Embedded LES modelling of atmospheric boundary layer in urban region using accelerated turbulence generation

A novel development of Embedded Large Eddy Simulation (ELES) model was proposed for urban wind field simulations, incorporating the synthetic volume force method integrated with the Graphic Processing Unit (GPU) parallel computing. By combining the advantages of RANS and LES, the proposed model enhances both computational efficiency and accuracy in comparison to conventional LES models. A validation simulation for the atmospheric boundary layer wind field was conducted to assess the accuracy of the proposed method. Subsequently, an ideal urban block was used to evaluate the influence of LES region sizes and the positioning of the RANS (Reynolds-Averaged Navier-Stokes)/LES interface on the simulation results. It was found that the positioning of the RANS/LES interface in the ELES model appears to have little effects on the mean wind field, while it does affect the fluctuating wind speed results. Comparative analysis with wind tunnel experiments and conventional large eddy simulation confirmed the efficacy of the proposed ELES with accelerated turbulence generation.