It seems that wind tunnels have been labelled bad, because they are much more costly to run than simulation software. So today’s invention is an alternative to them, but which retains their potential for precision.
Most tunnels rely on dynamical scaling: essentially using model cars and planes scaled down so that their aerodynamic behaviour stays the same (ie the same Reynold’s number -the ratio between inertial and viscous forces), as the full scale flow (Re = density, flow speed, size/ dynamic viscosity).
It’s common to model airflows by using water. Both the viscosity and density are increased. Given that the water flow speed is lower than say that of a car on a motorway, to keep Re the same, requires the use of a larger than life model.
This would be impossible in a wind tunnel, where size raises the costs enormously.
There are, however, many places where there are tide races…predictable, high speed tidal flows within channels.
If we put our big, low-drag car design in a tideway we could do a bunch of tests at very low cost.