I use COSMOS/M - the main reason I use this one is that when I upgraded to the latest version, none of the SolidWorks versions (or DesignStar) were capable of running composite models (multi-layer shells and solids). Each time you wanted to do so you had to go back to GEOSTAR and then do the number crunching in COSMOS/M anyway, outside of the SolidWorks or DesignStar environment. Since I've been using the user hostile GEOSTAR-COSMOS/M package from way back (version one point something - many years) and I use Rhino3D for modeling and design, I saw no compelling reason to go to SolidWorks.
I've also used NISA some years back and have had some experience with NASTRAN. At this point I plan on transitioning to FEMAP, as soon as I can afford it and can take the proper amount of time to learn the features.
Have you seen these? I know you are looking for composite stress analysis, and there are some listed. A lot of material to review there (87MB)
[SIZE=+1]Astronautic Structures Manual (On-Line)[/SIZE], NASA MSFC (Marshall Space Flight Center), 1975. http://euler9.tripod.com/analysis/asm.html
Lots of great stuff there, especially if one is designing his own but doesn't have the education background. Actually some of the stuff is great reference even if you do have the education. But many of the references are published in various book forms - for instance, some of the papers look to be the same information that is published almost verbatim in Roark's "Formulas for Stress and Strain".
As far as I'm concerned, COSMOS has a pretty good laminate analysis capability (actually it's one of the better packages for this) but what I don't like is its interface. The geometry module (GeoStar) is archaic and user hostile at best. Its limitations on selection modes and the poor ability to modify imported geometry is extremely frustrating, making each structural model a pain to develop, especially considering the capabilities in modern CAD systems. Yes, it has several automated features but in dealing with composites these are not useful due to the material coordinate system definitions. As such, it is necessary to define the mesh manually, which often means that the underlying structural geometry has to be configured in a very particular manner - something that the default mechanisms within the program will not do.
But to complicate things, I'm now in the process of looking to get back to doing all my work on a Mac platform. Unfortunately though, many engineering and CAD software suppliers have stopped porting their engineering packages to the Mac many years ago. The trend however may be reversing: my CAD system (Rhino3d) is currently in process of being ported to Mac and so is much of the higher end publishing stuff that I got used to working under Windows.
Now all I need is for the FEA software to be converted, as well as my MathCAD program. I'm not holding my breath though. I'll probably have to keep using much of it within Windows emulation.
Earlier you indicated you wanted to compare COSMOS notes - what did you have in mind?