CFD Simulation of a Racecar in a Wind Tunnel
Having recently built a Computational Fluid Dynamics (CFD) simulation of the Honda Wind Tunnel at Imperial College, it seemed only fitting to actually use it to perform a virtual test on a scale model. Clearly, given the wind tunnel's close association with motor racing, the test model had to be a racecar.
Velocity magnitude iso-surfaces
It just so happened that I had a Formula 1/Indy car open wheel model from a previous CFD study. I first scaled the racecar to be 40% of its original full size to match the maximum size specification for the wind tunnel. If the model was any larger then the results were likely to be compromised by the blockage effect of the model in the test section. I also added a new internal duct to the car to route air through a radiator which I represented as a porous volume. I only had to model half the car due to its symmetry and for it to be compatible with the symmetric virtual wind tunnel.
Symmetric half model
Given the modular multi-volume design of the virtual wind tunnel I can switch out various sections and insert new ones. In the case of this racecar study I constructed a new test section with the racecar embedded within it, and then I connected the new test section to the rest of the virtual wind tunnel.
Within the test section I replicated the moving road in the real wind tunnel with a linear velocity assigned to the ground locally around the racecar. To simulate the rotating wheels in contact with the moving road I assigned an angular velocity to both wheels of the half car.
The maximum velocity of the air in the real wind tunnel is of the order of 30-40 m/s, so I adjusted the rotational speed of the wind tunnel fan to be 6000 rpm, which resulted in a mean air velocity at the test section entrance of 29 m/s.
Streamlines colored by velocity magnitude
With CFD we can visualize the entire flow field around the racecar as shown below.
Arrows following streamlines colored by velocity magnitude
How do the wind tunnel results compare with an equivalent simulation performed in free air? Now that's a topic worthy of another post me thinks, so stay tuned.