I did some work along these lines when I was in college. We were studying Magnetohydrodynamics, and needed a way to test different venture shapes. Since we were working in a water medium, and using full scale shapes there wasn't a problem with scaling or density. Our problem was getting a straight flow of water through the tunnel.
The walls of the tunnel must be mirror smooth, and the test subject must be far enough away from the tunnel walls so it won't be affected by eddies and torrents produced by the boundary layer effect on the tunnel walls. For a 6 inch test subject we ended up with a tunnel 3 feet wide by 12 feet long. A system of vanes shaped like a honey comb were used at the front and rear of the tunnel to "straighten" the flow of water. This whole "Cluster !@#$%" was about 20 feet in circumference, and was pumped by a large (10" dia) muffin fan type pump on the opposite side of the "Apparatus".
So if you can get a good straight flow of water through this thing, then scaling and density problems should be the least of your worries.
Hope this helped some.... Smitty