Singular matrix for buckling but not static analysis

jte · Nov 7, 2006

Hello folks-

This is my first post in this forum (I'm active in several other eng-tips forums) so just for background, I've been working with Cosmos for well over 10 years (using the Geostar pre/post processor) and have recently started using SW.

I'm a bit perplexed by a model that I'm working with. If you can imagine an oversize 20# propane tank, thats about what I'm modeling. I'm running a simple gravity load on a model I built and meshed using shell elements. The issue I'm evaluating is how big of a hole I can tolerate in the support skirt on the bottom, so I'm running a buckling analysis. The models run fine in the stress analysis mode and result in about the stresses I'd expect. However, when I switch to the buckling study it kicks it out and claims that I have a singular matrix. I don't quite understand how this can be since the same mesh and boundary conditions work for the stress run. I figure I'm just missing some button to press or extra value to add that I'm missing due to my newbie factor with SW. Any ideas?

jt

gfbotha · Nov 8, 2006

Are you modelling the complete 360° of the tank?

Could it be that you actually do not have enough constraints and may be used something like soft spring stabilizing whilst running the stress study, but not so in the case of the buckling study?

jte · Nov 8, 2006

gfbotha-

Yes, I'm modeling the entire tank. Forgot the exact size but something like 10' diameter x 30' high with an 8' skirt. No symmetry. The bottom of the skirt is fixed (displacements and rotations). I should have identical loads and BC's on both the stress and buckling studies. Not using soft springs.

This is a fairly simple model and I might just go back to use the Geostar pre/post processor. Not as pretty, but it'll get the job done.

jt

gfbotha · Nov 9, 2006

Jte, just thought - not sure whether Geostar would change anything because depending what versions you have loaded, it is likely the solver used are still the same...

You could also try out/select the alternative solver (under study properties). Or, play around with the number of buckling modes requested...

jte · Nov 10, 2006

gfbotha-

I guess I'm leaning towards Geostar since although it uses the same solvers, it only works on one FEA mesh while SW seems to work on different meshes (though the parameters may be copied) for different studies. I've never seen a model in Geostar which worked with a static analysis but kicked out of a buckling analysis with a singular matrix issue.

In SW I tried both solver options. Didn't consider the number of modes requested. Although it doesn't make sense to me that it should have any effect, I'll give it a shot. I've learned long ago that not everything in any software makes sense! Thanks for the idea.

jt

gfbotha · Nov 12, 2006

Glad you appreciate why I mentioned the number of modes (of course it does not make sense, but as you said...) E.g. I have seen Cosmos DesignStar leaving the party when the analysis exceeded a certain size.

Please let us know if you find a way to succeed!

jte · Nov 13, 2006

Well, I cut my requested number of modes down from 5 to 3 and... wait, wait... it ran! But only on the first model I tried it with. The second one (different hole size) crashed out with the same error.

jt

gfbotha · Nov 14, 2006

Usually the lowest mode is sufficient because it is associated with the lowest critical load.

I just checked my e-manuals: the FFEPlus solver can calculate one buckling mode only...

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

Singular matrix for buckling but not static analysis

jte

Mechanical

Recommended for you

gfbotha

Mechanical

jte

Mechanical

gfbotha

Mechanical

jte

Mechanical

gfbotha

Mechanical

jte

Mechanical

gfbotha

Mechanical

Similar threads

Part and Inventory Search

Sponsor