Reliability based analysis of a railway temperature prediction model

Resumo

Rail temperature prediction plays a crucial role in ensuring railway safety, as extreme temperatures can cause local buckling and track instability. This study conducts a reliability-based sensitivity analysis of a previously developed prediction model using MATLAB and UQLab. Two analyses were performed: a global sensitivity analysis considering all parameters as random variables and a Data-Driven Sensitivity Analysis incorporating measured data for key variables to refine the model and enhance its practical applicability. Results indicate that uncertainties in convection and solar absorption are the most influential parameters affecting the response statistics of the rail temperature predictions. Furthermore, Monte Carlo simulation was employed to estimate the daily probability along one year of rail temperatures exceeding the threshold for local thermal buckling of a railway track.