Rayleigh damping 1st-2nd vs 1-3rd mode (simple frame geomet)

[Pablongo]

I am running a 2-story, 2-bay RC frame in 2D nonlinear time-history with both GM records and sinusoidal acceleration excitation (with a relatively small amplitude). 1st and 2nd mode are horizontal (typical for a 2-story building), 3rd mode is vertical (beam deflections, symmetrical mode about center column).

My input excitation is only acting along the horizontal direction. However, when I run the sinusoidal, if I assume damping on the 1st and 2nd mode (intuitive for a 2-story building) the solution converges to a large residual deformation (does not approach the elastic case). But if I assume damping on the 1st and 3rd I get a convergent solution (that matches that of an equivalent elastic frame). I thought the 3rd mode would not be excited in a horizontal excitation. Also, wouldn't a 1-2 mode Rayleigh damping over-damp the 3rd mode? Mass is lumped in the beam-column joints. I have some softening elements in a few locations throughout the model.

Any thought on why this is happening?

[fmk]

The 3r'd mode will get excited if the nodes start moving vertically. obviously, so why are they moving up and down? if using the force beam column element and you have a concrete section, this may be due to the forces (bending and now additional axial causing additional elongation/contraction in element which feeds back as axial forces on the nodes). try using a dispBeamColumn and see if the vertical response is as great. you could also look at the effect of removing mass from the vertical direction.