Safety margins for torsional eigenfrequencies of big shaftlines?

[cbrn]

Hi,
can anybody tell me if there are International Norms which prescribe safety margins as regards torsional eigenfrequencies of big shaftlines, wrt exciting frequencies ?
To make myself clearer: does any Norm specify something like "torsional eigenfrequencies must differ of at least xx% from any possible source of excitation, including grid frequency, grid first harmonic, ..." ?

In fact, I'm using internal prescription which have proven to be really valid throughout years and different machines / installations, but right now I need to debate with a Client about very strict specifications he's giving, and I fear that our internal rules may not be enough...

Thank you very much in advance,
Regards

 

[GregLocock]

Way out of my field, but I would be surprised if there was much variation, given the costs involved.

As a completely useless data point we use 15%, for engines, but we don't drive /through/ our first resonances.

Mind you, many of these rules of thumb were developed before it was possible to accurately analyse the systems, perhaps you could do enough dynamic analysis to prove your point.



Cheers

Greg Locock

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

 

[cbrn]

Hi Greg,
thank you for your attention.
As a matter of fact, as you say, it's not really a technical, mechanical problem. I'm about sure we'd be able to design the shaftline even with the very stringent (in my opinion "unusefully" stringent) specifications from the Client.
But it would not be cost-effective, neither from an engineering point of view, nor from a constructive one.
We'd like to make a comment saying something like "your statement that the eigenfrequencies must stand 20% apart from excitations is exaggerately stringent. We have long experience in designing these kinds of shaftlines with 10% margin, and the good behaviour can be prooved in the list of references attached. --> Moreover, the 10% margin is widely-used and obeys to Industrial Standards"...
Well, this would be the "optimum". Of course, only IF standards exist which have this 10% (or even narrower) as a margin, for similar applications... That's why I'm looking for them, otherwise I'd be very happy with our internal criteria 'cause I know they work very good! ;-)

Regards

 

[rob768]

Torsional vibrations calculations are very accurate, and a lot of is known on damping and excitation forces. Therefore, at least in the marine industry where I work, class does not give limit margin between excitation forces and natural frequencies, but limits on allowable resulting stress and vibratory torques. One practical reason for this is the large number of excitation forces generated by a Diesel engine, which makes it impossible to have an instalaltion without excitations near natural frequencies.
So, basically, it is allright if a natural frequency coincides with a resonance frequencies, as long as enough damping is present (some care with regards to coinciding of natural frequencies and excitations should be considered, regardless of this)

 

[cbrn]

Hi,
thank you Rob. In our case instead the excitation freqs are easily determined and very few (basically grid freq, grid 1st harmonic, and Blade-Passing Frequencies), and using simple margins can avoid performing an harmonic response(s) analysis and allow to use straightforward eigenmodes analysis. In addition, our shafts are almost "undamped" structures, at least tortionally (as regards bending, shaft itself may be almost undamped but the overall shaftline system is not, because of the bearings).
I keep on hoping that somewhere some Norm has written these kinds of "margins"...

Regards

 

[GregLocock]

The trouble with analysing lightly damped systems is that even a tiny bit of damping would have a big effect on the stresses near resonance. I guess the problem you face is that the standards would be proprietary.

The obvious resource that springs to mind is Dr JD Smith at Cambridge University Engineering Department. He wrote the book (and gave the lectures) on dynamics of prime movers.





Cheers

Greg Locock

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

 

[rob768]

Class margins with regards to lateral vibrations of shaft are 20%, which incluses a certain margin to acocunt for modelling inacuracies (for instance to allow for errors in the estimates for propeller inertia), so these may be appicable. For lateral vibrations, hardly ever a response calculation is made because there is litlle info on excitation forces and especially damping.

If you believe your calculation to be accurate due to good knowledge of the component properties, a margin of 15% is more than sufficient in my opinion.
In cases where we measure resonance in lightly damped structures, the actual band where slope excitation is visibile or noticable is very narrow, so 15% is certainly sufficient in those systems.

 

[cbrn]

Hi,
Greg,
you are perfectly right: "I guess the problem you face is that the standards would be proprietary".
That's the core of my problem. Our standards are reliable but they do are proprietary, so we're hardly allowed to "self-reference" ourselves!
I'll try to have a look to the reference you indicate. Thanks!

Rob768,
unfortunately I won't be able to use your types of references, because... they match what the Client wants! In fact, we know that in our specific case our internal criteria are sufficient (and our calculations well-tuned, we have tons of internal prescriptions, statistics and references), the problem is to demonstrate to him that these criteria are not there because we have dreamt of them tonight: in this sense, finding out an international Norm which includes such kinds of margins would be "the top". Otherwise, we'll try to convince him based on our "case-histories".

Thanks again for your attention,

Regards

 

[rob768]

offer to do vibration measurements to validate your assumptions?

 

[cbrn]

Hi,
Rob768,
you're right, that's an option. The only drawback is that we'd have to provide calculation AND corresponding measured data for some machines similar to the one offered to the Client, and these data are generally undisclosable... I mean, calculation data are ours, so this could not be a problem, but measured data are proprietary of the Customers who exercise the plants, and THIS may be a problem (we have of course free access to these data at least until the end of warranty period, but under the clause that we may not redistribute them...).

Regards

 

[electricpete]

From API 684-1996, APPENDIX 2A — "API STANDARD PARAGRAPHS
SECTION 2.8.4 ON TORSIONAL ANALYSIS"

2.8.4.2 Undamped torsional natural frequencies of the
complete train shall be at least 10 percent above or 10 percent below any possible excitation frequency within the
specified operating speed range (from minimum to maximum
continuous speed).

Note that even if there are excitation frequencies inside 10%, section 2.8.4.4 allows this to be accepted if acceptable stress analysis is provided.

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

 

[cbrn]

Thank you very much ElectricPete !!!
This is exactly the sort of reference I was looking for.

Thank you to everybody for the attention,

Regards

 

[VibFrank]

Perhaps I am a little bit late: API 610 (we have 8. edition I think) also gives 10%

 

[cbrn]

Hi,
API 610 is for heavy-duty pumps if I remember well, so it may be also applicable, but we preferred mentioning API 684 because apparently unrelated to a specific application field.
Thanks again.

Regards