Long stroke flywheel shear sensing

[tadd]

I've read over the forums on what I can find about the long stroke flywheel bolt shear problem and it would have to be a doosy of a harmonic. So would not one assume as the harmonic was approached that the case should be ringing like a bell, just like knock? Could one then not 'watch' for the bad vibe with a piezo element attached at the flywheel bearing (or there abouts)? A simple circuit with adjustable gain wired to a 'shift light' would seem to let you use the motor to the fullest without an arbtary rpm limit.
I'd like to abuse my future 2.8 with impunity!

Has this been tried? Thoughts?

[Flieger]

I had written a longer response but then the forum had a hiccup so I re-posted and now my first response shows up. Oh, well...

Skip to the next response, then.

I believe the vibration was torsional, not a bending wave (like "cracking a whip"). The 917 used a 70.4mm stroke on the larger engines like the Can-Am 5.4 liter. The 917, of course, used the central power take off gear to utilize the vibrational node at the middle of the long twelve cylinder crankshaft. I am not sure how that engine compares to the 911- it may not just be "halved". The 917 engine had two connecting rods on each crankshaft throw while the 911 has one throw for each rod. This may change the vibrational pattern and the firing order. If it were "half" a 917 engine, then the worst vibration would be at the end of the crankshaft. If not, then there is a node in the middle on a 911 as well. The factory eventually went to longer strokes for racing engines after the 2.5 short stroke was created to avoid the flywheel shear issues.

[Flieger]

I think this harmonic is a torsional vibration rather than a bending wave in the crankshaft. The torsional vibration puts lots of stress on the flywheel bolts due to the forces involved when the large angular momentum of the flywheel changes slower than the crankshaft's does. It also does not help to have the bolts near the center of the flywheel. Luckily, the forces are not directed out radially so the crankshaft's bearings inside the case should be fine. I think the flywheel shear is the most visible and most extreme effect of the vibration- that is the worst it will get. The factory seems to have decided it was alright and just secured the flywheel better.

I am also interested to hear any Engineers give their thoughts.

[racing97]

It happens in the range of 7200 RPM the factory treated it in the RSR And RS in 1973 by radiusing the fillet in the rod journal and in 1972 they (factory) actually tried to run a dampner
on the 2.8. The 6 bolt attachment was fine by bolt size but to tight of a bolt circle for the load.
Some people actually would weld the bolts to the torque plate, you could get it to stay in place if you increased the torque on the bolt by 50%. AT this time TIlton started making Aluminum flywheels to help the situation.

regards

[tadd]

Thanks for the replies... I understand why it happens, thus I dropped 3k on Ti rods (345 g vs 720 g for stock steel) and Ti wrist pins (45 g vs 95g) to cut the weight effectively making the crank stronger. I've thought about having a 2.0 crank reground to 70.4 with a thick fillet radii. Given that the SC does fine and is 2 mm smaller width than the 2.7 crank, it seems one could loose some bearing area and still be ok.

I have also been told that it takes some time for the resonance to build. In other words a short rip to 8k followed by a fast shift is ok, like a 2nd to 3rd shift, but a long hold, like reaching top speed allows the pertubation to build.

So even if the true force involved is a 'helical windup' of the crank shouldn't a fraction of that be transmitted into the case? Like I said, the whole point is to be able to use the motor to its fullest without making a guess - ie, keep it at 7.8K (or whatever). It just frustrates me as a chemist not to know, ya know?

Just like GI Joe, knowing if half the battle!

[Steve@Rennsport]

Originally Posted by tadd

So even if the true force involved is a 'helical windup' of the crank shouldn't a fraction of that be transmitted into the case? Like I said, the whole point is to be able to use the motor to its fullest without making a guess - ie, keep it at 7.8K (or whatever). It just frustrates me as a chemist not to know, ya know?

Yessir,..some of that energy is transmitted to the case and thats why mag cases will eventually crack around the #3 bulkhead when operated at high RPM for periods of time. Porsche aleviated the problem by switching to aluminum for the 3.0RSR and later race engine cases and eventually to 9-bolt crankshafts to secure the flywheels from coming loose.

One can minimize the issue by doing just what you've done using lightweight reciprocating components and spending extra time with precision balancing.

