Ran into this interesting problem and was wondering if anyone could shed any extra light:

I am running a linear dynamic modal time history simulation. My model has two fixed constraints. I am inputting a half sine pulse acceleration in g's. For uniform base excitation, I get zero velocity for all time at these constraint surfaces. However, if I repeat the study using selected base excitation and apply the same half sine pulse on the two constraints, I get a velocity profile that is non-zero. The velocity is exactly equal to the half cosine wave of this acceleration input, starting at zero and then reaching a maximum velocity, where it then remains constant.

Probing the non-fixed locations for their acceleration responses yields identical results for both scenarios.

So it seems that selected base excitation is actually solving for velocity through some form of integration of the acceleration input, but uniform base excitation is ignoring this integral. Just wondering if anyone could confirm and potentially elaborate on the difference in the background algorithms that Simulation is solving.

(This is SW Simulation 2017 if that matters)