Repetitive failures on Stripper Pumps (What's going on here?) 1

[REVenezuela]

Dear colleagues,

Looking for some advice here.

I'm new to this plant and have been assigned to solve a recurrent problem on a couple of pumps which MTBF is around 3 months combined, the characteristics and operating conditions as follows:

Pump Type; API OH4 (not very fan of these particular type of pumps by the way)
Rated Flow: 260 gpm
Normal Flow (Data Sheet): 237 gpm
Actual Flow: It varies from 154 to 176 gpm ( 56 to 64 of the BEP)
BEP: 275 gpm
Diff. Head: 237 ft
Fluid Temp.: 264 *F
Fluid: Waste Water (Traces of CO2, NH3, HCN, particles / soot fines)
S.G: 0.936
Nss: 9660
Ns:487
Speed: 3600 rpm
Impeller Diam. = 8.5”
NPSHa (Data Sheet): 13 ft
NPSHa (Calculated - Suction Gauge Pressure Method): 12 ft
Min. Flow: 70 gpm
Suction Pipe: 4.5 times straight run of pipe diameters.
Inlet Velocity: < 3.5 FPS
Volute Type: Single
Materials: A351 CF3M (Impeller, Casing Cover, Volute), Ni-Resist (Wear rings)
Mechanical Seal: JC 2800 / API Plan 74 + 13 (1 1/8"shaft diam.).

Symptoms / Findings:
I’ve been quite a bit in the field checking these pumps and a “mild” cavitation like noise could be heard all the times although may becomes worse and make all the system to shake badly (similar to water hammer), this have caused in the past cracks on weldings, supports to come loose, failed bolts, etc., slight pressure fluctuations at the discharge are evident too. I have seen this happening even when a new (spare) pump installed.
Vibration could be severe at times but in my opinion this is a consequence after the radial sleeve bearing (carbon) gets worn and the wear ring rubs and get unbalanced, high speed and a massive rigid coupling make things worse.
I have had the opportunity to inspect these pumps internally a couple of times after they failed, I will let the pictures to speak for themselves. One thing that caught my attention though was the condition of the casing compared to the impeller being made of the same materials (although I’m pretty sure they are not) which tells me sometime on the run they may have been changed (I’m tracking this with the OEM). This does not explain the problem I’m describing though, does it?. Erosion-corrosion, recirculation cavitation?.

Control / Operating / Process conditions:
The system operates on level control, waste water and high particle filtrate is fed to the tower while live steam 50 PSI is injected in the mid section of the column, normal level is 50% although most of the times is at 40%. Control of particle content seems to be an issue upstream (still working with Process eng. to understand the problem). Flow meter and one of the level controllers is not working either (working with E/I to see if it’s a process related failure) so basically the control is manual. Pump was originally provided with basket strainers at the suction but they were removed most likely because of previous failures were attributed to cavitation, a filtration system was installed downstream of the pumps instead. This filtration system operates with a battery of 3 filters in parallel and three other filters in standby, the control is completely manual, a crew of operators monitor the dP across the filters and switch to the standby filters (which is kept pressurized) when the differential pressure reach the alarm values. The mesh of the filters is 5 microns and they can be changed as much as 14 times per day although in average they are change 4 to 6 times.
The minimum flow control is kept by an ARC valve sized to allow the 70 gpm of minimum flow to recycle back to the column. I noticed though the Operators keep this valve closed on both pumps, I asked why (nobody seemed to know) until a Supervisor told me they decided to do that because the pump trips on overload when they keep it open. This did not make to much sense to me then (overload on min. flow condition?….keep reading).

Tests performed in the field
The conventional pressure gauge, amps reading, temps, vibration were done, herein the results:

First Test:
Pump “A”
Suction Press. = 22.5 PSIG
Disch. Press = 137 PSIG
Pump Casing Temp. = 232 *F (Max.)
Amps. = 17,7 (average)
Column Level = 40%

Pump “B’ (after shuting down “A”, “B” was coming from Maintenance – New Pump).
Suction Press. = 22.5 PSIG
Disch. Press = 134 PSIG
Pump Casing Temp. = 223 *F (Max.)
Amps. = 17 (average)
Column Level = 40%

A few minutes later the feed to the tower (waste water) was increased from 88 gpm to 176 gpm, the suction pressure increased to 25 psig and the discharge dropped to 131 psig. The amps drawn by the motor at that time was 22.2 (average). Aproximately 1 hour later the feed was cut back to 88 gpm and the pump was kept running. Ops were happy….

