High axial vibration on a 4 pole motor Vibration Analysis - Balancing - Alignment
Reply
Member 3 posts BMag34 Member 3 posts
BMag34
Dear members,
 
We received a 355kW, 4000V, 4pole, 3ɸ squirrel cage induction motor from a customer. 5 of the rotor bars upon arrival at our shop was visibly broken but still passed the growler test. That being said, we recommended the full rebaring of the rotor to the client. They insisted on only replacing only the broken ones since the motor was a spare and was a critical asset to their plant's operation.
All of the cracked bars were replaced and the unit was assembled. During no-load test run, the motor had an unusually high axial vibration on the drive end (3.7mm/s rms). As shown on the spectrum, the peaks are located at 222.5hz as well as 247.5hz. We are looking into the possibility of the 222.5hz coming from an electrical issue (rotor bar defect). during the refacing work after the rotor rebaring, an additional crack also appeared indicating that the rotor bars are already brittle (I assume this indicates short mean time before failure of the old bars). I tried comparing the 222.5hz with the ball pass frequencies, bar pass frequencies and nothing matches. I am looking to get your expert opinions on this matter, is there any other factors that I should look into and is there anything that I might've missed during my analysis.
 
The following might also be worth mentioning:
• All electrical tests passed (rotor growler, stator winding, surge and insulation resistance)
• Resleeving of drive end and non-drive end bearing housing was performed.
• Mechanical tolerances have been double checked and all is within tolerance.
• Motor bearings are 6224/C3 on both ends, both brand new bearings.
• Motor starts and stops 4x each day
• Dye penetrant crack testing was performed on the rotor bars, no additional bar defects was identified
• 58 total rotor bars
• rated slip speed of 1781 but we get around 1793 at no load conditions
 
I only use old spectrum analyzer, I can only upload photos of the screen. I apologize for the low-quality photo of spectrum.
viber_image_2025-09-04_17-56-32-349.jpg
Reply 0 0
electricpete Supporter, Moderator 1,770 posts
electricpete
At first glance I would have guessed 50hz supply frequency, in which case the 25.0 hz spacing between 222.5 and 227.5hz peaks would be 1.00 x running speed. I guess that's just a coincidence (and supply frequency is really 60hz)....
• rated slip speed of 1781 but we get around 1793 at no load conditions
If your supply frequency is 60hz, and sync speed of 1800rpm, how the heck do you end up with 1793 at no-load conditions!!?!! That could be a huge red flag of severe rotor problem... bad enough to degrade the torque capability so much that friction losses slow it by 7rpm !?! (which is the slip you wouldn't expect until the motor is loaded to 7/19 fraction of full load). But before jumping to that extreme conclusion, there are some other things to consider first
  • How was the machine powered during this no-load test? (motor generator, autotransformer, wound rotor motor used as autotransformer, electronic supply, other?)
  • Do you have an accurate estimate of the supply frequency to the motor (how close to 60hz?). If you don't have a record then look at the 2*LF vibration peak on log scale and tell us what is the frequency (should be 120.0hz, but 119.7hz would explain the 1793 noload rpm).
  • How accurately are the 1x peak and the 2*LF peaks measured in vibration?
    • what is the frequency bin width
    • Are you using frequency interpolation to estimate the peak frequency more accurately than bin center? (in Emonitor and AMS this is invoked with the peak label feature)
    • Ideally verify speed using a strobe
  • Assuming the actual speed is confirmed 1793 with supply frequency near 60.0hz, are there any pole pass frequencies present in vibration? (would be approx 28rpm above and below 1793).
  • Verify supply voltage magnitude during the test. If this was a reduced voltage test, what was the actual voltage during the no-load test?
  • Verify supply voltage balance (or current balance as a proxy... typically 5-10x more current unbalance than voltage unbalance at no load)
  • Has this motor been load tested since the rotor repair? If a rotor problem existed which was bad enough to cause that much slip at no load with balanced rated supply voltage, then that motor would definitely fail the load test miserably (and rotor problems would show as pole pass anomalies)
  • Have you done a single phase test for rotor condition since rotor repair? That is a test of rotor condition which can be done without needing to load the motor or disassemble it.
  • Do you know if the alloy of replacement bars was matched to the original rotor bars? (If I were the customer I would ask for documentation to prove the resistivity matched)
  • Is it practical to check airgap and axial alignment of iron and inspect for internal severe rubs? (reaching at straws for remotely possible causes other than rotor for severe torque capability degradation... IF that's what we're seeing)
Stepping back from the rotor to your original question of identifying these peaks.
  • Do you know how many rotor bars and stator slots?
  • I don't currently have access to my software... what are the fault frequencies for 6224 in orders?
  • If there is an electronic supply involved, please provide more details including the supply frequency at the input to the electronic supply
  • Maybe post a log scale spectrum with the highest of the peaks labeled to see if you can fit these two peaks into part of a larger pattern (the more peaks you can fit into a pattern, the more distinctive the pattern becomes imo)
 
ps - I would love to hear more about the rotor and any photos you can share. It seems unusual that a growler test would not detect broken bars, but maybe there are some lessons learned there in technique somewhere.
Reply 0 0
Danny Harvey Admin / Sr. Member / Supporter 2,720 posts
Danny Harvey
Pete,
 
I was wondering about the slip too, but from my normal viewpoint of ignorance when it comes to big motors. The few tests that I have witnessed have shown the motor running within 1 or 2 rpm of synchronous speed. I have seen a couple of tests that were invalid because of the difficulty in producing the required voltages in the shop.
 
