We had a pump shaft failure over the weekend and I am trying to find the cause of the failure. The pump/motor base is mounted on isolators. I do not know the proper procedure for adjusting these isolators but I do not believe that they should be backed off all the way as they are in this setup. The motor end is vibrating about .4 ips in the vertical direction. I suspect that the base may be flexing between the motor and the pump. The main vibration is around 6.8Hz and turning speed is 14.88Hz, any Idea's what might cause the 6.8Hz. Pump has been in service three months.
Thanks for your help,
Scott

shaft_failure_on__1_ehsc_pump.doc (386 Kb, 165 downloads)

What is the phase relation between the motor and the pump and the base?

The break looks "twisted" in the picture, it is not allowed to be free wheeling backwards when the power is applied is it? Surely not. Does it have a check valve to prevent this?

Scott,

What a strange vibration mount setup! Based on the full side view photo, it looks like the far left mount under the pump is very deflected and possibly bottomed out. The 6.8 Hz is probably the vertical natural frequency. If spring coils are hitting, then ocasional impacts could excite other natural frequencies. The base frame looks to be rather stiff, so I doubt that it is bending at 6.8 Hz, but I still would measure for operating deflection shape. There usually is not much excitation at 45% of shaft speed for a centrifugal pump, but possibly inlet recirculation flow turbulence or a rotor rub at 0.5xSS.

Shaft breakage can be from excessive torque during startup from:
a) Reverse flow rotation (already mentioned)
b) Restart before rotor comes to rest
c) Hard start (slurry material?) with full line current (no motor soft start?)

The pump may not be demonstrating noticeable operational or vibration symptoms when you are present.

Walt

Just some thoughts.

Slightly less than 0.5x.... could be a sybsyncronous rotodynamic instability? For example oil whip?

Does the pump have sleeve bearings? Does that frequency show up on the pump at all?

To investigate, do a coast-down test and check for high vib at 6.8 hz would tend to support a rotodynamic instability.

Possibly the strange base conditions (I agree you'd expect those top isolator nuts to be tightned) may contribute to dropping the rotor resonant frequency below 0.5x where it can become an instability problem.

Non-syncronous whirl of the shaft can be much more destructive than syncronous whirl. For syncronous whirl, the shaft would generally flex a certain amount and then hold that shape while rotating without accumulating fatigue cycles. For non-syncronous whirl, shaft flex changes continously and fatigue cycles accumulate continuously. Although in this case the failure point is at the edge of the coupling? Seems like there would to have been a lot of bending moment transmitted through that coupling (as if the pump shaft were whirling in cylindrical shape and motor was restraining it). What type of coupling? How was the coupling fit on the shaft?

What about that rust on the shaft? Was that ther before or only accumulated since the failure? Certainly surface imperfections and corrosion can contribute to failure.

For any failure at the keyway, you also want to look to make sure you had proper key fit and proper radiusing. Hard to tell but there doesn't look like much radiusuig. Also I notice the end of the keyway is rounded, but not sled-shaped.

The general appearance of the fracture appears somewhat similar to photo's identified as failure from "shaft bending" in a document that I have (EASA motor failure root cause analysis).

Also it always helps if you can have a metalurgist look closely at the failed shaft.

TIOMOAICBTWA

This message has been edited. Last edited by: electricpete,

I looked at the side view picture of the machine again and I can't figure out exactly what I'm looking at. Is the inboard pump bearing located in that blue housing? If so there doesn't look to be much support to that bearing except for that slim blue vertical leg. It doesn' look much like anything I have seen. The closest thing I have seen is on smaller overhung pumps with a "wobble leg" to be loosened after maintenance but before startup and tightened at full temperature supposedly to allow the bearing housing to move with the pump casing as the casing thermally grows. Is this an overhung pump (doesn't look like it) or between bearings? High temperature fluid? (doesn't look like it... no insulation)

How many horsepower? What diameter shaft? What bearing configuration on the pump?

Is there a single base for the pump/motor or some kind of seam right next to the pump? Seems like it would be worth it to measure vib on the base and see how high it is and if there is looseness at that seam. And do a bump test to check for resonance as suspected by Walt.

This message has been edited. Last edited by: electricpete,

Yes! What a strange set-up! With all the hold-down bolts slack, it looks as though the pump in "hanging" on its piping and therefore able to move any way a force takes it. The pump inboard bearing pedestal also looks very flimsy! The fracture would tend to support this with what looks like an old, quite deep, surface crack (rusty area) penetrated into the shaft.
The whole unit needs to be re-aligned, pump to piping and then motor to pump, making sure that the hold-down system is properly supporting the unit.
Are the pipes on "hangers" or only supported by the pump unit supports (I can't believe that)? If hangers then these must also be checked to make sure they are doing their job correctly as well as the alignment.

