∆P across the pump at constant speed is fixed based on fluid density. Any reduction in inlet conditions are going to reduce density and mass flow rate which will reduce pump power therefore amps.

Pumping power is calculated as the volume of the fluid per unit time (flow capacity) times the density of the fluid times the gravitational constant times the pumping head (vertical distance to be pumped). Anything that reduces volume pumped will result in lower amps.

"Speed began to decrease"??

How? A fixed speed motor is exactly that - fixed within a few rpm and if you noticed that happening then there is something seriously amiss, but I guess you know that.

On the sudden suction loss - kind of depends what type of pump. Some pumps increase power as the flow goes down - mainly axial or mixed flow so you really need to look at or post the pump curve.

If it's a "standard" centrifugal then yes, less flow normally means less shaft power. You need to add in pump efficiency to bimrs list of things as it is shaft power you're interested in.

But sounds much more "mechanical" than loss of suction.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

Yes, it was a significant reduction of RPM. We have prox probes and keyphasors on the high speed side which record speed. We can use the gear ratio to compute motor speed from that. The motor nameplate speed is 3551rpm (it's an old motor, these days most large motors have much lower slip). The inital speed was steady state at 3565rpm (71% load estmiate based on slip). Over the course of 2 minutes the speed dropped gradually to 3509rpm (185% load estimate based on slip) at which time the motor tripped. There are obviously a lot of scenarios we can look at including low/unbalanced voltage or motor problem or mechanical problems in the train or increased fluid demand, but in terms of the loss of suction scenario I take it the extreme change in RPM tends to confirm this is not a suction blockage problem?

I do recall for slow speed axial flow pumps, the bhp vs flow curve tends to have a negative slope, especially near shutoff. I realize this is the opposite of axial (pure radial) but I bring it up to mention fluid power is not the whole story, there are also potential losses in heat and possibly vibration. But I gather this type of behavior applies less to pure radial flow. I'm not sure why that is (why the shape of bhp vs flow tends to have negative slope on axial and positive slope on radial), maybe losses associated with internal recirculation play a bigger role in axial than radial? That's a little bit of a detour from original topic but just wondering if I have that right or off in left field.


=====================================
(2B)+(2B)' ?

It doesn't look like it (being a suction issue) and yes, basically you're correct about the axial units- usually high flow low head type machines with something approaching propellers, not a 10 stage unit. 10 stages is actually quite a lot - there must be some serious output pressure to that?

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

What power quality monitoring do you have? From your description either the load went up or the supply voltage went down. Do you have voltage and current measurements across all 3 phases?

You didn't read my red bolded words!!!

... haha, that's ok, the original question is answered so it's fair to move to the inevitable followup.

We collected a fair amount of data on this event and conducted what we call a fault tree. The fault tree approach tries to brainstorm and methodically list all remotely-plausible explanations and then categorize their likelihood and develop plans to evaluate or test each one until it can be ruled out or confirmed. That is what leads me to ask questions like this.

Unfortunately all we are left with on this particular event seems quite unlikely and improbable. I don't think there is any easy way to explain it other than to sanitize our writeup (to remove identifying info) and post it here. Maybe I'll do that in a new post.

=====================================
(2B)+(2B)' ?

All good. I made my first response with the red text in mind but then I started thinking about the problem and forgot about the red text for my second response.

Maybe better if you could post the pump Q-h-speed curve, and a process sketch which shows the control scheme for this pump. Is this a regenerative type high speed centrifugal pump? - the performance curves are unlike those for standard centrifugals.

> Maybe better if you could post the pump Q-h-speed curve, and a process sketch which shows the control scheme for this pump

In terms of control scheme, it is a non-vfd induction motor and there are no automatic flow control valves in the system. There are manually operated throttle valves.

I don't have Q-h curve handy. Maybe later (it's a little outside of my personal focus). The part that seems most relevant to me is that we suspect BHP will increase with flow over the range of interest.

