in a static study study on shell mesh (generated automatically from sheetmetal parts) with bonded contacts i am running, some parts keep "flying away", the bonds seem to not work. These parts are simple flat (no bends present) parts, quite small in relation to the surrounding parts, which do have bends. In the results those parts are not only some distance away from the rest of the geometry, but also greatly deformed, sometimes blown up to very large sizes and hardly recognisable due to strong deformation. Also they visually seem to be displaced only inline with their own plane.
The way i have bonded them to surrounding shells is equal to the way in which i have bonded similar parts that DO seem to work...
I have deleted the global bonded contact and have defined bonded contact sets between the edges (source) of the problematic and the faces of surrounding shells (target). The target faces are 90 degree flanges and inplane with the source part. I have done this for both faces of the source and target parts to make sure the face that is used as "fixed face" in the sheet metal definition of the source part is included. Mesh controls were defined on the problematic parts to make the element size equal to its thickness.
Also i have gradually built up the amount of components in the assembly, and have run a simplified study. All work, it is just that in the large study the parts keep flying away.
Any ideas what i am doing wrong?