I read about the 3-2-1 minimum restraint approach to boundary conditions on Roshaz homepage.
To compare this approach to more common ones, I did some tests in Solidworks.
Please see attached presentation (sorry for the swith between metric and inch units)
Mesh is identical in the 3 examples and the loads are added as bearing loads.
Boundary 1 is the "simple" approach where radial fixtures are used in the holes wich prevent the part to deform in these areas.
Boundary 2 is the "3-2-1" approach where reaction loads are used. (reaction force in the 3 fixture points are below 0,5N)
Boundary 3 is the "real" approach where the pins are modeled and no penetration contacts placed between them and the linkage.
The "simple" approach (BC1) don't show the high stress on the radius at the large diameter.
The "3-2-1" approach (BC2) shows very high stress on this radius due to free deformation of the large hole. (stiffness of the pin is not taken into account)
The "real" approach (BC3) are somewhere in between but with slightly higher deformation.
Is there another approach/solution to get results closer to BC3 without using pins and contact?