Hyjal
Mechanical
- Apr 20, 2012
- 2
Hi all,
I am fairly new to Abaqus. I'm having trouble with a rather complex structure consisting of slender, curved beams that are subjected to large deflections. So it seems reasonable as a first attempt, to model the beams using beam elements and constrain them together using kinematic coupling.
In order to get a better understanding about the different types of beam elements and their integration methods i modeled a simple one-sidedly clamped cantilever and then compared the results to the ones of a very fine solid mesh.
Here is the problem: In my simulations, beam element structures under large deformations seem to be quite unstable. I've tried everything, like using hybrid elements (B31H as recommened for this situation by the Abaqus manual). My loading case doesn't involve any structural instabilities (like surpassing the bifurcation point). My step is set to "nonlinear geometry" etc. I'm always getting tonns of negative eigenvalues. It seems like for a coarse mesh sometimes all works well, but the finer i make the mesh (still not "ridiculously" fine) it gets unstable.
Now if I model an initially curved beam (all done in MATLAB with correct normal definitions, so the automatic averaging algorithms won't induce initial twists or curvatures in the individual elements) and try to "straighten" it by pulling it apart, the numerics get even worse and in case of a rather fine mesh, I can't get any convergence after a few increments.
So if someone is experianced in modeling large deflections of curved beam structures I would be eternally greatful for some help on how to approach this.
Greetings from Switzerland,
Kay
I am fairly new to Abaqus. I'm having trouble with a rather complex structure consisting of slender, curved beams that are subjected to large deflections. So it seems reasonable as a first attempt, to model the beams using beam elements and constrain them together using kinematic coupling.
In order to get a better understanding about the different types of beam elements and their integration methods i modeled a simple one-sidedly clamped cantilever and then compared the results to the ones of a very fine solid mesh.
Here is the problem: In my simulations, beam element structures under large deformations seem to be quite unstable. I've tried everything, like using hybrid elements (B31H as recommened for this situation by the Abaqus manual). My loading case doesn't involve any structural instabilities (like surpassing the bifurcation point). My step is set to "nonlinear geometry" etc. I'm always getting tonns of negative eigenvalues. It seems like for a coarse mesh sometimes all works well, but the finer i make the mesh (still not "ridiculously" fine) it gets unstable.
Now if I model an initially curved beam (all done in MATLAB with correct normal definitions, so the automatic averaging algorithms won't induce initial twists or curvatures in the individual elements) and try to "straighten" it by pulling it apart, the numerics get even worse and in case of a rather fine mesh, I can't get any convergence after a few increments.
So if someone is experianced in modeling large deflections of curved beam structures I would be eternally greatful for some help on how to approach this.
Greetings from Switzerland,
Kay