IMPACT HAMMER TEST FOR BAYESIAN PARAMETER IDENTIFICATION OF RAILWAY TRACK SYSTEM

Abstract

The current design approach of railway sleeper is majorly based on some simplifications considered in the European guidelines. One example of these simplifications is how the distribution of the ballast under the sleeper is accounted for. When ballast is damaged, the stiffness it provided to support the sleeper is reduced and the dynamic behavior of the sleeper changes. However, these guidelines consider a uniform distribution of ballast stiffness, either along the complete length of the sleeper or limited to the rail seat. The embedding ballast conditions of railway tracks seemingly vary randomly from site to site and the ballast-sleeper interaction is neither uniformly distributed nor fixed. Therefore, there is need to evaluate the health status of the ballast under the existing railway track. A field test was carried out on an existing ballasted track, to obtain the time domain responses under impact excitation. The measured vertical vibrations at different locations on the in-situ sleeper were employed in time-domain Markov chain Monte Carlo (MCMC)-based Bayesian model updating for identification of the railway track system. MCMC was used to ensure that proposed method can be applied even when the problem is unidentifiable. The proposed method identified the distribution of railway ballast stiffness under low-amplitude vibration and the “equivalent†rail stiffness and mass. The model updating results confirmed that the ballast stiffness under the existing sleeper was uniform, which implies that there was no ballast damage under the tested sleeper.