Yes, I am aware of this. However, I expect (and will test of course) if I can manage to apply the concentrated mass to the mass matrix without changing (too much) the stiffness matrix. The best case scenario would be to transfer the forces, change the mass matrix, and not change much the stiffness matrix. Unfortunately, I could not test this since the coupling did not work in my few tries.
Since the gravity is acting vertically and the mass is located just above the end of the bar, I expect the eccentricity to not be a problem in this test case. This could be the next part of my investigation, but rotary inertia is not relevant to the work I was doing using this connection. Your test is exactly what I tried to achieve, maybe some confusion is made because for the vertical beam, the bending stress is Syy and you plotted Sxx. In your next post with the Mises equivalent we can see that the horizontal bar with a free end and the vertical one have the same stress at the T junction, that’s the expected output. As expected, the fixed horizontal bar has zero moment and should not have any bending stresses. Maybe this asks @Disla 's question about the abrupt stress end.
Thank you guys for pointing out these solutions. I need to understand better how the *EQUATION keyword works and will use this @xyont example as a reference. Could you please share your input file where you define the *DCOUP3D and *DISTRIBUTING COUPLING? I want to study this implementation as well. While reading the documentation, my first thought was that this keyword was not meant for mass elements connection. I’m glad it is 
. I’ll run some frequency checks to see how both approaches perform against some analytical examples.
Very interesting, Disla. I did not cite this before, but these bars are theoretically welded and belong to a scaffold, so unfortunately the release approach would not work for this specific case. But it is very nice to know that we can do this at CalculiX. Some time ago I saw similar approaches in Nastran.