The railway is an important mode of transport, due to its environmental friendliness, high safety level, and low energy consumption combined with a high transport capacity, among other factors. The Swedish railway network is old, there has been almost no expansion of the network during the past few decades, and more traffic is expected. Therefore, there is currently a demand for more track capacity and, in the short term, the existing network is expected to deliver the increased capacity. The railway operators in the network have a large impact on train delays, and wheel failures are one large contributor of delays. Delays destroy capacity and, therefore, capacityconsuming failures, such as abnormal wheels, need to be minimised. This can be achieved by using appropriate condition monitoring for the wheels on the track to find potential capacity consumers before failures happen. Therefore, the condition of the wheel-rail interface is important, since the state of the wheel influences that of the rail and vice versa. The monitoring of rail profiles is already being performed, but the monitoring of wheel profiles is still in the development phase. This thesis treats the applications and performance assessment of a wheel profile measurement system (WPMS), and presents case studies focusing on its system and measurement performance. The proposed applications concern how the information from the WPMS can be integrated with information from other data sources and with physical models to obtain a true current picture of the wheel behaviour. The thesis investigates the measurement performance of the WPMS by using a paired T-test and a number of quality measures, e.g. the reproducibility and repeatability, the precision-to-tolerance ratio and the signal-to-noise ratio. In conclusion, this thesis shows that the WPMS works well with an expected level of reliability in a harsh climate with respect to its measurement and system performance. By combining other data with the data from the WPMS, potentially abnormal wheels can be found in an early stage if the proposed new maintenance limit for the wheel parameter of the flange height is implemented. Furthermore, through adding a physical model to the process, the real contact condition of the actual wheel-rail interface can be evaluated and measurement deviations can be found. However, the wheel parameters, as well as the entire profile, need a high measurement quality with little variation, which seems to be an issue with respect to the measurement performance when advanced calculations are to be done. Therefore, a new approach for evaluating measurement performance has been developed using established statistical tools and quality measures with predefined acceptance limits; with the help of this approach, one can differentiate between the variation in the measurements originating in the different measurement units and the variation originating in the wheels. This new approach can be applied to judge the measurement performance of wheel profile condition-monitoring systems, and can also be implemented for other condition-monitoring systems to evaluate their measurement performance. Finally, this approach promotes the development of a condition-based maintenance policy by providing more reliable information for maintenance decision makers.
Luleå tekniska universitet, 2016.