[jcge]

Tadd - to address your initial question....yes, you can add an accelerometer (piezo type) to detect pretty much any type of vibration or resonance.....but that is only the tip of the iceberg

A sensing element (like a knock sensor) would work just fine (given some of the signal will be transmitted to the case structure)

BUT....then it gets really ugly

You need to
1) know the characteristics (frequencies, amplitudes etc) of the signal you're looking for
2) implement a filter (eg digital signal processor with fast fourier transform) to isolate that portion of the total signal
3) display the output
4) implement a preventative action or countermeasure

#1 the characteristics should be mathmatically modelled (lots of complex calcs), assuming you are able to determine even basic physical parameters like the polar and flexural moments etc, then verified experimentally (eg spectral analysis) to ensure you've isolated the right signal. (which may be heavily masked/aliased by other signals)

#2 & #3 are software and hardware (with expensive development)
#4 could be manual (lift your foot) or automatic (rev limiter etc)

This kind of spectral analysis is used in industry where the implementation costs are justified by the return on investment or risk (think $$$$ in machinery downtime (eg minesite conveyor) and lives saved or lost (eg aircraft)).

Or you could get lucky...just hook up a knock sensor, attach an opamp and just watch for total average signal rise as you get to the point of no return. (but I'd be surprised - and delighted - if this delivered anything more than a pseudo tachometer signal)

I think you can safely say that in the case of an automotive engine, preventative action (which you are already taking) is you best course of action

Feasible - yes, Pracitcal - probably not.

In any case, here's a link to some info on knock sensors and interfaces - the'yre not that expensive
Bosch Motorsport - Knock
Robert Bosch GmbH - Automotive Semiconductors and Sensors
SAE Paper 960614 "Knock Detection Using Spectral Analysis Techniques on a Texas Instruments Tms320 Dsp" Thomas G. Horner - Texas Instruments, Inc. Feb 1996 - provides a good overview of "how to".http://focus.ti.com/lit/an/spra039/spra039.pdf

[Walt Fricke]

Tadd

There is a very practical, and inexpensive, fix for this problem.

First off, I don't believe the bolts shear. Mine never have, and I haven't heard of this happening (not that I hear about everything, nor that maybe it hasn't happened). What happens is that they back out. You know this has happened because you realize you can't get the transmission engaged sitting in the paddock or otherwise at rest. Why? A bolt or bolts are rubbing on the clutch disk, creating a drag.

You can drive like this for a while, because shifting while driving sort of naturally has everything rotating close enough for this drag to be overcome by the synchros. Don't ask me for how long, because I don't know. But maybe for a 20 or 30 minute track session.

You still don't want this, because in addition to ruining your event you are ruining the flywheel and crankshaft due to metal transfer from one to the other, and maybe vice versa - the mating surfaces look really ugly. The second time this happened to me I knew what was happening and didn't keep going, and the end of the crank could be cleaned up. The flywheel's surface looked somewhat lunar. Still have it, may sometime see if flamesprayers or the like can bring it back.

Secondly, the fix:

I got this from Bruce Anderson, and I'm here to tell you it works: Torque the six flywheel bolts to 150 lbs/ft. Yes, that is maybe 50% higher than the factory spec. So what? It works. After cleaning bolts and crank threads carefully, I put the flywheel etc on and run the bolts in "tight." What's that? Doesn't matter, so there is some torque on them. Then I back one out and apply just a little red Loctite. And run that bolt in and torque to the 150. Repeat until all 6 are done. This procedure insures that no Loctite gets between crank and flywheel.

I purchased a short 12 point tool for this, so as to keep the torque wrench in a lot closer to the bolt head. When I have used a longer tool, I have put a jack stand under the head of the wrench to keep things lined up. It is kind of scary, especially those last few degrees before the wrench clicks.



I've been doing this on my 2.7 race motor (which I routinely ran up to 8,000 until I dynoed it and realized I should shift at 7,600 or so with my gears) since 1995. I've never had a flywheel come loose since, nor broken a bolt.

One race team, which wasn't doing this, tack welded the bolts to the flywheel (must have been to the funny 6 hole washer). Didn't work - the bolts broke the welds and backed out.

Somehow the effects of a 4th order harmonic node acting on the mating surface stretches the bolts enough that they lose clamping force for an instant, and then the inclined plane of the thread causes them to start rotating. Repeat this enough and you have trouble. (Confession - I only know about 4th order harmonics and such because an article in Upfixn said so, pointing out that the 917 12 cylinder engines took their power from a gear in the middle fo the crank, where such forces cancelled out)

I am also dubious about shearing bolts, because when mine came loose, all of them were loose. Didn't shear then, when I'd think the forces were larger due to the looseness and whatnot. Weren't even very damaged, so they sit in the spare hardware drawers being put to other uses, like bolting spring plates to banana arms.

My 2.7 has nothing fancy in the way of rotating mass reduction. I have used a full RSR flywheel/PP, which is good on that end. But stock rods (with good aftermarket bolts) and Mahle racing 90mm pistons, and a stock crank.

