I have a long cylindrical shell with dimensions length = 20 m , radius = 2 m and thickness = 3 cm. I want to do a geometric non-linear buckling analysis in Abaqus with no imperfections included. Riks method is used as a solver. However , the solution seem to diverges and after few steps my job is aborted. Any ideas why is this happening?
Note: I upload a figure of satus file. Does the appearence of first 1U at 0.766 LPF mean that buckling happens there?Also why load factor is higher than 1?
In the case that there are no initial imperfections (geometrical or load offsets) your long part under axial compression loads will lose its stability at some level of load.
At this moment of stability loss (Buckling point), the solver hits a singularity, at which it may continue and so gives no sense results (approaching machine infinity), or it may just collapse and labor (this is the one that happens to you).
In other words, starting from a perfect model (no imperfections), you can study the behavior of your long part for loads less than a critical load value (buckling load).
You can specify that critical load by conducting an eigenvalue (Buckling) analysis on your model.
In addition, if you are interested in the behavior of the model under loads beyond the critical value (also called the post-buckling analysis), you have to start with some tiny imperfection. The tiny small imperfection would not really change the behavior of the model beneath the critical value. But when the load gets close to the critical value the model response considerably increases.
***********************************************************
Please consult any book on buckling analysis, it will clarify more a lot.
You can also see the following paper which shows clearly the need for the initial imperfection and the effect of its size on the total behavior of the model. In addition, this article elaborates on the special concerns and the correct use of boundary conditions and symmetry modeling in buckling analysis, and more.
Article Modeling and Buckling Analysis of Rectangular Plates in ANSYS
Nihat Can
Ahmed Hassan Ahmed Hassan I applied the critical load which i took from a linear buckling analysis. The solver stops around 1.05 of lpf but i was expected to stop at 1 or a little bit less. Does this 0.05 difference occur from the fact that the solver hits the singularity there and try to deal with it?
Also from message file I saw that negative eigenvalues appear for first time at iteration 10,.Shouldn't the proccedure throw an error there , since the stiffness matrix is not positive definite?
I think this is the mean of instability. You start seeing non-sense results and numbers.
The problem doesn't need to happen only at the exact singularity; just being close to the singular point is enough to shake the numerical model.
As I said, you have to have some kind of imperfection in order to pass beyond the critical point.
- Badges
- Science topic