First excuse me because my English is not so good.

When I want to calcule force and moments in a pump nozzle (API 610) I should modelate pump nozzle as an anchor and the anchor default stiffness for translation and rotational defined in the Caesar configuration file are infinitely stiffness(1.2E+12).

Is it possible that I can modelate a pump nozzle as a flexible anchor? considering that there are gaskets in valves, strainers and the pump nozzle that they allow deformations.

Thanks for your comments

You should model the pump nozzle as a rigid, displaced point. Use the Caesar displacements fields to define, or simply define an anchor if there are no thermal displacements.

It is not uncommon to model the body of a strainer or a light wall valve with, say, twice the pipe wall thickness, rather than specifying a rigid body. However, if a flanged joint had any significant flexibility it would probably leak or blow out the gasket. API pump bodies are massively stiffer than any connected pipe. Any significant body flexibility would lead to failure by internal rubbing or misalignment. The weaker part is often the pedestal and frame.

Finally, 1e12 is by no means infinitely stiff, fortunately nothing is. As one of the worthy contributors to this forum so rightly points out.... all the world is a spring.

I understand your comments.

In this moment I am checking an existing pump distribution that has 14 pumps(end suction) conected to a common suction header, our client request us to install two pumps more, and checking the existing pumps they must have collapsed because the force and moments results in suction pump nozzles are 4 times the API 610 recomendations. These pumps have been operating during four years without any inconvenient.

I think that the explanation about this, is that the displacements by thermal expansion were liberated in the base plate or gasket.

regards,

Hello Angelo,

Have no fear for your English, you are perfectly understandable. Thank you for making the effort to communicate with us in English.

Some of the "natural" flexibilities that you mention will sometimes be there and will have a part in determining the magnitudes of the loadings that finally are applied to the pump nozzle. But these flexibilities are unreliable so we never "take credit" for them. Also, remember that the API limits on pump nozzle (and casing) loadings are not based upon gross structural collapse of the pump but rather they are based upon how much casing distortion can be tolerated before the bearing life is affected. This is something that we cannot see directly but which is manifested in pumps "losing" their bearings and in excessive vibration. It might be interesting for you to ask the maintenance people how often these pumps need attention. Arrays of pumps attached to common headers (especially very long headers with many pumps) are often problematical as the header expansion /contraction is not easy to manage. In fact there is an art to designing headers that will accommodate many pumps without "penalizing" the pumps that are located a greater distance apart.

With pumps we like to model them as rigid after the initial expansion/contraction movement. This is exactly as MoverZ describes above. Our models will return calculated forces and moments that are likely overstated but when you consider the vagaries of piping systems (those things that can never be modeled exactly) and the relative fragility of the pump casings, it buys us a little more bearing life most often.

Regards, John

Thanks John.

I will ask to the maintenance people how often these pumps need attention and change of bearings.

I am thinking to model considering the thermal growth of the pumps nozzles.

Regards,

John,

I have few questions.

1)Although the API 610 loads are based on allowable casing distorsion and shaft deflection , but will these two parameters be not affected by the stiffnesses that Angelo is referring to?

2) WRC Bulletin 449 also recommends getting the stiffness values from the vendor. Till date I have never seen any vendor providing these informations not even the nozzle displacements( I am personally against modelling the pumps as I believe that proper location of the pump point of fixity is difficult w/o consulting the vendor and most of the consultancies I have worked with follow the procedure of modelling upto the PUMP CENTER as defined in API 610 and considering that point as ANCHOR, something in my opinion is not correct). Has anybody seen any value of the pump nozzle stiffnesses furnished by vendor and share that information with us? This can give us a feel of the "actual stiffness" values that really exists. However, none of us should start using those values w/o consulting with individual vendors.

3) The reason why I mentioned 1 and 2 is that I feel that the pump nozzle loads that we actually compute are by and large overestimated and hence we provide all sorts of things, bad layout ( overlooking problem with flow induced vibration and NPSH related issues), queer supporting which are impractical, difficult to fabricate and on many occasiosn overlooking the access and mainenance problems. There can be several reasons for this overestimation: Ultraconservative Line designation table informations, ultraconservative temperature profile assumptions ,high restraint stiffness values, conservative assumption about ambient temp. ( 21C installation temp. may be ok with European site conditions, but not in tropics like India, Middle east)etc.
I have seen many real life pumping systems about which I have not heard major complaints from the maintenence personnels and if we model those pumps in a computer with the actual supporting as is on site , they are 4-5 times API 610 allowables but workinging perfectly fine.

4) As mentioned in a paper by L C Peng that when the loads on the pump nozzles increase, by and large the solution is to thicken the base plate and that in my opinion should not go for high cost escalation.So must we go for meeting the allowables or negotiate with the vendor for reduction in price and avoid excess flexibility,bad supprting etc ?

I would like to get your opinion on this.

Regards

As the pump piping stress analysis is done on an existing installation, one can consider other input parameters at resonable observed values - viz. maximum operating temperature from recorded past data instead of design temperature from LDT, standby pumps actual temperature instead of 21 C, correct friction coefficient, gaps in restraints etc.

With proper layout, low friction supports, properly placed restraints &/or spring supports one can control external forces & moment load on pump nozzles. As respected John Breen pointed out, when we exceed the recommended limits of vendors, maintenance problems arise. Owner pays - either upfront to the A/E following best practises or to the maintenance for the whole life.

What is the use of exceeding the allowables using FEA or justification - after all, we design pipings for rotating machines for good operability, not to test whether it fails or not at limiting condition!

regards,

sam

I'm not familiar with WRC449, but I would say it is bad advice to tell a stress analyst to get stiffness values from a pump vendor to include in a stress analysis of a pump piping systems.

As John points out, those allowables we get from API610 are not limits on the stress capability of the casing - they are load combinations that, by whatever black magic the pump manufactures have come up with, will insure that the shaft deflections are not excessive to the point of premature bearing failure.

I doubt the pump vendors had it in mind that stress engineers would be putting flexible casing in their models. If you think there is an issue with where we place the fixed anchor of the pump to account for thermal expansion of the nozzle, just imagine the can of worms that you would be opening if you tried to determine just how a particular pump casing would deflect under the loads you apply to the nozzle face.

Is there some conservatism is the approach of assuming rigid pump cases? Certainly. Ultimately we have to draw a line between getting a design completed in a time fashion vs. optimizing to the nth degree. In the real world, time is also money and our perfect computer layouts are going to be installed with no better than +/- 1/4" tolerances. They need to be robust enough to so that after they've been manhandled by pipefitters, welders, and millwrights the piping systems will actually perform its intended function.