Surely option 2 would better represent operational conditions and the additional stiffness would give you more accurite mode shapes/NF's?

Gary

It depends upon what you want from the test.

If you want to verify a FE model, a free support system of the impellor might be good.

If you want to know the impellor modes mounted stationary to a shaft, fix the hub. How you support the rotor, or if you have a rotor other than a hub may not have much importance.

If you need to look at rotor modes (stationary) to compare to a model, hang the rotor; generally it is easier to obtain a reasonable approximation to the free-free boundary condition. Then the rotor model can be checked.

Of course you are doing this without spinning. This will effect the modal frequencies in real life. Impellors can loose their shrink or part of it (often one end of the shrink will be lost to some degree). Ever see the Campbell paper to see how he modified his static results for spinning using Southwell's Theorem?

Thanks guys for your responses.

No, I am not doing an EMA in order to compare the results with FEA. I think EMA can stand on its own. I would like to use EMA to prove that the crack pattern developed in the impeller is due to the excitation of a specific modal frequency.

The drawback of an EMA test however is that it is static. A non spinning impeller will have some modal parameters affected by additional stiffness due to spinning. I guess Southwell's Theorem talks specifically about that.

I'll appreciate if you can provide a reference.

Thanks

You can find a brief entry in Hartog's book on vibration. Campbell's paper uses it.

As I surfed the literature, it appears that during the modal test one would want to replicate as close as possible the actual operating conditions of an object.

For instance, an aircraft should be suspended using soft ropes; on another hand, a heavy impeller should NOT be tested in soft suspenders, but rather with the shaft in the bearings which approximates fixed points condition. The softness of the ropes and stiffness of the attachment should meet certain condition for accuracy, of course.