Six hours later Ops switched to pump “A” as pump “B” (new pump) was not capable of keeping up the column level. As normally occurs this happened late at night when there was nobody around to troubleshoot, process data does not show much either.

Second Test (the following day – I didn’t give up…), asked them to switch pumps, the results:

Pump “B”
Suction Press. = 24 PSIG
Disch. Press = 136 PSIG
Amps. = 18,4 (average)
Column Level = 38%

Pump was kept running with no issues, Ops happy again but not quite convinced (neither I was). At this time was when I noticed the ARC valve isolated, I explained the Operator the importance of keeping this valve lined up and kindly asked to open it although he refuses because “always have been isolated” although he didn’t know why, I told him I would find out, didn’t give up……

Returned a few hours later and after obtaining the approval of the Plant Supervisor we decided to open the ARC valve in order to "keep the pump protected". The conditions on the pump before opening this valve were:

Suction Press. = 23 PSIG
Disch. Press = 136 PSIG
Amps. = 18,7 (average)

Let's what happened a few seconds after the operator started opening this valve? wala! discharge pressure dropped and surged badly, the whole system started shaking, pump vibrated, became noisy and obviously the level on the column increased considerably (besides I became not very popular in the control room). The operator quickly switched to pump “A” which also showed the before describe symptoms, the operator cracked opened a vent in the pump casing and the pump recovered slowly (this trick didn’t worked in pump “B” though).

I gave up keep doing tests…..until having something.


Additional info:
For some reason Operators seem to have preference for pump “A” over pump “B”, even when having “new” pump "B" (like in this case) the pump does not perform as well as pump “A” (which is not good either). In terms of piping configuration pump “B” is a bit different as you can see in the attached sketch of the ISO.

Sorry for the lengthy of this post but I wanted to bring as much information as possible, maybe I’m over analyzing the problem (Do you guys golf?, have you heard the term “paralysis by analysis” when facing an important shot?.). I know there are some deviations in the system (the filtration unit downstream, the suspicious material of the casing, the ARC valve being closed, etc.) still this does not explains the symptoms and some of the damages.

My thoughts:
1.- Operating off the BEP and moving back and forth on the curve when changing filters creates radial loads that progressively worn out the radial bearing and then a snowball effect comes: wear rings worn, internal recirc., unbalance, vibration, seal fails, etc….
2.- Particles in the fluid eroded-corroded the casing which is probably of a different material despite of what the data sheet says (I’ve requested a PMI but the casing may have been scrapped already).
3.- Why would the pump cavitate (or water hammered?) when the arc valve was opened. At the beginning I said to myself of course! we went to the very right of the curve and made it cavitate but hold on…....no flow was supposed to pass through this valve as we were not in a low flow condition! or it was?.
4.- I doubt of real cavitation as I could not see the typical signs on the impeller (yes I used a mirror and looked closely every single section of the vanes), just saying, I'm not an expert in cavitation.
5.- As of today I have seen the damages just in one of the pumps ("B" pump), I can't say for sure the damages are the same to the other pump.

Sorry I made you read all this but I wanted complains they were not enough info (pretty sure someone will ask for more details though…...working on it).

Please do not forget look at the attachment and kindly contribute to this diagnose.

Thanks!

PS: I apologize if my grammar is not that perfect….yo hablo espanol).

Regards

 

[DubMac]

I don't have time to read this in its entirety or reply at this point; I'm sure you will get plenty of good enough advice from others, BUT.......

I do however have time to thank you and give you a Gold Star for the day, due to the very thorough case you have presented here. Your question is a model of how to present a problem to others in asking for assistance.

I only read a sentence or two, but it seems like the pump is running fairly far to the left of where it was designed to run.

 

[JJPellin]

This may be the most well documented question I recall. Well done. I suspect multiple problems. I will address one of them. Wear on only one side of impeller wear ring suggests high shaft run-out. This style of rigid coupling is prone to this.

I am working on this problem with a similar pump. Even with a new coupling, new motor shaft and precise assembly, the best run-out we can achieve is 0.006 inch. Our specification requires 0.003 inch or less. We are designing a new coupling.

Assemble the coupling, pump shaft and impeller with no seal or bushing. Measure run-out on lower impeller wear ring. If it is high, make sure pump shaft is straight. Make sure motor shaft is straight. Check for interferences with coupling keys.

Johnny Pellin

 

[Artisi]

As others have commented congratulations on your posting, as for apologise re grammar - not needed, your English is better than a lot of native speakers who post here.

