rotor frequencies 1

[suviuuno]

Greetings,
I am getting peculiar results on rotor frequencies. Adding constraints to two bearing locations of the rotor lowers natural frequencies. One would expect to have higher frequencies as compared to non-constrained rotor. I am not considering rigid mody modes. I am using beam and shell elements. Could the problem be in mass matrix formulation, lumped vs. consistent? I am using Patran. Any ideas?

Thanks

 

[gwolf]

Suviuuno,

Without looking at your model I am not 100% sure but I think the reason is this:

Consider a beam with no restraints and find the natural frequencies - when you look at the mode shape of mode 7, the first bending mode you will see that there are two stationary points at about 0.25 and 0.75 of the beam length, not at the ends. The effective half wavelength is about 0.5xbeam length.

Now do the same with the beam pin-ended at both ends. The first natural frequency will be lower and will have a half wavelength of 1.0xbeam length.

By constraining the rotor at both ends you force it into a longer wavelength and lower the frequency. I surprised myself with this a few weeks ago on a compressor vane.

Maybe someone else will post and give a more mathematically rigourous explanation, or correct me.

 

[suviuuno]

gwolf,

Thanks for your input. I had actually tried with simple model but failed to notice detail you mentioned and it does make sense. Something still bothers, generally if I add constraints I am making structure stiffer thus higher frequencies. Of course, there are always exceptions to the rules.

Thanks

 

[gwolf]

Generally yes but in this case, and surprisingly, no. Your model is fine, it just didn't do what you expected - I was really surprised when my compressor vane did that, the secret is in looking at the mode shapes and considering the wave length. Anyway, problem solved I think.

 

[cloche]

As far as I know, adding constraints rises up the nat freq only if it doen't "cancel" a modeshape. It seems it's your case, so that you obtain something like two or more "uncoupled" systems where at least one now has a natural frequency below the previous ones. Most likely, this new modeshape has a lower (or very lower...) mass participation factor than the first of the "unconstrained" model...