Quality assessment of Large-Eddy Simulation of wind flow around a high-rise building: Validation and solution verification

• Published in: Computers & fluids 2013-06, Vol.79, p.120-133
• Main Authors: Gousseau, P., Blocken, B., van Heijst, G.J.F.
• Format: Article
• Language: English
• Subjects: Bluff body aerodynamics; Computational fluid dynamics; Computational Fluid Dynamics (CFD); Computer simulation; Fluid flow; Kinetic energy; Large eddy simulation; Large-Eddy Simulation (LES); Mathematical models; Turbulence; Turbulence modeling; Turbulent flow; Urban wind flow; Validation and Verification (V&V)
• ISSN: 0045-7930; 1879-0747
• DOI: 10.1016/j.compfluid.2013.03.006

•LES can simulate the wind flow around a building in good agreement with experiment.•This agreement is not always improved by computational grid refinement.•The numerical and modeling error partly compensate each other on the coarse grid.•With 20 cells per building side 80% of total turbulent kinetic energy is resolved.•The vortex method is suitable to generate neutral atmospheric inflow turbulence. When undertaking wind engineering problems such as urban pollutant dispersion or pedestrian wind comfort with Computational Fluid Dynamics, an accurate simulation of the flow-field around buildings is required. In this respect, the good performance of Large-Eddy Simulation has already been established but because the formulation and the use of this turbulence modeling approach are complex, the uncertainty on the results is relatively high. This implies the need for Validation and Verification (V&V) studies like the one performed in the present paper for the wind flow around an isolated high-rise building with aspect ratio 1:1:2. In the first part of the study, the numerical results are compared with measurements from a reference wind-tunnel experiment and the agreement is quantified by validation metrics. The vortex method to generate inflow turbulence is shown to provide accurate results. Unexpectedly, the best agreement with the experiments is obtained on the coarsest computational grid, with 20 cells per building side, while a finer grid with 30 cells per building side over-estimates the turbulent kinetic energy measurements. A similar result was also found by earlier studies for different flow configurations. In the second part of the study, solution verification is performed. The Systematic Grid and Model Variation technique is used to provide estimates of the modeling and numerical error contributions. The LES_IQ indicator shows that a grid with 20 (resp. 30) cells per building side allows resolving 80% (resp. 91%) of the total turbulent kinetic energy in the region around the building.