1/3 x running speed vibration in electric motor

[PaulDonnellan]

Hello all,

We have a 2400kW, 60Hz, 1185rpm motor with white metal oil-mist bearings that has a vertical radial vibration at 1/3 x running speed. The 1/3 x running speed peak is not sinusoidal and the horizontal looks more normal with only a small component.

It has no broken rotor bars (checked stator current and running speed sidebands).

Our experience is that mechanical looseness is normally integer harmonics or 0.5x integer harmonics.

The only thing I am left with is rubbing?

All thoughts / further tests are welcome.

P.

 

[skills]

There isn't anything within a revolving motor itself that I believe will cause a sub-fundamental vibration. Ideas, anyone?

 

[mcdm]

Hi Paul
technical associates of charlotte diagnostic charts suggest that mechanical looseness can often cause sub harmonics at 1/2 0r1/3rpm
some diagnostic charts may suggest oil whirl this can occur at 39%-40%. so depending on your data collection resolution this may be the case.
you could try a bump test to confirm structural resonance
regards mike cooney

 

[electricpete]

I have also heard that looseness can in theory produce vibration at 1/3 rpm. To prove or rule out, I would check with high resolution. If it's really 0.345x or something like that, then it's not likely looseness.

Vertical deep draft pump/motors often have approximately-sinsoidal vibration in the range of 0.3x-0.4x. We have it right now on our power plant condensate pumps. It can be due to operation at high shutoff pressure/low flow, wear of internal pump surfaces causing recirculation. Possibly something similar to whirl although in water-lubricated sleeve type bearings. I have read somewhere where these things were described in two categories (that can produce subsync vibration): hydraulic or hydrodynamic excitation.

See if vibration disappears when motor is run uncoupled (that would rule out all of the above items).

Also try a bump test.

Consider observing vibration under different flow conditions... increased vib with lower flow indicates pump is the source.

Can you describe some more what you mean by non-sinusoidal peak at 1/3x?

 

[Franko]

There could be a natural frequency of the system at 1/3 running speed - akin to resonant whip in hydrodynamic bearings - where subharmonic frequency stays constant with speed. But non-linearities can give rise to high amplitudes when the frequency is at exact fractions of speed. It's hard to believe that the first rotor/ bearing critical would be near 400 cpm - perhaps so if the bearings are loose. Can a pinch check of liners to retainers be done? We once had a similar case on a steam turbine we supplied where for a temporary fix, a large set-screw was threaded through the top of the bearing housing against a loose retainer for a tilting[pad bearing. The president of the user's company came to the platform to see it for himself.

Since it's six-pole, could it be to assymetry (air gap?)giving 1/2 of the six poles - perhaps giving pulsating torque? Pulsating torque is one of many self-excitation mechanisms discussed in a G.E. paper by F.F. Ehrich (ASME paper 72-DE-21). It may be in ASME transactions. Linda Law library should be able to get it.

http://www.lindahall.org/

(Sometimes old papers are the best)

 

[PaulDonnellan]

Hello again.

Just to confirm that the vibration we see is exactly 1/3 x running speed. I have a time waveform that shows that every 3rd revolution I get a spiky peak.

There is a piece of literature from Bently Nevada that investigates rubbing in bearings and has mathematics (scary) and experiments to support that you can get 1/2x, 1/3x, 1/4x etc. harmonics of running speed.

Cheers.

P.

 

[electricpete]

Hi Paul... got a link?

In my mind looseness and rubbing are very similar.
I have my own pet explanation for the phenomenon of 1/2x, 1/3x, 1/4x.. not sure if it’s right or wrong.

Imagine you have a shaker table moving up and down sinusoidally. Now drop a ping-pong ball onto the shaker table. Each time it hits it bounces back up in the air. But how much force/velocity is generated by each impact depends on what phase it hits in the shaker table cycle. When you first toss the ball the impacts will surely be random. Each time the ball pops up with a different velocity to a different height for a different length of time and lands again at a different point in the cycle.

Continue this for long period of time and one of two things will happen.
#1 – The activity will remain random. If viewed on time waveform the spacing of the impacts will remain random. The spectrum will not have any clear harmonics but only a raised noise floor. Similar to certain types of rubbing.

#2 – The activity may reach some kind of periodic steady state where the ball pops to the same height each time. Once it finds this rhythm it will stay there forever. It must hit at the same point (phase) in every shaker table cycle. But not necessarily once per cycle (1x). It may be once every two cycles or once every 3 cycles or once every 4 cycles. This would give rise to 1/2x,. 1/3x or 1/4x (plus harmonics).

Drawing the parallel from this exact analogy to a rotor is a little difficult, but I think the basic math is the same. In the case of rotor there is a periodic sinusoidal force like unbalance applied to the rotor. The rotor moves and impacts the stationary part initially in a random fashion. If steady state motion is reached the rotor likely will make an impact every 1, 2, 3, 4 cycles.

Just my way of looking at it. Any comments?

 

[Franko]

I can't envision the free ping pong analogy - perhaps if there was stuck in the bottom of a spring attached to a beam above the table, with a spring/mass natural frequency averaging between .3 and .35 times depending on the deflection (non-linear).

I don't have much experience with motors, but it shouldn't have tight seal clearances that could rub - bearing looseness would be my first check. Agree with Paul that 1/2 rpm is a sign of looseness, but 1/3'rd is also reported.

 

[electricpete]

One more comment about ping-pong ball analogy, there is no one-to-one correspondence between the ball/table and rotor/stator.

The important point is that IF steady state is ever achieved, then impacts must occur at the same phase with respect to the periodic forcing function. That only occurs when impacts occur at 1x, 1/2x, 13/x, 1/4x etc.

 

[alwayslearning]

Electric Pete,

A good imagination and a good example.. I can vaguely correlate this analogy to the problem that might be occuring (although in a much more complex way). It will be interesting to see if someone has worked the maths based on your analogy.

Regards