Scott,

In your photos of the springs, are you showing us one pump that has loose springs (broken shaft pump) and one which is not, because the photos appear to be from 2 different pumps. I first thought they were of the same pump. Does this clearance at the springs disappear when the pump gets hot while running?
Just wondering.

I cover one pump with this type of set-up. It never creates any problems, so I have never really looked at it that closely. It has a common base for the motor and pump with a solid bottom that is filled with grout, I assume to add mass to lower the natural frequency. Out of curiosity, I have monitored the spectrum and added my mass (about 240 lb counting the 50 lb data collector ) to the system and never noticed any change.

Looking at the shaft break, it occurrs at the weakest point from the bearing end, so I would suspect (uneducated guess) that movement in the pump rather than the coupling or the motor was the root cause. I would guess that looseness in the spring mounts progressed until 1/2 x running speed starting showing up. That excited the spring natural frequency, amplified the vibration and finished the shaft off. Somewhere along the way, the shaft cracked in the area where the rust is shown. It looks like there is Corrosion on the radial surface of the shaft-like maybe something sticky and corrosive was stuck to the shaft. There also apprears to be some surface damage in that same area. I would guess that the crack propagated until the shaft could not longer handle the torque.

If the springs are really being used as shown then I think something is very wrong. The spring natural frequency should be available from the manufacturer as should instructions on how to set this up porperly. I'm editing this to add that every time I have seen any type of vibration isolation device over-compressed, there have been major vibration problems.

You can measure the wire diameter, coil diameter, length and a few other things and estimate the natural frequency. The formulas are probably in the Machinery's Handbook. If they aren't, I worked with Schindler Spring Co. in New Orleans once and they were very helpful. Of course, I bought thousands of dollars worth of springs from them and that tends to make small companies very helpful.

Good Luck,

Scott,
This appears to be a large Gould tower water pump. Am I correct? If so back flow is not likely as it appears to located next to the tower.
The gap above the springs on the discharge side of the pump is likely from the failure of the lower springs. This is the expected direction of torque due to the location of the pump discharge and with the weight and torque of operation, the springs may not have been properly sized or may have simply "given-up" with time.
It appears by the fresher paint on the pump cartridge that the pump may have been rebuilt within the past year or two. Is that correct? (3 Months in service?) If so was anything noticed during repair. Was any shaft damaged induced during repair? Also the black tar like substance covering the pump volute appears to be somewhat corrosive. Could that have damaged the shaft?
Does the pump still turn freely?
I would expect for the vert data to be high with this style isolation.
One other note, sometimes shafts fall due to poor manufacturing. This may not be true with your case due to the obvious void in the outter surface, but it is possible.

Good luck,
Nicholas A. Martin

Ralph, This pump runs 24/7 and does not freewheel backwards. The picture of the springs is the same piece of equip just on the backside of the pump. Not sure about the change in the spring clearence from hot to cold.
Walt, The springs under the pump end are very close to bottoming out and may be periodically impacting. I took some phase data on the base at turning speed 14.9Hz and at 7Hz where subsynchronuos vibration is at just to compare.
Tried negetive averaging to identify natural freq but results were not very helpful(will try again)
Pete, 450Hp shaft is app 2 7/16 (shaft is not at hand). The down leg support on the pump does not seem very rigid. The pump is a goulds 3180 with antifriction brgs and yes the blue housing is the rotating element.
Cheddar, Piping is on hangers but I do not believe that they are supporting as much as they should.
Danny, We are checking on the correct procedure for the springs to be adjusted and are sending the shaft to be analyzed

Thanks for all the input,
will let you know any conclusions
Scott

_1_ehsc_pump.xls (18 Kb, 23 downloads)

Hi Scott,

quote:

Ralph, This pump runs 24/7 and does not freewheel backwards. The picture of the springs is the same piece of equip just on the backside of the pump. Not sure about the change in the spring clearence from hot to cold.

Good! Thanks.

Was the pump running when the pictures were taken?

Scott,

The 7 Hz vertical vibration is in-phase and the 14.9 Hz is out-of-phase along length of machine. Neither frequency and its amplitude and phase characteristics would indicate a cause of the shaft failure. So possible causes to consider:

a) a torsional natural frequency near shaft speed or vane-pass frequency
b) a transient overload operating condition that has not been observed or measured yet
c) a significant material flaw or manufacturing defect

I would probably rule out a shaft bending natural frequency, since lateral vibrations on the bearing housing are low. Measurement of shaft speed and comparing to motor nameplate and motor current compared to nameplate full load amps could indicate actual load (torque) on shaft.
Did the shaft design (stress analysis) assume that a keyway would be machined, or that a keyless coupling hub would be used. The shaft design may be very marginal.