I recently acquired a 2.8L motor (not an RSR motor, but same dimensions except for the heads), and reinstalled a lightened flywheel on it with the same methods. I have 18 hours on this motor (in my car) and it is running strong. My RPM data shows spikes to 8,500, which I think are due to a defect in my upshifts to 5th. Knowing this I am trying to work on that. At Miller I was winding it out close to 8,000 in 3d in spots to avoid an upshift. Flywheel still firmly attached.

Of course, if you take my advice and then shear all 6 bolts first time on the track, you are on your own.

Walt Fricke

[burgermeister]

In general, bolts don't carry any load in a typical bolted joint - they only keep the pars pressed together. All load is transferred through friction at the interface of the 2 clamped parts. Even loads applied along the bolt axis (not applicable in this case) usually just reduce the clamp load - the clamped parts are generally stiffer than the bolt, so clamp load gets reduced much faster than bolt tension increases.

So Walt's suggestion should not cause any bolts to shear - if they survived torqueing, they should be fine.

[304065]

Why not put a triangular piece of red electrical tape on the face of the tach instead?

[docrodg]

INteresting... would love to see some scientific data on this (plots of harmonics vs rpm etc.). In general, I agree about a higher torque, if the bolt is made to be torqued higher and the threaded hole can withstand it too then go for it. There is no gasket that will leak as a result.

[tadd]

Awesome replies, thanks! You folks are the best.
To be honest, I was hoping to measure and observe. Simulation has it's place but some times it's just easier to do and observe. Especially given that there is historic precedant as to know where to look. So the plan was to include a few sensors, data log and go digging for patterns. Not truly sophisticated, but maybe with a little luck, something useful might fall out.
Walt.. I agree that the 'brute force' approach makes sense here. I've taken a hard look at
SPS's web page and their fastners are a finer
pitch unforuntally. It would be interesting to know what Porsche makes it's flywheel bolts from. I've always been told from fastners designers that over tight is far better from a fatigue standpoint than under tight.
Y'all have given me much to chew on, thanks.
tadd

[Walt Fricke]

Tadd

All I know is that the bolts are DIN 12.9. Maryland Metric gave a max torque value of something a bit over the Porsche factory 110 lbs/ft, but well under the 150 I use, if I remember that correctly.

I think that the 70.4 RSR 2.8 engines also had a problem breaking cranks. Resolved (somewhat?) by the larger filet radius as mentioned, and a harmonic balancer. Fortunately for me, I haven't had cranks break.

Not sure just what one would do with data here, if harmonics have some relationship to the crank throw (the 66mm cranks don't have the reputation of shaking flywheels off), or the firing order/throw angles, or some other characteristic you couldn't really change. The 9 bolt cranks don't have the flywheel problem, either, at least not that one hears about.

There are guys who use Chevy rods (big end size, anyway) so they can undergrind the crank, use larger filets, and use Clevite bearings. I've got a 66mm crank with a damaged bearing I keep thinking I might have this done to it.

But we are getting into the engine building forum's turf.

Walt

[Flieger]

Could altering the piston pin offset towards the exhaust valve side on the 4-6 bank and to the intake valve side on the 1-3 bank make any difference? This would make more power, I think, through the decreased rod angularity at 14* past TDC- the most efficient place in the crank revolution to apply power. This would change the gas torques on the crankshaft and subsequently alter the torsional vibration inputs into the crankshaft. The crankshaft might then reach resonance at a different rpm.

Please see John Cramer's post here: JE pistons in original cylanders? at the bottom of the page.

[Walt Fricke]

Max

Yes for power, but unclear if worth the substantial extra cost.

Keep going in the referenced Cramer post, and you will get to another post, which says as much as I think can be said about this subject.

If I do say so myself, since I, amidst others, said it.

Note particularly discussion of Formula V, and why doing this pin offset is problematic for 911s, as they have asymetric piston domes.

So far, I haven't spotted discussion of how to calculate harmonics, resonances, etc. in crankshafts (nor could I even begin to do it, while some of the engineers amongst us might). But I have yet to see wrist pin offset mentioned as a way of dealing with crankshaft problems. Besides, simply moving a bad feature around in the RPM range might not help. Guys are building race motors to run at 9,000 rpm.

Walt

[Flieger]

I remember another thread that I was involved in where the wrist pin offset was discussed in regards to having different pistons for different cylinder banks. I think the conclusion was that Porsche used specific pistons in the earlier years such as when the RSR was racing in the 70's but that on the 964 the psitons became the same for ease of servicing. The offset was therefore power on one side and quiet on the other.

