Large Eddy Simulation for Train Aerodynamic Noise Predictions

It is known that for high speed trains the main noise source is the aerodynamic noise; however also at medium speed the sound connected with the air stream around the train body is certainly not negligible, especially concerning both operator and passenger comfort due to the relatively close location of the sources, and environmental impact. Therefore it is fundamental to assess the effects of aerodynamic noise both in terms of integral sound level developed and localization of acoustic sources. It is indeed quite difficult to separate the different noise contributions and to localize the sources in on-track measurements, anyhow decoupling of aerodynamic noise from the other sound generation can be achieved both in wind-tunnel test or with aeroacoustic numerical simulations. The purpose of this work is to develop a tool to perform aerodynamic noise predictions from a direct simulation of the flow field around the train surface. A computational method to predict surface pressure fluctuations aiming at evaluating aerodynamic noise sources was implemented. Exploiting hybrid unstructured meshes, Detached Eddy Simulation and specific acoustic post-processing tools, a CFD method was developed in the framework of the open-source code OpenFOAM. Validation of the procedure was performed on a simpler test case concerning a generic side mirror exposed to a uniform airstream above a flat plate for which detailed experimental results were available from literature. A 1:5 scaled model including one and a half wagon and two bogies cavities was simulated at the nominal speed of 70 m/s. Pressure fluctuations at selected points were monitored in time and analyzed in frequency while overall sound pressure levels were computed on the full mock-up body and presented as bidimensional maps. Comparison with experiments showed a good agreement both in terms of frequency dependency and of peak values registered.