Booster pump station breakdown 2

[TotemCruise]

Hello!
Before 300 m3/hr RO system 6 pump booster pump station used to work round the clock for three months. At the end of this period the customer informed us about periodical alarm disconnections of pump station. The message informed about the frequency converter fault and pump #2 short circuit. The service center engineers disassembled pumps and discovered that the impellers of all pumps except one that was switched off all the time are almost fully destroyed. Undamaged parts of the pump #2 check valve were found there in the output collector of the station.
In their conclusion they wrote that the reason of destruction was operation of the pumps in wrong duty point. According the project the station has to operate under the input pressure 2 bar and produce the changeable at any time output pressure from 8 to 9 bar. The customer is not ready to use the system in full power till now so they used to operate it most of the time in conditions as follows:

Inlet Pressure 2 bar
Duty point 6 bar.

At that the lowest point of the curve is at 5.8 bar.

They said there was an overload that lead to damaging of the impellers. They absolutely didn't take in account the check valve. Were there any overload that could be so dangerous for motors and pumps there. The input pressure was high enough to prevent the suction cavitation. The view of debris confirms this. The surface of the damaged impellers looks new. If somebody wants to have a look at them I can send some pictures.

We have our own version of this destruction. We think exactly the broken of check valve initiated this destruction. A high speed water flow run back through the broken check valve and pump #2 when it was switched off according to schedule off the PLC. Setting this pump
in motion lead to superhigh impeller loading that caused the damage of the impeller. The starting current because of the rotaition of the pump in opposite direction was too high and was interpreted by the frequency converter like short sircuit.
What do you think about this situation. Could this be a
reason. Has somebody ever faced with problem like this.

 

[TotemCruise]

I forgot to write that
1. the pumps are multistage centrifugal pumps with 304SS impellers.
2. the total pump station flow was right and stable, about 400m3/hr. If the discharge pressure would 7.8 bar the pump station worked in right operation point.
3. the first stage of each impellers were almost in working condition, other stages were seriously damaged, the last ones were almost fully destroyed.

 

[BigInch]

To "quote" Sherlock Holmes, "Investigate all the possibilities. Eliminate the impossible and what you have left, no matter how implausable, must be the answer." Backflow in this situation would appear to be the only possibility remaining. Have you done a broken check valve transient simulation to support your theory?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

A selection of photo's would help with the evaluation.

"Seriously damaged" or "almost fully destroyed" does not allow us to assertain the problem as they are meaningless as a description.

Phitsanulok
Thailand

 

[TotemCruise]

I have placed some photo’s of damaged impellers to the Web. Anyone can find them at

http://new.photos.yahoo.com/totemcruise/

I have not done any simulations. Is there any software that can help. Can you advise. I am afraid all of them simulate only normal operation mode.

I would like you to abstract your mind from this specific case and consider the questions that I am sure could be interesting for most of the members of this forum.

1. What could be the consequences of working of the multistage cenrifugal pump in the point below the extreme right point of the curve, below the lowest pressure depicted by the curve shown in the pumps specification. Let us assume that the inlet pressure is enough to prevent suction cavitation.

2. What could be the consequences of working of the 6 pump multistage centrifugal pump station if one of the check valves at the output of the pumps does not operate. Only 5 pumps operate simultaneously at any time and PLC exchanges the working pumps at regular intervals.
Thank you in advance.

 

[Artisi]

A quick look shows mechanical damage to all the components, is this typical of all pumps?

Dismissing the hydraulic problems at this point, you need to assertain where/ why / what initiated the mechanical damage, it is very unlikely to have resulted from the hydraulic problems you must be experiencing operating so far out on the curve.

Phitsanulok
Thailand

 

[Artisi]

It is not easy to trouble-shoot by looking at a group of photo's and written information due to the many variables and unknowns - all that can be done is to discuss the likely cause from the visual information presented, e.g., the photo's and your comments.

For the hydraulic consideration.

With 2 bar inlet pressure, it is very likely that the NPSHa is far greater that the NPSHr, therefore it is probable that the pumps will operate beyond the end of curve. However, they will be operating inefficiently, could be noisy and could exhibit higher than normal levels of vibration, motor size could be marginal (depending on head being developed across the pump)leading to an overload condition.

Operating under these conditions would not result in the mechanical damage shown in your photo's where you can see portions of 1 blade broken off, other photo's showing diffuser blades torn-out of the interstage plates and other views showing impact damage.

