As 'Fundamentals of Aerodynamics' by JD Anderson suggest
Today, for the most part, we do not attempt to simulate all the parameters simultaneously; rather, Mach number simulation is achieved in one wind tunnel, and Reynolds number simulation in another tunnel.
Is it done for accuracy? Or Does it need an extra big wind tunnel to do that?
It would be great if we could match all parameters at once. But that would require a tunnel large enough to accommodate the full size airplane and powerful enough to match the actual flight speed. The reality is that we cannot in most cases.
We have some answers here on windtunnel testing. They should cover the basics:
The most important similarity numbers are the Reynolds number and the Mach number. Both grow linearly with flow speed, so it helps to match flow speed. In case of the Reynolds number, the other parts of it are the length of the flow path and the viscosity of the fluid. If we use air, which is plentiful and free, we also need to match the actual size so that the Reynolds number matches the original. The Mach number also depends on the speed of sound in the fluid, which in turn depends on temperature.
It is possible to match both Reynolds and Mach number in super cold tunnels using a pressurised gas like nitrogen. This lowers the required gas speed so the tunnel needs less power to run and produces realistic results with smaller models. The expense of building and operating such a tunnel is immense, though.
It is extremely expensive to build a wind tunnel that can be both pressurized and handle high-velocity flow.
Such large-section wind tunnels do exist, but to maintain high pressure during a test run they require enormous pressure tanks which take ~hours to pump up to the required pressure, which can then be maintained for only ~seconds during a full blowdown run.
Digital simulation has gotten so accurate and fast that in some cases, wind tunnel tests of scale models are no longer strictly necessary for routine aircraft design tasks.
