I changed the load-bearing structure a little bit: got rid of the timber member and added a vertical UPE80 instead. Robot is able to calculate this nicely - no instabilities. However this is probably not the best way to model a bolt, because currently the effects of the washer on one side and the nut on the other side are not taken into account. There are very large stress concentrations where the bar element meets the shell element. The upper bolt, which has the highest tension should spread it to larger area, because the head of the bolt is much larger than the shank of the bolt. Any ideas how to factor in the effect of the bolt head?
AutoCAD screenshot:

I modeled all the steel elements (UPE80 & RHS - rectangular hollow section) as shell elements with correct thicknesses. The 3 bolts are bar elements with a thickness to 10mm. I also added a rigid link (x-direction only) to simulate the contact between the bottom edge of the RHS and the UPE80.
3D view of connection:

Side view showing the bolts and a rigid link (X-direction only)

Stress concentration near the upper bolt. REASON: bolt head diameter is not taken into account.