The question of operating these pumps with 1 check valve inoperative - this will result in the pumped discharge recirculating back to the inlet side of the system - this could have the effect of increasing the inlet pressure (and NPSHa) as well as reducing the developed head of the pump/s.
I cannot see how the service centre advise can be correct.

"In their conclusion they wrote that the reason of destruction was operation of the pumps in wrong duty point."
Operating at the wrong duty point will not give this type of damage.

As a fix-up

You need to address the mechanical damage first - why / what / when / how before addressng the hydraulic operation as it would appear they are unrelated.

If the pumps are required to operate at a lower duty, consider removing 1 or more impellers from each unit or restricting the discharge flow by either a throttling device or recirculate some of the flow back to the source.


Phitsanulok
Thailand

 

[TotemCruise]

The damages are absolutely typical for all the pumps. I surmise the small impeller debris could recirculate for a long time through the broken check valve and pump#2 when it was switched off. It was swiched on for only 1/5 of total pump station operation time. If the total time is 90 days only 18 of them were dangerous.

 

[BigInch]

I agree with Artisi. That was definately not the result of operating outside the normal duty area.

Do you have any more details of the event (actual Pin,Pout, Q, rpms, current draws, etc. for each pump during the event), or is the converter and short circuit message @#2 the only info remaining?


BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[TotemCruise]

Thank you for help.
I am not sure if I can mention the model number and the name of manufacturer at this forum. It may be considered as false advertising.
At the end the operation of station became unstable so they decided to increase the inlet pressure. This helped.

According to service protocol the parameters were as follows:
Pin= 2,5 bar
Pout=5,2 bar
Q= 430 m3/hr

The extreme right point of single pump curve is 5.8 bar, 120 m3/hr.

That is all we know. The ingress of loose iron to the input of the pumps is absolutely improbable because of the 5 mk grade microfilter installed there.

Just a quote from Fluide Design PUMP AND PUMP SYSTEM GLOSSARY I found:

http://www.fluidedesign.com/pump_glossary.htm

"Check valve: a device for preventing flow in the reverse direction. The pump should not be allowed to turn in the reverse direction as damage and spillage may occur."

 

[BigInch]

Ya, the location of the damage being in the upper stages, would kinda' indicate any forces hit from that direction.

Discharge check valve failure is something I routinely check in simulations of pump stations to determine the reverse flow rates, pressures and counter rotation as the fluid goes into reverse flow, but although I can imagine what effects they might have, which includes the damage you describe, I have not actually seen a pump that has been subject to such a load, nor any pump damaged like yours.

My e-mail can be found in the contact section on my MSN webspace, if you have something you might want to only wisper. I would like to know more details about the specific pumps involved for one.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[TotemCruise]

When the pump is switched off there is not any flow through the pump(if certainly the check valve is in working condition).
When the pump starts, first period of time it simply works in reciculation conditions, the check valve is closed until the pump head slightly exceeds the pressure in output collector. What if there are some debris there in the impeller at that time. It works like coffee-grinder and the debris will recirculate there in the impeller too. This is the only my version why the upper stages are the most damaged.

 

[BigInch]

Right, the impeller will obviously grind up anything it can and what it can't will grind up the impeller. But what might that have been? But the photos didn't look like grinding, more like some kind of impact blow hitting the sids of the vanes near the edge. (At least that what the photos looked like). If ground up, I think there would be more gouges and scrapes. If you found the check parts undamaged, what else was in there to do it? Nothing unusual caught somewhere downstream in a strainer, hardhat, wrench or something? But I don't think there was any grinding going on. Do you have a better naked eye view?

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[TotemCruise]

These pictures are all I have. If I was there I made some more. The customer said the check valve parts are too big to recirculate but the debris of the impeller are just right. You can see them at the photo's. Nothing else was found there. In conditions of check valve mulfunction it was enough even one piece of vane to travel through the pumps and crash more and more vanes. Just to remind, they have been working for three months.

 

[BigInch]

OH! I didn't know they were check valve pieces. I thought you found those in the "downstream collector". Neither did I realize they were operating the broken pumps for 3 months. That rivals the worst operators on the planet!

If you found discharge check parts in the pump, there's one thing for certain... There WAS backflow! And yes, hydraulic programs WITH transient capability can do backflow scenarios. Visit my webpage for more info on surges, etc.

