RALLIYO AI
← Case studiesIncident walkthrough · RAIB report 07/2021

Broken track screws and gauge widening

Sheffield station, England · 11 November 2020

Track screws holding rails and baseplates to wooden bearers broke over weeks or months without being picked up by inspections. The rails spread apart under traffic until a wheel dropped between them.

What happened

At 02:44 on 11 November 2020, 16 wagons of a freight train carrying cement powder from Hope to Dewsbury derailed at the north end of Sheffield station.

The train was coasting at about 12 mph. The leading right-hand wheel of the twelfth wagon dropped between the rails because they were too far apart (gauge widening).

Wagons and track were significantly damaged and the station was partly closed. No one was injured.

What the investigation found

  • A number of track screws securing the rails and baseplates to the wooden bearers had broken, letting the rails spread under passing trains.
  • The screws had probably failed weeks or months before the derailment, and the failures had not been identified by maintenance inspection activities.
  • The location had a potentially high derailment risk that was not recognised, because the guidance for identifying such risk had not been applied.

Timeline

  1. Weeks to months beforeTrack screws on the curve begin to break; the failures are not recorded by inspections.
  2. 11 Nov 2020, 02:44Wagon wheel drops between the rails; 16 wagons derail.
  3. 5 Oct 2021RAIB publishes report 07/2021 with four recommendations to Network Rail.

Where Ralliyo fits

The evidence below is the kind Ralliyo records and keeps together. It shows how the capabilities map to the findings, not a claim about what would have happened.

Observation historyMissing fastener · Fastener deteriorationEach broken or missing fastening becomes an observation tied to its asset, so a cluster of fastener defects on one curve is visible as a pattern rather than as separate notes.
Trend between inspectionsComparing successive inspections of the same assets shows whether fastener defects are increasing, the deterioration the RAIB says developed over weeks or months.
Condition score with reasonsFastener observations lower the section score and appear in its explanation, so a declining curve rises up the priority list with the evidence attached.
Engineer reviewEvery AI finding is a suggestion that an engineer confirms or dismisses; Ralliyo does not decide whether track is safe to run.
Limits. Ralliyo works from inspection imagery and records. It can only surface broken fastenings that are captured in images or recorded by inspectors, and it does not measure track gauge. Gauge and geometry still need dedicated measurement.

RAIB recommendations

  1. Implement processes for identifying locations with a high risk of derailment.
  2. Implement changes to safety-critical processes.
  3. Review standards governing the fitment of check rails.
  4. Improve track geometry data formats.