Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
electricpete

Sr. Member
Registered:
Posts: 451
Reply with quote  #1 

UNDERSTANDING THE VIBRATION FORCES IN INDUCTION MOTORS

By Michael J. Costello, President, Magnetic Products and Services, Incorporated Houston Texas, in PROCEEDINGS OF THE NINETEENTH TURBO MACHINERY SYMPOSIUM

https://pdfs.semanticscholar.org/7b40/d0cbe1504018c377b923138fe9f1cab19866.pdf

There's a lot in there, but two things that interested me:

=== === === ===

1 –  1 times slip speed sidebands (instead of pole pass sidebands)

We know that dynamic eccentricity can cause pole pass sidebands (poles times slip speed) in vibration and static eccentricity can cause 2*LF.  Apparently if you have both present, you can get sidebands at 1* slip speed.  It makes some sense if you think about it, and he demonstrates the logic in figure 4.   I’ve never seen it myself, but it is shown in his case study #5.   There you see full load is 3567 rpm.  Slip speed is (3600-3567)/60  = 0.55hz.  Pole pass frequency would be poles times slip speed = 2*0.55hz = 1.1hz.

In figure 14 bottom plot, we see a familiar pattern of 120hz and 2x = 118.9hz, the difference being pole pass frequency 1.1 hz.  BUT, there is a peak half way in between at 119.45hz -  one times slip speed away from the other two.

In figure 14 middle plot we see running speed 59.45hz. Slip speed above that (+0.55hz) is a peak at 60hz.  Another slip speed above that is 60.55hz.  There are also similar spacings below running speed although they’re not labeled.

== == == ==

2  - How the vibration transmission of 2*LF from stator core to frame can be tweaked to hide the 2*LF.

For 2-pole motors the stator core itself has a high 2*LF vibration due to the rotating elliptical wave.  The motor OEM’s have a lot of sneaky ways to isolate the core vibration so it doesn’t get to the bearing housings where we measure vibration (There might be some benefit in reliability to protecting the bearings from that vib source, but based on conversations with OEM's it's done more to keep the customers from complaining about the 2*LF that shows on their housing vibration measurements).

In large sleeve bearing motors it’s easy. The frame is a large metal box of thin cheap walls with a lot of empty space inside  The core is mounted to a broad area of  the BOTTOM of the motor frame.  The core is not in any way attached to the side or the ends of the motor frame, so it has very little influence on the bearing housing vibration.  The ends of the frame where the housings are supported are also supported from below, not relying so much on their attachment to the center frame. 

In smaller (NEMA frame) rolling element motors, it’s harder.  The endbells are supported from the center frame rather then supported from below. Since the frame is made as small as possible, there is no empty space.  If we tried to only attach the core at the bottom of the frame it’s not a broad enough support. The core has to be supported several places around the circumference by the frame. The interface between the core and the frame can be critical in influencing how the 2LF vibration is transmitted to the bearings.  For motors with simple cylindrical interference fit between these components, I've heard they are more likely to have 2*LF in general and more susceptible to soft-foot effects due to twisting of the frame which changes the interference pressure / pattern between these two components.  But I have a story about some small motors where the core was not set in the frame with a complete cylindrical interference fit, instead there were ribs on the frame supporting the core.... 

...In the early 2000's we bought seven 100hp 2-pole motors from a motor OEM factory that happens to be nearby, with a specification that they had to pass our vibration requirements during testing while rigidly bolted down.   The OEM was used to testing motors on a flexible pad, but accepted our rigid spec.  When we got to the factory test, they seemed very surprised to see high 2*LF (0.2ips+)  in the horizontal direction on their motors bolted down during the test run.  They spent several days trying to fix it, by getting the airgap perfect (even sleeved and rebored one of the endbells).  They planed the motor feet. None of it worked, the 2*LF didn’t decrease very much.   Then they had a phone call with an engineer at a remote location. They informed me they were going to cut some slots in the motor frame ribs supporting the core and that would reduce transmission of the 2*LF vibration from the core to the frame.  They didn't permit me to watch exactly what they did, saying it was some kind of secret.  I came back the next day for the run and the 2*LF was almost gone (<0.02 ips I think).  Their modification had worked, but it’s always been just a mysterious story I can tell, with no proof or specifics of what they cut..... Until now!  Figure 11 shows a modification where the frame ribs are notched to minimize transmission of 2*LF vibration from the core to the frame.  I'm not sure if it's exactly the same as for our motors, but probably similar.  It looks like the notches in the ribs are at each end of the ribs. Maybe by removing most of ribs at the end this transmits the core vibration efficiently to the center of the frame and keeps it away from the ends of the frame where the endbells are supported. Or mabye something much more subtle. Beats me. 

electricpete

Sr. Member
Registered:
Posts: 451
Reply with quote  #2 
Quote:
for 2-pole motors the stator core itself has a high 2*LF vibration due to the rotating elliptical wave.  The motor OEM’s have a lot of sneaky ways to isolate the core vibration so it doesn’t get to the bearing housings where we measure vibration

Here is another illustration of this:
https://pdfs.semanticscholar.org/d815/a65f12ea894f2f2b54187cf6d7d7ed7c8aeb.pdf

Just under figure 9, you'll see they measured the core back-iron at two locations and saw 2*LF was 4.5 times as high as on the bearing housings. 
George D

Member
Registered:
Posts: 30
Reply with quote  #3 
Thanks so much for sharing Pete. Always appreciate your insight, and willingness to share.

