Measuring coupling dynamic torsional stiffness 2

[jeyaselvan]

Hi All

I am looking for a test equipment, which could measure the dynamic torsional stiffness of couplings. If you have come across the same, kindly let me know.

Regards
Jeyaselvan

 

[rob768]

you could try to simulate it, using two known masses inertias and only the coupling as elastic element. Then calculate the natural frequency and compare with measurements.

 

[ivymike]

if you're using elastomeric elements, be forewarned that the stiffness likely varies with displacement amplitude.

 

[jeyaselvan]

Thanks for your feedback

I am doing quite a good amount of torsional analysis calculations. I am concerend that my results are only as good as my inertias and stiffness used in the drive train, particularly that of the coupling. Most of the times, the recommendations and failure reasons are attributed to this. Hence I feel it is high time to sample test the dynamic torsional stiffness at operating torques (static + dynamic).

I am interested to know that anything like that of a sweep test in vibration shaker is available for this sort of measurements.

Thanks in advance

 

[electricpete]

Some thoughts here:

http://maintenanceforums.com/eve/forums/a/tpc/f/3751089011/m/109108503?r=109108503#109108503

One tip - if you try a bump test, bump tangentially in vertical direction and measure tangentially in horizontal direction - tends to limit the radial response. But no matter what bumping for torsional resonance would be tricky.

The other possibility is shaker, but of course where are you going to find a torsional shaker without so much mass. If it's motor driven there may be ways to use the motor for variable torsional excitation.

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

 

[electricpete]

And of course Mike’s point is relevant for interpretation of results of bump or shaker – may depend on magnitude. Repeating the test at varying excitation amplitude I guess is a way to try to judge that.

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

 

[ivymike]

yeah, a bump test was specifically what I had in mind when I warned about the amplitude sensitivity (above).

 

[GregLocock]

Two torque sensors, a couple of torsional vibration meters, and a torque exciter are all you need to do the job in the most expensive way possible.

If you do a hammer test then use enough accelerometers that you can identify the mode shapes properly, then you can back-calculate the torsional properties. For instance, if you have 3 tangential accelerometers then you can calculate the pure torsional response in one plane.





Cheers

Greg Locock

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

 

[jeyaselvan]

Thanks for all of your feedback on bump test recommendations. I believe the bumps tests are for static torsional stiffness.

I use elastomeric elements, which ivymike rightly forewarned that the stiffness varies with amplitude.True.

Also I have seen coupling manufacturers providing static and dynamic torsional stiffness (some provide this at 10Hz) in their catalogs. We use the latter for all our design calcuations. Their ratio varies from 1.5 to 10 or higher.

Is there a way to deduce analytically or experimentally the dynamic torsional stiffness? I suppose, this depends on the torsional ampltiudes as well as on the frequency(?). For your comments

 

[rob768]

We do a lot of torsional vibration calculations for ship's propulsion installations as well as calculations of resilient mounts. We also have a lot of feedback on such calculations as we also do torsional vibration measurements.

Note that manufacturers often state that a margin of 30!% may exist on catalogue stiffness values. They also give guidelines for frequency factors and temperature influences in their catalogue. Note that these factors depend strongly on rubber hardness. a trend it the harder the rubberm the larger the dynamic factor
.
Bottomline: if you calculate using polymer components, use a safe margin, and don't try to calculate up to the 0.1 Hz

 

[electricpete]

I believe the bumps tests are for static torsional stiffness

In general bump tests can be used to predict dynamic stiffness. The fourier transform of the bump response can be viewed as a frequency response function which can be used to predict the response at any frequency … constant (static) or otherwise

Unless…
there is non-linearity (in which case you might try varying the amplitude)
or there is stiffening under torque load…
or there are speed-dependent effects …


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

 

[BobM3]

Tough part of measuring dynamic torque is that the torque measuring devices are accelerating which produces electrical outputs due to the devices' own inertial acceleration. If you decide to measure the dynamic torque (or have a lab do it), make sure they take this into account.

 

[FeX32]

With all due respect, how is static stiffness different from dynamic stiffness?


Fe

 

[ivymike]

for a simple spring element (massless w/o damping), they're the same.

see this article for more

http://www.bpb-co.com/articles/printableFiles/00understand_stiff_print.pdf

 

[BobM3]

The following older thread talks about the difference between static and dynamic stiffness for rubber:

thread335-102034

I suppose one could argue that the dynamic stiffness effect is actually due to damping.

 

[GregLocock]

"I suppose one could argue that the dynamic stiffness effect is actually due to damping." If we call it non linear damping, yes. The actual complexities are many, you need to consider the damping element is sensitive to strain rate, strain amplitude, temperature, preload, etc etc.

The 30% modifier that is being bandied about is actually the Holy Grail for complex bushings, I see figures of 80% routinely.





Cheers

Greg Locock

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

 

[jeyaselvan]

Thanks to all for your feedback again.

BobM3 : Thanks for the reference to thread on rubber spring rate.Interesting.

I am little confused with the use of "dynamic stiffness" terminology being used in this scenario. I am aware of using (K-mw^2+iwC)as dynamic stiffness for a conventional academic SDOF model. Does this hold good in the context of coupling dynamic torsional stiffness as well? The manfr often gives this value and depedency on frequency factors and temperature influences as rightly said by Rob 768.

Or is this a two step approach? Meaning,we get the dynamic torsional stiffness assuming coupling as a individual system (which is normally calculated and provided by manfr). Then we use this dynamic stiffness in the drive train torsional analysis.

electricpete: That is exactly my doubt.
In linear dynamic systems, in the (K-mw^2+iwC), when w-->0, the stiffness what we arrive is the static stiffness.

Say from 10 - 50Hz at low frequencies where the system's coupling torsional mode normally lies (where the coupling manfr usually gives the dynamic torsional stiffness data), is it good to assume that the dynamic stiffness is K+iwC (Greg, is that your argument?).

My problem is from the measured torsional stiffness, say from static test or by bump test, can I estimate the dynamic torsional stiffness at my excitation frequency of interest?

Can I say we can approach this in two ways?

1)Measure static torsional stiffness (measuring torque/deflection) using rotor1-coupling-rotor-2 setup. Then estimate the dynamic torsional stiffness as (K+iwC) at excitation frequency w and use that for torsional analysis calculation

2)with Bump test, get the dynamic stiffness directly at the excitation frequency from the frequency response function and use the same for torsional calculations

Only thing in both cases is that I can't see the amplitude and frequency effect of torsional stiffness?

Your suggestions and opinions, please

Jeyaselvan

 

[FeX32]

Thanks ivymike and BobM3.

To the OP: I would do a full dynamic test if possible modeling 'C' as a nonlinear function. (I assume you have 'C' highly nonlinear which is why you are in this direction)
Just my opinion,





Fe

 

[jeyaselvan]

My collegue is involved on a problem identification of a "belt loosening / slippage problem leading to failure" in a motor (11kW power) driven screw compressor after a few (20 - 30) hours of operation.
Is there anything else he needs to consider apart from the static design, contact angle, etccc? Say, whether a torsional analysis is required to see whether he is operating in some out of phase modes? Could this also be point of concern? Any other check points.

Thanks in advance

Jeyaselvan

 

[electricpete]

may I suggest a new thread?

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