Validation and application of pressure-driven RANS approach for wind parameter predictions in mountainous terrain

Computational Wind Engineering has undergone numerous developments over the past few decades and has been applied to solve a variety of wind engineering issues, but it cannot substitute wind tunnel testing and field measurements for accurately replicating wind characteristics in naturally hilly terrain. This paper is aimed at detecting the validity of the approach on numerical simulations by estimating wind flow behaviour at a bridge site in a Y-shaped valley and comparing the results from wind tunnel testing. Further, the terrain-induced effects on wind flow are investigated and the spatial distributions of wind parameters along the bridge structures are characterized. It is concluded that three Reynolds-Averaged Navier-Stokes (RANS) turbulence models, namely k-ε, k-ω and SST, make significant differences in the leeward regions and wind velocity, speed-up, and pitch angle profiles, with the k-model producing superior predictions within 20% error when compared to experimental results. The simulation results show that non-homogeneity of wind characteristics occurs frequently along bridge structures. One is significant rotations of the wind velocity profile up the height of both bridge towers, and then a practical equation for predicting wind deflection is proposed. The other is spanwise nonuniformities in mean wind speeds and pitch angles along the main girder, which necessitates the use of a modified terrain coefficient when securely predicting wind speed. The discussion of a few national bridge design codes reveals that most of them fall short of characterizing the wind load at the bridge site in mountainous areas.