There are very large wind tunnels out there. Wouldn't it be very efficient to test new airplanes flying them "on the spot" inside of these tunnels? That is, in free flight - not mounted to a pylon? One could presumably omit the test pilot and remote control them, get excellent visual feedback and have a completely controlled environment. Has this been done? If not, why?
There have been tests of real aircraft in wind tunnels, with a pilot sitting in the cockpit. But in all cases the plane had been tied in place with strings. Trimming the lift and especially the thrust in all phases of the test is really impossible.
The Messerschmitt 109 was tested in the French Chalais-Meudon tunnel in 1940, the Bell P-39 was tested at Langley, where the top speed could be improved from 340 to 392 MPH, the P-51 was tested at Langley in 1943 like most small military planes of that period, and lately a stealth demonstrator made by MBB was tested in 1984, in most cases with the pilot in place. These are the ones I know; there were certainly more.
It would actually miss the point of wind tunnel testing. A free flying model could only be subject to air stream in speed and direction so that the forces are balanced. A fixed model on the other hand can be subject to air stream of any direction and speed and the forces measured. It is important to measure the forces under those conditions too.
Could you? Yes, probably, but it would be very difficult due to the very limited space. You would have to have a huge airflow in excess of 50 knots for a small aircraft, and you could not simulate any motion fully, such as takeoffs rolls. You could also not test stalls and other manoeuvres. I think the main reason against this is that this is simple areas of flight dynamics that can just as well be done with computational fluid dynamics on a computer for a fraction of the effort and cost.
I don't think it would be safer either. It's a very confined space, and you couldn't cut the airflow very quickly, so it's probably more dangerous than flying outside with unlimited space and a stable airflow.
You certainly could fly a completely assembled aircraft in a sufficiently-large wind tunnel -- at least in theory (theory is an awesome place - everything always works there. When I retire I'm moving to theory).
In reality there are lots of drawbacks which have already been pointed out, but there's one that hasn't been yet: Wind tunnel testing is traditionally done with scale models, and even then it's not "cheap": It takes a lot of power to move wind through a tunnel at high speed, and scaling this up to a full size airframe - even small-airplane size like your typical Cessna or Piper trainer - is cost-prohibitive since you would have to build the entire plane and then pay to blow high-velocity air over it.
(There are wind tunnels at this scale though - at least one. NASA spent a lot of money on it, and then Boeing and Airbus came out with their new super-jumbo planes which don't fit!)
It's also worth pointing out that the kind of fluid dynamics modeling that used to be done with wind tunnels and smoke trails is now often done with Computational Fluid Dynamics (CFD) software. Cheaper, Safer (no risk of parts blowing off and bouncing down the tunnel), and more accurate in many regards -- but nowhere near as much fun to watch :)
Yes you can. Yes it has been done. And yes people are still doing it. The caveat being that most (all?) of the tests are done with scale models rather than full size planes.
Free flight wind tunnel tests are done for different reasons from static mounted tests. Where static mounted tests are done to collect measurements of forces acting on the aircraft/part, free flight tests are done to understand/confirm the stability of the aircraft.
Flying a model plane in such confined spaces is very difficult and typically the plane will tend to wander around in the wind tunnel. As such, the data that you can get from a free flight test is very limited.
Due to the difficulties of doing the test and the limited amount of data you can get from it most new plane designs don't go through such a test. Instead, what typically happens is that the design goes straight to a radio controlled scale model that is flown outdoors to study its stability. However, when trying out new ideas that has not been developed it is still worth doing the stability tests in a controlled environment - in a large wind tunnel.
Here are some old videos of free flight wind tunnel tests (these come from a time period where the behavior of delta wings were not yet fully understood):
Here's a more recent video of a free flight wind tunnel test of the Blended Wing Body:
No, you cannot fly a plane like that.
Problem is: propulsion is missing. Once you expose a plane model to an airstream, the airfoils (wings) start producing lift but also drag. In real world, an airplane requires propulsion to overcome the drag (induced drag,parasite drag,..) in order to move forward.
The wind tunnel airplane model would just lift-off and fall behind, there would be no chance to establish a stabilized flight attitude where propulsion equals drag and lift equals weight. Such a stabilization is required to seriously start measuring. That's why the models are strapped down to a pylon or whatever. Measuring results can therefore also be reproduced (certain fixed AoA's at certain airstream speeds etc.)