A probabilistic approach for estimating the behavior of railway tracks

Railway engineering is confronted with problems due to degradation of the rail network that requires important and costly maintenance work. However, because of lack of knowledge on the geometrical and mechanical parameters of the track, it is difficult to optimize maintenance management. In this context, this paper presents a new methodology to analyze the behavior of railway tracks. It combines new diagnostic devices which permit to obtain an important amount of data and thus to make statistics on the geometric and mechanical parameters and a non-intrusive stochastic approach which can be coupled with any mechanical model. Numerical results show the possibilities of this methodology for parametric studies (influence of the distribution and of the correlation of the parameters...). In the future this approach will give important informations to railway managers to optimize maintenance operations.