Characterisation and Understanding of Rail Steel Behaviour for In-Service Life Prediction

Rail steels commonly accumulate large amounts of subsurface plastic deformation due to repeated rail-wheel contacts. The accumulation of this plastic damage is known to be responsible for the generation of RCF (rolling contact fatigue) and significant instances of wear damage while in-service. An improved quantification of this plastic deformation development is important to allow better decisions regarding rail maintenance programmes and the selection of the most suitable rail steel grade, helping to support the installation of premium rail steel metallurgies. In addition, to providing potential cost savings through possibly helping to optimise maintenance programmes and the prevention of rail failures. Therefore, the focus of this thesis is to expand the knowledge and understanding of how plastic damage develops in rail steels due to repeated rail-wheel contacts.
The plastic deformation response of rail steels subjected to cyclic shear-compression loading conditions was characterised from experimental twin-disc samples for different rail steel metallurgies and contact conditions. This information collected provided the data necessary for deriving the rail steel shear stress-strain curve, ratcheting load conversion factor (conversion factor between ratcheting load and increment shear strain), and the shear strain at failure (limit of accumulative shear strain that can occur before the nucleation of voids and microcracks). The shear stress-strain curve, ratcheting load conversion factor, and the shear yield point were found to be insufficient information by themselves for understanding the wear performance of a rail grade. The comparison of results generated under different contact pressures showed that the shear stress-strain curve was insensitive to the normal contact pressure it was derived under and so would be classed as ‘material property’ data. The ratcheting load conversion factor, however, was found to exhibit some dependence with the normal contact pressure the data was derived under, which would class it as ‘machine behavioural’ information. The derivation of the rail steel shear strain at failure was also investigated via a new value optimisation process using an implementation of the brick-based plastic ratcheting simulation model. This value optimisation process looked at varying the material shear strain at failure until the simulation cyclic wear depth loss results matched up with the experimental twin-disc data. The results generated from this approach was successful for the conditions it was believed to be best suited for, however, it was less successful for the conditions believed at the time to be challenging to apply this approach to.
The development of a novel optical monitoring system for application to twin-disc testing was also explored in this thesis. The intention of this system was to photograph in-detail the surface evolution of a twin-disc sample’s running track during standard twin-disc experiments without the need to interrupt tests. The images collected using this system were able to clearly observe the development of wear flakes via the shadows they cast onto the sample’s running surface for dry twin-disc tests. The visualisation of the development of RCF cracks from water-lubricated experiments was not as a clear due to interference caused by the water film present on the specimen’s running track. A comparison of the wear flake shadow pixel count and the wear data showed a good correlation between the two datasets, showing the potential of understanding the wear behaviour of rail steels using this equipment. Investigating the wear flake morphological change using this equipment revealed a correlation between wear flake size and the undeformed hardness of the rail steel, with harder rail steels generating smaller wear flakes. A delayed initiation of the wear flakes was observed on the heat-treated rail steel grade R350HT compared to the air-cooled grades R260 and HP335, which is believed to be caused by the refined pearlitic microstructure of R350HT. The effect of the elongated MnS inclusions orientation on wear flake behaviour could also be seen, with elongated MnS inclusions orientated perpendicular to the running surface causing earlier initiation and faster development of wear flakes compared to elongated MnS inclusions orientated parallel. Tracking the development of individual wear flakes showed them occasionally growing and shrinking over time, but not completely disappearing, which points to material only being removed as wear debris from the tips of wear flakes. A comparison of the surface plastic flow measured using the optical monitoring system correlated well with the surface plastic flow predicted by the plastic ratcheting simulation model during the early stages of the experiment. This correlation between simulation and experimental results did break down later into the experiment, which is possibly caused by the 2D nature of the simulation model used and the fact that wear and plastic deformation counteract each other’s influence on the movement of wear flakes.