I looked into the model a bit more to make sure that I included all the effects and get the coupling right [more to come]

fig1: displacement(beam line direction) per unit acceleration on the mirror surface. X-axis represents the position on the mirror along vertical line. Each plot represents result from different support positions. For optimum point (1.17"), the sensitivity to vertical seismic is around 2.1x10^-12 m/(m/s^2).

fig2: This plot shows the result as in fig1, with the means removed. Typically we want the tangent line at the center to be zero for minimum tilt.

1. Now I use SI units in the simulation.
2. The surface displacement along vertical direction is ~ 5 x10^-10 m. The spot radius is 100 um, so the effect from translation along vertical direction is negligible. We can assume that the beam still hits and senses the center of the mirror.
3. Check the result between support length between 1.18" +/- 0.02" (+/- 0.5 mm). This tolerence is what I estimate to be our assembly tolerence.
4. I calculated the tilt by using the tangent line at the displacement on the vertical line in the mirror center. Note that my mirror surface in the simulation is flat instead of curve. At 100 um away from the center, the surface height will change by ~ 10^-8 m due to the mirror's RoC. Meanwhile, the result from simulation (flat surface) shows that the displacement around 100 um away will change ~ 10^-12 m. However, I don't think it will effect our model much, since most of the deformation occurs around the support point, not the mirror. It means all the tilting comes from deformation on the spacer, not the mirror. So the simulation for flat mirror should be a good approximation.