Numerical and physical investigation of roof corner vortex dynamics

The pressure distribution and the associated roof vortices on the roof of low rise buildings are investigated using computational fluid dynamics (CFD) and physical models. The study uses the Texas Tech University (TTU) building geometry and wind environment as the basis for all the investigations.; In the first part of the study the flow field around the TTU building is numerically simulated using Reynolds Averaging Navier Stokes (RANS) equations. The results of the simulations are compared with full and model scale pressure coefficients along the centerline of the building.; The characteristics of roof vortices and their implications to structural loads are investigated using proper orthogonal decomposition (POD) and pattern recognition (PR) methods. The study shows the peak structural loads and the associated roof pressure patterns for four wind angles: 90° for normal winds and 60°, 45° and 30° for oblique winds. The results also show the relations and differences between the POD and PR methods.; Unsteady inflow boundary conditions are necessary to perform Large Eddy Simulations (LES). The available wind data from full or model scale is limited to few locations at the inlet boundary of the computational domain. Various methods of inflow data generation are investigated and benchmarked against experimental results. A new inflow generation method, less expensive that produces reasonable inflow data is suggested.; LES simulations of the flow field around the TTU building are performed for 90°, 60° and 45° winds using a FORTRAN code developed for this purpose as well as commercial software, Fluent. The numerical solutions are used to explain the effects of inflow boundary conditions and length scale. Also for the 90° and 60° simulations the pressure coefficients along the center line of the building are compared with full and model scale data and show that LES better predicts full-scale data compared to RANS models. It is also shown that the inlet boundary conditions and the scale of the simulation are important factors.