Im modelling a clamped-clamped beam with a uniform load. In order to see when collapse occur Im increasing the load to see when Abaqus is aborting, this is working except that the collapse load is overestimated by 700-800 kN in this case. I have calibrated the time increment and the mesh so it should not be regarded to this.

Then I looked at the stress distribution in the cross-section and found out that the stress is actually increasing above the yielding point at the support (see picture), so wonder if there is a correlation here? I mean if the beam obtain more stress after the plastic limit then the collapse will be overestimated right? :)

So how to I tell Abaqus not to go over 355 MPa which is the yield stress.