...it seems I cannot. At first I tried to follow the best practices to avoid this kind of problem (not to put loads as well fixtures on points or edges, avoid re-entrant corner...) then I start running analisys with coarser mesh and then refining them to control the percentage of stress increasing.

My model is a simple truss with a vertical load, I run the analisys choosing solid elements with fixture on lateral holes where bolts where supposed to hold the structure. A circular split line for each hole with  no traslation normal to the face provided the washers.

I found irrelevant putting a fillet or a chamfer for the holes' edge because the behaviour was the same. However refining the mesh limited the maximum stress to the element just outside the above mentioned split lines while stresses in the near elements decrease pretty dramatically. I plotted the element values and so I can't isolate this peaks by iso-clipping (why this...?).

My concern is at least to know if these peaks are real or not, how can I evaluate the real maximum stress over there where I suppose it could be a crisi of the material.

By the way I tried the nominal load for my problem but I find this happen also with half the load.