Atmospheric Boundary Layer Processes To Mimic Peak Pressures on Low-Rise Buildings: CFD Versus Full-Scale and Wind Tunnel Measurements

Abstract

Realistic prediction of peak wind pressures is indispensable in a safe design of low-rise buildings. For several decades wind tunnel testing was employed to obtain wind loads on buildings and other structures. However, there is still doubt in the wind engineering community regarding the adequacy of wind tunnels to predict accurately full-scale pressures on low-rise buildings and small-size structures. The recommendations of the American Society of Civil Engineers (ASCE) 7-10 standard for external pressure coefficients, for roof components and cladding (C&C) design are also based on published wind tunnel data. Recent field measurements show significant deviation of full-scale pressures from wind tunnel results. To fully understand this issue, the current research focuses on appropriate processes of atmospheric boundary layer (ABL) flows by computational and experimental efforts to mimic full-scale peak pressures on roofs of low-rise buildings. Computational fluid dynamics (CFD) simulations were employed to create ABL flows at an open-jet hurricane simulator, Louisiana State University (LSU). Various turbulence closures of k-є, Reynolds Stress Model (RSM), and Large Eddy Simulation (LES) were implemented in CFD. The LES simulated wind velocity and turbulence intensity profiles are in a good agreement with target theoretical profiles and experimental measurements, compared to k–ε and RSM. Extensive experimental testing was conducted at the LSU open-jet facility on two scaled models of a benchmark low-rise building from the Texas Tech University (TTU), to examine how the turbulence properties of the approaching flow, scale issue, and open-jet exit proximity influence the flow pattern around low-rise buildings, and alter the length of the separation bubble on the roof surface. The results show that mean values of pressure coefficients are correlated with the horizontal distance between the test model and the exit of the open-jet. Closer the test model to the exit of the open-jet, shorter the separation bubble. Similarly, peak pressures in the separation bubble are highly sensitive to the open-jet proximity. The vortex method was used to computationally generate transient winds at the inlet boundaries. The purpose was to update/expand existing guidelines for the longitudinal extension of the computational domain via LES, as existing recommendations (e.g. COST and AIJ) are mainly based on the Reynolds Averaged Navier–Stokes equations (RANS) models. The findings suggest that the optimal location of the building from the inlet boundary, in the computational domain, can be different from that recommended by the COST and AIJ guidelines. The results of CFD with LES were compared