The pump team at our plant is working on a vibration problem. I am not involved because it is not a motor problem. But I wanted to ask you guys to help brainstorm about possible causes:

Here are the facts as they were explained to me:

Horizontal 7000 hp 3600 rpm sleeve bearing motor drives horizontal startup feedwater pump.

5 impeller blades, blade pass frequency ~ 18,000 cpm

Even though there has been no change in the range of flow conditions / fluid system lineups / temperatures and no maintenance on the pump, the pump suddenly exhibited a very large increase in blade pass frequency at the pump outboard (jumped up to around 0.8 ips). This was observed both at normal high flow and normal low flow (recirculation conditions).

Waterfalls of housing vibration readings are slides 1-11. Slide 11 shows the dramatic increase at 4V position. Slide 12 is shaft rider data expressed in ips (relatively high around 0.45 ips at 45 degree from vertical and 0.62 ips at vertical top dead center.) At this frequency (~ 18,000 cpm), 1 ips pk/0 ~ 1 mil pk/pk. Slide 13 is a bump test on the casing.

Pump pressure/flow performance has not degraded comparing the recent performance to the historical performance.

Any suggestions on what could cause a sudden increase in blade pass frequency (not associated with a change in flow)?

U2SUFP_BPF_VIBSmall.ppt (1,057 Kb, 144 downloads)

Pete,
Take dynamic pressure readings from the discharge side of the pump using a dynamic pressure transducer. This will confirm if the source of energy is coming from impeller vane interaction with the volute cutwater. If it is vane pass, you should pick up vibration in the discharge piping.
Not likely, but I have seen recirculation at vane pass frequency on light hydrocarbon pumps.

Is it possible the rotor has shifted axially?

I would check the discharge pipe for obstruction.
Is there a check valve in the discharge pipe? How is the operation of this check valve?
Does it open freely?

everytime I have seen this it has been a system problem. Something in line-up or a flow restriction, such as a broken check valve. It was never the pump.

Vib Institute book explains this as a result of shifted pump casing.

I can't explain it, but I once saw similar response to a plugged suction line.

Pete,

I have seen this before. As others have stated pump center changes will cause this. But I have also seen it when the recirc valve on the pump was damaged and leaking through. If the rotor has shifted axialy as Rusty says you have bigger issues than an increase in vane pass. We have had problems on Byron Jackson pumps where the thrust collar locking nut had came loose. It fretted the threads causing the pump to shift. The shaft had to be reworked because of the thread damage as well as impellers and volutes.

Pump will cavitate if it does not get enough water because of plugged suction pipe. I do not think cut water to be an issue here because this pump was running fine for past several years.

Was any repairs done on this pump recently? this might have resulted in elevated vibration levels.

Riyaz,
I mentioned the cutwater vibration as a symptom to a change and not necessarily the cause. Once you identify the source of vibration, one can start to eliminate or add other possibilities.

Erik,

I am planning on taking Cat 3 test this year. Is it tough?

El'Pete,

I have been working on nuclear feedwater pumps with high vane-pass vibrations. The historical monthly survey data (by plant) had significant variation in amplitude that was not correlated to anything. The shaft-drive lube oil pump at the outboard bearing was resonant at 7xSS (3-stage pumps with 5 and 7 vanes) in vertical direction. Another issue is measurement consistency, since no measurement point disks were used and points were poorly marked. A 2-pole magent base was attached to the rough curved casting surface. Different folks could get different measurements. as far as second hand information, just consider the source and its accuracy.

Walt

If valve is leaking or damaged,is that not possible to identify from the Discharge pressure gauge reading trend?.

Is that problem can be due to wear ring clearance/damage

E Pete,
The data definitely indicates a resonant condition at the base of the 5x peak. Change your Y axis scale to log and look at your historical data, one at a time (don't use log in the waterfall, it looks awful).
You may see the resonance more pronounced in your later plots and this could be related to a standing wave issue in the piping.

I appreciate the help. Those are a lot of good suggestions and I have been passing them on.

I tried Ron's suggestion to view the historical data on log scale. I don't particularly see evidence that the noise floor in the area of BPF is magnified as might be associated with resonance (what do you guys think?). But I did see something else interesting. At the time when the blade pass frequency jumped, the harmonics of blade pass frequency did not. (See attached slides in reverse chronological order) This would seem to suggest a change in resonant amplification at that frequency, rather than change in forcing function, wouldn't it?

There is quite a lot of variability in the historical spectra and I'm sure flow conditions have a big effect (I don't have flow data readily available to add to these slides). I am fairly confident that the historical spectra present a range of flow conditions (as do the recent spectra where BPF has jumped).

u2sufpbpfandharmonicsreversechronsmall.ppt (1,222 Kb, 47 downloads)

quote:

This would seem to suggest a change in resonant amplification at that frequency, rather than change in forcing function, wouldn't it?

Pete,
It all depends on the linearity of the system. You can have a forcing function such as a 2XLF vibration from a distorted motor frame and not have harmonics if the motor has no looseness.

Pumps with a resonance at vane pass typicaly vibrate at a constant amplitude. Your slides do show small harmonics with sidebands of running speed. Vanepass frequency changes amplitude over the years most likely from process changes. Taking dynamic pressure readings will varify the hydraulic pulsations at least identifying the source.

Thanks Erik.

I am trying to figure out what it would gain us to do that pressure monitoring.

I guess if we don't see abnormally high blade pass pressure pulsation on the pump, then we look for mechanical resonances.

If we do see abnormally high pressure pulsations on this pump (compared to the others), then I guess we rule out mechanical resonance and look at all the other things listed above? (within the pump or fluid system).

El'Pete,

Sidebands of 1xSS around the vane-pass frequency could indicate eccentric motion of the impeller possibly caused by unbalance or increase in wear ring clearance.

Measurement of pulsation is not needed to identify the source, since it is the pump. There are few if any severity guidelines for pulsation amplitude. If a standing wave were present, then the amplitude would vary along the length of pipe. One would have to measure the same exact location on two pumps if a true comparison was needed. Mounting a dynamic pressure transducer in existing pressure taps can be a challenge, since the ideal location is to have the diaphram flush with the pipe inside wall. I have done it one on a 10,000 hp nuclear feedwater pump.

Walt

Walt,
There is little doubt that the pump impeller is the source of vibration from the data presented. You also mention that measurement of pulsation is not required because it's the pump. But how does one know if it's a pump structure resonance, impeller (vane pass), acoustic resonance etc? In this case, one can say it's most likely vane pass but to be absolutely sure more data needs to be taken such as dynamic pressure readings or structure impacting. You also mention a standing wave. A standing wave is related to an acoustic resonance, which does vary in amplitude depending on location of the pipe. Vane pass is not a typical standing wave but rather a pulsation (forcing function) created from the impeller vanes. These pulsations are not a function of pipe length so you will actualy see a larger pressure wave closer to the source. Also, you are finding the source of the energy and not necessasy the severity, but of course if the valve handles are falling off and tubing is breaking you may want to reconsider.

Erik,

It is typically easier to confirm/eliminate a resonant structure than making pulsation measurements. I certainly wouldn't discourage you from making pulsation measurements.

"These pulsations are not a function of pipe length so you will actualy see a larger pressure wave closer to the source." Have you ever made measurements that would confirm this? We are not speaking with the same terminology definitions, but that is OK.

Walt