Non-Linear Buckling Analysis - Solution Convergence Issue

  • Last Post 27 January 2019
Sreenath1994 posted this 26 January 2019

Respected Sir, 

I am having an issue of solution non-convergence when trying to carry out the non-linear buckling analysis of two rectangular plates (10mm and 20mm) , simply supported on all 4 edges. As soon as the material gets fully yielded (250MPa) the solution has problems converging. Material plastic behaviour was defined as bilinear isotropic hardening. 

Can somebody please explain why the solver can't converge on the solution as soon as the entire material gets yielded although I have defined the tangent modulus.

I am on student license.

Thank you

Attached Files

Order By: Standard | Newest | Votes
jj77 posted this 26 January 2019

Ok, I mean you have passed and we can see your buckling/collapse/softening load on most of these curves, there is not much point pushing it down in the softening part because you can already see that it is softening (or going flat), thus collapsing and the solver needs to cut the loads back to get equilibrium. 


First plate buckles, the thicker one goes all plastic first (and buckling load/stress is the same as yield).


Many times (especially for these type of analysis), it can be easier to use an enforced displacement instead of a Force, and then just look at the reaction there to get the applied load.


Also arc length sub step technique can help if you have some complex type of snap through buckling, but I do not think it is the case here.


  • Liked by
  • Sreenath1994
Sreenath1994 posted this 26 January 2019

Respected Sir,

Thank you for the quick reply. 

My doubt is why the solver cant calculate beyond the yield stress even tough I have given the tangent modulus. The solver logic will bisect the load to give the appropriate increment right?

For completing my work I have to get the plate stressed beyond the yield point.

So instead of the giving a force should I give a displacement instead. Will the non-convergence issue be resolved then.

jj77 posted this 26 January 2019

You could try that.


If though all the plate is yielded (hinge), then it will be difficult for the solver to go past, since it is difficult to get equilibrium.


Try enforce displacement and gradually increment then we control the solution so even when you pass yield, go slow, but it might as I said be, that there is too much plasticity.

I will try this next week on Strand7 which I use, and see how it gets (for the 20 mm plate which is more tricky).

Sreenath1994 posted this 26 January 2019

 I will try to enforce displacement instead of the force and post the updates.

Thank You

jj77 posted this 26 January 2019

Also make sure you have a deformed shape for the 20 mm plate, so you trigger buckling, and so you are not only compressing it.

jj77 posted this 27 January 2019

Strand7 gives this pressure vs deflection curve, failing at 2.5E8 Pa (2.5 E6 N, times two since I have symmetry). So it looks as we said that it fails in buckling and the softens due to the yielding (hence down slope after failure).

Used arc length, but disp. as mentioned should be OK. Getting same results with ansys and arclen method (apdl command in wb to activate arc length, also used quarter sym, and line pressure load)