I’m going to read these two papers carefully, and contrast them to some thoughts I’ve been considering, lately, regarding a couple of motor related vibration tenets. I may have learned these from you? I’m not smart enough to come up with them myself. I’d be curious to hear your thoughts, and whether or not you believe these theories to be true?

1. We can accept a slightly higher threshold of 2xLF casing vibration versus running speed related vibration? This is because the 2xLF energy flow path travels from the stator directly to the bearing housing, rather than going through the bearing.
2. The Fmax for routinely trended motor casing velocities should be set below the rotor bar pass frequency. RBPF vibration is an expected condition of an induction motor, and should not be used to judge the severity of a motor’s vibration.

Like I said, still testing these theories? Looking for corroboration, or otherwise... from you or anybody else?
electricpete

Sr. Member
Registered:
Posts: 451
Reply with quote  #4 

I'll take the easy one first. 

Quote:
2. The Fmax for routinely trended motor casing velocities should be set below the rotor bar pass frequency. RBPF vibration is an expected condition of an induction motor, and should not be used to judge the severity of a motor’s vibration.

I agree. RBPF is often load dependent and may change with other factors.  The T.A. charts suggest it is useful for finding “rotor bar problem”.  I don’t know how well it correlates, given that it shows up in many healthy otors. I would rather look for pole pass sidebands in vibration and especially pole pass sidebands in current signature as indication of rotor bar electrical problem.

Quote:
1. We can accept a slightly higher threshold of 2xLF casing vibration versus running speed related vibration? This is because the 2xLF energy flow path travels from the stator directly to the bearing housing, rather than going through the bearing. 

I would say I generally am not as concerned about 2*LF vibration as other sources.  But I don't ignore 2*LF in the same way I tend to ignore RBPF. 

1 - What does it (2*LF) tell us about bearing load and how does that compare to other types of vibration? 
  • For simplicity, let’s say there are two sources of 1x (misalignment and unbalance) and two sources of 2*LF (rotating stator core deformation and unbalanced magnetic pull due to uneven airgap).  I would rank the ratio of force on bearings to vibration velocity as follows:
  • Misalignment > mechanical unbalance ~ magnetic unbalanced magnetic pull > rotating stator core deformation.
  • In other words misalignment can cause a lot of bearing forces without showing up much in vibration.  Rotating stator deformation causes a lot of measured 2*LF vibration but no/little bearing forces.  Mechanical and magnetic unbalance are somewhere in between.... for the most part we don't worry about their effect on bearings unless the vibration gets very high. 
2 - What else does 2*LF tell us? 
  • We also use vibration not just as an indication of forces on bearings, but more importantly as indication of something going on that we want to know about.  For 2*LF like other types of vibration, I would be interested if I see an increase.    In theory (worst case) increase in 2*LF might indicate change in airgap, which is something I would want to know about.  It is not very common for airgap to change in service, but we had a gradual change in airgap once on a vertical slow speed  motor due to degradation of the electrically-insulating hardware associated with upper bearing, that eventually led to a rotor/stator rub.  In that particular case, the signs in vibration (that we didn’t notice at the time) were increase in 1x, not 2*LF.  (even when in retrospect I checked in velocity, where we normally watch displacement on these slow speed machines).  That was a complicated case from a long time ago, I guess we must have had dynamic eccentricity on top of static. And slow speed motors act different than others.  I think for other similar scenarios this type of fault might easily have shown up as increase in 2*LF.
Curran919

Sr. Member / Supporter
Registered:
Posts: 412
Reply with quote  #5 
Quote:
Originally Posted by electricpete

They informed me they were going to cut some slots in the motor frame ribs supporting the core and that would reduce transmission of the 2*LF vibration from the core to the frame.



This just sounds like a tradeoff of vibration in a place that is measured to someplace where it is not. What is the more critical failure mode? 2xLF on the bearing housing or 2xLF on the core?

That is some shady shit...
John from PA

Sr. Member
Registered:
Posts: 779
Reply with quote  #6 
Just as an aside, other good articles on motor vibration can be found at https://www.industry.usa.siemens.com/drives/us/en/electric-motor/anema-motors/specification/Documents/approach-to-solving-motor-vibration-prob.pdf and https://oaktrust.library.tamu.edu/bitstream/handle/1969.1/163643/T1417-22.pdf?sequence=1&isAllowed=y


electricpete

Sr. Member
Registered:
Posts: 451
Reply with quote  #7 

Quote:
This just sounds like a tradeoff of vibration in a place that is measured to someplace where it is not. What is the more critical failure mode? 2xLF on the bearing housing or 2xLF on the core?

