Centrifugal Pump: Inadequate Performance. What happened?

[RefineryRR]

I have a vertically-submerged 5-stage centrifugal pump that suddenly yielded reduced head (and therefore reduced capacity). The pump had a normal operating differential head of 22.5 bar at a flow of 80 m3/h, but the pump was delivering only around 20 bar diff head and a flow of around 60m3/h.
We confirmed that the operating temperature was within spec. Speed was confirmed (despite being motor-driven). The recorded suction pressure was 0.7 bar (minimum for the pump according to the data sheet is 0.4 bar). We went through the entire system and matched everything (levels, tower operating pressures, etc.) to the run-period when the pump was delivering as per design - but the pump continued to not deliver.

Eventually, we discovered that there was an obstruction in the suction line of the pump. Once that was removed, the suction pressure rose to 0.75 - 0.8 bar, and the pump was delivering as per design.

What, exactly, happened in this system? What were the hydraulic effects at play to cause this phenomenon? No-one in my team can reconcile the findings with the pump's delivery issues.

 

[MJCronin]

The pump was operating at reduced flow and NPSHa .... There could have been some cavitation damage.

Multistage pumps are expensive ..... Certainly worth a little extra care .

Suggest that you pay for an inspection of the first stage impeller (or inducer) for damage.

What is the material for the impellers/casing ?

Was there severe vibration during the period of reduced flow ?

It is not a bad idea to develop and pay for a vibration signature when a multistage pump is new, so that as the pump ages, a comparison can be made after a significant incident.

MJCronin
Sr. Process Engineer

 

[RefineryRR]

I cannot, off-hand, recall the impeller & casing material, but the vibration steadily increased till we had a catastrophic failure of the pump. My thoughts is we had debris from the obstruction slowly block a greater proportion of the pipe ID and eventually cavitated the pump to destruction, or the debris moved through the pump and damaged the internals causing the catastrophic failure.
We do have a 5-year vibration signature for the pump and we know where it should run. When it was running at reduced performance the vibration was still satisfactory. We didn't see significant cavitation damage though.

 

[bimr]

Regarding:

"Eventually, we discovered that there was an obstruction in the suction line of the pump. Once that was removed, the suction pressure rose to 0.75 - 0.8 bar, and the pump was delivering as per design.

What, exactly, happened in this system? What were the hydraulic effects at play to cause this phenomenon? No-one in my team can reconcile the findings with the pump's delivery issues."

Operation with a blocked inlet means that the pump is operating off the pump performance curve. Operation of pump at either the extreme right or left of the curve may cause damage to the pump. Down thrust can occur when a pump is installed and run in a situation in which it will produce HEAD in the range of shut off pressures (far left side of the performance curve). If this condition continues possible damage to the bearings in the pump and motor can occur.

http://petrowiki.org/ESP_centrifugal_pump

 

[1gibson]

Check your motor amps against "normal" operation.

Low = cavitation due to blockage in front of the pump.
High = reduced efficiency due to debris within the pump (or potentially but unlikely, after the pump, acting like an orifice.)

Same = little bit of both?

 

[LittleInch]

Where was the recorded suction pressure?

Do you have any sort of drawing or sketch relative to the pump suction? Showing pressure tapping point.

Sounds very much to me like the first stage was cavitating.

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

 

[Artisi]

Seems to me that you have supplied a lot of conflicting data.

Eventually, we discovered that there was an obstruction in the suction line of the pump. Once that was removed, the suction pressure rose to 0.75 - 0.8 bar, and the pump was delivering as per design.

I cannot, off-hand, recall the impeller & casing material, but the vibration steadily increased till we had a catastrophic failure of the pump.
How about editing your posts into one post so we have a chance to comment intelligently.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[RefineryRR]

bimr, LittleInch

The suction pressure reading was at the pump suction flange. The actual suction pressure is higher as the inlet is a further two metres below this (within the sump). I would agree with you to say it is first-stage cavitation, however, if the stated minimum suction pressure was 0.4 bar and we were seeing suction pressure readings of 0.6 bar, either the data sheet is incorrect or the pump was not cavitating. Also, the obstruction that was found was UPSTREAM of the suction pressure gauge.

1gibson,
The amp readings were lower than when running as per design.

Artisi,
There is no conflicting information. More articulately:
Event 1: Poor performance. Vibration steadily increased until catastrophic failure.
Event 2: Pump repaired & re-commissioned. Poor performance prevails. As before.
I do not want to steer the conversation; I want to see what our peers think.

 

[Artisi]

I repolished the crystal ball and things should now become much clearer.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[Artisi]

What was the catastrophic failure, impeller destroyed, support bearing or thrust bearing failure, shaft breakage?

