Numerical simulation of crack growth and wear in rails

Doktorsavhandling, 2012

Wear and rolling contact fatigue (RCF) of rails are huge problems for the railway industry with increased world-wide occurrence during the last decades. It is, therefore, of high interest to develop tools for the prediction of wear and RCF which can be used to estimate rail life. This thesis is concerned with the numerical simulation of rail deterioration due to wear and RCF. The main focus has been on the modeling of crack growth in rails, which has included development of a numerical tool in terms of a 2D Finite Element (FE) model of the rail that can be used to simulate the growth of (short and long) surface cracks in rails. Wear is taken into account through the process of crack truncation (partial removal of the crack). Furthermore, the change in geometry due to crack growth and wear is considered through a remeshing procedure. In the numerical framework, the concept of material forces is adopted from which a crack driving force can be derived. % Based on this crack driving force, numerical procedures for simulation of crack growth under monotonic, cyclic and typical RCF loading have been developed. Different propagation laws have been proposed and used to study crack growth in rails under various loading conditions and crack geometry. The influence of the highly deformed (anisotropic) surface layer, often present in railway rails, on the crack growth direction has also been studied. Results from the numerical simulations indicate that anisotropy has a large influence on the crack growth direction and needs thus be accounted for in order to be able to simulate the growth of surface cracks in rails more accurately. Furthermore, a numerical framework for simulation of rail degradation due to wear and plastic deformations has been developed and implemented. The procedure includes simulations of the wheel--rail dynamics using a Multi Body Simulation (MBS) software, together with detailed FE simulations of the plastic deformations. The procedure was applied for the calibration of a wear coefficient from experiments on a full scale wheel--rail test rig. Quantitatively good agreement was obtained between simulations and results from the full-scale tests in terms of worn-off area and shape of the worn profile.