A bottom-up numerical model for wheel-rail contact to predict rail-way noise emission

Pass-by noise emission models are often based on measurement databases, yielding engineering models that are fast enough to calculate noise maps or to do predictions based on existing infrastructure. However, these cannot predict the effect of innovative components, either in the track or in the rolling stock. Changing properties of a single component, e.g. damping of rail pads or the profile of a wheel, leads to a chain of coupled effects including the dynamics of the individual component, the contact force, the vibrational velocity, and finally the emitted noise. These intricate interactions can only be captured by detailed numerical models. We present a step-by-step implementation that allows to predict the effect of changes on the component level on the pass-by noise emission. The model includes the finite element calculation of the mobility of track and wheel separately, and uses these to derive the contact force from the combined roughness and train speed. Next, the vibration velocities of rails, sleepers, and wheels are used as input for a boundary element calculation of the noise emission. Pass-by measurement results validate the models, and show good agreement if wheel-rail contact noise is dominant and secondary sources can be ignored.