Inconsistent side decoupler behaviour

[lpgagnon]

An old issue, I think; but I'd never pinned it down to a specific case.

2 payload fairings with identical shapes obtained 2 different ways: cylstart==0 and cylstart==cylend. Everything else is identical; in particular, both have ejection power 100%, torque 1%:

Ejecting the cylstart=0 version (bottom) behaves as one would expect from those settings; sides go out mostly straight sideways. The cylstart=cylend version (top) clearly gets much less sideways impulse (it seems it does get _some); enough that the torque effect is much clearer

In a real flight, with an engine pushing, the cylstart=cylend version with default forces (32%/1%) tends to fail to get the sides out of the way of the stage fast enough

cylstart=cylend doesn't seem to be the only necessary condition; the Size being smaller than the base diameter seems to also play a role. A straight cylinder behaves closer to expectation.