1. I would agree with Johhny re wear on one side of impeller, probably the result of coupling flex and operating left of BEP which gives uneven radial loads on the impeller.

Yes, many problems by the look of it. I would opt for sorting out the mechanical problems firstly but at the same time look at the operating conditions and try to smooth out any flow upsets etc, if the actual and constant flow is below the rated / designed flow maybe you need to look at an impeller diameter reduction to get the pump back to somewhere near BEP. This will help in reducing shaft deflection and vibration.
A lot of erosion corrosion by the look of it - but that's a separate problem to be addressed.

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]

I echo the others in congratulations for a most comprehensive post.

Some things that jump out are noted below, but like said on a similar post, problems with pumps often revolve around the system not the pump and that is a factor in many items here.

1), At 264F, your vapour pressure is above your inlet pressure and without seeing your calculation it is difficult to comment, but any loss of level in the column will cause cavitation. Also be aware that onset of cavitation is normally 1 to 3 m above the NPSH line on the curve. The fact that your column is supposed to be 50%, but normally now operating at below 40 will not help when you have such a small margin before cavitation.

2) I can now understand why they moved the filters as you have virtually zero head loss available on the inlet side, but this is the root cause of your issues. Damage in the casing is typical of erosion damage. I suspect your ARC valve (I had to go look up what this was and how it works) is completely eroded on both the plate and the internal orifices so as soon as you open the isolation valve it's like a full bore release back into the vessel and no forward flow, hence a rise in level as your inlet continues, but you've gone off the end of the curve on your pump in full recycle, hence the vibration etc.

Next time one of the pumps goes down take the ARC out and inspect it and post the photos. Should be interesting....

3) Pump A might have a slightly better flow path, but that's probably just the normal variation in pump performance.

Basically your pump is operating right on the limit of inlet pressure, has gritty stuff going through it and isn't particularly well matched for head and flow, but probably ok. Fix the ARC, though it will probably erode fairly quickly once more, but without cleaning up your inlet fluid there's probably not much more to do, but the pump guys here have offered some good advice and will probably add some more. Try and get them to keep a higher level in the column to stop the cavitation, which won't be doing the wear ring any good.

Let us know how you get on and hope we've added some info for you.

LI

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[JJPellin]

I would like to add to my previous post. As noted by myself and others, you likely have a number of problems

This is possibly my least favorite configuration of pump. When I do pump training for new project engineers, I tell them that there are two reasons to purchase a vertical in-line pump: You are cheap or you are stupid. I know this is overly harsh. And these pumps work well for many users. I don't feel that they are up to the challenge of the typical refinery service I deal with. A vertical in-line, rigidly coupled, single-volute pump running at the left of BEP is a bad combination. I would purchase an upgrade to add a true bearing housing and flexible coupling. This is a standard upgrade package offered by many pump manufacturers.

The hydraulic design is poor. For water services, we require a minimum 5 foot NPSH margin calculated with the vessel level at bottom tangent for a vertical drum. I am not able to open your attachment at this station, but I believe you have much less than that even at normal liquid level. You are experiencing cavitation. Keep in mind that cavitation does not necessarily leave classic damage patterns in the impeller. The damage can occur at a number of locations or not at all. There is no good solution to this problem. At bottom tangent level, you probably have less than 10 feet NPSH available. With a requirement for a 5 foot margin, you will have a hard time finding a pump to meet these conditions.

I agree that there is a materials problem, as well. The case may have been replaced with a lesser material. But, the material shown in the pictures is not holding up well. Evaluate the materials of construction and select a better option. If you need a replacement case anyway, see if the manufacturer can provide it with two volutes rather than one.

The minimum flow valve is probably washed out and leaking by. As noted above, remove and repair the valve at the first opportunity.

When I get to work, I will try to find a few minutes to open your attachment and review it for other issues.l


Johnny Pellin

 

[REVenezuela]

DubMac, JJPellin, Artisi and LittleInch.

Thanks for your comments. Much appreciated.

I agree with most of your comments.

LittleInch this is how I calculated the NPSHa=
NPSHa= Suction Pressure + Bar. Pressure + Pressure Gauge Height Correction + Velocity Head - Vapor Pressure
NPSHa= (67.91 + 34.05 + 1.4 + 0.059 - 91.27) = 12.15 ft
NPSHr as per the curve for the flow range the pump normally operates at is between 8 - 8.5 ft.

