Modal analysis on nonlinear material

[dshaffer1001]

I have a certain amount of experience doing modal analyses and other vibrational analyses on metal structures using NX NASTRAN. However, my company is evaluating a proposal to do modal analyses on some composite parts.

Their material is highly nonlinear, basically plastic. Can anybody enlighten me about the sorts of issues this might cause ?

It seems to me that the material's vibrational modes would also be nonlinear, and would depend on amplitude as well as frequency. Am I wrong ?

 

[EdR]

I don't believe mode shapes for nonlinear materials (structures) have any meaning......i.e. as you said they are dependent on "displacements" which are nonlinear.

If you are interested in the dynamic response of this type of structure I think you will have to do a nonlinear dynamic analysis using direct integration methods to obtain meaningful results......

Ed.R.

 

[spongebob007]

The eigenvalue soultion methods do not support a nonlinear stiffness matrix.

 

[GBor]

I have done modal analyses on composite materials, but you have to assume that displacements are sufficiently small to stay within the linear region of the stress-strain curve. Basically, you can run it as an orthotropic plate "smearing" the layered properties into a single panel. If it is a cored composite...it becomes a little more difficult.

 

[JoshPlumSE]

When the vibration is happening at very low amplitudes (and is essentially elastic), you will have something aking to a a natrual frequency and mode shape. But, that mode shape and natural frequency would become invalid as the magnitude of the displacement increases. Or, more accurately, as the material begins to behave non-linearly.

I mention this only to point out that a modal analysis can still have something to offer about the behavior of this struture when subjected to these lower force levels. It's limited, but it can still be useful.

 

[gwolf2]

As others have said, it's OK to use the linear assumption for the hookian part of the plastic's stress-strain curve, ie if it is only moderately loaded.

If you are trying to predict frequencies to avoid then this approach is perfectly acceptable.

If you get into a scenario where there is some resonance, however small, then watch out. Even if the resonance is small to begin with because the damping is good, the material will begin to heat up. When the material heats up, the material will soften and it's modulus and natural frequency will drop.

If you are lucky the softening will take its frequency away from the driving frequency and it will calm down and cool down. If you are unlucky and the reduction in modulus takes it nearer the driving frequency then it will vibrate more, heat up more, soften more and actively hunt down the driving frequency. Soon after it will dissintegrate from either excessive strains or thermal degradation.

gwolf

 

[dshaffer1001]

Thanks to everybody for the info.

My problem is further complicated by the fact that some of these panels are lying on a rigid floor, and are being deflected out of plane. Even to do a static analysis, I'd have to do it nonlinear with contact surfaces, etc.

It looks like a plain old Eigenvalue analysis is out the window.

 

[gwolf2]

Try to simplify it by making conservative assumptions, you may be able to drag it back to a more simple analysis which is not reality but definitely conservative.