The MFI pistons have symmetrical domes, correct? I do not see how the dome shape interferes with wrist pin offset.

[Walt Fricke]

Max

You will see what the issues are when you contemplate it a bit more.

If a piston has a symetrical dome, so nothing about the piston top dictates which part faces which valve, then offsetting the wrist pin is pretty straight forward (that is, installing them and getting the desired effects is straight forward). You can just mark the dome or something so you can remember which way to install them. The flat top old VW pistons are a case in point - they have offset wrist pins, and arrows on the top of the pistons which are supposed to point to the flywheel no matter which side you install a piston on. This reduced piston slap, but for more power guys point them the other way.

When you have valve pockets (and intakes are larger than exhausts), or shapes like the CIS, and you want wrist pin offset, you have to have one set for the right bank, and another for the left, if you want the offset to do the same thing on each side.

Offset effects are relative to the direction of rotation, so it has to be on the top on one side, and on the bottom on the other (or vice versa). No can do with different size valve pockets and the offset only on the intake side (or vice versa) for the whole set of six.

The big mystery with Porsche is that for some period of time it offset the wrist pins, had asymetric pistons, but only made one flavor of piston. You will not find reference to part number X for right side pistons, and Y for the left, as one would expect. So you had one effect on one side of the case, and the opposite on the other.

The best post is the one suggesting that on one bank you install things so the exhaust is on the top, and the intake on the bottom!

Walt

[burgermeister]

Wrist pin offset is not likely to change crank/flywheel resonances. The resonance is a function of the inertias/masses and the stiffness between the inertias/masses, and none of that would change. Reducing flywheel inertia would seem a better approach for altering the resonance frequency.

[jcge]

Originally Posted by Walt Fricke

... I haven't spotted discussion of how to calculate harmonics, resonances, etc. in crankshafts .....

If you want to look into the mathematics....(not for the faint of heart !!!!)

Internal Combustion Engine in Theory and Practice: Vol. 2 - 2nd Edition, Revised: Combustion, Fuels, Materials, Design
Charles Fayette Taylor
The M.I.T. Press

More practically, try some Finite Element Analysis (most packages can determine natural frequencies and modes of vibration).

Another interesting approach is to support an object (like a crank on knife edges or hanging by a thread), strike the object and record the "ringing" sound (edit - sound = noise & vibration) it makes (like a bell or tuning fork), then do a spectral analysis of the sound recording to establish its natural modes and harmonics. (noted that journal lubrication will provide some level of damping)

John

[Flieger]

Originally Posted by Walt Fricke

Max

You will see what the issues are when you contemplate it a bit more.

If a piston has a symetrical dome, so nothing about the piston top dictates which part faces which valve, then offsetting the wrist pin is pretty straight forward (that is, installing them and getting the desired effects is straight forward). You can just mark the dome or something so you can remember which way to install them. The flat top old VW pistons are a case in point - they have offset wrist pins, and arrows on the top of the pistons which are supposed to point to the flywheel no matter which side you install a piston on. This reduced piston slap, but for more power guys point them the other way.

When you have valve pockets (and intakes are larger than exhausts), or shapes like the CIS, and you want wrist pin offset, you have to have one set for the right bank, and another for the left, if you want the offset to do the same thing on each side.

Offset effects are relative to the direction of rotation, so it has to be on the top on one side, and on the bottom on the other (or vice versa). No can do with different size valve pockets and the offset only on the intake side (or vice versa) for the whole set of six.

The big mystery with Porsche is that for some period of time it offset the wrist pins, had asymetric pistons, but only made one flavor of piston. You will not find reference to part number X for right side pistons, and Y for the left, as one would expect. So you had one effect on one side of the case, and the opposite on the other.

The best post is the one suggesting that on one bank you install things so the exhaust is on the top, and the intake on the bottom!

Walt

Yes, I remember you saying this in several other threads. I am not building a CIS engine- or an engine. I am just thinking of the theoretical dynamics of these engine tweaks.

The early 911s had a correct piston pin offset (up on the left, down on the right). Do we know how much it was? .8mm?

Are you saying that a dome that has a larger valve cutout for the intake valve than for the exhaust valve will negate the low-noise piston pin offset through uneven thrust on the piston crown? If so, then single spark plug engines should have a similar issue because of the offset ignition.

Regardless of sideloading and noise, the small offset of the piston pin should move the connecting rod into a more favorable orientation relative to the cranshaft in order to make slightly more power. This would be achieved by offseting barrels 1-3 towards the intake side and barrels 4-6 to the exhaust side.

I understand the effect of the offset would hardly be measurable and really only of interest to engineering students like me.

Thanks for any information.