I am modelling a simply supported I beam in Abaqus using shell elements and I am considering a load case with end moments. I would like to apply an end moment as accurately as possible. Until now, I have applied shell edge loads (with a linear distribution along the web), which worked quite well. It is not entirely valid however once the distribution of stresses is no longer elastic. I have also tried applying a bending moment or rotation, using kinematic and distributing coupling, with a reference point outside the beam, but without success. At the moment, I am appying a displacement with a linear distribution along the height of the beam (similar to the stress distribution mentioned earlier), but I was wondering if anybody would have a more accurate / simple idea.