You may find this interesting. Cited below is an excerpt from an expert opinion posted at Reliabilityweb.com tips:

Here is another convenient check (for electromagnetic problems resulting in 2xLF). First of all, note the amplitude at 2xLF in the horizontal position. Instead of leaving the accelerometer in the normal horizontal or vertical position, secure it at the edge of the motor in a tangential position. Now note the amplitude at 2xLF. If there is an electromagnetic problem affecting the stator, the amplitude at 2xLF should be considerably higher. This will help to confirm a stator problem.

The reason for this behavior has to do with torque. The 2xLF is actually a “torque pulse” frequency. When there is an electromagnetic problem, it will affect the torque. Torque produces a twisting action in motor. Therefore, when the accelerometer is positioned in a tangential location, it will be respond to a change in the twisting action much better that it will, if positioned radially. It will be more sensitive to the torque since it is in a tangential location. The amplitude will be considerably higher at 2xLF if there is a stator problem.

What is Board's opinion on torgue pulsation at 2xLF?

Thanks

The torsional pulsation would be at running speed, or a multiple of running speed. I used to use this trick on synchronous motors to verify that the rotor field windings were bad when customers would not allow us to shut down the machine to do voltage drop checks. (and, if the fields were bad enough for the rotor to drop out of synchronization).

The 2XLF is an related to the airgap fields and, as a result, will not translate into a torsional vibration but cause a change to the interaction of rotor and stator fields radially (but at a slight angle).

I think the author (I read the rest of the tip) may have mistaken the additional 'sway' (yep, just made that one up) at the top of the motor as it will move more (the base is fixed and the top moves).

The attached is an animation (modal analysis) of a large pump motor that we analyzed last week. This is the horizontal movement at 2LF on a 3600 RPM. Note that the movement is less at the base and greater at the top. (right click to save and then put your .avi player on repeat to watch the motion).

This is a twisted frame condition coupled with an airgap problem (opposite front to back).

Sincerely,
Howard

Feedwater_Pump_Hor_LF.Avi (550 Kb, 95 downloads)

Howard,

That is exactly what I had in mind.

Dynamic force at 2xLF is directed radially ( a slight tangential component is still present but it is not likely to cause measurable torsional vibration). Rocking motion, when an accel is positioned at the top tangentially, may be mistaken for torsional if one does not move it 90 deg in order to verify it.

I guess not all tips on Reliability site are reliable ....

(Thanks for the ODS)

David

The condition described in the second posting ("Penrose Sway") is exaggerated when using a triaxial probe that is mounted on top of the motor. I have found that when monitoring small HVAC motors, invariably mounted on flimsy flexing bases, it is quite common to see a much higher Tangential reading from the triax seemingly because of this torsional flexibility. You can also sometimes pick up a torsional surge visually with a strobe light, but this is at a much lower frequency, and I suspect it is due either to a hunting VFD controller or a flexing cycle of the belts. Any thoughts?

Rich Wurzbach
Maintenance Reliability Group, LLC

quote:

Originally posted by Rich Wurzbach:
You can also sometimes pick up a torsional surge visually with a strobe light, but this is at a much lower frequency, and I suspect it is due either to a hunting VFD controller or a flexing cycle of the belts. Any thoughts?

Rich,

Assuming this vibration attenuates, isn't this a manifestation of torsional resonance due to torsional "impact" at the start?

David

When I observe this, it appears to be a steady-state condition. If it occured at startup and dissipated, then perhaps I could believe it to be resonance. But I think the forcing function here is related to an operational condition, not a startup condition.

Rich

David,
What a confusing replies so far to your interesting kick-off
Let me try to shed some light, hopefully without adding more confusement.
The twisting action of the stator requires at least two accels in order to prove anti-phase.
I have done this succesfully with thyristor fed DC motors that generate a lot of 6*Fline (300 hz or 360 hz) ac torque superimposed on the average dc torque when loaded. Action on rotor = reaction on stator. With one accel on one motor foot and a second accel on a motor foot at the opposite side of the motor, both in vertical direction, one can observe 180 degrees phase shift at 6*Fline.
With AC induction motors I have not been so succesfull yet in this way. Of course not. On one hand torsional vibrations are not so common with AC induction motors, even VFD applications. And certainly one should be very, very carefull with 2-pole motors when trying to link an electrical flaw and torsional vibrations when investigating 2*Fline rad/ax/tang signatures.
In my opinion torsional vibrations on AC machines should not at all be detected with accels. Torsional pulsations are basically angular speed variations and therefore you need to couple something like an encoder (often readily available with VFD and DC drives) and than FM demodulate the high frequency pulse train of this encoder (512 pulses/sec * 1500 rpm = 12.8 khz carrier, as an example). Or - if you love accels - pass the HF RBPF signal through a FM demodulator. BTW, nice method to detect defect rotor bars too as the demodulated signal will have PP pulsation frequency, this is a true LF torsional vibration!
Regards,
Arie Mol
Rotating Equipment Consultant, NL

That's a great discussion. I think there have been good points raised. I'd like to add my two cents.

We all know that 1x for mechanical vibration can have a large number of possible causes. We can guess that usually unbalance or misalignment with some other clues, but we all know that there are a wide variety of causes of 1x.

I think 2*LF is the electrical analogy to 1x vibration, from the standpoint that there can be many different causes of 2*LF. We have an idea of the most likely and common causes which would include airgap asymmetry and rotating oval deformation of the core in 2-pole motors even with perfect airgap, but in general there can be many causes.

