Natural frequency analysis: please help

[encocrema]

The structures are correctly bounded but the message is:

Error: Eigensolver failure.
Possible rigid body modes.

In some analysis I have this message if I choose a number of modes to calculate greater than "n".

For example:
-Model 1: n=12
-Model 2: n=9
-Model 3: n=1
-Model 4: n=0!!!

 

[UcfSE]

Rigid body modes means you don't have enough supports to keep your structure from moving freely or rotating freely. You may be missing a restraint about a longitudinal axis somewhere. That is a common error in 3D models (for structures). Check all your boundary conditions and supports to ensure that no member is free to move or rotate.

 

[NS4U]

Rigid body modes result when the stiffness matrix is singular. (det(K)=0).

Are there restraints in your system that you could stiffen? So that rigid body motion is restrained.

 

[271828]

UcfSE is right of course.

I do not understand what your "n" represents. If it's the number of modes, why is it outputting "Mode 1: n=12" etc?

 

[encocrema]

Model 1, model 2...are different structures

 

[271828]

So n=0 indicates how many modes it found?

It can also find no modes if there is no mass that can be in motion. For example, the other day I had a shell structure that I forgot to automesh. All the shell mass was attributed to the corners which wouldn't move.

 

[encocrema]

Why does the analysys finish correctly for n° of modes<"n"?

 

[Castigliano]

Several reasons could be causing the program to behave as you've described. As mentioned above, unrestrained rigid body modes are most often the cause. For the models that give you some number of modes before giving the error message, you may have asked for more modes than there are DOF's. This would apply if it was a relatively small model. Do the results for the outputted modes make sense?

Castigliano

 

[271828]

I don't understand: "Why does the analysis finish correctly for n° of modes<'n'?"

Are you saying that you tell it to find 12 modes for your first model and it works ok, but then you ask for 13 and it doesn't? Like Castigliano typed, this could be due to number of possible modes for your particular model. For example, if you have a flagpole model with all the mass at one point near the top, you could only have a small number of modes. If you ask for 12, you'd probably get an error.

 

[Qshake]

What is the program that you're using and has it been checked thoroughly for convergence matters?

If you're using a commercially available program then finding higher modes should not be a problem and should not give an error message.

It will give an error message for rigid body modes though.

Regards,
Qshake

Eng-Tips Forums:Real Solutions for Real Problems Really Quick.