A comparison of major steady RANS approaches to engineering ABL simulations

Abstract

Engineering atmospheric boundary layer (EABL) simulations performed using Computational Fluid Dynamics (CFD) can give key insights in successfully addressing a variety of topics in environmental and structural aerodynamics. To improve understanding of this complex topic, a computational evaluation of the shear stress-driven (SSD), pressure-driven (PD) and body-force-driven (BFD) EABL flows was performed using the open-source CFD code OpenFOAM®. The EABL simulations were computationally evaluated by using the successor domain technique (SDT) and precursor domain technique (PDT). The boundary conditions applied in the CFD algorithm using OpenFOAM® to computationally model the SSD, PD, and BFD EABL flows were reported and discussed. A developed CFD approach may also be satisfactorily used in other relevant CFD codes, including but not limited to Ansys CFX®, Ansys Fluent®, STAR−CCM+®. The impact of the EABL models on surface pressure distribution on a low-rise cubic building, the total building drag force and the wind velocities in pedestrian-level areas was computationally evaluated by using major Reynolds-averaged Navier-Stokes (RANS) turbulence models, i.e., the standard k−ε, re-normalization group (RNG) k−ε, realizable k−ε, Wilcox's k−ω, and Menter's k−ω shear stress transport (SST) turbulence models. The obtained computational results indicate that the SSD and PD EABL flows may be successfully modeled in cases where additional modifications of the officially released CFD code are not required in regard to the PDT. The PD EABL flow may be successfully modeled by using all studied RANS turbulence models in combination with the PDT. The EABL flow throughout an empty computational domain is homogeneous even without using a body force because the employed CFD code is capable of creating the required pressure gradient that drives the flow along the computational domain. While the surface pressure estimation on the windward body surface and in the recirculation zones was generally proven to be an issue when using steady RANS two-equation turbulence models, surface pressures in the stagnation zone on the windward cubic building surface and pressures on the top and the side building surfaces in the present study agree well with the experiments when using the RNG k−ε and Menter's k−ω SST turbulence models. The pressure distribution on the building surfaces, the total building drag force and wind characteristics in the vicinity of a lift-up building are not considerably affected by the choice of the EABL model, while they are substantially affected by the type of RANS turbulence model. The best agreement with the experiments was achieved when using the RNG k−ε and Menter's k−ω SST turbulence models, so these models may be recommended for future applications.