Regarding the "lumped mass" input:" In most cases you should model the weight of the system properly - therefore you won't need to specify lumped masses. This is more of an academic input value.

Active BCs: Spectrum dynamics are linear in nature, because of the natural frequency extraction - even if the model includes non-linear restraints. In CAESAR II, we use the referenced static load case to set the non-linear boundary conditions - based on the output of that specific (static) load case. If in that (static) load case, the gaps don't close, those restraints are removed completely from the dynamic model. If the gap does close, the gap is removed from the restraint. Similarly, if you have a +Y in the model, and it lifts off in the referenced static load case, then that restraint is removed completely. If the +Y doesn't lift off in the referenced static load case, it is changed to a full Y support in the dynamic model.