1
$\begingroup$

I am currently trying to model the buckling behaviour of a cylinder using sol 400 nonlinear static analysis using Nastran. I am using the arc-length (modified Riks) method as apparently it can increment load and displacement in both positive and negative directions, as opposed to Newtons method which only increases the load or displacement increment. However, my problem is that the analysis always stops after a certain load due to convergence problems.

I have tried changing many of the parameters such as MINALR/MAXALR ( min/max arc-length ratio between increments) and MXINC(max increments), however I still don't get any good results.

I feel like I am missing a parameter that enables the load to decrease after a maximum load has been reached.

I am aware that load control and displacement control is possible, but my understanding is that arc-length method uses neither of these, and is sort of a combination of the two. My model has a force applied without any enforced displacement. Is this maybe the problem?

Any other ideas which might help me?

Thanks in advance.

$\endgroup$
1
$\begingroup$

If your model is very symmetrical (e.g. a circular cylinder with an axial compressive load, modeled with a regular grid of elements not an automatically generated irregular grid) you can hit problems because when you reach the instability, there are several possible unstable paths and no way for the program to choose between them. (In the circular cylinder example, the cylinder could theoretically buckle in any lateral direction.)

One solution is to make the model unsymmetrical, so it prefers to buckle in one particular direction. For example, apply a small side load in the direction you want it to buckle.

| improve this answer | |
$\endgroup$
  • $\begingroup$ What you say makes sense, but I still get convergence problems adding the side load. However, I can at least now see load decreasing for certain values side load and increasing MAXBIS (maximum number of bisection per increment). So thats a step closer I guess. $\endgroup$ – Jeroen Jan 24 at 11:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.