I am dealing with a medium sized direct coupled overhung fan unit.  85KW motor.

The unit is roughly 7 years old (PDM program is new here, I have 6 years experience).  Unit was checked for the first time on 3/30/2017 with an overall amplitude of 0.08 in/sec (0-peak) at 1200 RPM on the Motor non drive end horizontal.

Unit checked again on 5/12/2017 had an amplitude of 1.14 in/sec  (0-peak) at 1500 RPM on the motor non drive end horizontal. Velocity spectrum showed a very high 1X RPM peak.  A natural resonant frequency problem was suspected.

Today I was able to run a Coast down peak and phase test with my CSI 2140 on the unit.  The results are:

The orbital plot is very odd, and the phase changes aren't as apparent as what I've normally seen. 

This is the same test run on the fan unit directly next to the one in question.  The orbital plot looks normal, and the phase shifts are very definite.

I am now questioning my original diagnosis of the unit.  Any help and further insight into possible problems is greatly appreciated!





Images (2)
Original Post

You indicate that the fan with the high vibration over 1.14 in./s at 1500 RPM has an unusual orbital plot. The full-scale amplitude on the plot is about .26 in./s, so either you're not showing the correct frequency (1X SS) for amplitude/phase versus Speed, or there some other issue.


The orbit plot was taken during the coast down test.  The fan drops speed very rapidly.  So I believe with the rapid drop in speed there isn't enough time at the critical speed for the vibration to get up to 1.14 in/sec.

This was taken during the last route on 5/12/2017. 

Only the motor DS and NDS Horizontal readings are high, both showing a high 1XTS.  Also the amplitude is doubled at the NDS compared to DS.  This information (in my opinion) rules out Fan unbalance, and any coupling problems.  A resonance is the only thing I can come up with, but the coast down and orbit plots just don't seem right.


Anyone have any experience with worn-out spring isolators?  This could be a cause of the odd plots.



Images (1)

.08 ips seems about normal to me.   1.14 ips is VERY HIGH and it should be obvious even to the touch.  Can I assume this is controlled by VFD since the only difference you indicate is going from 1200 RPM to 1500 RPM?

That orbit sure looks odd. Have you done those before?  Was it setup correctly? You may have tried to take too high a resolution or too many averages, especially if this motor coasts down rather quickly.

If so, why dont you simply take some steady-state spectrum plots at say 1200 RPM, 1300 RPM and then at 1500 RPM and look to see if the amplitude takes off? That would be a good indicator that you may have a resonance.  If that turns out to be the case, you might look to take a simple bump test to confirm.  Seems a bit odd only showing up at the non-drive end though. You don't mention how its mounted but If its on a skid, you might look to see if the structure is weak at the free-end location (that was an issue I ran into some years ago).


Jim P


Got this reply through email, but doesn't show up on the page.  Possibly a glitch?

Yes, the unit is run off a VFD.  The fan will run at a steady speed during production.  The speed is only changed with product and batch changes.  I've only been working with the 2140 for 2 months.  I was a Pruftechnik guy for 6 years at a previous job.  So it is possible that I did something wrong.  However on the second fan the orbit plot came out normal.  They are almost identical fans, and were setup the same way.

The entire fan unit is one "skid" mounted on 8 spring isolators.  I can run further tests on our next down day.  Not sure what you mean by a steady state spectrum plot?




Trying to get a good orbit during fast coast down is difficult; you need to have low # lines of resolution with just one average.

Since your original post stated difference between 1200 RPM and 1500 RPM, is it safe to assume this is on a VFD? If so, why dont you just simply take spectral data at say 1200/1300/1500 RPMs and compare horiz and vertical (vs. coast down plots)?

I also suggest you take a simple bump test (e.g. analyzer on peak hold) and look at differences in response at both ends of motor.

I had skid-mounted motor pump assy where my non-drive end had high vibrations in horiz direction only; very similar to yours. A bump test revealed the resonance (root cause) at the end of the skid. This was a good case for doing an ODS (Operational Deflection Shape) which showed the rear end wagging like a dog's tail. 

Looking more closely, the base support frame was just too flimsy/flexible in that direction. We stiffened the structure and the problem was solved. Don't know if this might be an issue with yours, you might want to send a picture.


Jim P

p.s. Your plots are hard to read.....can you make them a little more clear? 

New comment --> Skid on spring isolators?  Need to make sure they are properly sized for the range of speeds and weight of your machine.    

I apologize for the pictures.  If I click on the photo attachment at the end of the posts they look clearer.  Not sure if you tried that?

