On 17 October 2000, a GNER passenger train was derailed at Hatfield with the loss of four lives. In the early part of the investigation it was seen that one of the rails was broken. Visual examination of the rail showed the presence of a slow growing crack, starting from the surface where the rail and wheel are in contact, and a final failure area that was brittle. The slow growing crack showed signs of long-term exposure to air, whilst the final fracture was comparatively bright and shiny. Thus the problem, denoted gauge corner cracking, was essentially one of fracture mechanics.

Firstly, cracks are much more difficult to initiate than they are to propagate. This means that if a crack is large enough to be seen by visual inspection its rate of crack growth will be much greater than in its earlier life, when it is very small and unable to be located by visual inspection. It is known that:

• crack propagation becomes faster with increasing stress
• crack propagation becomes faster with increasing crack size
• final crack size before brittle fracture will be a function of applied stress, with large applied stresses producing a small final crack size

Thus the life of the rail will be largely controlled by the applied stress with large applied stresses producing short life rails. Clearly the inspection interval of the rails must be substantially shorter than the expected time that a crack will take to grow from being just visible to being of such a size that it will break in brittle fracture.

The applied stress will be a function of many factors, including the exact geometry of the wheel and rail, the lubrication between wheel and rail, the mass and velocity of the trains and the radius of curvature of the track. Deterioration of the track can be by wear, as well as cracking, and it may well be that when track is steadily worn that any small cracks are ground out before they become fast-growing long cracks.