BigInch-born in the trenches.

http://virtualpipeline.spaces.msn.com

 

[Artisi]

The statement in the link regarding check valves and pumps running in reverse can be true as well as untrue.

http://www.fluidedesign.com/pump_glossary.htm

"Check valve: a device for preventing flow in the reverse direction. The pump should not be allowed to turn in the reverse direction as damage and spillage may occur."

Generally a centrifugal pump can run backwards without any risk of damage, however, some mechanical seals are designed to operate only in one direction and if a pump unit is on auto start, unless some provision is made to sense direction you could end up with problems attempting to start while the pump is running in the reverse direction.

Damage:

My initial thoughts were damage from the check valve parts -but this now seems a little remote -- but you have to locate and closely inspection for impact damage ALL the check valve components - this will confirm if it was a check valve component that has started the chain-reaction.

What is interesting is that the damage is on all pumps - why ?
what foreign material has been in the system other than broken pump parts?
are all impellers in each pump damaged?

Picture PA 120038 11/12 shows that looks like a clean break on the leading edge of one blade - what impeller is this one - is it the first impeller in the stack? and there doesn't seem to be any other damage evident on this impeller.

I will stay with my initial comments, this damage is from foreign material in the system and not the result of running off the cump curve

Phitsanulok
Thailand

 

[TotemCruise]

I found some corroborations of my assumption.

1. First corroboration is indirect. It's Goulds AquaForce Pumping Station Typical Specification

http://www.goulds.com/pdf/AQFSPEC.pdf

Page 6, Pump Check Valves, last paragraph.
The quotation:
"When pump is retired, valve shall function to close tightly before flow is reversed, and reducing the possibility of water hammer or shock"

2. Second is dedicated to this problem directly.
It's Lawrence Pumps Inc. newsletter "Preventing reverse rotation".

http://www.lawrencepumps.com/newsletter/news_v03_i3_mar06.html

So it seems that the chain of causes and effects is as follows:
1. Falling out of the defective check valve's parts(immediately at pump station start up procedure or some time later)
-->
2. Damage of pump#2 impeller during start in reverse rotation backflow condition, more and more.
-->
3. Recirculation of impeller's debris through the broken check valve, switched off pump#2 and other pumps for 1/5 of total pump station working time. With already known consequences.

 

[poppeye]

As a service and repair manager, I am asked to diagnose this type of failure on a regular basis. With the first stage basically undamaged and the remaining damaged and more damage added to each stage. I would be looking at insuficient flow through this pump. Each stage adds energy. If minimum flow is not maintained the water inside the pump will boil and cause thermal expansion damage as parts grow and make contact. You mentioned that the pump starts in recirc mode. Does the pump recirc back to point of origin or just back to the inlet side of the pump? On a multiple stage unit the recirc must be back to the point of origin. If not you will heat the liquid excessively raising the fluid above the allowable inlet vapor pressure. All the parts in this pump are cooled and lubricated by the fluid. If the fluid is not maintained at proper temp and flow, this pump will fail. I would suspect that the discharge check valve failed due to debris damage leaving the pump.

 

[TotemCruise]

I would suggest to read all my messages more carefully.
"According to service protocol the parameters were as follows:
Pin= 2,5 bar
Pout=5,2 bar
Q= 430 m3/hr"
It's a reverse osmosis booster pump station so the flow could never be unstable and less then above mentioned. The pump station is equiped with smart controller. It takes decision how many pumps to switch on at once to maintain the prescribed output pressure.
The frequency converter is used to start each pump smoothly. The rotation rate of the latest started pump is controlled by frequency converter all the time. Even if a saboteur turned off the valve at the output of pump station nothing dangerous could happened. The controller simply would switched all the pumps off.
What insuficient flow can we talk about. The case is exactly opposite.
The pumps were working in the point below the extreme right point of the curve, below the lowest pressure depicted by the curve shown in the pumps specification.

 

[poppeye]

If I am reading right, you are operating the pumps at the far right hand side of the curve and not at the BEP. Operating at the far right side of the curve requires the greatest NPSHR. If you don't have the proper NPSHA compared to NPSHR you will flash the liquid in stages two on up. This could be why you see the damage you are seeing. You could also be seeing cavitation in 2nd stage on up with water slugging into each impeller. This causes severe radial loading on this type of pump. Radial loading causes wear in the alignment bushings at the bottom of the stack and starts taking out stages.