> Is this a regenerative type high speed centrifugal pump? - the performance curves are unlike those for standard centrifugals.

It is not a regenerative pump, it is a 7 stage centrifugal pump (I was incorrect to say 10 stages earlier).

=== ==== === ADDITIONAL INFO ==== ==== ====

As promised, here is additional info, starting with facts, then a table outlining the possibilities, and finally the actions to investigate further. I tried to remove identifying info.

For various reasons beyond the scope of this thread, we we don't want to do intrusive disassembly of the pump and we do not want to run the pump for troubleshooting purposes (we already did run motor and gearbox uncoupled from the pump). We do plan to send the motor out for off-site load test. Personally I think the problem will be found to be degradation of the end ring of the squirrel cage rotor in the motor. It would be surprising for degradation to proceed that quickly (the motor was running the same speed as all the others normally do just two minutes before the trip), but that's still where my bet is.

=====================================
(2B)+(2B)' ?

• https://files.engineering.com/getfile.aspx?folder=8fa79fdc-632b-4abf-828e-4

Either that or some solid material got into the pump. Your fault tree examines a potential foreign material suction line blockage, not pump ingestion.

Still - take a shop vacuum and block the suction side. The RPM will rise and the output will drop, but so will the amperage as the motor is no longer accelerating as much material (less m * delta-V).

A shop rag from some previous service perhaps, It would be unlikely to cause much vibration as it is a braking element. The remaining fragments might have been released during the slowdown when the power was removed.

I saw a relatively expensive diesel engine wiped out because someone working on the radiator pushed a rag into the radiator hose to stop it from dripping. Little bits of it were distributed throughout the engine cooling jacket, also about 400 HP. I think that water pump had more clearance than the one you are using.

A drop in pump shaft speed from 3565rpm to 3509rpm isnt much so that would explain why flow and press remain more or less constant during this event just prior to overcurrent trip.
Gearbox G (acceleration) increasing? - presume this means negative acceleration ?
Doesnt seem like there is a suction restriction. My guess is some problem with the electric motor, since the report indicates no problems with the speed increasing gearbox.

> Gearbox G (acceleration) increasing? - presume this means negative acceleration ?

You lost me on that one. I don't know what negative acceleratIon is. We have an acceleromter on the gearbox. It gives a bandlimited "overall" acceleration ("overall" for us means sqrt2*rms... don't blame me for the the terminology) as I recall up to 10khz. Our computer recording system "clamps" the value at 10g's for whatever reason... it probably went a lot higher.

=====================================
(2B)+(2B)' ?

Sounds like an accelerometer, which is typically for vibration detection - high input or output shaft vibration at the gearbox?

> high input or output shaft vibration at the gearbox?

We have four pump prox probes (2 on each end) which did not change until the pump tripped. We have one accelerometer on the gearbox casing which did increase during the event. We have no other vib sensors on the gearbox or on the motor.

=====================================
(2B)+(2B)' ?

So something has failed at the motor to gearbox coupling or the electric motor output shaft ? Sounds like you've got some kind of flex coupling between the gearbox output shaft and the pump drive shaft, since the pump vib sensors at the drive end dont show high vibration.

> Sounds like you've got some kind of flex coupling between the gearbox output shaft and the pump drive shaft

We have a diaphram coupling between the gearbox and the pump. We have a gear coupling between the gearbox and the motor. Yes they are both flexible couplings.

> Sounds like ... since the pump vib sensors at the drive end dont show high vibration.

Yes the fact that we have NO VIBRATION CHANGE recorded on any of the 4 pump shaft prox sensors but hefty acceleration on the gearbox is an interesting fact.

