Snap-thru in I-DEAS 10 NX

[iam92008]

Hello All,

I'm modelling large deflections of a thin-walled cylinder (0.25m dia, 0.1m long, 0.005m thick) under symmetric 'pinching' forces (as if the cylinder was squeezed between two platens radially). The nonlinear (geometric and material) statics analyses I've attempted seem to get hung up (ie. stops at load step 25 of 40) as the cross-section deforms into a flat elliptical shape.

Question: Could the problem be that the geometry is experiencing a snap-thru?
Follow-up question 1: If yes to question above, should I perform a linear (I_DEAS 10 NX does not have non-linear buckling capabilities) buckling analysis to find the critical load and buckled shape?
Follow-up question 2: How do I save the geometric and FEM deformed model resulting from the buckling analysis so that I can continue the non-linear statics analysis?

Thanks for your help.

Regards,

Peter.

 

[feadude]

In IDEAS there is no analysis type called nonlinear buckling , but you can use geometry nonlinear
and watch the stiffness parameter. When it approaches 0. you are at a critical load.

You can also use the arc length solution method to solve post buckling or snap thru problems.

a geometric nonlinear analysis use the almansi strain definition ( up to 10% strain, so watch for ur strain not to go over 10%)

 

[iam92008]

feadude,

Thanks for the response.

I'm interested in the post snap-thru behavior of the structure described above. A colleague suggested using the linear buckling analysis of I-DEAS to find the first mode shape, save the geometry and fem (I haven't figured out how to do this) of this deformed shape, then perform the nonlinear statics on this deformed model.

However, according to the I-DEAS help, linear buckling assumes that deflections before buckling are small. In this case, deformations are large before buckling. From this, I assume my colleague's suggestions may not work.

Please tell me if I understand your suggestion correctly: I should use nonlinear statics with the loading method 'Adaptive Load' and watch the stiffness parameter as I increase load. When this parameter is close to zero, I should switch to the loading method 'Arc-Length Control' and this will get the structure past snap-thru and into the post-buckling regime. Is my understanding correct? Also, what is this stiffness parameter and how do I monitor it?

Thank you.

 

[feadude]

arc length control, adaptive control, load control, etc are only solution options. They only determine how you increase the load on the structure , ie load step, load increment. I use the load control loading method.
Watch your stiffness parameter numbers in the IDEAS list window when you are solving.
hope this helps

 

[izax1]

I would be very careful in trusting FEA buckling analysis answers. Since this is an instability, very small variations in boundary conditions or geometry will have large impact on the real behaviour. And those small variations you will not be able to model with FEA. Your manufacturing procedure will never be as perfect as your FE model. We have experiemced 10-15 times lower buckling loads from test than what we get from FE analysis.

The best you can do is to do your analysis with a band of tolerances on geometry and boundary conditions.

 

[feadude]

izax1
I guess one of the geometry conditions u are talking about is eccentricity. Did u do linear or nonlinear analysis?
Can u not do linear and then nonlinear to improve/verify geometry , boundary conditions, etc. If I understand u are saying if analyses gave you say buckling critical load of 10-15 lbs, experimental buckling load was only 1 lb?
ASME CODE SECTION VIII gives buckling critical loads or stresses based on cylinder eccentricity ie being out of round, etc. I do not think they are orders of 10-15. It has been several years since I have usded that code. Know you got me wondering because I have seen I beam compression flange failures (possibly buckling) at lower loads than FEA results.
Can you or have u modeled eccentricity into unloaded geometry?
thank you

 

[izax1]

Yes, eccentricity is one. Thickness variations is another.

Our early analysis was linear. Now we use non-linear analysis with ABAQUS, but still the initial FE-geometry is perfect. In fact, as far as I understand, DnV (Det norske Veritas) will not accept buckling results from FE analysis. Only using their own standards. And ASME VIII is extremely conservative for buckling (external pressures). We have checked ASME Code against FE results, and again we see huge descrepancy.