We only have one thermal load in the case (ie cold to T1), so we were looking at an operating case, and because we had big axial displacements (over 100mm) when we plot the start and finish locations of the node we can see the pipe is stopping at the support gap as expected (but then rotating, resulting in a larger lateral displacement than the support gap at the end destination of the node.

But Good point about the thermal loadcases.