Uncertainty assessment of a model to predict the vibration induced by train traffic in tunnels

The uncertainty assessment of ground-borne noise and vibration predictions is important to reduce risks when decisions are made based on the simulation results. In addition, it provides robustness to the prediction framework that could potentially be used for virtual validation of proposals to reduce noise and vibration in railway infrastructure. In this work, a general methodology for prediction uncertainty assessment based on the guide to the expression of uncertainty in measurements is applied to a numerical model dedicated to predict the vibration induced by train traffic in tunnels with slab track with isolated blocks. The standard uncertainty of the predicted acceleration on the tunnel wall is obtained by combining the standard uncertainty of the model inputs: sprung and unsprung axle mass, primary suspension and rail pads or fasteners stiffness and isolated blocks mass and stiffness. The input uncertainty is defined according to the guidance given in international standards, published work or by experience judgement. The sensitivity factors are obtained as the slope of the function that fits the results obtained by running the simulations for a given range of the input quantities. The proposed methodology provides the uncertainty of the result and the contribution of each input to that uncertainty.