In a buried piping system you're most concerned with the thermal strain. A properly prepared trench will eliminate any weight stress, which is what CAESAR II assumes. The only way to achieve this in a discretized "node/beam element" flexibility program is to zero the densities.

You're correct, with zero mass you can't run dynamics. But think about the situation for a moment: in a seismic event, the pipe will either move with the surrounding soil (i.e. no induced deflections or stresses), or fail in shear (two pieces) across a fault. Why is a flexibility analysis required in this situation: the pipe is either ok, or destroyed - there is no in between?