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[LittleInch]

What do you mean by minimum suction pressure? Is this NPSHR?

What was the vapour pressure of the fluid compared to however this pressure was set? NPSH is measured in metres head, not pressure.

Be aware that cavitation can occur at heads above NPSHR.

How about you give us some pump data and data on the fluid you were pumping. My current thought is that this min suction pressure hasn't taken account of temperature and hence value pressure, but without data we can't work it out.

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

 

[Artisi]

Initially it would be of more interest to know a bit about the installation, at the moment we know nothing and are floundering around in the dark making assumptions on things that only the OP knows.
I repeat what was said before, too much conflicting or missing information, ie,

Vertical 5 stage centrifugal pump, what type/ configuration
Blockage in the suction line, what suction line????
2 separate events are they connected?
Mode of failure????
Etc. Etc.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[RefineryRR]

The pump is a Flowserve 400WUC-4H-5 pump.

The failure included: shaft fracture, seal failure, bushes destroyed, all wear rings damaged, bearings were in a good condition, impellers were still in a good condition with no obvious cavitation indications such as pitting, etc.

The recorded suction pressure was 0.6 bar (equivalent to an absolute pressure of 1.5 bar) at 106C. The vapour pressure at the selected pumping temperature (105C) is 1.225 bar.

As the pumping medium is very similar to water, a vapour pressure extrapolation does not indicate that the fluid would vaporize at the actual operating temperature of 106C.

 

[TenPenny]

To put it simply, there was an obstruction in the suction, so there were more suction losses than you accounted for, which meant the pump could deliver a lower flow because there was more restriction to oversome.

 

[LittleInch]

From your data, and I assume you mean 1.225 bara Vapour pressure, and assuming your density is close to water, you have around 4m NPSH at the inlet to your pump. Not a lot. The data sheet would be nice to see though.

what was the obstruction - something very solid or did it start to let flow past with hard gritty bits? The damage you note sounds like it's been ingesting stuff it wasn't designed for.

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

 

[1gibson]

If the shaft broke, you locked up (at lease one set of) the wear rings and the inertia of the motor rotor caused the break.
If the wear rings locked up, there were particles in between them or maybe just high enough vibration over long enough time to cause a rub and eventually galling. Once the rings grab enough to spin the case wear ring (overcome the tack welds or set screws holding them to the case) then it's a very short time before it locks up.

Torsional shaft failure with any ductility, will be at an angle. Top side of shaft still spins, and due to the angle of the break it bounces up/down on the broken surface of the bottom of the shaft. This causes the top shaft to jump up and down, and will smash the seal faces to pieces.

Check your motor bearings as well.

 

[RefineryRR]

TenPenny,
The measured suction pressure of 0.6 bar was downstream of the obstruction - the reading was at the inlet flange of the pump. Hence my confusion regarding the obstruction as the rated suction pressure of the pump is 0.4 bar. This should have still been sufficient.

LittleInch,
The obstruction was a solid (leak-sealing compound protruding into the suction line).

1gibson,
That's a good analysis. Unfortunately, a lot of the evidence was lost due to the "run-down" damaged the pump incurred so things such as the shaft fracture angle could not be assessed. But you bring up valuable points I would have liked to have assessed.

My speculation is that the rated suction pressure of 0.4 bar on the data sheet is actually incorrect (due to potential variations over time of the process medium). With the obstruction we saw suction pressures of 0.4 bar and the pump failed to deliver as per design. Once the obstruction was identified and removed, the suction pressure rose to 0.75 - 0.8 bar and now the pump is delivering optimally, as per design.

 

[bimr]

Regarding "Once the obstruction was identified and removed, the suction pressure rose to 0.75 - 0.8 bar and now the pump is delivering optimally, as per design."

This scenario is easy to identify because the pump will not be operating on the performance curve.

 

[georgeverghese]

Could it be that the suction pressure at 0.6bar is not sufficent to keep the seal chamber fully liquid flooded and adequately remove frictional heat ? ie the seal chamber may have vapor locked?

Presume the shaft seals are not plan 52 or higher ?

Temp in the seal chamber would be several degC higher than 106degC even during normal operation with adequate suction pressure.

 

[Artisi]

I fail to see how a reduction in inlet pressure that probably results in a shift of the operating point to the left on the performance would result in the failure as described "shaft fracture, seal failure, bushes destroyed, all wear rings damaged,", and with no signs cavitation indications such as pitting, etc.
It would be fair to say without any further data, the mechanical failure and a drop in hydraulic performance are unrelated, unless of course the shift of the operating point to the left was to near shut head and then maybe, but that's a maybe, some mechanical failure was likely, but without a lot of additional data it is unknown.




It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)