Dear all I am simulating a solid welding process and so far I got nice results. I started from the worked example on abaqus documentation for inertia friction welding and I added several new features to fullfil my research project. Now I am facing an issue. If I use some process parameters, that causes highly localised contact shear stresses I start to have warning about negative eigenvalues and I see some element deforming, badly, oud of plane. I consider this due to the fact that my model despite is axisymmetric, is allowed to consider the twist. However, this deformation tends to introduce nonlinearities that the solver cannot cope, and even remeshing does not help at all. Does anyone had similar experience, not specificly applied to this process? I tried to improve convergence control parameters, remeshing more, refining the mesh more, tuning differently time increment and temperature allowed increment. Nothing seams to sort out this issue. If you have any suggestion I am keen to know.