That's why I'm bringing up the adaptive meshing technique.
Make a large sized mesh and let the solver fill the areas of high
stress with a finer mesh. This might speed up your solving time.
"no penetration contact" is the worst to solve. If there is a face
that is bolted to another face I will actually make two studies for
this. One that bonds the faces together to get a quick highly
accurate stress in the metal and then another study that has the no
penetration with a rough mesh to solve for stresses in the bolts.
Hope this helps?
My trust comes from using a adaptive meshing others will use other techniques (protoyping, engineering...) 10% accuracy is good enough for me with adaptive meshing but It might be different in your situation. The colors you see are everything to this program. Strong understanding of the Physics/Enginnering behind FEA is critical.