There will always been a point in the pipe that acts a virtual stop where thermal growth in opposing directions prevents that pipe from moving or rotating along a specified axis.
Your mission, if you choose to accept it, is to estimate the locations of those virtual stops.
I hesitate to use the term "anchor" because that indicates a translational and rotational stop in all 3 axes, which may not always be true.
Generally speaking, this point lies in the middle of the pipe. However, this causes us problems on substantially long runs of unguided pipe. A substantially long run of unguided pipe will permit cumulative friction forces (and branch lines) to hold the pipe back, compress the pipe axially, and even bend the pipe similar to slender column buckling, which CAESAR does not calculate without sufficient prodding to do so.
If the system is existing, you may instead be able to infer how far it currently travels through observation in operation, or through evidence of wear at support locations, and instead utilize those values in your analysis.