I made some benchmark on simply supported composite sandwich panels under uniform pressure using the SHELL SECTION, COMPOSITE.
Considering standard values for carbon skins with G13=G23~4GPa and core shear stiffness G13=G23=20MPa, I could not get similar results to other software (i.e. displacements are lower, error not negligible), whilst applying the same core shear stiffness (or lowest stiffness) to the skins gave me really good results.
I cannot explain why this happens. I guess it is related to the element formulation?
Thanks
I recommend comparing the solution to a solid element model, not just to shell elements in other software because classic laminate shell elements can perform badly on sandwich panels in shear.
Maybe the solution is sensitive to the normal stiffness (E33) that CCX requires. Other software typically doesn’t include this.
Linear or nonlinear? Start with linear to make it easier.
Make sure the mesh is refined enough, not just the same amount as in the other software.
Look at the curvature of the deformed shape to confirm the boundary conditions are right. Rotational DOFs in CCX can be weird sometimes.
hi vicmw,
thanks for the reply.
I tried benchmarking against solid models as well and shell/solid models compare really well for other software.
Analysis is linear.
E33 does not influence results. I tried different values. I guess it is because for this type of load/geometry does not change the internal energy.
Boundary conditions seem ok looking at the deformed shape
Rotational dof should not be present if the model is made with 3d elements only?
Even with thick skins/thin core it works good enough (using the core shear stiffness everywhere).
I would be curious to run the same model in abaqus which should be quite similar to ccx so I could try exactly with the same element type, but I do not have any license for it…
reading lecture notes in composite material from Dassault (2009), Abaqus has three types of composite element named: laminated shell (thick), continuum shell (solid) and layered solid. which element do you use and compared?
refers to documentation in CalculiX, composite shell element are expanded to solid element with some formulation. may it similar to the last ones from Abaqus (layered solid) and not too far with shell continuum (solid).
actually i’m rare using composite keywords for anisotropic material. only isotropic, used to enhance performances by layering of shell element through thickness. it’s required to reduce locking phenomena as suggested.
the element I am referrring to is the layered solid. Plies are made with orthotropic material type. I cannot compare against abaqus not having any license though.
also, i have not access to Abaqus. so i’ll looking out to another document with similar known problems and solved.
corrected to my previous post, i read again. solid layered element in Abaqus may differs with CalculiX shell composites. as seen only one element through thickness, with additional thin layer inside. did not recommended since it less accurate as noted (p.58&59).
latest works from Farid Abed-Meraim, Université de Metz (2019) expanding and improve these Abaqus element type which look similar to CalculiX. if i should and interested in futures to study, i will took these report as a benchmark.