B31.1 Appendix II has a subtle note - if "...the safety valve installation can be idealized as a one degree-of-freedom system...", then you can use their approach.
You said you calculated the natural frequencies of this (sub)system. If you believe more than one of these natural frequencies could be "activated" by your applied load, then your system is not behaving like a single degree-of-freedom (DOF) system. Look at your mode shapes, if more than one mode shows that elbow rocking up and down, this Appendix II method may not be appropriate for this system.
But... should only one mode be activated by this thrust load, then use THAT frequency to calculate the DLF. Maybe that will help.
The idea here is that the mode shape activated by the jet load will be similar to the shape produced by a static analysis of the (amplified) point load on your piping.
One more issue - you say you started your subsystem with an anchor at the PSV branch. That's good of that point cannot move in the field. But should there be deflection on the main, then the Appx. II method is questionable as you probably do not have a single DOF system. The lead in to the B31.1 quote above says "If the run pipe is rigidly supported," .
_________________________
Dave Diehl