Rail and Infrastructure Failure Investigation

The wheel-rail contact patch is typically 1 to 2 centimetres across under a loaded freight wagon or express passenger vehicle. Within that tiny area, contact pressures exceed 1,000 MPa, well above the elastic limit of the rail steel in the near-surface layer. Each wheel passage plastically deforms the surface slightly, creating a thin layer of highly work-hardened metal with a different crystallographic texture from the bulk steel beneath it. This is called the white-etching layer (WEL), and its presence on a fracture surface tells an investigator that the rail saw very high contact stress.

Micro-cracks initiate in or near the WEL, often at angles of 20 to 30 degrees from the running surface in the direction of traffic. In the gauge-corner region of a curve, where wheel flange loads push the contact patch toward the corner, the shear stress component is highest. Cracks here are called gauge-corner cracks, and their geometry is treacherous: for a long period they grow parallel to the surface, causing harmless surface flaking, but they can turn and begin to propagate transversely through the rail cross-section. A transverse crack that reaches critical length causes a rail to fracture suddenly and completely under a passing train.

By the summer of 2000, the rail at the Hatfield site had visible surface cracking that track inspection records confirm was noted but not acted upon. Replacement was scheduled but deferred. No speed restriction was imposed. On 17 October, the GNER InterCity 225 service from Leeds to London passed over the section at line speed and the rail fractured. The investigation recovered broken rail sections and found multiple transverse fractures propagating from heavily cracked gauge-corner zones. SEM fractography of the fracture surfaces confirmed rolling contact fatigue with no evidence of pre-existing manufacturing defects. The failure was entirely a product of service loading combined with inadequate maintenance response.

The post-Hatfield investigation by the Health and Safety Executive, carried out through an Independent Investigation Board, found that Railtrack had no systematic programme for monitoring and acting on RCF defects across its network. Ultrasonic testing records were inconsistently kept. The maintenance contracting structure created ambiguity about who was responsible for tracking defect severity and scheduling replacement. The findings were systemic: multiple organisations had failed together in a way that no single actor had fully recognised.

A railway axle is a solid or hollow steel shaft with wheel seats, disc brake seats, and journal bearing seats pressed or shrunk onto it. The wheel seat is the most critical region. An interference fit generates a circumferential compressive pre-stress in the axle beneath the wheel, which is beneficial under bending loads. But the fit also creates a micro-slip zone at the edge of the contact, where fretting fatigue damage accumulates with each wheel revolution. Corrosion pits forming within the fretting zone act as stress concentrations from which fatigue cracks can initiate.

• Fretting fatigue: micro-motion at the press-fit edge under cyclic bending loads damages the surface and creates pits. Cracks initiate at pit roots and propagate under rotating bending stress.
• Corrosion-assisted cracking: moisture ingress under the wheel seat accelerates pit formation and can cause hydrogen-assisted cracking in high-strength axle steels.
• Overload fracture: derailment loading can fracture an axle through a single overload event. Fractographic distinction from fatigue is essential: overload fractures show fibrous cup-and-cone or flat crystalline surfaces, not beach marks.

Investigation procedure for a suspected axle fatigue fracture begins with careful photographic documentation in the wreckage, followed by sectioning the axle perpendicular to the fracture plane. The fracture origin is located using low-power stereomicroscopy, and the origin region is examined under SEM to identify fretting damage or corrosion pits. Metallurgical sections through the origin region establish whether the steel composition and hardness were within specification. The maintenance record is reviewed to determine when the last ultrasonic axle inspection was performed and what it found.

Track geometry describes the spatial relationship of the two rails to each other and to the vertical. The key parameters are gauge (transverse distance between rail heads), cross-level (height difference between the two rails), twist (rate of change of cross-level along the track), alignment (horizontal deviation from the design centreline), and surface or longitudinal level (vertical deviation from the design profile). Abnormalities in any of these parameters generate dynamic wheel-rail forces that can exceed the flange-climb threshold or cause a vehicle to oscillate unstably.

In an accident investigation, the geometry records from the most recent measurement-train run before the incident are retrieved first. If the site shows a peak in one or more parameters close to the accepted intervention level, that peak becomes a prime candidate for investigation. The records are then compared against previous runs to determine whether the defect was developing over weeks or appeared suddenly, which helps distinguish a chronic maintenance failure from a sudden geotechnical movement.

Railway signalling systems are designed with multiple layers of redundancy and fail-safe logic: a failed relay is supposed to drive the signal to danger, not to clear. When a signalling failure contributes to an accident, the investigation must determine whether the fail-safe logic actually failed (a hardware or software defect), whether the logic was overridden or bypassed (a maintenance or operational procedure failure), or whether human operators made decisions that brought about a collision or derailment despite the signals functioning correctly.

Modern signalling systems generate extensive event logs: every signal aspect change, every track circuit occupation, every points movement, and every driver acknowledgement on ETCS-equipped lines. Retrieving and correlating these logs is one of the first actions in any major rail incident investigation. The logs establish a precise timeline that is independent of any driver or signaller account, anchoring the investigation in objective data before witness interviews begin.

• Track circuit failures: a broken rail or contaminated ballast can prevent a track circuit from detecting a train, showing a clear signal when the section is occupied. Hatfield's immediate signal state is one example of the consequences of missing a defect that affects track circuit performance.
• Points failures: a points mechanism that moves partially or returns to the wrong position after a train has passed can route a following train onto a conflicting path. Mechanical wear, contamination, and control-system software defects are all documented causes.
• SPAD analysis: a Signal Passed At Danger is recorded by on-board data recorders and trackside logs. The investigation determines whether the signal was displaying correctly, whether the train brake performance was adequate to stop at the signal, and whether driver workload or visibility contributed.

When a train derails inside a tunnel or on a viaduct, the forensic engineering response has two simultaneous components: the accident investigation that determines why the train left the track, and the structural assessment that determines whether the tunnel or viaduct is safe to use. These are legally separate but technically intertwined, because the structural damage observed after the accident may help establish the severity and sequence of the derailment.

Derailed vehicles in a confined space typically impact tunnel walls, portals, or viaduct parapets. Impact marks on the tunnel lining record the height, angle, and sequence of contact as vehicles yawed, rolled, or overrode each other. Engineers use these marks to reconstruct the derailment progression. Structural assessment then evaluates whether the tunnel lining retained its structural integrity under the impact loads, whether cracks propagate beyond the immediate impact zone, and whether any drainage or waterproofing systems were breached.