With respect to boundary conditions, your task is to either:
A) Simulate accurately the loads and reactions of the boundary condition,
or
B) Conservatively simulate said loads and reactions,
and
A) Ensure "your" scope doesn't excessively impact "their" scope,
and
B) Ensure "their" scope doesn't excessively impact "your" scope.
This is difficult to capture in a concise manner to direct information collection.
With that said, I've come up with instructions for my opinion for adequate boundary conditions:
1) Hard anchor points, such as piping anchors, ground penetrations, and equipment,
2) Midpoints of extremely long runs of rack piping,
3) First limit that's axial to the tie point,
4) First piping guide after a 90 degree turn beyond the tie point,
5) Midpoint of "expansion loop" resembling piping configurations.
Exceptions for 3) and 4) is that if the piping configuration is imbalanced - i.e. if beyond the limit or guide is another perpendicular straight run of pipe that is much, much longer than that which is represented between the tie point and the guide or limit, then we must pursue the next hard or soft anchor point.
Existing springs that are potentially impacted and requiring modification (or other in-line equipment or attachments similarly at risk to fail) by the addition of "your" piping are considered "in scope" and you must take your boundary conditions further as though the existing piping between "your" pipe and "their" spring is treated as "your" piping for purposes of modeling scope.