It's a really interesting application, and I definitely agree with Dave that at some point some shoes became anchors. Let's revisit the steps of what we think happened:

1) The line grew hot. (normal analysis)
2) The line displaced axially/laterally due to expansion.
3) The line fell off its supports in some locations (displaced vertically). This act could have bent the pipe.
4) The line presumably cooled down, but remained vertically displaced in some locations, but is now moving axially/laterally in the reverse direction.
5) The line shrank, but encountered new "anchors" in a displaced condition. Some shoes sheared off. This also could have bent the pipe, or made any previous bend worse. It's noteworthy that these anchors will apply a moment and a force.
6) Additional shoes in other locations may have contacted their supports after the previous shoes sheared off as the line continued to cool. They may have sheared, as well.
7) Present day configuration.

And what's being requested:
Try to estimate what current stresses are and will be if it is turned off, presumably possibly shearing off additional shoes or rupturing the pipe. Verifying the local disruption of supports on this loop does not cause other loops to react similarly.

I don't recall seeing anything regarding pipe guides, but if there are expansion loops, one should conclude that some kind of stress analysis was done. Perhaps the pipe was allowed to walk off its supports (repeat thermal cycles causing the pipe to displace over time, little by little until...).

I look forward to hearing how you ultimately proceed.

Thanks.