I have a model of a needle with the following boundary conditions: fixed at one end and constrained in the y and z directions at all 5 faces in the tip. I have also applied a total of 1 N in the five faces of the needle (see Figure attached). I am trying to do a mesh convergence study by starting with a stardand mesh with 0.18 mm element size and reducing it with a factor of 1.5 until convergence. However, as I decrease the element size, the maximum Von Mises stress increases and convergence is not attained (see graph). Even with mesh control in the edges with high energy norm error, the graph obtained is the same. Does anyone have a suggestion on how I should proceed?