Research on Wheel Wear Coefficient Considering Wheel Profile and Operating Route Based on Multi-Rigid-Body Method

Wheel wear has always been a problem in the railway industry, and wear prediction simulation is an important approach to address this problem. However, the use of simulation requires the use of wear coefficients. The existing wear coefficients are based on laboratory rolling test rigs, which have different operating environments and materials compared to actual wheel-rail systems. Directly applying these coefficients to wear calculations can result in significant errors, so they need to be calibrated accordingly.

In this paper, a wheel wear prediction model that considers wheel profile and operating route is established. The model consists of a vehicle dynamics model and a wear rate calculation model, and the prediction method takes into account the track roughness, curve ratio, and different wheel profiles. In the dynamics model, the Hertz-Fastsim wheel-rail contact model is used to calculate the physical quantities such as normal force and tangential force in the contact patch, which are then inputted into the wear rate calculation model to obtain the wear amount. During the calculation process, the influence of various curve radii and the proportion of straight tracks is considered. Based on the proportion of different curve radii and straight tracks in the actual operating track, the wear amount is weighted to obtain wear prediction results that are consistent with the actual track conditions.

The wear model adopts the Archard local model, which has four wear regions (i.e., seizure mild I, severe and mild II). Simulation and experimental results show that when the two endpoint values in the Jendel wear coefficient diagram are selected, there is a significant prediction error. For the study in this paper, the initial values are set to the median values of each region in the Jendel wear coefficient graph, that is k₁ = 350 × 10⁻⁴, k₂ = k₄ = 5 × 10⁻⁴, k₃ = 35 × 10⁻⁴. The wear of the wheels is predicted based on the dynamic model and wear rate calculation model. This prediction is then compared with actual measured data at different operating distances. The differences in the comparison results are used to narrow down the range of values for each region in the Jendel wear coefficient graph. In the end, the wheel-rail wear coefficient was obtained.