hi, wide flange (I or H) section is commonly used in steel structure but is not available and supported unfortunately. It seems can be really useful when CalculiX beam section capable to analysis this type of beams, thank you.
also, double current number of integrations seem required to improve result accuracy (section force and stress) and eliminating hourglass in large deformation and plasticity.
Maybe this is related to the fact that there are not really beam elements in CalculiX. The developers would need to generate the corresponding 3D meshing for such profiles, which might be difficult, considering how thin the wedges and flanges tend to be (since shell elements are also not really shell elements). Ideally we could have the usual beam elements, where you could just define the moment of inertia, area and all needed parameters.
I actually just learned today how beam elements work in CalculiX (just starting with this program), and I was wondering if there was something fundamental about the implementation that makes the definition of beams and shell elements impossible, or is it just a decision from the developers? Is it maybe not possible to easily adapt the routine for assembling e.g. stiffness matrix due to the rotational degrees of freedom, or something like that? I tried to look for hints in the source code, but it turns out that I suck at fortran 
.
beam element in CalculiX have three approaches in modeling, classical formulation improved Timoshenko (user defined by area, inertia and torsional constant), expanded to solid with appropriate integration scheme (rectangular, circular, box, and pipe) and the last is composed models of duplicate element (patched by rectangular section with offset).
each of them has limitations, advantages and disadvantages. Classical beam (U1) is limited to linear elastic and small deformation similar to classical shell element (US3). Current approach of expanded solid may not work properly in large deformation and plasticity due to single element and number of integrations being used. Composed beam took advantages of all but complicated in definition at input by large duplicate element, generate knot make over stiffening the models, outputs section forces not properly calculated.