Hello

I am working with Abaqus for some time already, but I never simulated models with residual stresses.

However, I want to explain my current problem. Maybe someone can offer a solution or guide me in the right direction.

When steel columns, such as I-sections, are welded, the material heats up locally and then cools down non-uniformly. This results in a residual stress field, so that the column is already in a stress state without exterior loads applied. If an exterior load is applied to the columns, the compressive residual stresses and the compressive stresses introduced by the external load would add up and reduce the load-bearing capacity significantly (which would be overestimated without the residual stresses). However, in real life, the stresses would stay inside the column even though it is not somehow constrained or loaded.

I managed to induce residual stresses as desired via the SIGINI subroutine. They are correct in the initial step and a quick Excel sheet calculation shows, that they are already in equilbrium. However, it is recommended in the user manual to proceed with a general static step after the initial step, so that Abaqus can compute an exact equilibrium of the initial stress field. The problem is, that Abaqus doesn't only make minor changes to the stresses just to get an equilibrium, but changes them completely. There are differences up to 100Mpa. Now I was told, that this is because I didn't activate any boundary conditions to avoid the model to reduce the stresses.

Constraining all nodes in the equilibrium step is possible, but not reasonable, because as soon as I changed them to the real boundary conditions in the next step, the stresses would again start to reduce. How can I include my residual stresses in such a way, that they would not be reduced even if there was no exterior load applied? Is this possible with Abaqus at all?