(NX Simulation) Linear Static Compression using PCOMP elements

[HarambeWasAnInsideJob]

Hey there,



I was hoping someone with experience using PCOMP elements could help me out with my project.



So I have a small trapezoidal 'stiffener' sample (about 2.25" tall by 2" wide with a material thickness of about .45") that is to be compressed in unaxial load (strictly uniaxial)... The material used is orthrotropic (its a composite)... It's 3 layers of composite: outer 2 layers are carbon fiber and the middle is fiberglass...



So first I decided I would model the cross section as a line (no thickness) and then extrude that, and mesh it with shell elements... For the material, I decided I would use AUTODESK HELIOUS; you can input lamina values and ply layups and it will output the laminate's mechanical properties. So once I got the mechanical properties of the laminate I put those into NX as an orthrotropic material (i was not aware of NX Laminate or PCOMP when doing this at first)... Regardless, for whatever reason everytime I ran the simulation I kept getting error due to the Jacobian (2x2 error)... I believe it was a combination of the material and not a good mesh. Who knows.



So after more research online, I decided to use the PCOMP elements since it would allow me to build up my laminate by defining each ply (lamina) properties all in the NX simulation environment... It worked pretty smoothly except the results. I had to turn BAILOUT to -1 so I could tell the simulation to output results regardless of errors; and then once I saw the results, the Displacement was in the ballpark.. but the stress was not. Maybe I am looking at it wrong, but it only gives me stress PER PLY.. So do I take the average of each ply's stress to get my total stress calculation???



I just feel like I might be doing this whole thing wrong.. Can anyone give me advice or a tutorial if I am missing something..



OR is there another efficient way to do this because I am getting so frustrated!! lol.



Please let me know guys!

Thanks

 

[whitwas]

The Jacobian is a function of the element's geometry, not the material, so maybe your elements were inverted or very badly shaped.

What do you want "total stress" for? Each ply can have different stress and averaging them would lose the high stresses that might be important for failure.

What were the other errors that prevented it producing a solution? I think you should solve those instead of trying to ignore them.

Also, it's probably not accurate to use shells for such a thick object. Solid elements are generally better. Just make a mesh with a layer of elements for each ply.

 

[HarambeWasAnInsideJob]

Hey whitewas! Thank you for the response.

I actually wrote down the thickness wrong earlier. The thickness of the model is .045 inches, so therefore I was thinking shell elements are better??

Also the total stress versus ply stress: the stress of the overall FEM (stress at each element), would describe the stress better than just the maximum right?

I got the model to run eventually. I had to use the ply layup builder in NX.

Thanks

 

[whitwas]

I see. Shells are never really "better" in terms of accuracy. They're just easier or faster. Especially for composites so that's probably a good choice.

I'm not quite sure what stresses and what averaging you mean. Averaging over the plies sounds like a bad idea. But averaging over the elements connected to a node is common and usually improves accuracy when all the elements involved have the same properties. Averaging over the nodes on an element would reduce accuracy.