The "shells are not part of 2D geometry" is a red herring... not a problem. The error should actually say something more like "some surfaces are not defined for meshing and will not be included". The curves I needed to define the tank boundaries are the outer edges of surfaces that I created as "Fills". So the "warning" occurs because of my construction technique (which is highly stable for design and change purposes). All of the surfaces that I DO want meshed as "solids" represent stiffener beam cross-sections which you'll see at various places in the model.
I spent ~ 6 hours on this yesterday, can't sleuth anymore to help (go on vacation next week), BUT... The problem started to occur when I took OPT1 (whose "design limits" had all of my "shell" measures for "failure index") and copied it to OPT2 where I ADDED all of the "BM/beam" measures so that I could optimize the beam cross-sections and reduce the shell thicknesses. After the copy corrupted, I deleted it and created another optimization by the exact same name... but manually added all the required measures and parameters fresh. That still corrupted. I can't spend more time. I sure hope this helps.