From there I think everything gets more complicated. We know that establishing flat feet on the machine often improves the vibration level. What is the mechanism by which soft feet can increase 2*LF vibration? I think there are a number of possible explanations. One often suggested is that s the distorted frame distorts the airgap. Another is that tweaking the feet conditions can tweak the machine support resonances. Finally, I think that an under-appreciated mechanism is that the vibration coupling from the oval rotating deformation of 2-pole core to the frame depends very heavily upon the interface between the frame and the core. I have seen a manufacturer "cure" 2*LF on a 2-pole motor by tinkering with the frame ribs to change that vibration coupling.... it didn't change the rotating oval deformation of the core... only how much of that core oval deformation was transmitted to the frame and then measured on the bearing housing. For some NEMA frame motors that have the core pressed into the frame, if the feet are not flat then the frame is twisted and the fit between core and frame becomes tighter and the frame responds more to that inevitable (on 2-pole motors) rotating oval deformation of the core, then of course when the frame vibrates we see it show up on the bearing housings. I have heard two different OEM's explain similar versions of this last explanation about coupling between the core and the frame. It rings a little truer to me than the other explanations.. but again there may not be just one explanation for how the feet interact with the 2*LF vibration..

There is a difference between eccentric airgap and rotating oval deformation in terms of where the forces act. Off-center airgap produces a force on the rotor transmitted through the bearings and vibration will be higher at the bearings and likely lower in the center of the stator frame. Rotating oval deformation does not cause any force on the rotor or bearings and in some cases (depends on coupling mechanism from core to frame) you may find vibration higher at the center of the stator frame than at the bearing housings. And it seems plausible there may be difference in tangential vibrations for these two types of vibrations, depending on frame support.

Now what about torsional oscillations? We collectively as vib analysts know a lot less about torsional vibration because we don't typically measure it. If it existed, we often wouldn't even know it.

Can torsional oscillation at 2*LF occur on an induction motor? Absolutely. Theory tells us that a simple single-phase motor running with no aux winding (imagine that centrifugal switch has cutout the start winding) has extremely high 2*LF torque oscillation. (in fact, torque can go to 0 twice per cycle when the current passes thru 0). For a three-phase motor the situation is not as dramatic, but a similar thing can occur if the voltages are unbalanced. If you subtract out the balanced portion, the remaining unbalanced portion can look very much like a single-phase winding because the reverse rotating components o the field no longer cancel. Krauss' "Analysis of Electric Machinery" section 9.5 provides a numerical simulation of the 2*LF torque oscillations expected for a 3-phase induction motor operating on unbalanced voltage.

Now the question of whether torsional vibration shows up in lateral vibration measurements. I think that sometimes it may or may not... depends on the geometry in complicated ways which are mostly beyond my grasp.

But one easy way to imagine one situation when there will be coupling between torsional and lateral vibration is something David G pointed out to me in the context of discussing broken rotor bars. The same torque seen by the rotor is also seen by the stator (a unique fact for motors... does not apply for other machinery). If the support of the stator is asymmetric (H vs V), then the oscillating torque on the stator will show up as lateral movement of the stator. Imagine a horizontal motor standing on a tall flimsy stand (very low H stiffness). Turn on the steady torque in the CW direction on the rotor and CCW direction on the stator. The stator moves slightly to the left but does not vibrate because the torque is steady. If the torque oscillates, then the motor will vibrate horizontally at the same frequency.

For Rich mentioning movement at low frequency, one remote possibility is that it could be torque oscillation from the load or from rotor asymmetry / broken bar.

I'd like to hear that author explain why he made those statements. There may be some nuggets of truth or knowledge hidden behind there that we just don't appreciate. I suspect if we all had a ton of time, there may be a lot more intelligence that can be gleaned by carefully studying 2*LF tangential measurements, 2*LF phase relationships and O.D.S, including comparing 2*LF vibration at the center of the stator frame compared to the bearings to visualize what is taking place. But who has the time. In general I treat 2*LF on 2-pole motors as a less-severe condition than a comparable level of 1x because I believe in most cases it is that rotating oval deformation that does not necessarily load the bearings as much as a force on the rotor transmitted through the bearings

This message has been edited. Last edited by: electricpete,

Arie:

In 1997 we had an in-depth discussion related to torque measurements in vibration at the Chicago Vibration Institute Chapter. The method the presenter used actually did use an 'encoder' of a sort. It was a specially lined tape with a laser with the tape wrapped around the shaft of the motor and the laser emitter/receiver positioned perpendicular to it.

I have used the method that was initially presented in this string, but only to determine if such vibration existed. Then would use a strobe to see if the shaft was stable (unstable would indicate severe torsional vibration).

More recently, I have used Electrical Signature Analysis/Motor Current Signature Analysis to look at torsional issues with motor loads.

Howard

As Pete has noted, it is important to realize that torsional vibration at 2xLF, as oppose to the lateral one, is caused by a completely different mechanism - electrical voltage unbalance in the stator. To detect torsional vibration in motors one has to position an accelerometer in tangential direction on one side of the motor (viewing in axial direction)and later on the other side. As Arie mentioned, once there is a 180 deg shift, a torsional vibration is present.

But torsional vibration at 2xLF can be easyly mistaken for that of radial if one positioned an accelerometer radially in horizonatal direction on the motor. In this case, if torsional vibration is present, the top point of the motor will be moving at higher amplitude then the bottom one (where motor is fastened to the base) and in opposite direction. Having the accelerometer positioned radially in the middle of the height will create a false impression of radial excitation force. In this respect the ODS at 2xLF presented above is quite confusing.

David