[aerohead]

You might consider the book,'Low Speed Wind Tunnel Design,' by Alan Pope.It will walk you through all the different types,their merits and shortcomings.
Wolf Heinrich Hucho also covers various tunnels in his 'Aerodynamics of Road Vehicles.'
Conservatively speaking,GM,Lockheed Marietta,and NASA Ames are the only tunnels in North America with test sections large enough to satisfy full-scale model criteria.
The Pininfarina tunnel at Turin,Italy would be the absolute bare minimum.
--------------------------------------------------------------------------
*You'll first need to select the 'type',then design from there.
*The test vehicle cannot occupy more than 5% of the test section cross-sectional area,so you largest vehicle will determine the test section size.
*A minimum airspeed of 20-mph is required to produce a turbulent boundary layer/proper Reynolds number .
*If say,you were going to test a full-size SUV of 36 sq-ft frontal projected area,your test section would need to be at least 720 sq-ft.
*At 20-mph,the test section airspeed would need to be a minimum of 1720 ft/min.
*And flow volume produced by the fans,1,267,200 CFM,and at a static pressure high enough to overcome all the surface friction losses from the ductwork,turning vanes,stator vanes,fan(s),radiators (if your recycling your air),turbulence screen,etc..
*And you'll need your force measuring equipment (Sting,load cells,knife-edge balance,with boundary layer control,with capability for maybe a minimum of 8-degrees of yaw.
*A moving floor has been deemed 'statistically insignificant' by some researchers who did back to back investigations.So you could probably skip that expense.
*You are looking at millions of dollars,even if it's measured in sweat equity.
*Hundreds of thousands in materials.
*"A trial by fire measured by any standard "says Ayla of the Clan of the Cave Bear.