Would like to post pictures of the unit, however this is highly against company policy.  I've only been at this company for 2 months, but from what I am told these fan were originally installed with 65kw motors.  During a speed upgrade (to get more m/min out of the line) done by the OEM, a 85KW motor was placed on the existing base.  I'm assuming the fan wheel was changed as well, but have not confirmed this.

Out of 9 blowers on this line, there is one other that is identical (upgrades and all).  Both run same speeds.  The second unit has had low amplitudes.

While recording constant velocity readings:  The unit shows steady amplitudes over 1.00 in/sec.  If I (roughly 250 lbs) stand on one of the NDS corners of the skid, the amplitude drops to 0.30 in/sec range.

Still not sure if this is a worn isolator problem or resonance problem.

I will check for cracks in the structure, and any signs of wear in the springs.  I can not vary the speed of the unit until our next down time.  At that time I will record spectral data a different speeds.

Did I miss orbits?  I saw some polar plots, were there orbits, too?

Fast coast downs may prevent acquiring good data for polar or Bode plots.  With a VFD, you may be able to obtain excellent data.  

It may be worth determining how the isolators are helping or hurting.

Polar/Orbit = Terminology issue!

If the new motor is heavier than the OEM motor, then the isolators may be close to or actually "bottoming out" which changes the mounting natural frequencies. There are 6 natural frequencies for a rigid mass on flexible/isolation mounts. Perhaps the motor upgrade is indirectly causing/contributing to the vibrations.


The phase portion of the Bode plot is quite erratic because it passes through "zero".  One can see a rapid downward direction toward "zero" every time it gets to 360 deg, the maximum on the vertical scale.  I'm not that familiar with CSI plot formats, but that same problem may be causing the spiral effect to the polar diagram, what the OP is calling an orbit.

I apologize for the terminology error, and appreciate all of the responses.  I was unsure of how accurate a coast down peak and phase test would be if the VFD slows the unit down, rather than coasting.  If this will give me better results I might have a window to perform the test next week.

For now I am speaking with the process engineers to possibly have these resonant speeds locked out in the PLC.

The spring isolators still have a lot of travel to them, so I don't believe they are bottomed out, but could possibly be over loaded?  Would the change in motor size lead to a change in natural frequency of the entire skid?  If a frequency lock-out in the PLC is not granted, adding structural bracing to the unit is an option.  Would the addition or more, or stiffer spring isolators also be a solution?



Quotes: "For now I am speaking with the process engineers to possibly have these resonant speeds locked out in the PLC."    AND   "adding structural bracing to the unit is an option.  Would the addition or more, or stiffer spring isolators also be a solution?"

This whole thread discussion was to confirm the root-cause problem. Although resonance is suspected, it is not clear (to me anyway) that is the case. So going off making these changes seems premature. 

1. Since you can adjust speed via VFD, my suggestion is to simply adjust speeds to 1200/1300 and 1500 RPM and take simple spectral data at both ends of motor in hor and vert directions. In fact, you may want to acquire data on the spring supported skid.

2. The fact that you indicate ONLY the NDE motor is running high would suggest that the spring isolators may not be the root cause of the problem. If they were, why wouldnt all the points on the machine be high (unless there is some twisting mode perhaps)?

3. A bump test of the machine (while idle) can help support findings and suspected causes. My previous comment about stiffening the rear end of the skid structure was only after I was confident that was the root-cause problem of the machine I was working on. I don't know yet what your specific problem is.....so lets not go off stiffening anything till we know for sure

Does this help clarify?




As others suspect, I agree a possible isolation concern. One method that I have used in the past is a couple of wooden wedges, large enough to "Lock Out" the springs. Using the analyzer in monitor mode, place the sensor on you motor point and push the wedge in to remove the spring at one location at a time, going around the non-drive end of the machine. This has shown where the weak or incorrectly adjusted isolator was. I have had several large air handlers in the past have original heavy cast motor frames replaced with sudden changes in vibration levels also. So even more likely that these may be part of the problem. 

The next time the unit is running at 1500 RPM (the problem speed) I will try the wooden wedges.  I took readings on the unit again today as it was running at a different speed.

The unit exhibits the same characteristics.  The highest point of vibration is Motor ODE Horizontal.  Next is Motor DE Horizontal.  The amplitudes have dropped significantly with a delta of 100 RPM.  The spectrum still shows only a high 1X peak.

During the next downtime I will conduct a second coast down test.  This time using the VFD to gradually slow the unit down to confirm a resonance problem.  I will also do a bump test at that time to confirm.

I have been looking at requesting the CSI 3lb modal hammer.  Anyone have good/bad experience with it?

I will update this post when I get more data collected.  THANK YOU EVERYONE FOR THE REPLIES!


Images (2)

Add Reply

Link copied to your clipboard.