Methodology to evaluate the uncertainty of train exterior noise prediction

Abstract

Noise emissions play a key role in the development of environment-friendly rolling stock. Noise limits given by EU directives for interoperability purposes but also by internal directives of the different countries where any kind of rolling stock operates have to be respected for train approval. Exterior noise predictions are used nowadays to validate the design of new rolling stock. These predictions are used within the companies to evaluate the different design options, to optimize costs, to assess the risks and for virtual certification in cases where existing products are modified. Potentially predictions could be used in the future for a complete virtual certification of the train allowing to decrease or even to completely avoid the noise tests that are currently carried out by the train manufacturers. As a consequence, the cost of the certification process will be reduced. One of the challenges for a complete virtual certification is the assessment of uncertainties in the predictions as probably this will not be accepted by the stakeholders without a defined and validated uncertainty assessment. This work presents a methodology to estimate the uncertainty of the predictions of train noise emissions. An example of the application of the proposed framework for uncertainty evaluation is provided for a typical suburban rolling stock showing the feasibility of its use for railway exterior noise predictions but potentially also for interior noise calculations and for different applications other than railway.