Evaluation and modeling of the flow in a slotted wall wind tunnel

Sammanfattning: As vehicle manufacturers strive to shorten the development time of new models, an increasing share of the aerodynamic development work is shifted from wind tunnel testing of prototype vehicles to numerical simulations of virtual models. However, comparing measurements from the wind tunnel with numerical simulation data is not straightforward due to several interference effects occurring between the wind tunnel and the tested vehicle. The objective of this thesis is to improve the understanding of the properties of a slotted wall wind tunnel used for automotive aerodynamic testing and enhance numerical simulation accuracy by providing validation between the simulations and physical wind tunnel measurements. This is done by investigating the empty test section flow, as well as both local and global interactions between the wind tunnel and the test object. Using standard instrumentation that was calibrated in-situ to achieve low measurement uncertainty, it is shown that a swirling flow angularity pattern seen already during the wind tunnel commissioning is still present in the test section. By ruling out two alternative hypotheses on the cause of this pattern, it is concluded that it most likely originates from the fan. It is also demonstrated that the measured levels of flow angularity are unlikely to have a significant impact on the forces measured on a vehicle. Scale-resolving numerical simulations of the flow in the empty wind tunnel show good agreement with the measured longitudinal pressure distribution in the downstream region of the test section but deviate upstream of the turntable. This is attributed to shortcomings in the modeling of the distributed suction system used to limit boundary layer growth on the floor upstream of the car. Investigations of the tangential blowing system used to fill in the boundary layer behind the belts in the moving ground system show that the blowers effectively reduce the displacement thickness of the boundary layer as intended and that this can be well represented in numerical simulations using a simplified representation of the blowers. It is also shown that the force differences measured between different configurations of a vehicle can be significantly affected by the tangential blowing. Simulations of vehicles as tested inside the wind tunnel test section are shown to improve the prediction quality compared to open road simulations. This requires that the process of non-dimensionalizing the forces and pressures are done in the same way for the in-tunnel simulations as in the physical wind tunnel. Furthermore, lift predictions are significantly improved when including the lift acting on the wheel drive unit belts used to rotate the wheels of the test object.