• Does no vibration change at all seen by he pump shaft probe indication during this extended (90-120second) event rule out things like mechanical rub in the pump (which would include effects of foreign material getting entrapped between rotating and stationary parts)? I tend to think so (but interested in your opinions). We've gone through extra lengths to verify we have proper indication (did an operational checks of the prox probes, reviewed the computer recordings in all available formats to make sure the event symptoms weren't removed/supressed by a deadband / data compression algorithm).
• Review of fluid system shows no plausible way the flow could have increased significantly without showing up on the flow indication. There are only small pipe penetrations for vents, drains and indications between the pump and the flow indicator. Any significant leakage there would be captured in the same room as the pump and abundantely obvious by visual inspection.
• If we rule out pump rub and pump fluid loading, we've ruled out the pump altogether.
• So what is the high gearbox acceleration telling us... problem in the gearbox? In the fault tree attachment I had reported gearbox rotated smoothly by hand, had proper axial float, and also ran fine with motor(uncoupled from the pump) and coasted down over a period of 3 minutes. Well I have some NEW INFORMATION THAT FURTHER EXONERATES THE GEARBOX: We pulled the top half cover of the gearbox for better inspection. Radial bearings and thrust bearing (a Kingsbury tilting pad bearing on bull gear) are all in good shape.

So what's left (since I seem to have ruled out pump and gearbox)? The motor meggered and bridged fine. I tend to think it is torque oscillations of the motor either due to sudden rotor circuit degradation (causes motor torque oscillation at pole pass frequency) or less likely (based on our troubleshooting) due to voltage unbalance (which causes torque oscillation at 120hz). The gearbox provides a mechanism for cross-coupling torsional oscillation to radial vibration (reaction force from on the gearbox bearings resulting from torque transmission would be steady for steady torque causing no radial vibration, but it would oscillate for oscillating torque, causing time-varying reaction force and vibration). That is where our main efforts will focus (continue to investigate voltage unbalance and remove the motor for load test to investigate rotor circuit).

Is there anything else that you guys think we've missed?



=====================================
(2B)+(2B)' ?

This is now getting well beyond my skills set, so its time for rotating machinery engineers and electrical engineers to continue with this thread support.
This vibration sensor that has picked up high vibration "on the gearbox casing" needs clarification - it cannot be on the casing; it must be sensing some rotating component within the gearbox casing.

Vibration sensors, in my experience, have always been on the gearbox casing. I have two experimental gearboxes in my fleet right now and they have accelerometers mounted on the casing near every accessable bearing. The manufacturer is remote monitoring to develop a condition based maintenance approach but I don't have access to any information beyond that.

Foundation stiffness and differential temperature measurements between each component are things to consider. If the problem persists, displacements of the machines can be measured with lasers while on-line. My parent company is troubleshooting a gearbox at this time by measuring housing displacements.

You mentioned gear and diaphragm couplings. These both tolerate substantial angular misalignments but don't have much offset capacity unless they're used in pairs.

What is the bearing type/material in the pump. I assume they're rolling element bearings but composite bearings can introduce their own set of problems depending on the specific conditions they are subject to.

> I have two experimental gearboxes in my fleet right now and it has accelerometers mounted on the casing near every accessable bearing. The manufacturer is remote monitoring to develop a condition based maintenance approach but I don't have access to any information beyond that.

> This vibration sensor that has picked up high vibration "on the gearbox casing" needs clarification - it cannot be on the casing;

It should not be there but it is. We have only one sensor on the gearbox. It is not well situated. It is neither near a bull gear bearing nor a pinion bearing. We just call it casing. That is the data we have to work with.

> What is the bearing type/material in the pump.

Babbit lined tilting pad radial and thrust bearings.

=====================================
(2B)+(2B)' ?

In my experience with centrifuges there will be an acceleration switch mounted on the machine. The machine is resilient mounted and acceleration exceeding a certain value will trip the unit. The mounting location isn't specific.

With that said, acceleration on a gearbox can be exaggerated by features. A lifting eye may resonate with more displacements than a bearing housing and may be a better indication of trouble when mounting an accelerometer.

Accelerometer diagnostics rely heavily on baseline measurements. There aren't strong correlations dictating position or nominal readings.