Determination of Balancing Speed for Flexible Body

[connectionman]

Hello All,

Recently I have been searching how to determine the balancing speed of flexible body. Let me give an example;

There is flexible assembly which normally operates at 28.000 rpm. Until that speed, there is no critical mode. The nearest critical mode is around 33.000 rpm. Normally assembly was balanced at 2500-3000 rpm (Around G2.5 according to ISO 21940) However when used on the platform with this balance condition, it causes much more than force on bearing with respect to the calculated force on bearing.

Does anyone has any idea how to determine balancing speed for flexible body.

Thank you.

 

[drawoh]

connectionman,

Your out-of-balance force is a function of RPM squared. Can you measure centrifugal force at 2500RPM, accurately enough to get reliable predictions for 28000RPM? If your resonant frequency is 33KRPM, your out-of-balance forces are amplified anywhere up to 47KRPM.

--
JHG

 

[GregLocock]

With long flexible shafts such as steam turbines I believe they used to balance at a low speed and then incrementally higher- I think they ran at above first resonance.

At 28000 rpm you are operating rather close to your first resonance and so might see some dynamic bending, and hence vibration that cannot be balanced out using a 2 plane low speed procedure. Again with flexible shafts they balance at many planes, not just two convenient ones.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[electricpete]

Further to what Greg said, I think you have two options (or maybe both)

1 - in the shop disassemble and low-speed balance each piece separately, assuming pieces can be assembled to tight positional tolerances. The idea is that for a low speed balance of a rotor that will later flex, you need to try to keep good balance in each plane because you can't rely on out of balance in one plane cancelling out of balance in another plane in any predictable way like you can for rigid rotors.

2 - trim balance at full speed (sometimes this can be done in the field).


=====================================
(2B)+(2B)' ?

 

[Tmoose]

"it causes much more than force on bearing with respect to the calculated force on bearing."

I'm guessing the "calculated force" might be based on G 2.5 claimed by the balance shop ?

Is the assembly dis-assembled after balancing before installation in the machine?

Things I'd be thinking about.
1 - In severe cases ( a long hollow roll, with a significant heavy spot in the center) deflection/whip several can begin at 50% of the first bending mode critical speed. Stated by Hartog, and I have witnessed first hand.

2 - See attached image "mode shape vs bearing stiffness". The flexibility of the bearing/supports always lowers the frequency of first mode bending mode compared to that calculated with rigid supports. Sometimes just a little. Sometimes a LOT.

3 - To maintain the G 2.5 balance condition centering of the assembly must be better than .000 03" compared to the rotating center of the bearings used for balancing.
Similarly, if the assembly is ever disassembled and re-assembled the components must be centered better than .000 03" compared to how they were assembled for balancing.
Those are MIGHTY tough requests. Without interference fit I think it will never happen.
And interference fits can cause distortion of the assembly.

 

[electricpete]

> To maintain the G 2.5 balance condition centering of the assembly must be better than .000 03" compared to the rotating center of the bearings used for balancing.

Yes good point. As machine speed goes up the idealized eccentricity (i.e. the e defined by e = m*r/M... corresponding to converting the entire mass into a lumped mass and running far above resonance where system becomes mass controlled) goes down as shown on pages 3 and 4 here

https://irdproducts.com/assets/balance_quality_requirements_of_rigid_rotors.pdf

While that idealized e is somewhat abstract I think it does reflect the reality that as machine speed increases we need tighter and tighter positional tolerances to establish a given ISO G level

This is nothing new to Tmoose. He's the one that first tipped me off to this years ago on one of these forums. But since I took the time to recall that and make sense of it, I repeated it here on the forum (partly to help me organize my thoughts and partly in case it helps someone else)

=====================================
(2B)+(2B)' ?

 

[Tmoose]

"Those are MIGHTY tough requests."

I should have said " MIGHTY tough REQUIREMENTS."

