PREDICTION OF WHEEL PROFILE WEAR: METHODOLOGY AND VERIFICATION

A methodology and a simulation tool to predict changes of wheel profiles due to wear have been developed. The methodology is based on a load collective concept, where a load collective determines a set of dynamic time-domain simulations that is chosen to expose the wheels to contacts with the rails that they are likely to encounter in the application at hand. In the load collective the railway network is primarily discretisized based on the curve radius distribution. Key parameters are found by performing parametric studies, where the input load collective is varied. Track design geometry parameters and rail profiles are most important to vary in the load collective. The vehicle-track simulations are performed with the MBS (Multi-Body-System) tool GENSYS and the wheel-rail contact mechanics modelling is done with the Hertzian theory in combination with the simplified theory by Kalker (FASTSIM). Comparisons have also been performed with more advanced contact mechanics methods. Material removal, or the profile updating, must be performed iteratively since the wheel profiles affect the wheel-rail contacts and the vehicle-track response. A wear step is defined as one loop in the main programme and corresponds to a maximum wear depth on the wheel profiles and the pertinent simulated running distance of the vehicle. The maximum wear depth in each wear step has been set to 0.1 mm (perpendicular to the wheel profile) or alternatively 1500 km of running distance. Archard's wear model is used and the wear coefficient has been determined by laboratory measurements by others, using disc-on-disc and pin-on-disc machines. All laboratory measurements have been performed under dry, room tempered conditions. The effect on the wear coefficient from man-made lubrication has been estimated by comparing rail wear for lubricated and non-lubricated rails in curves of about the same radius. The effect on the wear coefficient from natural lubrication (weather etc.) is also estimated. Wear due to disc braking has been estimated by introducing a factor that increase wear due to straight track running. The tool is verified by comparing simulated and measured wheel profiles on a X10B commuter vehicle trafficking the commuter railway network in Stockholm. Simulated and measured wheel profiles agree well. Simulation times are quite long, it takes about 48 hours to predict a turning interval for a wheel. (A)