If I use the whole system for the calculation the results are quite different:

NPSHa= Suction Surface Press. + Suction Static Head - Friction Losses - Vapor Press. Since the liquid inside the vessel is bubbling I assumed the pressure on the liquid would equal the vapor pressure (is that a correct assumption?), hence;
NPSHa= (91.27 + 20.01 - 0.6 - 91.27) ft
NPSHa= 19.4 ft

This is a big difference when compares to my previous calculation, I'm missing something or the actual suction losses are way more than what I've calculated.

As per the ARC Valve I have included an attachment showing the type of valve we have, as I mentioned in the original post the valve was selected and is sized for the Min. Flow.

I forgot to mention this in my original post but the ARC valve on pump "B" (the one that made the pump to cavitate badly when open) was a brand new valve so I ruled out erosion issues. I have asked Maintenance to locate the old valve to conduct an internal inspection, not many hopes on this request though since it may have been scrapped already.

JJPellin I agree with you on upgrading to a pump with an actual bearing housing, I have done this quite successfully in the past. That would be my mid-term solution though.

Thanks again.

 

[LittleInch]

There's something wrong with your assumptions and workings for npsh. You need to know what the surface pressure is in the vessel nor guess it. How have you worked out the vapour pressure that accurately?

I can't understand why the new ARC valve would pass fluid, but that's what it has done, but also look at the back pressure on the return line and the actual data sheet or stamps on the valve to check what it was designed for.

There's something going on here I can't fathom, but you're going the right way to investigate it and you will find out what it is eventually. Good luck.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[JJPellin]

The difference between the two methods of calculating NPSHa is interesting to me. I would be most suspicious that you have some restriction in the suction piping. I would check for a strainer. Even of the operators are not aware of one and the drawings do not show one, I would still check flange by flange. We occasionally find old start-up strainers that were accidentally left in place and not documented. A restriction in a strainer would certainly contribute to your cavitation problems.

Other suction line restrictions that I have seen include accumulated heavy solids in the bottom of the pipe. We have also seen gate valves that had broken stems. But, this would probably result in a much greater restriction that what you are finding. We have found issues with the stand-pipe, outlet strainer or vortex breaker in the bottom of the vessel. We have had vessels with internal features (trays or packed beds) that had collapsed and were partially blocking the outlet from the vessel.

It is often an appropriate assumption that the vapor pressure and the static pressure in the suction vessel are the same. I work in an oil refinery. Most of the services that I analyze are coming from separators, fractionators or other vessels where this is true. If the liquid and vapor are in equilibrium in the vessel, this is valid and appropriate. It greatly simplifies the calculation.


Johnny Pellin

 

[LittleInch]

Maybe I'm being stupid here, but when I work out the suction method I get 0ft npsha.

Translating 91ft absolute head to pressure I get 22psig based on the density of 0.935.

As your suction pressure is basically the same, it means you're losing 20 ft head somewhere in your system. I think that JJP has it right and somethings blocked in the pipe our the vessel that shouldn't be there. Your inlet strainers might not be a empty as you think.... As you say there are particles in it there is ample opportunity for problems.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[REVenezuela]

LittleInch,
I got the Vapor Pressure from a Process Engineer I can double check with him on this. Did you consider the atm. pressure on your calculations? (maybe I'm being stupid now, in our site the atmospheric pressure is 13.8 PSI. I will review the data on the valve as you suggest.

JJPellin,
I have asked Maintenance to review upstream the pump, I'm betting on something there.

Today I ran another test in the "new" pump (B). Shutoff pressure and differential pressure match the performance curve, but after a few minutes the pump started cavitating so we had to switch back to "A" (which is an older pump). This made more suspicious about restrictions in the suction. I have asked them to keep running "A" until fails and then installed the "B" pump in the "A" location....they haven't buy it yet but I have no more to offer.

I'l come back to you whenever I have something.

Again thanks for your comments.

 

[LittleInch]

For an enclosed vessel you only need to know the vessel pressure. Can you find out what the feeder vessel pressure is foot the gas space above the water? My calc says 22 psig, but would be good to know what it is. Vapour pressure I agree with for water at that temperature as an absolute pressure.

As jjp says, start with the easy stuff such as looking at every flame to see if there is an extra metal piece which shouldn't be there our take the flange off or open the "empty" strainer and have a good look inside.

Given your other instatement issues are you confident that you have an accurate level transmitter / guage? Blockage or partial blockage of the impulse lines could have a serious impact on the accuracy or delay in responding to a change in level if the level control is manual. Can you drain / flush the impulse lines?

You'll get there but it will take time. LI

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way