torque as the system passes through resonance during startup and shut down. Checking slip torque is part of annual inspection.
Yes, that's the friction clutch. The overload clutch is different. They have made different statements over the years about the design intent of the two clutches. All of the Rotax factory manuals for these procedures are free and online, and some sites have the older manuals which in some cases are a little more informative about internal details.
 

Hot Wings

Well-Known Member
HBA Supporter
Log Member
Joined
Nov 14, 2009
Messages
6,765
Location
Rocky Mountains

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
I believe the shut-down shake on a 4 cylinder Lycoming is from torsionals. It passes through the bad range more quickly on start-up.
Most of this thread, we have been talking about torsional vibration of the rotating parts. Lycoming's (the direct drive ones anyway) rotating parts along with the prop are a stiff system. In stiff systems, the lowest natural frequency in torsion is significantly higher than 2x the firing frequency from the engine. In a stiff system, everything vibrates together.

Now the engine is connected to the engine mount with rubber devices so that the natural frequency of the engine shaking on the mount is significantly below firing frequency. This is a soft system, designed to isolate the vibration from the things it is connected to. Usually, the torsional frequency is significantly above firing frequency at cranking speed, but significantly below firing frequency at idle, so most of the vibration over the entire running speed range does not get fed into the fuselage. Well, if it is there between idle and cranking, when you shut down, the engine has to go through resonance on its way to being stopped.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,314
Location
Saline Michigan
See the drawings at docusearch.flyrotax.com/files/pdf/d03185.pdf They refer to it as a "torsional load absorber" incorporating a "dog hub" with "helical dogs." It's on every 912 and breaks free between 600 and 800 N-m. Many 912s also have a "slipper clutch" which is a friction disk pack that breaks free at 15 to 45 N-m.

The 2-strokes have a rubber giubo.

Yes, that's the friction clutch. The overload clutch is different. They have made different statements over the years about the design intent of the two clutches. All of the Rotax factory manuals for these procedures are free and online, and some sites have the older manuals which in some cases are a little more informative about internal details.
OK, the comments before about "dog clutch" had me confused. A dog clutch is a completely different device. Almost all motorcycles engage the various gears with a separate dog clutch for each gear. This type clutch is just two sets of teeth that can be brought into engagement or slid out of engagement.

Now the overload clutch has a test specification with a minimum and a maximum torque at which the clutch is supposed to slip. This stuff in the manual about avoiding damage in a prop strike, well, it can do that. The manual specifies a min and max slip torque and requires checking it annually. That does not sound like the level of diligence required to protect stuff from the relatively low likelihood of a prop strike... I suspect the clutch is actually in there to set a max torque that the system can amplify to if the engine is operated close to resonance.

Now the other device is the "vibration absorber". This device has features called dogs in the manuals, but is NOT a dog clutch. This device is a rotational spring with a little friction in it. When you apply a little torque, it just carries it with no movement. Get to a certain torque and it can slide 15 or 30 degrees (depending on serial number of the engine). Once that sliding travel is up, the the dogs slide on ramps, compressing the Belleville springs with travel and additional torque related. So we have springs. The friction between the ramps serves to take off some vibration. The sliding torque allows you to check that the spring force is right. If the spring has the right force at the sliding part of the device, it likely has the right force at higher travel too.

So, the 912's have a soft system with the soft element being the ramp-spring device called a "vibration absorber" that also has some friction in it. Cool. And because the resonant frequency they get is close enough to idle speed, they can get a bunch of amplification, so they also have a clutch to limit the amplification to non-damaging levels...

The vibration absorber is the soft spring we have been talking about throughout this thread. The clutch helps...

Do other designs use these things? Yes to the springs, I do not know about the clutch.

Billski
 
Last edited:

Autodidact

Well-Known Member
Joined
Oct 21, 2009
Messages
4,513
Location
Oklahoma
There is a balance weight on the front of the eccentric shaft and a designed imbalance in the flywheel.
I see it now, the manual flywheel has it built in, but the automatic trans version has a pretty substantial counter-weight that mounts to the crankshaft with a taper and key. That makes the quill shaft too complicated I think; what is needed is a bolt on solution that is tailor made for the application and yet uses simple materials and OTS components. Sort of a cross between the peripheral spring clutch centers and the dual-mass flywheels but without the complexity of the DM flywheel.
 
2
Group Builder
Top