The OP is apparently in the Phillipines where 60 hz is the standard.
 
6224 can have 9, 10 or 11 balls. I don't see any connection, either.
 
My guess is that the rotor work that the OP suggested to the client should have been performed.
 
A better look at the spectrum from an analyzer (if you have access) might reveal some sidebands around the big peak that could point you in the direction of the cause. Can the IRD Balancing machine be zoomed in on the area of concern and have more lor applied?
 
 
[190759%2F9462383%2Fmceclip0]
Reply 1 0
electricpete Supporter, Moderator 1,770 posts
electricpete
Thanks for the bearing frequencies. Nothing jumps out at first glance.
 
Yes, for a large motor anything more than 1rpm slip at no-load catches my attention, but there are a variety of other factors mentioned in terms of frequency resolution, line frequency etc.   In my part of the US it would be rare to see grid frequency at or below 59.9hz for anything more than a brief infrequent transient, but there may be more variability in other countries now that you mention Phillipines. In any event, whichever of the questions can be readily answered might help narrow things down (I tried to prioritize my many questions within each category so the ones I considered the most relevant are near the beginning of the list and the longshots are last)
Reply 1 0
HuskerTim Sr. Member 169 posts
HuskerTim
ps - I would love to hear more about the rotor and any photos you can share. It seems unusual that a growler test would not detect broken bars, but maybe there are some lessons learned there in technique somewhere.
 
Could depend on the growler and the process used, one needs a growler big enough to show both continuity and discontinuity of individual bars...Not many rotor designs allow removing the bars radially but they do exist (inserted axially down the slot length or aluminum cast, typically) this could be a clue on why the failure occurred, combining loose bars with serial starts. Pic of the rotor would be helpful if available.
Reply 1 0
Danny Harvey Admin / Sr. Member / Supporter 2,720 posts
Danny Harvey
I'd like to see all that happen. I've never witnessed detailed testing like you guys are describing and I think it would really be helpful to the vibration analyst to have seen first hand what is involved beyond pole pass frequency sidebands around harmonics of running speed.
Reply 0 0
Joshua1981 Member 32 posts
Joshua1981
Food for thought, it is less than 1% off from the 2nd harmonic of BPFO. Deep groove ball bearing unloaded, might be a slight deformation in the newly sleeved bore or manufacture defect in the bearing. a harmonic cursor would be nice on that first little peak.
Reply 0 0
BMag34 Member 3 posts
BMag34
At first glance I would have guessed 50hz supply frequency, in which case the 25.0 hz spacing between 222.5 and 227.5hz peaks would be 1.00 x running speed. I guess that's just a coincidence (and supply frequency is really 60hz)....
 
 
If your supply frequency is 60hz, and sync speed of 1800rpm, how the heck do you end up with 1793 at no-load conditions!!?!! That could be a huge red flag of severe rotor problem... bad enough to degrade the torque capability so much that friction losses slow it by 7rpm !?! (which is the slip you wouldn't expect until the motor is loaded to 7/19 fraction of full load). But before jumping to that extreme conclusion, there are some other things to consider first
  • How was the machine powered during this no-load test? (motor generator, autotransformer, wound rotor motor used as autotransformer, electronic supply, other?)
  • Do you have an accurate estimate of the supply frequency to the motor (how close to 60hz?). If you don't have a record then look at the 2*LF vibration peak on log scale and tell us what is the frequency (should be 120.0hz, but 119.7hz would explain the 1793 noload rpm).
  • How accurately are the 1x peak and the 2*LF peaks measured in vibration?
    • what is the frequency bin width
    • Are you using frequency interpolation to estimate the peak frequency more accurately than bin center? (in Emonitor and AMS this is invoked with the peak label feature)
    • Ideally verify speed using a strobe
  • Assuming the actual speed is confirmed 1793 with supply frequency near 60.0hz, are there any pole pass frequencies present in vibration? (would be approx 28rpm above and below 1793).
  • Verify supply voltage magnitude during the test. If this was a reduced voltage test, what was the actual voltage during the no-load test?
  • Verify supply voltage balance (or current balance as a proxy... typically 5-10x more current unbalance than voltage unbalance at no load)
  • Has this motor been load tested since the rotor repair? If a rotor problem existed which was bad enough to cause that much slip at no load with balanced rated supply voltage, then that motor would definitely fail the load test miserably (and rotor problems would show as pole pass anomalies)
  • Have you done a single phase test for rotor condition since rotor repair? That is a test of rotor condition which can be done without needing to load the motor or disassemble it.
  • Do you know if the alloy of replacement bars was matched to the original rotor bars? (If I were the customer I would ask for documentation to prove the resistivity matched)
  • Is it practical to check airgap and axial alignment of iron and inspect for internal severe rubs? (reaching at straws for remotely possible causes other than rotor for severe torque capability degradation... IF that's what we're seeing)
Stepping back from the rotor to your original question of identifying these peaks.
  • Do you know how many rotor bars and stator slots?
  • I don't currently have access to my software... what are the fault frequencies for 6224 in orders?
  • If there is an electronic supply involved, please provide more details including the supply frequency at the input to the electronic supply
  • Maybe post a log scale spectrum with the highest of the peaks labeled to see if you can fit these two peaks into part of a larger pattern (the more peaks you can fit into a pattern, the more distinctive the pattern becomes imo)
 