I believe one of the ISO 1940 Standards carries notes that the tighter G grades ( maybe 2.5 and below ??) may require balancing the assembly in its very own service bearings. To me that is quit logical, since phenomena like the runout of even high quality ball and roller bearings' inner races/rings being significant compared to the "e" eccentricity for the respective G grade.
Change in centering = change in (un) balance.

 

[connectionman]

Sorry for the late response.

I guess there is some misunderstandings about the problem. Normally bearing force due to the unbalance can be calculated according to ISO 21940. As far as I know this calculation for rigid rotor. On dynamic or flexible rotor system, this can be multiplied by some factor. (I don't know how!?) Let us turn back my first question, I don't think that all flexible body must be balance at service speed. There should be a formula or graph to determine where unbalance value at specific speed converge to the service speed's unbalance value. That's why I asked because regular aviation balance machine can be turn up to 8krpm but almost all of engine or accessory connection more than 10krpm. So that normally we would do a run up measurement and get a graph like attached file. Regarding all updates. Does anyone has any idea how shouldI proceed?

 

[connectionman]

Attached file.

 

[snoopnoon]

How do you prove out overspeed if you didnt balance up to at least rated speed?
And that is not a correct blanket statement on aviation balance facility only to 8 kRPM.
Even our relatively larger "small" steam turbines can run upwards of 13kRPM on Ammonia or Ethelyne compressor trains and we still run a high speed balance and spin up to 120% rated speed. This would be 3rd critical for reference on shaft flexibility and L/D ratios.
In a previous company for larger electric hybrid motor rotors, we have balanced in excess of 20 kRPM for with a outside magnet pitch diameter of 41" and a bearing span of 50" or so.

 

[Tmoose]

"There is flexible assembly which normally operates at 28.000 rpm. Until that speed, there is no critical mode. The nearest critical mode is around 33.000 rpm. Normally assembly was balanced at 2500-3000 rpm (Around G2.5 according to ISO 21940) However when used on the platform with this balance condition, it causes much more than force on bearing with respect to the calculated force on bearing. Does anyone has any idea how to determine balancing speed for flexible body.."

"There should be a formula or graph ....."
No there ain't, because of the huge variation in rotors and machine arrangements and foundations and bearing supports and ........

How many times greater is the bearing force than you expected ?

Was the rotor dis-assembled after shop balancing in order to install it in the machine?

How many balance correction planes does your rotor offer, and where along the rotor are they?
Are there correction planes available in the middle 25% of the rotor?
A picture is required here.

is the slow speed balancing being done in a "soft bearing (or "hard bearing machine' ?

Is this a first of a kind product or installation?
If mature, the manufacturer may have a formal procedure. That process will be much like E-Pete's #1, and may also need to include provisions for "trim balancing" the assembly after installation. As I recall several API Standards have detailed procedures for building a well-balanced rotor.

If yes, you are on your own for developing a method of low speed balancing that will give good results at operating speed.

Consider a long rotor with a significant heavy spot midway between the bearings. Low speed balancing correcting in a plane near each end bearing usually is incapable of keeping the rotor straight at operating speed. The heavy spot bows the rotor, and the offset mass = unbalance that disappears when the rotor slows down.

https://www.youtube.com/watch?v=8n8f-rOahNI

https://www.youtube.com/watch?v=HeKDZJt8Ihc

starting 2:34 here-

https://www.youtube.com/watch?v=A84W69xFlto

If the rotor design permits it ( multi-piece stacked rotor) E-Pete's recommendation 1 is the logical first step.
If there is a machined feature near the middle of the rotor that is VERY round and VERY concentric with bearing journals then monitoring that feature might allow "whip balancing" at slow speed balance in a hard bearing balance machine. That center plane correction may significantly reduce the deflection at operating speed, that //may// be causing your reported increased bearing forces..

 

[Tmoose]

Maybe this would do the trick.

 

[GregLocock]

That's me out of a job then.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[drawoh]

That's me out of a job then.

