Critical Speed below natural torsional frequency? 4

[vc66]

Hi All-

I have a simple shaft, at the end of which is a disk that rotates about the same center as the shaft. Simple problem to find the natural frequency, I know. I have the natural torsional frequency solved as 1050 rpm; however, when we run the assembly (ramping up speed slowly), it starts to shake violently as we approach 350 rpm. If we ramp it up quickly, and pass 350 rpm, there is no problem.

I'm not too familiar with resonance and vibration, so can someone school me as to possible reasons why this would vibrate near 350 rpm if this is only 1/3 the natural frequency?

Any and all help, is appreciated.

V

 

[GregLocock]

It could be a bending mode.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[btrueblood]

How is the shaft driven, is there compliance on the opposite end? What is the wheel driving, is there compliance on that side?

 

[vc66]

btrue-

What do you mean by a compliance?

V

 

[vc66]

Greg-

I just solved for the lateral natural frequency. That comes out to 1413 rpm.

V

 

[vc66]

I should also state that the shaft and wheel are vertical. The shaft is driven by a motor, and the wheel doesn't drive anything.

V

 

[GregLocock]

Are your bearings really rigid?

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[vc66]

The bearing are fairly rigid, and they are preloaded.

V

 

[GregLocock]

Well, that is not really an answer, is it?

Also btrueblood's point is good. You can't assume that the shaft is locked at one end, usually.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[electricpete]

One tool for checkingt resonant natural frequencies is a simple bump test. Whack the parts with a 2x4 or your favorite whacking device and measure the frequency of the response. Crude, but usually fairly effective.

I think part of what Greg and btrueblook are getting at is that flexiblity of of the bearings and everything that supports them can lower the natural frequency far below what you calculate if you assume a fixed bearing. Especially if you have flimsy structural support.

Do you happen to have a picture of this beast? Preferably with a ruler in the photo to judge dimension?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[vc66]

It wasn't really a question, Greg!

What do you mean by really rigid?

Thanks for all the help so far, everyone.

electricpete-

I'll see if I can find a discreet picture.
In the mean time, here is the basic setup.

V

 

[vc66]

Assuming that the bearing structure is not rigid enough to us them as stiff supports, what would be the correct way to go about factoring that into my calculations?

Are there any books that anyone can recommend on the subject, so that I can do some extra research?

This is pretty new to me, so it's interesting (and frustrating).

Thanks, all.

V

 

[BobM3]

If you suspect the bearing supports, you would need to calculate their stiffnesses. Then, instead of rigid bearing supports, you would support the shaft with springs at the support locations. The stiffness of the springs would be the stiffness of the bearing supports.

Did you try running the calculations assuming the bearings have no bending moment stiffness (simple pivots)?

How far off center is the disk mounted?

 

[vc66]

The disk is mounted directly centered (sorry about the picture). The center of mass of the disc is .015" off center.

I will try running the calculations with the bearings as simple pivots.

Thanks for the help. Keep it coming.

V

 

[BobM3]

Did you include the offset in your equations ("equivalent" mass > actual mass)? It seems like a small number so I doubt it is causing such a big difference.

As mentioned earlier, did you hang some mass on the driving end also (for your calculations)? I assume you have a coupling to a motor. How about the stiffness/mass of the driver shaft?

For the compliance (stiffness) of the bearing supports, all it takes is one thin mounting plate for the bearings to ruin an otherwise high overall stiffness.

 

[vc66]

We buy the entire drive/spindle from a vendor, and I just received the step file from them. As you can see from the attached picture, the bearings are not where I originally thought they were. What bearing (no pun intended) is this going to have on my calculations. Please forgive the ignorance, I'm trying to learn as I go.

V

 

[btrueblood]

Sorry, vc66, I should have been more clear.

What I meant is - how much torsional flexibility is there in the thing that supplies torque to your shaft? I.e. if driven by belts, how much stretch per unit torque load do the belts allow; if driven by gears, how much lash/windup is there in the system per unit of torque; if a flex coupling, how much...etc. etc. It is usually a good idea to make such measurements in both directions, i.e. +ve relative to the running load, and then reversing and getting data in the -ve direction; this will tell you if there is a dead band (e.g. "lash") in the system which can really goof up your frequency calculations...

Sometimes, the load paths are pretty clear in terms of the boundary conditions they can be modelled by (there was a guy on here a few weeks back doing a spin test of what essentially was an induction motor core); other times (such as with belts, gears, etc.) the boundary condition may be a combined force/torque and displacement/angular boundary condition, and thus not (as) amenable to the "textbook" cases for vibration frequencies. You will end up making educated guesses, but those are generally better than uneducated ones...and hopefully Greg and some other dynamics gurus can come back with some arcane techniques to handle the goofy real-world boundary conditions and dead-band springs...

A really simple way to answer the questions in my first post above is to take a torque wrench and a protractor, and go find out.

The picture you posted looks a lot like a clutch pressure plate/flywheel. I also see what looks to be multi-row ball bearings, located too close to the c.g. of the assembly, so Greg's suspicion regarding lateral modes is a likely one. Again, a good test to answer the "how compliant are the bearings" is to measure force-vs.-displacement (I have done this with a coffee can filled with scrap bolts, some nylon thread, and a dial indicator, to good effect) in both the +ve and -ve directions, so you can see any "dead band" in the middle.

Good luck. Oh, and Merry Christmas!

 

[vc66]

Thanks, btrue.

I'll take heed to your advice, and do some more research. Merry Christmas!

V

 

[electricpete]

I agree most likely simple lateral resonance as others stated.

If you have a vib analyser, a bump test is a very quick easy way to check. (doesn't account for things like gyroscopic effect, but gets you in the ballpark).

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[GregLocock]

The bearings are very close to each other, so reducing their 'coning' stiffness. I don't think you've told us how the drive end is driven, the chances are that there is a huge torsional compliance/inertia that hasn't even featured in your calculations yet.

To be honest you are going to be lucky to calculate a helpful answer, I think measurement is the way to go.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.