I was wondering if you had any experience comparing Abaqus vs. Nastran linear eigenvalue results (buckling). I have noticed that with something like a stiffened panel of seemingly fair mesh density, Nastran will capture all the pocket buckling modes while Abaqus will not. The elements seem to be the same formulation (i.e. S4 or S3), and Lanczos solver is being used for both. The Nastran results show many pocket modes before what would be considered a global mode. Abaqus results show no pocket modes at all. Any thoughts?
Actually, I think there is a difference in the CQUAD4 and S4 element formulations that is causing this. I just cant figure out what that difference is based on the research I've performed so far.
Again, I don't know about Abaqus. In Femap/Nastran it is about both the element shape (CQUAD4) and property - which can be basically shear or plate (membrane). A plate aproach will show "pocket modes" for unstiffened patches; membrane will not.
The name S4 in Abaqus might suggest Shear? In Femap generally plate is used.
But this is mostly guessing by now. Show some pictures of eigenmodes, property settings and maybe you will get a more qualified answer. And the most important, do the frequencies of the global modes correlate? They should, no mattter whether shear or membrane ...