Virtual Wind Tunnel and Free Air CFD Comparison Part 1/2
Often the ultimate aim of performing a wind tunnel test on a scale model is to provide data on how the full size equivalent will behave in free air. This is especially true in motor racing (e.g., Formula 1) where teams use a combination of wind tunnels and Computational Fluid Dynamics (CFD) to guide what they hope is a winning racecar design. As a follow on to my project to build a virtual wind tunnel to test an open wheel racecar, here I start a two-part project to compare results between the virtual wind tunnel and free air using a series of CFD simulations.
Free Air Configuration
To better simulate the racecar in free air I took the wind tunnel test section and embedded it within a larger volume with boundaries positioned approximately 5 times the maximum length of the racecar in each Cartesian direction (except upstream where it was approximately 2 lengths) away from the racecar. In this new free air configuration the outer boundaries should have a negligible effect on the final results around the racecar. Also the entire ground plane is assigned a linear velocity to simulate the moving ground under the racecar. Note that in the wind tunnel model, the moving ground is localized around the racecar model.
The Original Test Section from the Wind Tunnel is Highlighted
I will run three CFD simulations in free air to compare with the wind tunnel results:
- 40% scale model in free air matching the wind tunnel conditions, Re = 3x106, U = 29 m/s
- Full size model in free air with matching Reynolds number to wind tunnel, Re = 3x106, U = 12 m/s
- Full size model in free air with matching velocity to wind tunnel, Re = 7.5x106, U = 29 m/s
Any differences between case 1 and the wind tunnel test will likely be attributed to the relative close proximity of the wind tunnel walls. Most wind tunnels provide a wind tunnel wall correction to modify results to be equivalent to free air, assuming the model is not above a threshold blockage area, usually about 5%. In this case the wind tunnel blockage due to the 40% model racecar is approximately 2.4%, so we should not expect a large discrepancy.
Adjusting the air speed for case 2 to match the Reynolds number of the wind tunnel test should produce similar results to case 1. There is a simple linear relationship between the racecar length and air speed, so if you scale up the racecar length by 2.5, then you need to scale down the air speed by 2.5 to maintain the same Reynolds number, assuming the density and viscosity of the air remains constant.
Increasing the air speed for case 3 to match that of the wind tunnel test should provide an indication as to whether the results are sensitive to a Reynolds number increase. Ultimately the difference between case 3 and the wind tunnel results provides a good indication as to whether the wind tunnel can accurately guide the aerodynamic design of the full size racecar.
Tune in next time to see the results and conclusions in the final episode of this two-part project.
Recent blog posts
- External Aerodynamics with CFD
- A Case for Renaming the Navier-Stokes Equations
- Ludwig Prandtl: Real Fluids Explained
- Osborne Reynolds: A Giant in Fluid Dynamics
- Sliver Treatment Strategies for CFD
- How to Fix Small Acute Angles for CFD
- Small Feature Removal for CFD
- Fluid Visualization in Nature
- Design is Compromise
- CFD Is Not Enough