If you want to get accurate aerodynamic data from a scale model the reynolds number of the model must be the same as the reynolds number of the actual car ( or whatever funny shaped two wheeled thing you want to model). To ensure this the model must be an exact scale replica of the original, even the smallest deviations can cause major disturbances in airflow. If you happen to have the rescources to make such a model you then must scale one of the other factors (velocity, viscosity, or density) of the renoylds number to make up for the smaller model.
For example: if you want know what happening at 200 mph and you build a 1/4 scale model you need to multiply one of the other factors by 4 or double 2 of them. This means that if you leave viscosity alone and had a tunnel capable of doubling the air density you would need to flow the air at 400 mph!!!
Of course there are ways around that, you could use a different fluid with a lower viscosity and lower the temperature which also lowers the viscosity. You could also take data at lower velocities which would not be as accurate but could still be useful.
Its just a question of time and money, kinda like everything else