Assessment of various turbulence models (ELES, SAS, URANS and RANS) for predicting the aerodynamics of freight train container wagons

This study presents an assessment of the capabilities of various turbulence modelling approaches -ELES, SAS, URANS and RANS-to predict the aerodynamic flow around a double-stacked freight wagon, both in isolation and within a train. The numerical predictions are compared with experimental measurements at the same Reynolds number to determine the accuracy of each model. Specifically, aerodynamic drag, front and rear surface pressures, planar velocity fields and skin friction lines are validated against the wind tunnel results. In particular, predictions from the ELES and SAS models show good agreement with the wind tunnel data, both qualitatively and quantitatively. Indeed, ELES predicts both the range and distribution of the rear-face surface pressure very closely, indicating that the separated flow is also likely to be well predicted. Both SAS and ELES predict the pressure drag of the multi-wagon configuration to within 2% of the experimental value. However, the steady RANS model predicts the trends in pressure drag in line with the experiments as the front and rear gaps are varied, even though individual drag predictions are considerably worse. Overall, the set of results establishes the benefits and deficiencies of using particular turbulence models to capture different aspects of freight train aerodynamics.