Practical CFD simulations of wind tunnel tests

Computational fluid dynamics (CFD) has the potential of replacing wind tunnel testing in many wind engineering applications. Validated CFD software could enable engineers to calculate wind effects on buildings for which no aerodynamic information is currently available. However, the use of CFD for structural engineering applications is limited mainly because of its prohibitive computational resource requirements. This is due in part to the need to simulate the imperfect spatial coherence of the low-frequency turbulent fluctuations in the incoming atmospheric boundary layer (ABL) flow. A methodology is needed that would remove this barrier. In addition, it is desirable to develop software capable of readily incorporating features specific to the simulation of aerodynamic effects on bluff bodies. For this reason NIST's open source Fire Dynamics Simulator (FDS) is being adapted for wind engineering applications. FDS numerically solves the spatially filtered form of the Navier-Stokes equations appropriate for incompressible flow, a technique known as large-eddy simulation (LES). In a first phase of this research, FDS has been used to compare data obtained from University of Western Ontario (UWO) wind tunnel tests on a 1:100 scale model of the Texas Tech University's Wind Engineering Research Field Laboratory building (prototype dimensions: 9.1 m (L) xs× 13.7 m (B) × 4 m (H)). Pressure taps or ports along four lines on the model are chosen for comparison with FDS output. FDS is used to simulate flow in the wind tunnel and compare pressures on the building at several locations for varying angles of attack. Mesh refinement is also varied in the numerical simulation. Overall, the results from the FDS simulations fit the experimental data well. The paper discusses in detail the validation of wind pressure results, as well as issues being addressed to incorporate simplified ABL flow simulations in FDS. Additional issues such as modeling of nonorthogonal building surfaces and numerical issues associated with modeling flow in the wind tunnel are discussed.