I have a few thoughts:
1 - Double check your boundary conditions to make sure that you do not have unconstrained degrees of freedom. This seems to give lots of different errors including "Singular Matrix". Does the buckling analysis have the same restraints as the static analysis?
2 - If I remember correctly, a fine mesh increases stress accuracy, but has little affect on deflections. I believe that a course mesh would be fine/preferable for a buckling analysis.
The buckling analysis has the same constraints as the static analysis that ran.
I have tried the analysis with a finer and courser mesh and that did not help.
I am going through the model to try and see if I can identify a particular surface that is the problem. I have tried to simplify the model as much as possible to establish what is causing the error.
Are there any limits on model size for shell buckling? A limit on number of elements/nodes?
I have never tried to do a buckling analysis with shell elements, but your idea of checking the surfaces sounds like a good idea. Along similar lines, you might try making an experimental model with the same element type, and boundary conditions, but with grossly oversimplified geometry. If it works, then you know the problem is geometry related, but if it still doesn't work, then you will know that you need look somewhere else.
Maybe someone else will be more helpful.