Large Eddy Simulation study on the structure of turbulent flow in a complex city

• Published in: Journal of wind engineering and industrial aerodynamics 2018-06, Vol.177, p.101-116
• Main Authors: Tolias, I.C., Koutsourakis, N., Hertwig, D., Efthimiou, G.C., Venetsanos, A.G., Bartzis, J.G.
• Format: Article
• Language: English
• Subjects: ADREA-HF; Boundary-layer wind tunnel; CFD; Computational Fluid Dynamics; Grid resolution; Large Eddy Simulation; LES validation; Michel-Stadt; Semi-idealized urban canopy; Urban turbulent flow
• ISSN: 0167-6105; 1872-8197
• DOI: 10.1016/j.jweia.2018.03.017

Large Eddy Simulation (LES) of atmospheric flows has become an increasingly popular modelling approach within the last years, as it has the potential to provide deeper insight into unsteady flow phenomena. LES can be improved and validated using specifically designed and well documented wind tunnel datasets. In this work, we evaluate the performance of LES against a wind tunnel experiment in a semi-idealized city (“Michel-Stadt”; CEDVAL-LES database) and use the LES results to study the structure of the turbulent flow at the particular urban area. The first, second and third order statistics are presented, as well as velocity frequency distributions and energy spectra. The results compare well with the experimental values. Information about special features of the flow field is also provided. A particular focus of this work is put on the influence of grid resolution on the results. Five different grids are examined and the required resolution for turbulent flow within the canopy layer is evaluated. This study reveals the strong potential of LES for urban flow simulations. It is shown that LES can assess highly non-Gaussian flow behaviour in street canyons, which has implications for urban ventilation, wind comfort assessment and urban design. •LES assessment against a wind tunnel experiment of a semi-idealized city.•Extensive grid sensitivity study is performed.•Thorough analysis of flow characteristics is made.•Evaluation of first, second and third order statistics.•LES successfully predicts bimodal velocity distributions.