dik,
What do you mean by "modeling" LTB? You can enter unbraced lengths for wide flanges, etc. and it uses that to design the beam per the code...taking into consideration the LTB limit state.
But as a "stick" member, the program doesn't model the behavior (tendency to buckle) based on its shape. It just solves a stiffness matrix based on the member properties.
I guess you could model a wide flange, for example, using plate elements, and then with a PDelta analysis see about the buckling then. But I thought that the code LTB forumlations were based upon a small initial out-of-straightness that the FEA analysis wouldn't have built into it.