A Hybrid RANS-LES Dataset for Data-Driven Turbulent Mean Flow Reconstruction

This paper presents results for a set of parameterised periodic hill geometries using the hybrid RANS-LES approach based on the Spalart-Allmaras improved delayed-detached eddy simulation (IDDES). These results are compared with data from direct numerical simulations. The dataset is intended to be used for data-driven Reynolds-averaged Navier-Stokes (RANS) closure modelling for aerodynamic shape optimisation. As a demonstration, the IDDES data for the baseline geometry is used for turbulent mean flow reconstruction with the k-omega shear stress transport model. Flow reconstruction results demonstrate considerable improvements compared to the baseline RANS model.

Springer Nature Switzerland