Keywords: random rail surfaces, stochastic numeric, extended covariance analysis, critical vehicle speeds, overcritical damping.

This paper investigates the dynamics of railway vehicles running on random track surfaces with constant velocity. The vehicle model of interest [1] possesses two degrees of freedom describing the vertical vibrations of the wheel and car mass. Both masses are connected by a linear dashpot and an elastic upper spring with nonlinear progressive characteristics. This system is linked to the rail surface by a second lower spring with linear characteristics. According to extended measurements the rail surface profiles are described by stationary processes with linear power density distributions in double logarithm scaling [2]. In a first model, they are derived from white noise by means of low-pass filter equations which are transformed from way to the time domain using way and time increments of the constant vehicle velocity and the corresponding noise relation.

In the linear case, stationary vehicle vibrations are studied by means of covariance analysis and classical spectral methods. The calculated root-means-squares related to the rms-values of the base excitations are one for vanishing velocity, they become resonant when the vehicle velocity times the railway frequency reaches the natural frequencies of the two-degree-of-freedom system and vanish when the speed tends to infinity. Singular vibrations are generated when the upper damping is vanishing or increases infinitely, respectively. In between these two extremes there is a critical damping where the vehicle vibrations become minimal.

To avoid frequency-limited simulations like harmonic analysis, higher-order time-discrete schemes are introduced to the drift and diffusion terms of the vehicle and excitation equations applying iterated integrals [3] and their approximations by time and noise increments. Systematic errors of the applied discrete integration schemes are discussed in dependence on the vehicle speed and compared with strong solutions calculated by an extended covariance analysis. Typical results show that first order Euler schemes are not applicable in case of stiff stochastic differential equations when the upper vehicle frequency is much higher than the lower one. The vehicle models are extended to nonlinear wheel suspensions with cubic-progressive springs. Introducing relative coordinates, rmsvalues of stationary vibrations are numerically computed using higher order simulation schemes and associated nonlinear covariance equations. It is shown that increasing nonlinearities reduce the vehicle vibrations in case of weak damping. For strong overcritical damping, however, both increasing mechanisms, the system damping and the suspension nonlinearity, lead to magnifications and destabilizing effects of the stationary vehicle vibrations.

1

W.V. Wedig, "Resonances of High-Speed Vehicles on Random Surfaces of Roads or Tracks", CD Proceedings of EURODYN 2011, 8th European Conference on Structural Dynamics, Leuven, Belgium, 4-6 July, 2011.

2

K. Sobczyk, D.B. MacVean, J.D. Robson, "Response to Profile-Imposed Excitation with Randomly Varying Transient Velocity", Journal of Sound and Vibration, 52(1), 37-49, 1977.

3

D. Talay, "Simulation and Numerical Analysis of Stochastic Differential Systems: A Review", in P. Krée, W. Wedig, (Editors), "Probabilistic Methods in Applied Physics", Lecture Notes in Physics, 451, 54-96, Springer, 1995.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £110 +P&P)