Investigation of railway curve squeal using a combination of frequency- and time-domain models

Paper in proceeding, 2016

Railway curve squeal arises from self-excited vibrations during curving. In this paper, a frequency- and a timedomain approach for curve squeal are compared. In particular, the capability of the frequency-domain model to predict the onset of squeal and the squeal frequencies is studied. In the frequency-domain model, linear stability is investigated through complex eigenvalue analysis. The time-domain model is based on a Green's functions approach and uses a convolution procedure to obtain the system response. To ensure comparability, the same submodels are implemented in both squeal models. The structural flexibility of a rotating wheel is modelled by adopting Eulerian coordinates. To account for the moving wheel‒rail contact load, the so-called moving element method is used to model the track. The local friction characteristics in the contact zone is modelled in accordance with Coulomb's law with a constant friction coefficient. The frictional instability arises due to geometrical coupling. In the time-domain model, Kalker's non-linear, non-steady state rolling contact model including the algorithms NORM and TANG for normal and tangential contact, respectively, is solved in each time step. In the frequency-domain model, the normal wheel/rail contact is modelled by a linearization of the force-displacement relation obtained with NORM around the quasi-static state and full-slip conditions are considered in tangential direction. Conditions similar to those of a curve on the Stockholm metro exposed to severe curve squeal are studied with both squeal models. The influence of the wheel-rail friction coefficient and the direction of the resulting creep force on the occurrence of squeal is investigated for vanishing train speed. Results from both models are similar in terms of the instability range in the parameter space and the predicted squeal frequencies.