Continuous welded rail is subject to the forces of thermal expansion, for which sufficient restraining force is provided by rail anchoring against the crossties, which in turn are restrained by a robust ballast section - including the shoulders. The "adjusted temperature" of the rail at the time it is laid is a critical measure; it is also a required matter of record. Typical track geometry readouts will detect variations only after displacement has begun. A buckled track failure often appears as a sudden event, triggered by ordinary traffic at the instant the built-up strain is released.
The lessons of maintaining temperature-adjusted CWR within acceptable limits has been learned at a price.
In the earliest years of Conrail, the massive backlog of rail replacement was urgently addressed with year-round rail replacement. The rail on the Youngstown Line was replaced in the dead of winter, with multiple buckled track events detected (without derailment) and adjusted over the course of the next few summers.
"Patch" rail replacement, by its nature, was a year-round necessity and the rail was heated when necessary prior to anchoring.
From the Youngstown experience and for other good reasons, the decision was made to otherwise suspend rail replacement as a matter of productioin (such as dual-rail) in the winter season.
The BN derailment was sadly preventable as the conditions at this site were plain to see on the basis of BN's own inspection records.