Rail squats are a rolling contact fatigue defect that have plagued the railway industry for decades. A defect superficially similar to a rail squat is the rail stud which is thought to initiate thermally (Grassie et al. 2011). This initiation occurs through the creation of a thin layer of martensite on the rail surface known as a white etching layer (WEL). Once formed, a WEL is significantly harder than a standard head hardened rail, thus compromising the delicate balance between fatigue and wear and allowing cracks to initiate. These cracks have been observed to grow subsurface which makes their detection quite difficult. If left untreated, cracks have been known to reach significant lengths, which can lead to catastrophic failure of the rail.
One of the criticisms towards the proposed initiation mechanism of rail studs is obtaining temperatures high enough to form austenite. Although the austenite temperature does reduce somewhat due to the large hydrostatic compression caused by contact with a railway wheel. Currently, two heating methods have been suggested: adiabatic shear and frictional heating. The fact that rail studs have been observed to initiate where vertical track excitation involving large contact forces and increased wheel slip suggests that frictional heating may be the more likely method of forming a WEL. As such, this thesis focuses on modelling the contact force and rail temperature over three rail excitations likely to excite wheel slip: a dipped weld, a rapid reduction in available traction and a poorly supported sleeper. Additionally, an estimate for the pressure reduced austenite temperature is used to assess whether there is a thermodynamic driving force for the formation of austenite and thus a WEL.
A simplistic vehicle model is used consisting of half a wheelset and a quarter of a bogie with motion restricted to two dimensions; lateral vehicle and track motion being neglected. The rail is modelled using a five mode vertical description at the point of contact with forces transmitted at the wheel-rail interface being modelled in both the normal and tangential directions. These forces were calculated using nonlinear Hertzian contact theory, accounting for loss of contact, in addition to use of the FASTSIM algorithm. The frictional temperature rise at the rail surface was determined analytically (Ertz & Knothe 2003). This model was implemented using an explicit finite difference method in MATLAB.
From this investigation, all three excitations caused the austenite temperature to be exceeded at certain speeds. Additionally, the dipped weld and poorly supported sleeper were found to give a dynamic amplification of the vertical contact force of up to 4 and 10 times respectively at the worst vehicle speeds. Although the latter is potentially an overestimate, both suggest that proper track maintenance is critical if the formation of a WEL and potentially a stud is to be avoided.