ps - I would love to hear more about the rotor and any photos you can share. It seems unusual that a growler test would not detect broken bars, but maybe there are some lessons learned there in technique somewhere.
  • The motor was run with a generator setup with softstart at 1.9kV switched to 3.8kV after it reached rated RPM
  • The generator supplies 60hZ plus or minus 0.5Hz if it drops below or speeds up more than its rated speed, the generator will trip and will display a code corresponding to that failure. Also, other motors did run at their synchronous speeds with the same test setup, which do make me realize it was quite unusual for this motor to exhibit a significant slip in its speed without load.
  • I'm not really sure how to answer this question 😅, I'm not that familiar with the terms but if you're asking about the setup, I set the analyzer with the following parameters.
    • Lines: 400
    • Num. Averages: 5
    • Avg Type: Exponential
    • Detection: RMS
    • *We measured the motor speed with the analyzer as well as with a separate laser tachometer and both displayed the same 1793rpm speed.
  • No there was no present, (I will attach spectrum of most recent no load test run)
  • No i did not see any pole pass frequencies in the spectrum
  • Yes, this is verified, also multiple no load test runs were conducted weekly with the same test setup, and none exhibited the same anomaly which led me to eliminate the possibility of the axial vibration coming from the test setup.
  • Yes, the second run, the slip actually increased to 1781rpm which strongly suggests rotor problems. Please see additional spectrums as well as time waveform.
  • No, we did not do a single-phase test but I believe surge test of the motor will fail if there is a significant rotor defect. Since the inductance of the rotor will affect the surge waveform of the stator, any major defect will greatly affect the ringing of the waveform. But I will still perform the single phase test. this might prove to be useful when trouble shooting SCIMs.
  • We did measure the resistivity of the copper bars that was removed and it matched the resistivity of the bars that we replaced however. I believe we should've done x ray fluorescence spectroscopy to have a more comprehensive examination of the alloy, but we do not have the equipment for that test. This is a room for improvement on our side in future projects but for now, I acknowledge that we did not have a proper evaluation of the alloy upon replacing it.
  • We tried checking the airgap however there are no provisions to see between the rotor and stator since there is an air baffle guarding the stator coil overhang. This also prevented us from inspecting any hotspots on the rotor after the no-load test run.
Stepping back from the rotor to your original question of identifying these peaks.
  • 58 rotor bars and 48 stator slots
  • SKF 6224 bearing fault frequencies (i do not have a software, thank you to Mr. Danny Harvey for providing the fault frequencies)
    • FTF: 0.410x
    • BSF: 2.688x
    • BPFO: 3.690x
    • BPFI: 5.310x
  • No electronic supply
  • Is log scale spectrum same as the waterfall spectrum, if so, my equipment does not have that function but I will keep this in mind and try to compare spectrums side by side to see if there are any larger pattern.
Thank you for your response, this is a really great help for me. Some of the photos did not upload from my original post but please see the attachment here for reference photos of pre-existing rotor defects as well as the spectrum from our most recent no load test runpdfviber_image_2025-09-06_13-15-06-778.pdf. viber_image_2025-07-08_14-19-25-517.jpgviber_image_2025-07-08_16-07-30-418.jpg
Reply 0 0
BMag34 Member 3 posts
BMag34
This is a photo during the growler test. The rotor bars were indeed removed and inserted axially from the rotor core. With regards to any looseness, we re-swaged all of the rotor bars (including the ones that was not replaced) just to make sure they are secured in place.  [190759%2F9465263%2Fmceclip0]
 
Could depend on the growler and the process used, one needs a growler big enough to show both continuity and discontinuity of individual bars...Not many rotor designs allow removing the bars radially but they do exist (inserted axially down the slot length or aluminum cast, typically) this could be a clue on why the failure occurred, combining loose bars with serial starts. Pic of the rotor would be helpful if available.
Reply 0 0
Reply