With reference to the debate on trackbed technology (NCE 26 April) I am not sure that rail cracking has much to do with trackbed design. Rail lines behave in much the same way as the race of large roller bearings: the fatigue phenomena is well understood.
While Hertzian shear stresses are highest at the contact surface of the rail, there is a second subsurface stress peak that causes 'core crushing' in roller bearings. The width of the wheel contact patch on the rail is narrow and contact stress is not uniform across the width.
This increases subsurface shear stress and reduces fatigue life. Once a crack forms (usually at a 45infinity incline), it propagates first to the surface.
It is then quite long and can rapidly propagate downwards under conventional beam action until brittle fracture occurs when a large load passes.
Ultimate failure is more likely at high than low speed and this needs to be allowed for in risk analysis of potential maintenance philosophies.
This is because on a bend the rolling contact is reduced by uneven wear, and bending stresses in the rail are speed dependent so that fracture would be more likely at high speed.
Also, when the core shear stress reversal is very fast the brittle fracture transition temperature of the steel is raised 10infinityC or more, increasing fracture risk, especially in winter.
For busy railway lines, it is essential to grind all rails periodically. This helps to maintain optimum contact shape, remove the cracks that start at the surface and to progressively lower the surface of the rail to move the core stress peak down to new locations before it has had time to cause cracking.
R W Brewerton (M), Natabelle Technology rwb@natabelle. co. uk