That is some shady shit...

Yeah, I thought you guys would be surprised by that part.  Likewise it might be surprising as reported in the second post that the 2*LF on the back of the core (that you can’t measure) can be 4-5 times as high as what you can measure on the bearing housings.   

2*LF due to the normal rotating oval pattern (not eccentricity) at either location (bearing or core) is not particularly critical.  We already talked about the bearing. The core is not particularly critical as long as rotor stator / contact is avoided.

In general, I think the core vibration could be reduced a little at a big expense of increase in frame/ bearing vibration if the core is more tightly connected to the frame.  OEM’s often seem to err on the side of reducing what can be seen (the bearing vibration). That’s my general impression and I’m probably out on a ledge saying that because there is a lot of variability between designs. In the end I haven’t heard of any core problems or rotor/stator rubs attributed to lack of frame support to the core.   The closest I can think of is that the core interlaminar insulation resistance can degrade over time under the influence of vibration and temperature, resulting in hot spots and necessitating expensive rework. But the most important vibration-related aspect is said to be the tight axial clamping of the core to prevent relative motion of laminations, not the radial support of the core around it’s circumference.

On a related subject to hiding vibrations… a lot of vertical motor designs hide the bearing housing deep inside the motor where you can’t get close to measure vibration, making it tougher to detect early stage bearing defects.. In this case it’s not intentional, there are good reasons.  On the top bearing, it is generally the result of embedding the bearing in a large oil reservoir, or else there is a box-like shroud which directs inlet airflow over the bearing housing for added cooling, but that same shroud prevents access close to the bearing. On the bottom, the bottom bracket has to be wide enough to accomodate the mouting flange.  Unless the bottom of the housing is extended below the bottom bracket, you can’t get close to the bottom housing.

Quote:
just as an aside, other good articles on motor vibration can be found at 

The Siemens article (analytical approach to motor vibration problems) is a good one, often referenced and worth reading. On the RBPF +/- 2*LF, they explain why it is a normal pattern in induction motors and they don’t believe the associated stresses or forces are cause concern. They also refute the claim that it is associated with loose rotor bars. 

The Corey article I was not impressed with. I noted a slew of terminology that was a little off and sentences that suggested to me he was not particularly familiar with the subject.

 “any phase imbalance in terms of resistance of voltage will add very significantly to the 120 Hz vibration force with harmonics.”

What the heck does he mean by “phase imbalance in terms of resistance of voltage”…. did he mean to intentionally exclude unbalances due to unbalanced source voltage (which of course would also cause 2*LF)?  He  could’ve just said “current imbalance” instead and would have been more succinct and more accurate.  I could go on, just a lot of small things that irked me. 

His piece de resistance is more interesting. After spending a lot of time explaining the theory of 2*LF vibration in terms of magnetic forces, he presents the case study where 2*LF varied with load.  It was 0.4-0.5 ips under load, but only 0.05 ips at no-load.  The conclusion: “The air gap was found to be 20 percent eccentric, due to the stator core offset, with respect to the bearing bracket to housing headfit. Removal of the condition cured the problem.”  The same behavior was mentioned prominently in the front page.

He does not treat this as unexpected, nor does he offer any explanation for this load variation.  A reader might logically conclude that this load dependence of 2*LF follows somehow from the theory he presented earlier in the  paper.   It does not.

Load dependence of 2*LF does not follow in any straightforward way from the theory without some other factors at work.   Unlike RBPF pattern vibration (which is expected to increase with load because it depends on rotor current), 2*LF is not expected to increase with load in general.  The 2*LF (from either rotating elliptical stator deformation or from unbalanced pull when eccentric) arises from the fundamental working flux in the machine.   That flux is approximately constant vs load, and if anything the portions of it crossing the airgap decrease with load due to the stator leakage reactance. 

I don’t doubt the author is reporting his observations correctly. There can be reasons that 2*LF would go up with load.  One reason would be distortion of the core or frame due to the electromagnetic torque on the core which is transmitted to the frame. This can change their relative positions in at least two ways: 1 - a horizontal core supported from below can move horizontally under the influence of torque even if the frame doesn’t move. 2 – force transmitted from core to frame can cause the frame to distort affecting bearing position and rotor position, which provides second way for load to affect airgap eccentricity.   

That doesn't explain everything that went on in his case study. In particular why did 2LF go away when the eccentricity was corrected, if presumably the load related changes remained?  Maybe the static measurable eccentricity looking at airgaps with the machine secured was in the same direction as the eccentricity induced by load-related changes..... one factor alone was tolerable but the combination is not (as he mentioned unbalanced magnetic force is a nonlinear function of eccentricity).  Or maybe the process of disassembling and reassembling the machine changed something else.   If the author had realized the load dependence didn't follow the theory he might have weighed in on some of this, but noooo. 

Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.