Not getting into the last paragraph and taking about expectations, I'd like to grab your attention to some other points:
1) suppress all fillets, they have minor effect on your structure and their only job is wasting your mesh.
2) How's your mesh resolution? Are these results independent of mesh resolution?
3) The stress pick is concentrated at a specific area...
Thank you Siavash !!
The last paragraph was about the case when two cylinders are pressing against each other, since they are not perfectly rigid bodies, there should be an contact area with a width and a length. And with the higher rigidity of both solids, the smaller the contact area. The length L is the thickness of my lever = 4.32mm excluding the fillets. The width b was assumed to be very small for both metal, i.e. 0.1mm.
Thus the expected stress is the applied force 206N / (4.32mm*0.1mm) = 4e+8 N/m2.
1. But my actual piece does have the fillets, shall I remove them for the simulation ?
2. That study was performed with the default mesh solution, that is the middle resolution. Choosing finest mesh and halving the mesh size did increased the stress for about four times...!! But what resolution should be the appropriate one for correct result?
3. Do you mean the stresses are reasonably distributed in the part? I observed that, but the stress value was not as expected.
Let's discuss those points (1,2,3) and later we may/may not need to get to your points.
1) Yes this is called simplification step in pre-processing. We do it to save computational costs.
2) There you go! 4 times escalation in stress! a highly potential for 'singularity'. The following article would help; many other great ones can be found online:
3) please refer to 2.