include mass of fluid in pump nat freq calcs? 1

[electricpete]

We are attempting to do a quick simple estimation of change in natural frequency of a vertical pump as a result of adding mass along the stationary parts below the deck mounting level (to demonstrate no significant change in natural frequency from small fractional addition of mass).

The structure is treated as cantilevered about the deck mounting level. The envisioned modeshape is simple cantilever shape with natural frequency wn = 3.53*sqrt(E*I / [mu*L^4] where mu is mass per length (Den Hartog). Therefore the simplistic calculation is that the natural frequency will change inversely as the square root of mu.
i.e.: Ffinal / Finitial = ~ sqrt(mu_initial/mu_final).

The question: should the mass-per-length parameter mu include weight of the fluid, or only dry weight of the pump?


=====================================
(2B)+(2B)' ?

 

[GregLocock]

if the casing and rotor contain the fluid, yes. If the pump is an immersed one in a large tank then the answer is somewhat.

Fluid structure interaction is a warmish topic in the dynamics community, mostly handled by multiphysics simulations at the moment.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[electricpete]

Thanks Greg. The pumps we are interesting in include water enclosed within casing (not a large tank).

Just to check for understanding, in this case it is a relatively straightforward conclusion to simply include the mass of the water in the calc (lump it with the casing in our situation which does not distinguish rotor from casing).

What if we were attempting to model rotor, casing, bearings.... would we still include water mass with the casing?


=====================================
(2B)+(2B)' ?

 

[ivymike]

I'd say that depends on what modes you're interested in. If only "external" ones, then it probably doesn't much matter.

Some people seem to like modeling pump bodies as completely rigid in a modal analysis of pump supporting brackets - that does not always give good results.

 

[izax1]

And what is the water mass? You need to include only the mass attached to the structure (from the boundary layers) during that particular mode shape. (Effecive mass) which will change for every mode shape. I would say, only complex multiphysics calculations (fluid structure interaction) can solve (or approach) the problem.

Some years back I was involved in a project of subsea air guns. Very much dependent on the correct modelling of FSI.

 

[GregLocock]

True, but in the simple case of a water filled pendulum bob you don't need to fret about it, and from what electricpete has written it sounds like that is a reasonable approximation to make. If the cg of the fluid participates in the mode then it can be lumped.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[electricpete]

You need to include only the mass attached to the structure (from the boundary layers) during that particular mode shape

The mention of boundary layers brings to mind a casing vibrating in the axial direction (is that what you meant?). The vibration of interest to me would be perpendicular to the axis of the casing.

=====================================
(2B)+(2B)' ?

 

[izax1]

OK. If we are only talking about water within the casing and you are only interested in the casing fundamental natural frequency, then you could assume the weight of the total enclosed water is contributing to the natural frequency (transverse or longitudinal) For a quick, simple estimate, that will do.
On the other hand, if you want to find the natual frequency of the rotor, bearing etc, that would require a FSI approach. (Vibration of structure entrapped in a fluid.)