Prediction of Railway Induced Vibration by Means of Numerical, Empirical and Hybrid Methods (Voorspelling van trillingen ten gevolge van treinverkeer door middel van numerieke, empirische en hybride methodes)
Prediction of Railway Induced Vibration by Means of Numerical, Empirical and Hybrid Methods
Verbraken, Hans; S0110994
Ground vibration induced by railway traffic is a problem with a large environmental impact. Vibration generated at the wheel/rail contact is transferred through the soil and excites the foundation of nearby buildings where it causes disturbance of sensitive equipment, hindrance to inhabitants, or structural damage. An accurate prediction of expected vibration levels is required when designing new tracks or when changes are made to existing tracks. The study of possible vibration mitigation measures also requires accurate models.This work addresses the numerical, empirical, and hybrid prediction of railway induced vibration. For the assessment of these methods, an elaborate measurement campaign has been carried out at a site in Lincent, Belgium. First, a numerical model is presented that allows for a detailed representation of the track geometry. The influence of commonly made simplifying assumptions regarding the track and soil geometry is investigated for the site in Lincent. The dynamic track and soil characteristics are identified based on preliminary tests at the site. The predictions are compared to measured results of the transfer functions and during train pass-bys and show a reasonable agreement. Second, empirical prediction methods are discussed with a focus on the procedure presented by the U.S. Federal Railroad Administration (FRA), where the source and the propagation path are characterized experimentally. The experimental determination of these terms is critically assessed. Furthermore, the limitations of the FRA procedure are investigated by means of numerical simulations.Third, a hybrid model is presented according to the empirical FRA procedure by replacing the experimental result of either the source term or the propagation path term by a numerical prediction. The hybrid model is applied at the site in Lincent and compared to the experimental and numerical results.