I think that you have to draw the floor (a rectangle of a significant thickness will be enough and a very low density so as not to interfere with the weight of the whole or a surface) to be able to apply the force on all the beams.
If the surface solution does not work, a very flexible plate should be added (use a material with a Young's modulus of 1000 MPa for example) in order to distribute the reactions without distorting the overall stiffness of the structure (do not analyze the results on this part of course).
Now we should be sure that the load of 500 kg/m2 is well distributed over the entire surface of the beams. In a floor structure, the beams often only work in 1 direction.
Why rush headlong into finite elements? Take a step back. take out a sheet, a little Rdm will simplify your life! only then do you check by Elts Finis! So no need to model the floor anymore you will have already made an assessment of the mechanical actions applied to each beam.
In your case, the mesh should be fast. Use the "beam" model for all your HEA profiles
When it comes to beams, I think that an Rdm calculation is faster and more reliable than the big artillery of finite elements!
But your method is justified to check your manual calculations and size your plates and gussets in detail.
For the assembly calculation you will have to make a mixed mesh!
- beam mesh for HEA
- Volume mesh (or maybe even shell mesh) for your plates and gussets.
Then we have to add the contact constraints between the different meshes so that the solver faces the links. otherwise it will tell you that your mesh is incompatible or that your model is unstable.