Modal Analysis of Pump (bearing modeling)

[fer72]

Hi,
I have done modal analysis of vertical pump. The problem becomes when I compare the results of my model with the results provided by the manufacturer 15 years ago.
Basically in my model the 1st frequency of the shaft is twice the frequency given by the manufacturer. I can get this result modelling the shaft supported in two points, but Idon’t think that this is a good approach. I model the shaft inside the pump using beams and them attach the beam to the bearings using two points per bearing, one at start and another one at the end, releasing the displacements and rotations in these points in the axial direction except for the trust bearing.
Has anyone model a journal bearing this way? How were the results compared to frequencies measured in service? Any other options to model the bearing without taken into account tribology?

 

[Tmoose]

Housing and support stiffness is often so low that it overpowers bearing stiffness.

 

[zjliang]

fer72:

Actually I did some frequency analyses for aircoolers in the past 3 years by modelling the shaft, bearing supports, cooling fan assembly and motor. The FEM results are quite close to the hand calculation for the base lateral modes.

My experience is boundary condition modelling is very important. Besides, before the FE analysis you have to estimate/predict what kind of mode shapes will dominate, what is the dominant mass. In my analysis, motor is much heavier than other parts, so the C.G of motor has to be simulated as accurate as possible. The most flexible part is fan shaft, so the span (or supporting distance) has to be correctly input.

Besides, before the analysis you'd better know, at least make some reasonable guesses.

If you can let me know more details of your calcs, maybe i can give you more advice.

Hope it will help.

 

[fer72]

zjliang,

the characteristics of the pump shaft are:
Shaft length between bearings: 3 m
Shaft diameter: 250 mm
Bearing length: 260 mm
The electric motor shaft (10.000 kW) is coupled to this shaft by a joint which allows deflections.

Thanks for your comments.

 

[zjliang]

fer72:

As per your dimension info, I estimate that the effect of the modeling bearing support length on the frequency shall not be neglected. So, point support without modeling the exact contact length will cause substential erro.

Besides, the D/L ratio is so high that ignoring the later shear deflection could make the frequency calculation results far apart from the actual value, could be too rigid to model it as a beam. Please check the element you are using in your FE calc, make sure the shear coefficient has been taken into consideration.

Good luck.

 

[electricpete]

10.000 kW - is that 10 kw or ten thousand kW?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

I'm guessing 10,000 from the dimensions. I'm just not used to used to that notation.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.