Walt

I am missing something. I don't understand what vibration readings are labeled.

Does A/B' represent a magnitude/phase reading (A ips at B degrees) ?
It looks like you have labeled the following readings from top to bottom:
- Bearing housing 1x vertical readings/phase
- Base 1x vertical magnitude/phase
- Base 1x opposite side vertical phase only???
- Base 0.46x vertical magnitude/phase.
Are these correct?
Are there any bearing housing 0.46x vertical magnitude/phase?
Is the base vibrating at 0.4-0.6 ips under the motor? Or am I reading this wrong?
How does the axial look? (any phase change?)

This message has been edited. Last edited by: electricpete,

Is it possible that the operating characteristics changed. Not knowing your process is a disadvantage, however, if the pump was operating out on the curve above BEP, you can experience cavitation, high flow, high head pressure, high radial loads and shaft deflection.

Something to think about.

Scott,

I asked earlier was the pump running when the photos were taken, The reason is, from the photos, one side appears to be lower (where the nuts are not touching the spring retainer) than the other side. Is it possible that the torque from pumping is pushing that side down? Is the "low" side still low when it is not running?

Good comment Ralph, I never thought about that. I agree with the possibility of torque from pumping being transmitted to the base. It would be transmitted through the motor to the base. Also if there were oscillations in the torque, it might cause that vibration (maybe this would be similar to the torsional vib suggested by Walt). Might be interesting to take a current signature and check for sidebands spaced at 6.8hz around line frequency which would signify torque oscillation at 6.8hz. I know it's a long-shot, but I have been racking my brain to thinking what else in a motor can generate 6.8hz... and I can't come up with much else.

The only other possibilities I can think of:

1 - maybe it originates in the pump and for some weird reason only shows up on the motor. I agree with the Mick's comment that off-normal flow configuration can cause subsync vib of pump (although I have only seen it on verticals). Check how the vib changes with flow.

2 - a rotodynamic instability... something like a resonance of portion of the mounted rotating system at 6.8hz which gives rise to subsyncronous instability with machine running at 14.9hz.

Sorry for not getting back sooner,
Mick, this is a area of the mill that I am not very familiar with, I will check on the operating characteristics to see if anything had changed.
Ralph, the pump was running when the picture was taken. The sister pump is showing the same looseness at the same point of the springs and also has the same subsynchronous freq. Cant find anyone that has noticed the springs when the pump is down.
Pete, sorry about the confussion with the phase readings. The high amplitude was on the 7Hz readings.
With the subsynchronous showing on both of the machines would that point to something in the process or a resonant freq?

Thanks Scott

Pete, How would I go about testing or ruling out a rotodynamic instability problem. I do not have any experience with this.
Scott

quote:

this is a area of the mill that I am not very familiar with, I will check on the operating characteristics to see if anything had changed.

Is this system in a papermill Recovery area, maybe the cyclone area?
If this is the first time a shaft has broken on this pump, there may have been some type of flaw in the shaft rather than a cause that will strike again, especially if both pumps have basically the same signal and "looseness" on the spring mounts. If this is a reoccurring problem of shaft breakage, then indepth analysis is a must.

I also don't have a lot of experience with rotordynamic instabilities either. I have read a little bit. The part that I focus on is that rotordynamic instability just below 0.5x is much more likely to occur when that frequency correspnds to a critical speed of the rotor. So I would think that if you see resonant behavior at 6.8hz machine speed during coastdown test, you would look closer at rotordynamic causes (possibly attempt to model the machine train). If you don't see resonant behavior at 6.8hz, than focus on the other causes (fluid or torque pulsation). Maybe someone else can comment further?

The other thought that I have on the subject is that the motor shouldn't have any rotordynamic instability unless you have some kind of severe looseness within the bearings (I'm assuming the motor has ball bearings). If the machine is uncoupled at some point during troubleshooting, it would be worth to check to make sure there is not unusual axial or play of the shaft extension. Normally I don't expect to see more then a few mils of movement any direction when you push and pull on the shaft of a ball bearing motor. We have seen severely worn bearing which give 20-30 mils. I did see one motor that was an exception to the rule 200hp 1800 rpm which had 50 mils and the motor OEM told us it was normal for this particular motor (although I don't know why).

This message has been edited. Last edited by: electricpete,