PA Fan peculiar vib problem


I am attending on a vibration problem in a PA fan-1 (Primary air fan). I have attached the report and vib signatures for your reference. Motor rating is 870 KW. Max rated speed of the machine 1480 rpm and it’s a variable speed drive. Depending on the load the fan speed varies from 500 to1480 rpm.

The problem is there is steep increase in vibration levels when the machine reaches 1000 rpm on Fan DE horizontal. A predominant 33.4 Hz (2XRPM) was seen in the spectrum with amplitude of 15 mm/sec. When the speed goes to 1100 rpm again there is a decrease in vibs (5.2 mm/sec).

I can see 2X even at 800, 900 and 1100 rpm. But what is causing only 2X to excite at 1000 rpm is what I am just wondering…???. I did the bump test on fan shaft, coupling cover, fan blade, fan and found all frequencies are well above 33.4 Hz. Not able to do bump test on coupling internals and also run the motor in solo condition because of production pressure

It was observed that the vibrations on the motor cooler (Heat exchanger) just above the motor was also vibrating at 33.4 Hz predominant frequency at 1000 rpm with amplitudes reaching 77 mm/sec. So I took the cooler out and made measurements on the Fan DE. Again the same vibs are repeating. So I concluded that there is no influence of the cooler on Fan DE.

Phase measurements indicate the following at 1000 rpm,

Fan NDE: H-0.42/118 deg, V-0.068/146 deg, A-0.24/185 deg
Fan DE: H-1.05/302 deg, V-0.36/220 deg, A-0.18/55 deg
Motor DE: H-0.22/84 deg, A-0.10/255 deg

There is a phase difference of around 200 deg between Fan DE and motor DE in axial direction. But I was thinking how the coupling becomes unstable only at 1000 rpm and not at higher speeds. Infact the theory says the machine fault amplitude increases with increase in speed whether it is unbalance or alignment problems.

I made one more measurements at 1000 rpm. Here at 1000 rpm I slowly loosened both the bolts of top half cover of the Fan DE plummer block. The vibs dropped from 14.5 to 12.2 mm/sec (at 33.4 Hz). When I put my hands on Fan DE I felt some impacting inside bearing. But the bearing is in good condition and no bearing fault frequencies seen. You can see the time waveform plot given in the last page of the report.

There are no such high vibs on PA fan-2 which is running in good condition at all speeds. (Readings and signatures attached). PA Fan-2 is running next to PA fan-1

The snap of the spare coupling being used to connect motor and fan assembly is shown in the report. It is 2 halves of the coupling connected by 2 springs. Are these 2 springs trying to distort or making the coupling assembly unstable at 1000 rpm which shows predominant high 2X on Fan DE and Fan NDE and also low amplitudes (again 2X) on motor DE
and motor NDE. Your inputs will help me...???

But one question still remains... why only at 1000 rpm...??, why not at higher speeds…???



Original Post
I don't know that I would worry so much about why the big increase at 1100rpm (likely a resonant frequency) I would fix the 2x exciter, which with the information given, does look to me to be a coupling issue.
I hope that's not too simplistic an answer!
You have the answer, resonance. Additional data like start up coast down or just running from 500 rpm to 1480 rpm will also be useful.

You should have a distinct rise in amplitude AND 180 degree phase shift if there is truly a critical/natural frequency present. I assume the vibration levels you show in your report are at 33.4hz frequency? What type of fan bearings?

As you stated at 1000 rpm is the worst but yet your vibration is at 2004 rpm/33.4hz? So imbalance does not appear to be the issue.
Your vibration data shows the horizontal direction is always the highest no matter what the shaft speed. The ratio from H to V is over 5:1, this too indicates resonance/critical speed.

Keep in mind the frequencies you show in your bump test are steady and there all the time. As the speed of the machine increases so does 2x, 3x, 4x etc...

If you have double row roller bearings then something is loose. Bearing base plate, bearing feet, bearing cap, bearing inner race, bearing outer race, something is loose. "Fig showing frequency spectrum recorded during bump test on fan shaft near Fan DE bearing in horizontal direction", did you have sensor mounted on the bearing or the shaft when you did the bump test?
Thanks Dave,

Coast up data recorded from Fan DE horizontal is attached which shows a linear curve.

Also as seen from the spectrum only 2X freq component is high from 900-1100 rpm while 1X is low.

The bump test was done on the coupling cover and not on the coupling internals. I did not get a chance to do the test on the coupling internals due to production pressure. But with a simillar type of coupling PA Fan-2 is running well. Both PA Fan-1 and 2 were commissioned during the same time supplied by same manufacturer.

The fan bearings are single row roller bearing. I did the bump test on the bearing and found nothing. I also did the bump test on the fan shaft and results given in my earlier attachment.

One observation was the vibrations (2X) is high on Fan DE compared to Fan NDE which is on the other side of the fan. The coupling/fan assembly is closer to Fan DE than Fan NDE.

As you had suggested there is no unbalance in the system.

If we assume that there is a resonant frequency at 33.4 Hz on one of the internal components of the coupling will it be so sharp (narrow band frequency)...?? because for an increase/decrease of 100 rpm (1.66 Hz) the vibs are dropping by 3 times...!!!

The problem here is no increase in 1X but 2X at 1000 rpm. This is what is confusing...!!! Since the coupling has 2 springs which will bring the 2 halves together i am suspecting something is happening at 1000 rpm... just guessing...??



Was the coast up data taken in horiz, vert or axial direction? Have you bump tested each axis on the motor and the fan bearings?
You have a "similar" fan that does not show this issue so the coupling should not be the issue.
The coast up data shows a rise in vibration at 1450 rpm is there also a 180 degree phase shift? In your previous data that you posted, the waterfall plot at different speeds, shows mechanical looseness not matter what speed you operate the fan, with the highest vibration at full speed. Looseness will shift the natural frequency lower, the high 2x is most likely due to something moving while the fan is in operation. If the alignment has been checked and it is within in tolerance and there is a 180 degree phase shift across the coupling in either the horiz or vert directions then something is moving while the fan is running. The impacting you see in the time waveform is due to looseness. Take a long bar, block of wood and a dial indicator, put dial indicator at top dead center and "lift" the shaft. If all components are tight you should only see maybe .005" of movement. How about a waterfall plot with motor and fan readings displayed with the fan under normal operation? Does the motor and fan NDE bearing also show looseness? How are the fan bearings mounted? Concrete, steel plate, bearing housing, etc...
The "Bump" test was not very good, since the problem frequency was at 1000 cpm and the F-max was 30,000 cpm. The vib-vs-speed data indicates a resonance, but the bump test method is not confirming it. Improve the bump test by selecting better locations, lower F-max, and heavier hammer with soft rubber impact tip.


Add Reply

Likes (0)