Validation of Ceasar II results (Pump Nozzle Overload) on under-operation pumps 1

[Hadi-Kh]

Dear All

We are operating a pump house for about 1 year. We have some maintenance issues but I can not say its numbers are abnormal.

Pump House includes:
Five pumps (each 1 MW - 60m Head) run in parallel.
The pipe is made of GRP
Fluid: Seawater 17-37 C

Recently We have received a revised stress analysis report (by CAESAR II) which depicts a very high nozzle load on the pumps (the permissible amount given by Vendor). But there is a dispute that if this analysis is right (although checked by 2 other people), based on the fact that why it has not shown any clear consequence on the pump's performance so far. Out of five pumps, we have had recently some vibration issues but they are rectified and one can associate these vibrations to causes other than high nozzle load.

In API 610, it states that pump nozzle load should be limited to ensure shaft displacement is less than 250-380 micron.

Question: Is there any way to solve our dispute by measuring the displacement of the nozzle/shaft/pipe to verify if stress analysis results are correct or not? Any method like strain gauge measurement on GRP pipe or displacement of the flange by dial indicators?

Please note that according to many references, stress analysis of GRP pipes has lots of unknowns which may result in the uncertainty of CAESAR II outputs, and this makes our case a bit difficult. If we had not constructed this pumphouse, then we could have changed the pipe layout/supports type with a conservative approach. But for the time being, it needs enough reason to make expensive changes and costly downtime.

Thanks in advance for your replies.
Hadi

 

[MJCronin]

In my opinion ... Even GRP piping systems need some flexibility

Your system has near Zero Flexibility ... Your piping designer should know better than this !! (is he what, 19 or 20 years old ?)

Is the problematic pump the one on the far left ?

Put a piping loop in the middle of the system, between the two red supports.

(As an alternate, you could investigate the installation of an axial expansion joint at the same location)

Consider an axial restraint near green flange on the far left of the system. This will force all expansion (and vibrations) into your new piping loop or expansion joint. Discuss these options with an expansion joint vendor and ask for his recommendations ....

Please keep us informed about your final solution and your reasoning behind it

Best regards

MJCronin
Sr. Process Engineer

 

[dik]

MJC... something like

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?

-Dik

 

[Hadi-Kh]

Hi MJCronin

The reason that the pipe system was re-investigated is as per your message: the pipe is not flexible.

The location and situation of the pipe have limitations but we could find ways to re-engineer and resolve the issue with the cost of downtime. But the problem is that some of our team believe that we don't need to change the current state because

1- Any change has considerable downtime and cost
2- As there is one year of operation with few maintenance cases, there should not be a problem.

The middle pump has had a recent "work order" from high vibration but inspection of the bearing did not show any damage. Other minor problems of pumps were considered as the cause of excess vibration and therefore the vibration got normal following the repair.

Thus I wonder if there is a way to see if the pump nozzle is under severe load or not. I heard somewhere, that we can put a dial indicator over the pump's flange and measure before and after pump startup; If the difference is more than 2 mils (0.05 mm) then the pump is distorted too much while operating. Is this way an acceptable approach?

 

[NovaStark]

Yes that would be one way. Another way is if you have tooling balls, you can measure thermal growth in that way as well.

Are your pumps running efficiently or are being operated in the correct region of their pump curve? (I assume they are sized for parallel operation).

You may be able to contact the vendor with the query about then nozzle loads as they may have just limited the loads to the API values as a measure of conservatism.

Depending on the type of loads you are seeing, you probably may be able to install some kind of additional pipe support like a spring in order to alleviate some of the nozzle loads (if they are actually higher than your vendor recommends).

 

[LittleInch]

GRP is strange stuff and has a high Poissons ratio which means a lot of time it is actually in tension when pressurised. it normally needs vendor specific information to be inserted into the model so that it reacts properly. Was this done?

The other factor which may answer your question - why is this not impacting the pumps - is that stress analysis programs tend to treat flanged joints the same as welded in terms of lack of flexibility and pumps to be absolute immovable objects incapable of bending or flexing.

Neither is true in reality and for something like this, even 1mm of movement of a nozzle could relieve the high stress/ force values the program is finding.

Pump suppliers do tend to give very low nozzle forces which in reality can be exceeded significantly without issue in may cases, but if issues develop on a pump and it can be shown that you didn't comply, then any warranties are null and void.

The system does look rather stiff in the connections to the pumps and could do either with some flexible elements or greater lengths and ability of the pipe to expand and contract by means of longer pipe lengths on the branches.

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

 

[Hadi-Kh]

Hi LittleInch

The value given by pipe vendor has been entered in the model. But as you know there is limitation in GRP "stress intensification" coefficient even by GRP manufacturers.

Pump vendor ,as usual, declined to give stiffness value of pump nozzle so the default value of CAESAR is inevitably applied.

Do you have experience of validating or checking vulnerability of pump nozzle overload condition (casing distortion and then shaft displacement) by the "field" method which is explained in above messages (dial indicator, strain gauge, ...). How reliable are they?

Thanks in advance

 

[StressGuy]

Caesar is not going to give you a useful result. In terms of thermal loading on the pumps, you have none at the temperature range given. It's really just a support problem.

As you've found, since Caesar's default stiffness for a restraint is 1e12 and you have a close coupled system, it's very easy to generate highly exaggerated values.

It doesn't surprise me at all that your system is not having any issues in reality. This is one of the all too common cases where a stress analysis will predict a problem that doesn't actually exist.



Edward L. Klein
Pipe Stress Engineer
Houston, Texas

"All the world is a Spring"

All opinions expressed here are my own and not my company's.

 

[Hadi-Kh]

Hi Edward

Thanks for your comment.

Do you think installation of a dial indicator on the pump's flange is a reasonable way to verify default stiffness value of CAESAR. That is, to compare value of flange displacement predicted by the software (0.2 mm) with actual displacement value measured by a dial indicator?

Cheers.
Hadi

 

[LittleInch]

I think though that your analysis will show zero movement as your pump connection is very stiff.

So sure - you can measure actual movement, if measurable, but still have no data to compare it to as the vendor won't give you any acceptable limits.

Impacts on the pumps tends to show up as high vibration, high seal wear, high bearing temps, things like that.

If you don't have any of those you seem to be trying to solve a problem which doesn't exist / solve a theoretical issue.

Or next time make the connections a bit more flexible....

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