Non-linear buckling 2

[loom]

I am doing an eigen buckling analysis of a plate followed by non-linear buckling. I am using upgeom to provide an imperfection. The problem is with different amplitude factors for upgeom, my buckling load changes significantly. The thickness of my plate is 10mm so I thought a unit factor is good enough. However, when I increase the factor to 2 (upgeom,2) buckling load increase by about 50%. Do you have any ideas on this issue? By the way, stress stiffening is off. I think that, with more imperfections, the buckling load is supposed to decrease. Thanks in advance.

 

[TGS4]

Please provide more information regarding your non-linear buckling problem:
1) nlgeom - on or off
2) plasticity - on or off
3) Are you using the arc-length method?
4) Any other information about your problem - I'd rather wade through unnecessary information and discard it rather than keep asking for the information that would help.

 

[loom]

Sorry about that.

1. nlgeom is on. Non-linear buckling.
2. No plasticity.
3. No arc-length method.

autots is on.
lnsrch is on.
shell 63 used with keyopt(2)=1 with consistent tangent stiffness matrix.

Thanks again.

 

[TGS4]

Does the same effect happen if you use SHELL93?
I would also turn stress stiffening on.

Let us knwo hwo this turns out - I'm very interested.

 

[Drej]

I think, generally speaking, you're correct in saying that if you increase the imperfection, the critical load should reduce. But to see 50% increase in the load suggests something wrong either in your assumptions or in the UPGEOM transfer or...

More questions:

[ol]
[li]You say you're using UPGEOM to provide the imperfection, but where are you taking this imperfection from? For example, did you run your eigenvalue analysis and use UPGEOM from the eigenvalue "displacements"? Or what?[/li]

[li]When you issue UPGEOM, does the "undeformed" structure in your non-linear analysis make sense? Have you used the same db for both models?[/li]

[li]How is the plate loaded? Is it a pure in-plane load?[/li]

[li]What is the geometry of the plate? Is it continuous? Rectanglular? [/li]

[li]Why no plasticity involved in the non-linear buckling analysis?[/li]

[/ol]

Cheers.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com

 

[loom]

Thanks for the reply Drej.

1. Yes, I ran the eigen value analysis and used the displacements to update the geometry.

2. Yes. The buckled shape is as expected. The loading and end conditions are a bit complex but for the same plate and simpler loading the eigen value results compares favourably with Roark's formulae.

3. It is compression and shear in the plane of the plate.

4. Plate is a simple rectangle of size 1.7m*1m with non-uniform loads. It is simply supported on three sides, plus other restraints on the other sides.

5. Well, simply I am not allowed to go to the plastic range.

After further work on the problem, I think that the factor for upgeom should be of the order 0.002-0.05. These values give consistent results. When I apply a small perturbation load instead of upgeom I get similar answers to upgeom,0.002 or 0.05. Furthermore, eigen value solution also gives results in the same ballpark with upgeom,0.002 or 0.05, though slightly higher (which is expected).

With upgeom,1 , the plate is a shallow arch and will not buckle the way I expect. I guess this explains the discrepencies.

Please let me know what you think. Thanks in advance.

 

[Drej]

Yep, makes sense. I wouldn't want to use the actual eigenvectors for the deformed structure (I guess these are probably normalised to the mass matrix anyway, so they're pretty meaningless), but I would be okay with using scaled values of these to some sensible level like you say.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com

 

[loom]

Just to update the followers of this thread. The in-plane rotational stiffness is creating singularities. This can be fixed by using Allman stiffness or switching to 93's. This way the solution is not very much dependant on the upgeom amplitude as well. Thanks Drej, TGS4.