It depends on how bad you are for the ozone layer.

--
JHG

 

[electricpete]

Normally bearing force due to the unbalance can be calculated according to ISO 21940. As far as I know this calculation for rigid rotor. On dynamic or flexible rotor system, this can be multiplied by some factor. (I don't know how!?)

(To add to what others have said) There's more to it than a factor. You can no longer apply rigid rotor balance methodology to a rotor that is flexible at operating speed.

=====================================
(2B)+(2B)' ?

 

[GregLocock]

Here's a paper with some nice data in it.

https://arxiv.org/ftp/arxiv/papers/1704/1704.08530.pdf

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[connectionman]

@electricpete I understand your offer but as far as I know there is no such balance machine that can perform at 33krpm with reliable results on 0.1 g.mm sensitivity. Moreover when we search on benchmark studies we have a same assembly concept but they balance their assembly at 7krpm. That's why I am asking how did they decide this rpm.

@Greg thank you Greg, I will read that article. It seems good for now.

Thanks everyone who has answered properly.

 

[electricpete]

> @electricpete I understand your offer but as far as I know there is no such balance machine that can perform at 33krpm with reliable results on 0.1 g.mm sensitivity.

ok but I wasn't suggesting anything like a 33krpm balance machine. In my first post I was suggesting to consider an at-speed balance while assembled, typically installed in the final location in the field. It requires you to start up the machine to normal speed for vibration measurements. Then shut down the machine to a condition where you can add balance weights. Then start up again and get more vibration measurements. Repeat as needed. If you're so inclined you could try estimate residual unbalance based on remaining vibration and calculated effect vectors. But there are of course limits to how well you can balance and estimate, considering that there may be other sources of 1x vibration in an operating machine, besides unbalance.

Procedures like that are sometimes used for steam turbines. Spinning the turbine at speed requires steam and a condensor vaccum. Adding balance weights requires removing steam, breaking vaccum and removing an access cover. It's a lot of work but sometimes it makes sense. I imagine most equipment does not have similar access covers positioned to add balance weights without substantial equipment disassembly. The targets for balance in that situation are likely based more on measured vibration levels than an estimate of residual unbalance.

=====================================
(2B)+(2B)' ?

 

[Tmoose]

"Normally bearing force due to the unbalance can be calculated according to ISO 21940. As far as I know this calculation for rigid rotor."

Looks like there are more than 20 different parts/sub volumes of ISO 21940. Some are specifically for "rigid" rotors and some are for "flexible" rotors.

2140-1 - "to give guidance on the usage of the other parts of the ISO 21940 series"

1940-2:2017 defines terms on balancing.

21940-11:2016 establishes procedures and unbalance tolerances for balancing rotors with rigid behaviour.

21940-12:2016. Mechanical vibration - Rotor balancing - Part 12: Procedures and tolerances for rotors with flexible behaviour.
This nationally adopted international standard presents typical configurations of rotors with flexible behaviour in accordance with their characteristics and balancing requirements, describes balancing procedures, specifies methods of assessment of the final state of balance, and establishes guidelines for balance quality criteria. Can also serve as a basis for more involved investigations, e.g. when a more exact determination of the required balance quality is necessary. If due regard is paid to the specified methods of manufacture and balance tolerances, satisfactory running conditions can be expected. Is not intended to serve as an acceptance specification for any rotor, but rather to give indications of how to avoid gross deficiencies and unnecessarily restrictive requirements. Structural resonances and modifications thereof lie outside the scope of this document. The methods and criteria given are the result of experience with general industrial machinery. It is possible that they are not directly applicable to specialized equipment or to special circumstances. Therefore, in some cases, deviations from this document are possible

21940-23 for protective enclosures for the measuring station of balancing machines

 

[Tmoose]

"Moreover when we search on benchmark studies we have a same assembly concept but they balance their assembly at 7krpm. "

Can you proved a link to that document, or a copy of it?

thanks,

Dan T